A Practitioners Manual On Monitoring and Evaluation

A Practitioners’
Manual on Monitoring
and Evaluation of
Development Projects
A Practitioners’
Manual on Monitoring
and Evaluation of
Development Projects
By
Kultar Singh,
Dharmendra Chandurkar
and Varun Dutt
A Practitioners’ Manual on Monitoring and Evaluation of Development
Projects
By Kultar Singh, Dharmendra Chandurkar and Varun Dutt
This book first published 2017
Cambridge Scholars Publishing
Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK
British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library
Copyright © 2017 by Kultar Singh, Dharmendra Chandurkar

and Varun Dutt
All rights for this book reserved. No part of this book may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording or otherwise, without
the prior permission of the copyright owner.
ISBN (10): 1-4438-8992-X

ISBN (13): 978-1-4438-8992-6
TABLE OF CONTENTS
List of Figures........................................................................................... viii
List of Tables ............................................................................................... x
List of Abbreviations .................................................................................. xi
Preface ...................................................................................................... xiv
Chapter One ................................................................................................. 1

Understanding Projects
1.1 Defining a Project
1.2 Project Cycle
1.3 Project Goal and Objective
Chapter Two ................................................................................................ 9

Designing Projects
2.1 Results Chain
2.2 Project Designing Tools
2.3 Logical Framework Analysis
2.3.1 Situation analysis
2.3.2 Problem analysis
2.3.3 Objective analysis
2.3.4 Alternative analysis
2.3.5 Risk and assumption analysis
2.3.6 Development of Project Planning Matrix
2.3.7 The Logframe
Chapter Three ............................................................................................ 27

Understanding Monitoring and Evaluation: Basics and Key Concepts
3.1 Understanding M&E
3.2 Monitoring vs Evaluation
3.3 M&E Levels
3.4 M&E Typologies
3.5 Indicators
3.6 Design Indicators
vi Table of Contents
Chapter Four .............................................................................................. 37

Developing a Performance Monitoring Plan
4.1 What is Performance Monitoring Plan?
4.2 How to develop a Performance Monitoring Plan
4.3 Ensuring Quality of Monitoring Data
Chapter Five .............................................................................................. 62

Approaches to Monitoring
5.1 Progress and Process Monitoring
5.2 Participatory Monitoring and Evaluation
5.3 Monitoring without Indicators using the Most Significant Change
technique
Chapter Six ................................................................................................ 72

Choosing Evaluation Designs
6.1 Experimental Design
6.2 Quasi-experimental Design
6.2.1 Difference in Difference
6.2.2 Propensity Score Matching
6.2.3 Regression Discontinuity
6.2.4 Instrumental Variable
6.2.5 Reflexive Comparison
6.3 Non-Experimental Designs
6.4 Comparison Matrix
6.5 Practice Exercise- Choosing the appropriate Impact Evaluation
Design
Chapter Seven............................................................................................ 90
Approaches to Evaluation
7.1 Utilisation Focused Evaluation
7.2 Feminist Evaluation
7.3 Transformative Evaluation
7.4 Real World Evaluation
7.5 Evaluation Approach Using Appreciative Inquiry
7.6 Realist Evaluation
7.7 Equity-Focused Evaluation
7.8 Developmental Approach to Evaluation
A Practitioners’ Manual on Monitoring and Evaluation vii
of Development Projects
Chapter Eight ........................................................................................... 102

Analysing M&E Information
8.1 Data Basics
8.2 Univariate Analysis
8.2.1 Measures of central tendency
8.2.2 Measures of variability
8.3 Bivariate analysis: Correlation and Regression
8.4 Quantitative Data Analysis using Computer-aided Analytical
Software
8.4.1 Performing analysis using MS-Excel
8.4.2 Overview of data analysis using Statistical Package for Social
Science
8.5 Qualitative Data Analysis
Chapter Nine............................................................................................ 175

Executing Monitoring Evaluation and Learning
9.1 Finalising the Theory of Change
9.2 Finalising the Logic Model
9.3 Finalising the Purpose of Evaluation
9.4 Finalising the Evaluation Question
9.5 Finalising the Key Areas of Enquiry
9.6 Finalising Indicators
9.7 Finalising Research Instruments
9.8 Pilot Testing
9.9 Preparation for Electronic Data Collection
9.10 Recruitment of the Field Team
9.11 Preparation and Implementation of Data Collection
9.11.1 Development of the field manual
9.11.2 Training of field survey teams
9.11.3 Roles and responsibilities of the data collection team
9.12 Quality Assurance Protocols
9.13 Data Analysis
9.14 Ethical Approvals
9.15 Learning Framework and Structure
9.16 Evaluation Report
Bibliography ............................................................................................ 211

LIST OF FIGURES
Figure 1.1: Project Cycle

Figure 2.1: Results Chain
Figure 2.2: Results Chain - Example 1
Figure 2.6: Problem Analysis - Example
Figure 2.7: Objective Analysis - Example
Figure 2.8: Alternative Analysis - Example
Figure 2.9: Risk Analysis
Figure 2.10: Logframe Format
Figure 2.11: Logframe - Example
Figure 6.1: Difference in Difference
Figure 6.2: Regression Discontinuity - Example
Figure 7.1: Steps in Real World Evaluation
Figure 8.1: Data Types
Figure 8.2: Insert Function
Figure 8.3: Insert Function in Statistical Category
Figure 8.4: Dataset - Example
Figure 8.5: AVERAGE Function
Figure 8.7: Insert Pivot Table
Figure 8.8: Creating a Pivot Table
Figure 8.9: Selecting Pivot Table Fields
Figure 8.10: Pivot Table Output
Figure 8.11: Analysis using the ToolPak-Data Analysis
Figure 8.12: Selecting Descriptive Statistics
Figure 8.13: Correlation using the Analysis ToolPak programme
Figure 8.14: Regression using the Analysis ToolPak programme
Figure 8.15: Regression Output
Figure 8.16: Data Editor Window in SPSS
Figure 8.17: Variable View in SPSS
Figure 8.18: Opening Data File in SPSS
Figure 8.19: Option of Opening an Excel File
A Practitioners’ Manual on Monitoring and Evaluation ix
Figure 8.20: MS Excel File Opened in SPSS

Figure 8.21: Using Compute to Create New Variables
Figure 8.22: Splitting a Data File for Analysis
Figure 8.23: Coding Missing Values
Figure 8.24: Reliability Analysis
Figure 8.26: Analysing Frequencies
Figure 8.27: Crosstab Statistics
Figure 8.28: Regression
Figure 8.29: ATLAS/ti Main Window
Figure 8.30: NVivo Main Window
Figure 9.1: Data Collection Process using PDAs
LIST OF TABLES
Table 3.1: Monitoring vs Evaluation

Table 3.2: M&E at Various Levels of the Results Chain
Table 4.1: Project Monitoring Plan Format
Table 4.2: Developing PMP - Example
Table 4.3: Developing PMP-Example
Table 4.6: Developing PMP Example
Table 6.1: Computation of DID
Table 6.2: Comparison Matrix of Evaluation Designs
Table 7.1: Gender Approach vs Feminist Evaluation
Table 8.1: Data Classification
Table 8.2: Calculating the Standard Deviation - Example
Table 8.3: Frequently Used Functions
Table 8.4:Independent Samples t-test
Table 9.1: Review of the Indicators with Respect to the Logic Model and
Theory of Change
Table 9.2: Example Validation Check
Table 9.3: Example Range Check
Table 9.4: Example Consistency Check
Table 9.5: Snapshot of Evaluation Guidelines
LIST OF ABBREVIATIONS
AHS Annual Health Survey

AI Appreciative Inquiry
AIP Annual Implementation Plan
ANC Antenatal Care
APL Above Poverty Line
American Standard Code for Information
ASCII
Interexchange
BPL
Below Poverty Line
BPY
Beti Padhao Yojana
CAPI Computer-Assisted Personal Interviewing
CASI Computer-Assisted Survey Interviewing
CATI Computer-Assisted Telephone Interviewing
CAWI Computer-Assisted Web Interviewing
Cost Benefit Analysis
CBA
Christian Commission for Development in
CCDB
Bangladesh
CEA Cost Effectiveness Analysis
CHC Community Health Centre
CSPro Census and Survey Processing System
CSR Corporate Social Responsibility
DID Difference in Differences
FE Fixed Effect
FGD Focused Group Discussion
FLW Frontline Health Worker
GAD Gender and Development
Hb Haemoglobin
HH Household
HU Hermeneutic Unit
IDI In-depth Interview
IE Inclusive Education
IFA Iron Folic Acid
IGA Income Generating Activities
xii List of Abbreviations
IHHL Individual Household Latrines

International Organisation for Cooperation in
IOCE
Evaluation
IV Instrumental Variable
JSY Janani Suraksha Yojana
LFA Logical Framework Approach
M&E Monitoring and Evaluation
MDG Millennium Development Goal
MEL Monitoring, Evaluation and Learning
MIS Management Information System
MLE Monitoring Learning and Evaluation
MMR Maternal Mortality Rate
MSC Most Significant Change
NGO Non-Governmental Organisation
ODBC Open Database Connectivity
ODF Open Defecation Free
PAPI Paper- Assisted Personal Interviewing
PAS Project Annual Survey
PDA Personal Digital Assistant
PHC Primary Health Centre
PMF Performance Measurement Framework
PMP Performance Monitoring Plan
PPM Project Planning Matrix
PSM Propensity Score Matching
RCT Randomised Controlled Trial
RD Regression Discontinuity
RMNCH Reproductive, Maternal, Newborn and Child Health
RWE Real World Evaluation
SBA Skilled Birth Attendants
SC Significant Change
SPSS Statistical Package for Social Science
TE Transformative Evaluation
ToC Theory of Change
UFE Utilisation Focused Evaluation
A Practitioners’ Manual on Monitoring and Evaluation xiii
VHC Village Health Committee

WHO World Health Organization
ZOPP Zielorientierte Projektplanung
GOPP Goal Oriented Project Planning
PREFACE
This book is an effort to provide handholding support to professionals

working in areas of monitoring and evaluation (M&E), programme
management or other related domains in the development sector.
A Practitioner's Manual on Monitoring and Evaluation of Development

Projects attempts to take development professionals through the life cycle
of the M&E process and explain its fundamentals and concepts. Practical
examples have been used throughout the manual to assist professionals to
visualise and internalise the key concepts and processes that need to be
followed for monitoring and evaluating development interventions.
Chapter 1 of the manual aims at developing a fundamental understanding

of projects which is a prerequisite for understanding how to monitor and
evaluate the same. Next, the steps of the project cycle are explained
sequentially, after which the critical concepts like project objective and
goal and the difference between these terms are clarified.
Chapter 2, which focuses on understanding how to design projects, begins

with a discussion on the results chain or the series of end to end causal
pathways that need to be formulated to understand how the short- term and
the long-term project results are to be achieved. This is followed by
enumerating the key tools that are used in designing projects, after which
the reader is taken step by step through the logical framework analysis, a
popular tool used in designing projects, illustrated with an example.
Chapter 3 explains the key concepts, the fundamentals of M&E and also
the key differences between them. The key typologies and different types
of M&E are explained, in addition to the critical concept of indicators and
how to design good indicators.
Chapter 4 takes the readers through the process of developing a Performance

Monitoring Plan (PMP) which acts as an overarching framework for
monitoring and evaluating a development project. PMP is helpful in
monitoring the performance of the project and the achievement of its
A Practitioners’ Manual on Monitoring and Evaluation xv
intended results. Subsequently, the key activities that should be done to

ensure the data quality are explained.
Chapter 5 describes different approaches to monitoring including progress,

process and participatory monitoring. It also explains the most significant
change (MSC) tool used to monitor projects without using indicators.
Chapter 6 delineates in detail the various evaluation designs by explaining

the fundamental concepts of evaluation as well as elucidating the different
types of evaluation designs by using practical examples. Randomised
Controlled Trial (RCT), Cluster RCT and Pipeline design are explained
under experimental designs, followed by different designs like Difference
in Differences (DID), Propensity Score Matching (PSM), Regression
Discontinuity (RD), Instrumental Variable (IV) and Reflexive Comparison,
all of which fall under the category of Quasi-Experimental designs. The
non-experimental evaluation designs are explained next, followed by
different types of impact evaluation designs compared through a matrix to
understand when any of these evaluation designs should be used and the
advantages and disadvantages of using the same. Chapter 6 ends with an
exercise which has different situations in which the practitioner is required
to choose an evaluation design which will be suitable for each situation.
Chapter 7 explains the fundamental concepts of the different approaches

which can be adopted for evaluating projects based on the context and the
objective of a developmental project. It should be remembered that these
approaches are at the macro level philosophy of evaluations and they will
use the evaluation designs explained above as part of their methodology.
Chapter 8 delineates data basics and how the data may be analysed
usefully for monitoring and evaluating a project. Univariate and
multivariate analysis are covered as part of the data analysis. Basic
quantitative data analysis using MS Excel and SPSS software has been
demonstrated using screenshots. The fundamentals of qualitative data
analysis and the popular software that can be used to conduct the same
have also been touched upon.
In Chapter 9, practitioners are taken through a step by step process of

executing the Monitoring Learning and Evaluation (MLE) function. This
chapter focuses on the operational dimension of executing the MLE
process. Concepts discussed in the earlier chapters are summarised
through an operational lens. Steps like pilot testing, developing data
xvi Preface
collection and quality assurance protocols, getting ethical approvals,

developing a learning framework etc., are covered in this chapter.
To summarise, this book will act as a hands-on manual for professionals

working in the development sector to understand the fundamentals of
M&E from the point of view of a practitioner. Attempt has been made to
sequence the book in such a way that it covers all the key concepts related
to M&E in a simple, lucid and coherent manner. However, it is quite
possible that despite our best efforts, certain shortcomings may have
remained and we look forward to removing these in the future versions.
We will be grateful to our readers for their valuable suggestions and
feedback.
CHAPTER ONE
UNDERSTANDING PROJECTS
The mainstay of work in the development sector is in the form of projects,

which are targeted at various areas like health, education, sanitation,
livelihoods, child rights, climate change etc., depending on the mandate
and the objective of project implementation and funding organisation. In
this chapter, the fundamental concepts related to projects are explored and
a project-based approach is discussed, which is imperative for an in depth
understanding of M&E.
On completion of the chapter the reader will be able to:
" Describe what is a project

" Understand and explain the elements of a project cycle
" Define the project objective and goal
1.1 Defining a Project

As we embark upon our journey to understand how to monitor and
evaluate projects, it is important to first understand the fundamentals and
underlying concepts of projects and project management. At the outset of
every project, it is envisaged that several activities will be performed over
the course of the project's implementation. These activities constitute the
work that will be done during the project and they form the mainstay of
the action that will take place.
Every project has a specific objective and it is envisioned that through

these activities the project will achieve its objective. The example of a
five-year project of making its target villages open defecation free (ODF)
is used to illustrate this point. To achieve this objective, the project
engages in several activities like construction of household (HH) and
community toilets, conducting awareness campaigns to motivate people
not to defecate in the open, educating people about the technologies that
should be used for toilet construction etc. The types of activities performed
2 Chapter One
as part of the project vary depending on the project objective and the
implementing organisation’s capacity. These activities form the key work
that is done as part of the project implementation.
Another important aspect of every project is that it has a specific start date
and a specific end date i.e., a specific time period within which it has to be
executed. The project is expected to achieve its desired objective within
this specific time period, which in the example quoted above, is a duration
of five years. Last but not the least, it is very critical to understand that
each project is allocated a limited set of resources. Resources, which may
be financial, human and physical, are allocated to a project so that its
activities may be implemented and its objectives achieved within a
specific time period. Accordingly, the example project is also allotted a
fixed budget, human resources and fixed physical resources with which its
activities may be implemented and its objective achieved in a specific
period of time.
Hence, a project may be defined as:

“A set of activities implemented within a specific period of time and with
specific resources to achieve a specific objective.”
1.2 Project Cycle

Usually, social development interventions are formulated and implemented
in the form of a project and follow a cycle or a sequence which is known
as the project cycle. From its inception to its closure, every project has its
unique cycle of operation though the fundamental project cycle remains
the same. Therefore, it is essential to understand the project cycle in order
to better conceptualise, design, plan and implement it and also to monitor
and evaluate it effectively.
From the beginning till the end of the project, the project cycle comprises
of various phases or stages. All the stages in the project cycle are
delineated and implemented successively in a phased manner. Each of
these stages are defined by their objective, information requirements,
responsibilities and key outputs. The various stages of a generic project
cycle are:
Understanding Projects 3
Situation
Analysis
Gap
Evaluation
Analysis
Implementation Programme
and Monitoring planning
Figure 1.1: Project Cycle
Stage I: Situation Analysis

It is a well-known fact that a project does not exist in vacuum. It is
formulated to respond to a negative situation or condition. The stage of
identifying and understanding this existing negative situation which needs
to be responded to through intervention is called situation analysis. This
stage consists of understanding the prevalent situation and identifying the
cause(s) of this situation. Situation analysis is useful in the later stages
when the strategy and subsequently, the specific activities to target these
causes are defined. A good situation analysis serves as an entry point for
the project by throwing light on what needs to be done to address the
negative conditions in each context. For example, in the sanitation project
referred to earlier, the situation of open defecation (OD) and toilet usage in
the target area are assessed, apart from identifying the key reasons why
people are defecating in the open.
4 Chapter One
Stage II: Gap Analysis

A project always works towards achieving its desired objective. By the
time it is completed, the project envisages reaching the intended or desired
situation as opposed to the situation from where it had started. The project
works towards bridging the “gap” that exists between the present and the
desired situation. Gap analysis is thus done to identify the gap between the
current situation and the desired situation. In the example, the project aims
to make the target area ODF from the current situation, where it is
assumed that 35 per cent of the target population defecates in the open. If
in terms of actual numbers, 35 per cent works out to one lakh families, the
gap is 35 per cent or one lakh families. Therefore, the intensity of project
activities, the requirement of resources and the implementation plan are
made keeping in mind that the target population the project must reach is
one lakh families.
Stage III: Project Planning

Project planning follows the gap that needs to be bridged through the
project that has been identified. During the project planning stage,
objectives are defined, strategies by which to achieve this objective are
formulated, activities are identified, timeline based targets are set and
resources are allocated to the project. A detailed implementation plan with
the activity schedule and milestone timelines is also prepared as part of the
project planning. During this phase, a project monitoring plan (PMP) is
also devised to assess its achievement.
Discussing the project planning phase in reference to the example, the first
step is to define the project objective, which should be specific and
realistic. For this project, the objective is to make the project area ODF in
the next five years. The second step is to identify the activities that are
undertaken as part of the project to achieve its intended objective. The
project can undertake activities like building HH and community toilets,
creating awareness in the community about the ill effects of OD, tracking
people who defecate in the open and counselling them, and providing
information and technical support for building the right type of toilet etc.
After identifying the activities, targets are set, i.e., the number of HH and
community toilets that could be built within each year, and the number of
communication campaigns that need to be undertaken to motivate people
to not defecate in the open. This is followed by deployment of resources
for the project, primarily in the form of finances available for implementation
of the project. The money is utilised to recruit human resources (project
staff) based on the defined roles and responsibilities. Physical and

infrastructure resources like office, equipment etc., required for the project
are also purchased. Timeline targets are set for communication campaigns
and constructing toilets in a phased manner.
Stage IV: Implementation and Monitoring

The next stage is project implementation during which the formulated plan
is executed. Monitoring of project activities is done concurrent to their
implementation to ensure that the project is on track and as per the
formulated plan. Monitoring helps to identify deviations, if any, from the
project plan and also to introduce mid-course corrections. While executing
a project, its quality, time, cost and risk management needs to be
considered to ensure that it is successfully implemented within its
predefined resources and timeline.
To take the case of the ODF project, the various activities that were
identified in the planning stage are executed during this stage. The project
staff is recruited and deployed, awareness campaigns are conducted about
the ill effects of OD, and subsidy is provided for toilet construction etc.
These activities are also simultaneously monitored to assess whether the
toilets are being constructed as planned, both in terms of quantity and
quality and whether the communication campaigns are being conducted as
planned etc.
Stage V: Evaluation
After project activities are completed, many stake holders like project
implementers, policy makers, the government, and the external audience,
among others, want to know whether there is any change in the ‘situation’.
The stake holders also want to know whether this change is due to the
project intervention or other external factors. An evaluation, helps to
systematically assess the impact, effectiveness and the contribution of the
project. Mid-term evaluations are helpful because they provide timely
learning which helps in course correction. Post project evaluations help in
getting insights that are helpful in formulation of other similar projects.
Various techniques or designs are thus adopted for different projects in
different situations. These evaluation designs are explained in detail in the
following chapters.
Taking the current example of the ODF project, an evaluation should be

conducted to know if the rate of OD has reduced in the project area. Also,
6 Chapter One
in case it has reduced, would it be right to say that it has reduced due to the
activities undertaken by the project?
To consider another example of a project which aims to increase the rate

of institutional delivery, and examining its project cycle through the M&E
lens, the first question at the start of the project would be to inquire about
the proportion of women who had opted for institutional delivery. At the
end of the project, the same question is asked once again regarding the
proportion of women that had institutional deliveries. Also, whether it
could be confidently asserted that the institutional delivery rate had
increased only because of the project activities and not because of any
other external factor. The change from the initial status (e.g., 65%) is
measured and whether the desired situation (say 80%) has been reached or
by what margin the gap has reduced.
After the various stages of the project cycle are examined, the practitioner
needs to assess the stage at which M&E needs to be conducted and the
specific M&E activities that should be performed at each stage of the
project cycle.
1.3 Project Goal and Objective

One of the most critical and fundamental steps in designing a project is to
define its objective and goal. Many people often get confused and use
these terms interchangeably without realising that they are two different
though interrelated concepts. Poorly defined goals and objectives cause
ambiguity in project planning and implementation. Therefore, it is
essential for any project to lucidly define its goal and objective and make
sure that the entire project planning and implementation is aligned towards
achieving them.
Objective
The objective of a project is the specific condition that the project targets
to achieve and that too by itself. An objective is derived from a goal, has
the same intention as a goal, but it is more specific, quantifiable and
verifiable than the goal (SMART, Characteristics of Good Objectives,
2016). Usually, the project tries to address the inverted image of the core
problem. For example, if the core problem in a specific village is the ‘high
rate of OD’, then its corresponding objective would be to ‘reduce the rate
of OD from 35 per cent to zero per cent in a period of five years’.
The SMART criteria is widely used to judge project objectives. A project

objective is said to be SMART if it fulfils the following criteria
x Specific
x Measurable
x Attainable
x Realistic
x Time-bound
Using the SMART criteria to examine the objective to ‘reduce the rate of
OD from 35 per cent to zero per cent in a period of five years:
x Specific: The objective is said to be specific as it specifically

targets the issue of OD.
x Measurable: It can be safely stated that this objective is measurable
using the indicator ‘rate of OD’. The data for this indicator can be
collected using the HH survey data collection tool.
x Attainable: To assess if the target of reducing the OD rate from the
current 35 per cent to zero per cent, for one lakh families is
attainable or not the availability of resources, time in hand and
other relevant factors are considered.
x Realistic: While setting an objective, the project team and other
experts analyse whether the project would realistically be able to
achieve its objective, keeping in mind the resources available for
the project and the external factors related to the project.
x Time bound: It needs to be ensured that the project has defined
timelines. In the case of the current example, the project is expected
to achieve its objective in a time frame of five years.
If the project objective fulfils all the five criteria as part of the SMART
framework, it is said to be a good objective. Another question which may
be considered is whether a project should have a single objective or
multiple objectives? By its definition, an objective is supposed to be
precise, specific and definite. A project should thus simply have a single
objective. In common parlance, the term 'project objectives' is often used.
However, this term is incorrect and this usage is against the spirit of a
‘project-based approach’. A project is essentially about the breaking down
of a large problem into its smaller constituent parts, such that each
problem is addressed one at a time in a single project. It is not about
tackling all problems at once by trying to find a general solution for the
various aspects of the problem. A project which has more than one
8 Chapter One
objective is likely to lose focus, while it renders the objective open to

ambiguity and subjectivity.
Therefore, a well-designed project should ideally have one objective

which is clearly articulated.
Goal
By now the practitioner knows that the planned activities have to lead to
fulfilment of the project objective. The objective also has a higher purpose
which extends beyond the precincts of the project. This higher purpose is
the goal which the project aims to contribute towards. Each project, by
achieving its objective contributes towards a larger goal. Thus, it can be
rightly said that while a project 'achieves' its objective, it 'contributes'
towards the goal. In other words, the goal is the macro-level change that
the project contributes to at the micro-level.
Thus, the objective of our example project, ‘reducing the rate of OD from
35 per cent to zero per cent in a period of five years’, at the same time
contributes to the higher goal of 'improving the sanitation conditions in
rural India’. Another current example would be of a project which aims to
contribute towards the goal of improving the standard of living of the rural
poor. Increasing the HH income of the rural poor, improving their access
to social schemes, improving the sanitation conditions, improving their
health, providing safe drinking water, increasing the HH assets etc., could
be a few of the project objectives which would contribute towards the
above mentioned larger goal. Thus, a goal is the intended change we seek;
it is expansive in its scope and all-encompassing in its vision. In this way,
many projects may refer to and contribute towards a single goal while
achieving their respective objectives.
Hence, it is of paramount importance to define the objective and goal

lucidly to ensure clarity in project planning and implementation.
CHAPTER TWO
DESIGNING PROJECTS
Having understood what is project and project cycle, it is important to

understand what project designing is in order to ensure effective and
useful M&E. M&E per se is not an activity that starts at the time the
project is nearing completion. Monitoring starts from the day the project is
rolled out, therefore, the M&E system needs to be conceived during the
project design itself.
At the completion of the chapter, the practitioner will be able to:
" Develop the Results Chain for a project

" List and briefly describe Project Designing tools
" Describe Logical Framework Analysis for designing projects
" Describe the Logical Framework Analysis and Project Planning
Matrix
2.1 Results Chain

During the project designing phase, it is essential to specify all activities
and objectives that are to be achieved through the project. The results
chain helps to manage projects, and at the same time, to understand the
causal linkage between project intervention and its desired impact rather
than managing the project based solely on activities. It helps to formulate a
roadmap to the envisioned change, while highlighting the necessary
conditions and assumptions required for ushering in a change in each
situation (Foundations of Success, 2007).
Operations are based on the 'if-and-then' logic. For example, if we put fuel
in a car’s fuel tank, then only can we drive and go somewhere in it. This
'if-and-then' logic is the means-to-an-end relation or a cause-and-effect
connection between the system components. So, what implication does
this system model have for projects?
10 Chapter Two
Every project, as we know, has its own rationale of intervention, one that
clearly addresses the nuts-and-bolts of the problem of 'what', 'when', 'why',
'how', 'who' and 'where'. The clearer a project is about the logic of change
underpinning its project activities or processes, the better it can deliver the
results or achieve the objective it has in mind.
A results chain thus describes the causal pathways of the activities

translating into expected results i.e., the outputs, outcomes and impacts of
a project. The results chain helps to track the progress of the project from
its more immediate results (outputs), to a result more proximate to the
achievement of the objective (outcome) and finally to a long-lasting result
or goal (impact). A basic results chain has the following components:
Inputs Activities Outputs Outcomes Impacts
Figure 2.1: Results Chain
Inputs: This includes the resources that are available or allocated for the
project. Input resources may be natural, human, and financial, depending
upon the nature of the project. For example, funds allocated, human
resources deployed, laptops allotted etc.
Activities: Activities are actions undertaken using the resources. In

simpler terms, this is the work performed that converts inputs into outputs.
For example, the training of frontline health workers (FLWs) on the
counselling of women, building of separate toilets for girls in schools etc.
Outputs: Outputs are the immediate effect of the activities of a project.

Outputs are also defined as the short-term results and often form the
deliverables of the project. For example, counselling of mothers on
institutional delivery is the output achieved from the activity training of
FLWs on counselling. Also, the increased attendance rate of girls is an
output of the activity of building separate toilets for girls in schools.
Outcomes: The mid-term results likely to be achieved from outputs are

called outcomes. Outcomes are generally the objective which the project
aims to achieve. For example, ‘increase in the rate of institutional
delivery’ is an outcome achieved through the output of ‘effective
Designing Projects 11
counselling of women on institutional delivery’. Also, ‘increased female

literacy’ is an outcome achieved through the output of ‘increased female
attendance rate’.
Impact: The final desired goal or the macro level goal that the project
envisages to achieve is defined as its impact. Impact is what the project
aims to contribute towards rather than trying to claim that it is what it
would achieve by itself. For example, ‘decreasing the Maternal Mortality
Rate (MMR)’ is the impact which the project aims to contribute to by
providing the outcome, which is, ‘increase in the rate of institutional
delivery’. Also, ‘increase in the empowerment level of women’ is the
impact which the project aims to achieve through its outcome of
‘increased female literacy rate’.
Governing the interrelationships between inputs, activities, outputs,

outcome and impact are several assumptions or enabling pre-conditions
that are necessary for the delivery of project results and achievement of the
project objective. They provide the necessary if not sufficient
preconditions without which the project cannot hope to achieve its results.
These assumptions are the causal inferences that govern the change
processes in a project and lay the groundwork based on which correlations
between the results, chain of inputs, outputs, outcome and impact are
sought to be made explicit (Dharmendra Chandurkar, 2014).
Although generic in character, this framework can be fine-tuned to

understand and unpack the non-linear, multi-contextual and multi-layered
nature of change that defines and determines the landscape of a project. In
general, it captures the project's broad canvas of change in one sweep,
while it sheds light on the causal relationship among the various levels of
change termed as outputs, outcome and impact.
Examples of the results chain are presented below in Figures 2.2 and 2.3
respectively.
12 Chapter Two
Exercise: Can you think of other examples of results chain? Fill the results
chain given below with your example.
Usually, in a project, there are multiple results that are envisioned and
therefore the project has multiple chains. The example of a project with
multiple results chain is presented below:
2.2 Project Designing Tools

Various tools are used to structure and facilitate the process of project
designing. In the development sector, the most commonly used tools are
the Logical Framework Approach, Zielorientierte Projektplanung/Goal
Oriented Project Planning (ZOPP/GOPP) and the Result Framework.
Logical Framework Approach or LFA is a simple project design tool

that can helps to organise and structure our thinking while designing a
project. As the name suggests, it is a logical approach for designing an
efficient and effective project. On the one hand, it facilitates optimal
resource allocation, while on the other hand, it sets performance measures
and standards that provide a framework for M&E. All this helps in
efficient project management.
ZOPP or GOPP is an adapted form of LFA that is suitable for the

development sector. It also uses the same logical approach of LFA, but it
is more flexible in accommodating the qualitative and subjective nature of
issues inherent in the development sector.
Result Framework is more output oriented, as it focuses on the ‘things

that would be on ground’ after completion of the project. These are
basically the results that we want to achieve and the underlying
14 Chapter Two
assumption is that achievement of these results would lead to achievement

of the envisaged objective.
All the above are tools for facilitating project development and each one
has got its strengths and limitations. LFA is discussed in the following
sub-chapter in greater detail.
2.3 Logical Framework Analysis

LFA is an analytical and project management tool which is widely used by
funding agencies, international nongovernmental organisations (NGOs)
and many government agencies for the designing and management of
development projects. It was developed in the late 1960s to assist USAID
to improve its project planning and evaluation systems (Republic of
Serbia, 2011). LFA supports objective oriented planning and management.
It can be used to perform systematic and structured analysis of a project or
programme. This analytical process consists of a set of tools or techniques
which can be used in managing development projects. Logframe matrix is
the documented product at the end of conducting the logical framework
analysis. On one hand, it facilitates optimal resource allocation, while on
the other hand, it sets performance measures and standards that provide for
a framework for M&E. It also takes into account the assumptions and risks
envisaged while implementing the project (NORAD, 1999).
The steps that are followed in conducting an LFA are explained below. An
example is used to reinforce the reader's understanding of this concept.
2.3.1 Situation analysisA project, as the practitioner is aware, does not

exist in a vacuum, rather, it is a response to a situation or negative
condition that it attempts to change. In the first stage of LFA, which is
known as situation analysis, attempt is made to understand the existing
situation or condition to identify what is ‘wrong’ with the given context
that needs to be addressed through programme interventions. Along with
this, the causes for the situation are analysed so that appropriate strategies
may be adopted and specific activities designed targeting the causes. A
good situation analysis serves as an entry point to the project by throwing
light on what needs to be done under the project to address the negative
conditions in each context.
2.3.2 Problem analysis

After the situation analysis, it is necessary to understand the problem at
hand. Development projects are usually proposed as a response to address
and overcome the identified social development issues. Problem analysis
involves identifying the key problems and then establishing the cause and
effect relationships between these problems. The primary purpose of this
analysis is to try and ensure that it is not just the symptoms of the
problem(s), but the ‘root causes’ that are identified and subsequently
addressed (Kari, 2004).
For example, if a person goes to the doctor with high fever and the doctor
prescribes medicine for high fever without diagnosing the cause for the
fever, then the doctor is simply treating the symptom but not the cause of
the problem. If the real cause of the fever is not discovered, it is quite
possible that the fever will persist after the effect of the medicine wears
out. Similarly, projects which only address the effects of the problems and
not the underlying causes are unlikely to bring about sustainable benefits.
A common method used to identify the core problem is the problem tree,
which helps to visually depict the connection between the various possible
causes of the problem. This is best explored in a participatory workshop
with key project stakeholders, who are asked to write down both the
causes and effects of the problem in a structured way. This procedure
makes it possible to clearly visualise the causes of the focal problem and
its effects and to find out how different problems are related to each other.
The next step is to plan activities and implement them within the
framework of the project to treat the causes of these problems.
16 Chapter Two
A clear and comprehensive problem analysis helps to build a sound

foundation on which relevant and focused project objectives can be built.
A simplified example of a problem tree for a project which envisages to

increase the HH income and thereby contribute to improving the standard
of living in rural areas is presented below. The two key causes for the
problem of low income in rural areas are identified and stated, viz., non-
remunerative agriculture and lack of income generating activities (IGA).
The further causes of each of these problems have also been listed in the
problem tree.
Figure 2.6: Problem Analysis - Example

18 Chapter Two
2.3.3 Objective analysis

After identifying the problems and developing the problem tree, the next
step is to conduct the objective analysis. Objective analysis is done to
formulate the objectives of the project by developing an objective tree.
The objective tree uses the same structure as the problem tree, but with the
problem statements (negatives) turned into objective statements
(positives). Based on insights from discussions with key stakeholders, the
priority problems are identified. This means that the original problem
statement does not need to be translated into objective statements.
While the problem tree shows the cause and effect relationship between
problems, the objective tree shows the means to the end relationship
between the objectives. This leads directly to developing the project’s
narrative description in the form of the Logical Framework Matrix.
The objective tree needs to be developed for the problems which the
project aims to solve through its intervention. The objective tree delineates
a logical sequence or cause and effect relation that needs to be followed to
solve the problems as presented in the problem tree.
The objective tree of the example project intervention is presented in the

diagram given below. In this case, ‘Increasing HH income in rural areas’
becomes the key objective/result which the project wants to achieve, and
this is called the project outcome, which further contributes towards the
intended impact i.e., ‘Increasing standard of living in rural areas’. The
short-term results i.e., the outputs and the activities planned to be
undertaken in the project are the branches of the objective tree.
Figure 2.7: Objective Analysis - Example

20 Chapter Two
2.3.4 Alternative analysis

The objective of alternative analysis is to identify the possible alternative
options, assess their feasibility and select one project strategy. The
different means-to-end branches in the objective tree, which constitute of
alternate options, are then identified. It is advisable to have a set of
assessment criteria against which the alternative strategies are ranked or
compared. Criteria like benefits to the target group, resource availability,
sustainability of the intervention, technical feasibility, relevance,
likelihood of success, implementability etc., can be used for the
assessment of the alternatives. Based on the assessment, the most feasible
and appropriate alternative is selected which would further constitute the
project strategy (NORAD, 1999).
Alternative analysis is a challenging stage as it requires synthesising a

significant amount of information and then making complex judgements
about the most suitable implementation strategy that should be pursued.
In the example used, after assessing the two alternative strategies

available, the project management team selects ‘Increasing IGA
opportunities’. The project activities, thus, will focus on the objective of
‘Increasing IGA opportunities' instead of the objective of working on
'Making agriculture more remunerative', assuming that the other
stakeholders will work on the same.
Figure 2.8: Alternative Analysis - Example

22 Chapter Two
2.3.5 Risk and assumption analysis

After the project intervention is selected based on the analysis of the
alternatives, it is very important to assess and address (to the extent
possible) the potential project risks. Projects are always subject to
influence by factors beyond the control of the project implementation
team. This is applicable in the context of development projects too, where
among other factors, the cooperation of various stakeholders, behaviour
change or the target group's willingness to participate are essential for the
project to fulfil its desired objective. Analysis of possible critical external
and internal factors/risks gives the opportunity to assess the conditions that
the project is working under.
Based on the risk analysis, the project management develops a risk

management plan i.e., a plan of how to avoid potential risks. Some
logframes prefer to use the term ‘assumptions’ instead of ‘risks’, the
distinction being that while risks are negative statements about what might
go wrong, assumptions are positive statements about the conditions which
need to be met to ensure that the project activities are being implemented
as planned. Whether the term 'risks' or the term 'assumptions' is used, the
purpose is the same, i.e., to assess and adapt the external impacts of the
project to improve where possible, the robustness of the design.
After identifying the key risks associated with the project logic model,
they can be plotted on a 2*2 matrix consisting of Impact and Likelihood
on its axis. The project management should aim to mitigate the risks which
lie in the top-right most quadrant i.e., risks which can have high impact
and high likelihood as these risks can jeopardise the project in achieving
its objective. The 2*2 matrix used for plotting potential risks is presented
in the figure given below.
Figure 2.9: Risk Analysis
Also, in the case of the example used, several risks associated with the
project can be identified. It is a good practice to identify risks, which can
be both internal and external, at each level of programme implementation,
i.e., from activity to output level, output to outcome level, outcome to
impact level. The risks identified with the example project are listed
below:
External Risks:
1. There is not enough demand of the goods produced under IGAs

taken up by the target group.
2. Raw material cost is very high making the IGA unviable.
3. Shortage of availability of raw material.
Internal Risks:
1. Enterprises are not able to develop market linkages for selling their
products.
2. Banks do not easily provide loans to individuals or self-help groups
(SHGs) supported by the project.
3. The project target group is not self-motivated and enterprising
enough to adopt new IGAs.
24 Chapter Two
2.3.6 Development of Project Planning Matrix

Results of the situation, problem, objective, alternative and risk/assumption
analysis are used as the basis for developing the Project Planning Matrix
(PPM) or logframe. PPM or logframe, which is said to be a snapshot of the
project intervention, is explained in detail in the next sub chapter.
2.3.7 The Logframe

The logframe or the PPM is a four-by-four matrix that details the logical
connect between the various components of the project as well as the
framework for assessing the performance.
The basic philosophy of the logical framework approach (LFA) is the

logical approach in achievement of the impacts, where inputs given
through activities leading to outputs and further to outcomes, which finally
contributes towards impacts. This hierarchy forms the rows of the PPM
matrix and the logical connect between various levels is termed as the
vertical logic (FAO).
Performance measures called indicators are set for each hierarchical level,
with sources of information listed for each indicator. Finally, assumptions
that make activities translate into outputs, and outputs to purpose, or
purpose to goal are also listed. This forms the horizontal logic of the PPM.
Columns and rows of the PPM are given below.
Horizontal Logic
Narrative Objectively Means of Assumptions/
Summary Verifiable Verification Risks
Vertical Logic
Indicators (OVI) (MOV)

Goal
Outcome
Outputs
Activities Inputs
Figure 2.10: Logframe Format
The logframe of the example project which aims to contribute towards

increasing the HH income in rural areas is presented below. Objectively
verifiable indicators (OVI) and means of verification (MOV) are discussed
in detail in the following chapters.

Horizontal Logic
Narrative Objectively Means of Assumptions/

Summary Verifiable Verification Risks
Indicators
Goal x HH income Structured Goal to Outcome
Increasing Interview Non-profitability
the HH Schedule of the enterprises
income in due to reasons like
rural areas no market
linkages, low
demand and
competition from
the external
market.
Vertical Logic
Outcome x Proportion of Structured Outcome to

1.Increasing targeted Interview Impact
IGA beneficiaries Schedule There is not
opportunities engaged in sufficient uptake
micro- of IGA
enterprises opportunities by
target
beneficiaries
Outputs x Amount of Structured Activity to
1.1 credit taken Interview Output
Improving from financial Schedule Financial
access to institutions institutions are not
capital at x Percentage of supportive of
reasonable people providing finance
rates acknowledging to target
1.2 improved beneficiaries
Improving access to
skill level in finances
people x Percentage of
people
acknowledging
access to credit/
finance through
financial
26 Chapter Two
institutions
x Rate of interest
for credit
Activity: Inputs Project
1.1.1 x Number of MIS/Reports
Forming SHGs formed
SHGs under the
1.1.2 project
Performing x Number of
all pathway SHGs for which
activities in all pathways
formation activities were
and ensuring performed
quality of x Amount of
SHGs credit provided
1.1.3 through
Linking linkages with
SHGs with banks
financial x Number of
institutions people given
for access to vocational
credit skills training
1.2.1 x Number of
Providing people trained
skill based in life skills
vocational
training to
women
1.2.2
Providing
life skills
training to
women
Figure 2.11: Logframe - Example
The logframe should be used as the basis of the funding application and
then throughout the project lifecycle to track progress and adapt to
changing situations. It can be used to review assumptions and
implications, and to keep donors and other stakeholders informed of
significant changes related to the project.
CHAPTER THREE
UNDERSTANDING MONITORING
AND EVALUATION:
BASICS AND KEY CONCEPTS
In this chapter, the practitioner will develop an understanding of the

fundamental concepts of M&E, its importance for project management and
learn the difference between the two. The practitioner will also understand
how M&E is juxtaposed with the results chain of a project. Finally, at the
end of the chapter, they will understand about indicators and learn how to
design them. At the completion of the chapter, the practitioner will be able
to:
" Define M&E

" Understand the key differences between M&E
" Define indicators and the important characteristics of a good
indicator
" Learn how to design good indicators
3.1 Understanding M&E

Monitoring is defined as the concurrent process of tracking the
implementation of activities of the project and attaining its planned
outputs. It helps to provide real time information of the progress of the
project in terms of completing its activities and achieving its immediate
outputs, both in terms of quality and target. Monitoring, thus, is an activity
to see if an ongoing project is proceeding on track. It involves the process
of systematically collecting data to provide real time information for all
stakeholders (managers, funders, participants) on the progress of
implementation and the achievement of desired outcomes.
The critical functions of monitoring are: to gather feedback from the

participants; collect data; observe the implementation of activities of the
project; analyse contextual changes; and provide an early warning system
of potential challenges. Analysis of monitoring data is critical to ensure
28 Chapter Three
that the project is being implemented in the right direction for it to achieve
its intended outcomes. In case the project is not moving in its intended
direction, midcourse correction should be done. Monitoring is applicable
to all programme levels (from input, process, output and outcome). Most
commonly, the focus is on output data, although it is also important to
track the goals and the objectives. Monitoring should ideally be an internal
function of the project management team. Monitoring, thus, plays a critical
role in the success of a project.
Monitoring of results helps to:
x Improve strategies and targeting. Enabling decision makers to focus

the project resources on areas where they can get the maximum
output.
x Understand project implementation barriers or challenges in real
time and suggest course correction measures.
x Ensure that the project is more effective and result oriented. It also
focuses on impact level changes throughout the project, rather than
just at the end of project evaluation.
Evaluation is defined as systematic research to see if a programme can

achieve its intended outcomes and impacts. Evaluation is done firstly to
see whether the envisaged objectives and goals have been achieved or not,
and secondly, to see whether the achievement is because of the project
interventions. It should assess the magnitude of change in the outcome and
impact and whether the change in the outcome or the impact can be
attributed to the project intervention. Evaluation assesses if there is any
deviation from the goals and the objectives, and whether it can confidently
be said that the objectives are achieved only because of project
intervention. Evaluation, then, is a type of causal research that establishes
the cause-effect relationship between the activities and the outputs on the
one hand and the objectives and the goals on the other.
While monitoring facilitates mid-course correction in attainment of project

outcomes, evaluation helps analyse variances from envisioned objectives
and goals. By providing feedback to the project functionaries, M&E
facilitates learning by doing. Development and enhancement of in-house
capacities to anchor the M&E functions is, thus, a prerequisite for learning
organisations.
Understanding Monitoring and Evaluation: Basics and Key Concepts 29
3.2 Monitoring vs Evaluation

Having understood the definition of M&E, the practitioner can now list the
key differences between the two.
Some of the distinctions between monitoring, which is to see ‘what we are

doing’ and evaluation, which is to assess ‘what we have done’ are given in
the matrix below (KEPA, 2015).
Table 3.1: Monitoring vs Evaluation
Monitoring Evaluation
Definition Concurrent analysis of project Assessment of the magnitude
progress towards achieving of change in the results
the planned results with the proposed by the project that
purpose of improving may be attributed to the
management decision making project.
(Aquaknow, 2016).
When is it Systematic activity should It should be done only at
done? be done regularly throughout specific points of time like in
the project implementation. the middle of the project, at
the change of phase, and at
the end of the project etc.
Scope Focuses on activities, Focuses on delivery of
outputs and indicators of project outcomes and
progress and change. impacts. It assesses the
progress towards the project
objectives and goals.
Who does Ideally, it should be an Ideally, it should be an
it? internal activity. This should external activity to avoid
be done by project staff or conflict of interest. It should
its target beneficiaries. be conducted by external
evaluators while involving
donors, project staff and
project users.
Why is it It is done to report project It is done to ensure
done? progress to the management, accountability of the project,
to identify the bottlenecks, learn broad lessons and
take remedial action and provide recommendations to
modify the project similar projects. It highlights
implementation plans. the potential and the
achievements of the project.
30 Chapter Three
3.3 M&E Levels

The matrix in the sub-chapter above makes it clear that monitoring is a day
to day activity of assessment of project progress, whereas evaluation is the
episodic assessment of overall achievement.
By juxtaposing M&E with the various stages of the results chain, the
practitioner should be able to gauge at which stage of the results chain or
theory of change of the project, should the M&E be focused on.
Table 3.2: M&E at Various Levels of the Results Chain
Stages in the Results Chain M&E

Impact Evaluation
Outcomes
Outputs Monitoring
Activities/Processes
Inputs
With respect to the Logical Framework, impact and outcomes fall within
the domain of evaluation, whereas outputs, activities and inputs fall within
the domain of monitoring.
3.4 M&E Typologies

Some popular typologies and terminologies commonly used in M&E are
discussed below.
Monitoring Typologies
Monitoring is a task that is inherently undertaken by doers or

implementers of the project themselves. Therefore, when monitoring is
performed by the project team itself or internally by the project
implementing team, it is called internal monitoring. Sometimes, when
projects involve parties or organisations external to the project for
facilitating the monitoring functions, it is classified as external
monitoring.
If the project implementers restrict the monitoring process to themselves,

it is called non-participatory monitoring. Project stakeholders, including
the communities remain mere information providers and have no role in
analysing the information and providing inputs for project implementation.

When functional participation of key stakeholders of the project, including
the target community for the project is solicited, it is called participatory
monitoring. In the case of the sample project about working to make its
target area ODF, a committee from the local community can be involved
in formulating the monitoring plan, collecting information and analysing
it. The community monitoring committee can keep a check on activities
like communication campaigns and toilet construction (both in terms of
quantity and quality).
Evaluation Typologies
While monitoring is inherently an internal activity, evaluation is an external

activity usually done by those external (individuals/agencies/ institutions) to
the project. Generally speaking, evaluation is external evaluation. However,
when project implementers choose to undertake evaluation by themselves,
it is called internal evaluation.
Evaluation can be best defined based on the timing of conducting the

evaluation. Evaluation per se is a less frequent activity generally
undertaken at the completion of a project for assessment of attainment of
the project objective. This is the post-project or post-facto evaluation. In
the case of many long duration projects, evaluation is also conducted
midway through the project implementation. Mid-term evaluations help to
ascertain the level of achievement of long duration projects half way
through the project. When the programme funders or the management is
interested in a more regular assessment of the achievements of the project,
the outcome and impact level assessment is done at a six monthly or a
yearly period also. This time series design of evaluation is commonly
known as concurrent evaluation.
To relate the achievement of objectives and goals directly to the project, it

may also be necessary to compare the status in the project area with an
identical non-project area, which forms the control group while the project
villages are the treatment group. This kind of evaluation design is called
the control-experiment design. Depending upon the time when the
evaluation is implemented, it can be concurrent, mid-term or post-facto
and internal or external.
32 Chapter Three
3.5 Indicators
The concept of indicators is pivotal to M&E. As per its dictionary
definition, an indicator is defined as a sign or a signal. In the context of
M&E, an indicator is said to be a quantitative standard of measurement or
an instrument which gives us information (UNAIDS, 2010). Indicators
help to capture data and provide information to monitor performance,
measure achievement, determine accountability and improve the effectiveness
of projects or programmes.
Designing indicators is one of the key steps in developing an M&E

system. As mentioned above, indicators are units which measure
information over time to document changes in the specific conditions.
With respect to the various M&E levels and the result chain of the project,
specific indicators need to be developed for each stage of the results chain.
Thus, there should be a different set of indicators at the impact level, at the
outcome level as well as at the output, activity and input level. Also, for
each level, there can be more than one indicator.
An indicator may be quantitative or qualitative based on the characteristics

of information that it provides. Those that deal with information that can
be expressed in numbers are quantitative indicators, while those dealing
with information units expressed in any form other than in numbers, e.g.,
statements, are qualitative indicators. Another important attribute of
quantitative indicators is that arithmetic functions can be applied to its
corresponding data while this is not possible in the case of qualitative
indicators. For qualitative indicators, their count or frequency may be
considered. Income measured in rupees, the weight of a baby measured in
kilograms and the number of toilets built are examples of quantitative
indicators. If the same information of income or weight is collected in
categories of high, medium and low, they are qualitative indicators.
SMART and SPICED Indicators
In order that development interventions are more result oriented, projects

must be made SMART. The acronym SMART for developing project
objectives is mentioned in Chapter 1, sub-chapter1.3. At the same time, it
is important to make sure that the indicators or the performance measures
also fit the SMART criteria. SMART indicators play an important role in
results based project management as they ensure accountability (MDF
Training & Consultancy, 2016). They have the following characteristics:
x Specific
x Measurable
x Attainable
x Realistic
x Timebound
Another school of thought advocates qualitative indicators represented by

the acronym SPICED, which stands for the attributes listed below:
x Subjective
x Participatory
x Interpreted and communicable
x Cross-checked and compared
x Empowering
x Diverse and disaggregated
3.6 Design Indicators

Indicators are essential instruments for M&E, thus, practitioners need to
keep in mind some of the critical points while designing or formulating
them.
Firstly, creating new indicators or reinventing the wheel should be avoided

till the time it is absolutely required. Over the years, development
professionals have worked to provide M&E practitioners with sets of well
tested and proven indicators and these sets of indicators should be referred
to while formulating indicators for any project.
Secondly, while designing indicators, it should be made sure that they

fulfil either the SMART or the SPICED criteria. Indicators can document
change therefore, any indicator finalised should essentially be able to
capture change in the condition that is being assessed using the indicator.
A good indicator is therefore:
x Simple: As all the good things in the world are

x Measurable: Provides a measure for depicting change
x Precise: Has a definition so that it can be defined in the same way
by all
x Consistent: Has consistent measurement results. On measuring the
same thing, its value remains consistent and does not change over
time
34 Chapter Three
x Sensitive: Can capture the smallest amount of change in the

indicator value
x Action Focused: Captures information that is eventually useful for
stakeholders and leads to some action.
While designing indicators, it is very important to collectively brainstorm

to identify candidate indicators for a specific condition. Once several
indicators are listed for a given specific condition, the next step is to assess
each of the indicators using the characteristics of a good indicator to find
out whether the candidate indicator is simple, measurable, precise,
consistent and sensitive. The source of data for the indicator and the
reliability of the sources is also considered. The cost incurred in collecting
data for this indicator is also considered while finalising the indicators.
Candidate indicators that satisfy the criteria are then taken as indicators for
assessment of that condition. Candidate indicators are also modified till
they acquire the characteristics of a good indicator.
For example, in the case of the project which aims to make its target area
ODF, the output level indicator is ‘The number of individual household
latrines (IHHL) constructed’. As constructing toilets is one of the key
outputs expected from the project, this indicator helps in measuring the
same. Similarly, as creating awareness about sanitation is another key
activity, ‘The number of village level meetings conducted to create
awareness about sanitation’ is another output level indicator.
Considering another example of a project which aims to improve maternal

and child health (MCH) in its target area, ‘Maternal Mortality Ratio’
(MMR) and ‘Infant Mortality Ratio’(IMR) are the result level indicators
for this project. At the outcome level, the indicators are, ‘The number of
women with incidences of serious health problems related to child birth’
and ‘The number of women consuming iron fortified food or iron
supplements during pregnancy’. At the output level, the indicators are
‘The number of deliveries conducted by skilled health professionals’, and
‘The number of women receiving at least three antenatal care (ANC)
visits’.
Defining Indicators
After selecting suitable indicators, it is very important to fully define them.

No indicator should be deployed without fully defining it and making sure
its essential components are lucid and concrete (UNAIDS, 2010). Each
indicator definition should have the following components:
x Title: A brief heading that captures the summary or focus of the

indicator.
x Definition: A lucid and to the point definition of each indicator so
that everyone can interpret it in the same way.
x Source: The source i.e., the tool used for getting this indicator value
and the respondent from whom this information is collected is also
defined.
x Data Collection Frequency: The frequency at which the data is
collected to derive at the indicator value is defined. This could be at
quarterly, half yearly or annual intervals etc.
x Numerator: The variable that is included above the line in a
common fraction.
x Denominator: The variable that is included below the line in a
common fraction.
x Calculation Method: The method for calculating the indicator value
is defined.
For instance, the complete definition of the indicator, ‘The number of

deliveries conducted by skilled health professionals’, is stated below:
Title: The number of deliveries conducted by skilled health professionals
Definition: This indicator measures the number of deliveries conducted by

skilled health professionals. The term skilled health professional refers
exclusively to people with midwifery skills (auxiliary nurse midwives
(ANMs), doctors and nurses) who are trained in the skills necessary to
manage normal delivery cases and diagnose, manage or refer obstetric
complications
Source: The sample survey conducted as part of the baseline, midline and
endline survey
Data Collection Frequency: In the first quarter of the first year and in the
last quarter of the third and in the fifth year of the project.
Numerator: The total number of deliveries attended by skilled birth

attendants (SBA) as reported in the sample survey.
Denominator: The overall sample size of the number of women who had
deliveries in the last two years.
36 Chapter Three
Calculate: This indicator value is calculated by dividing the number of

births attended by skilled health professionals by the total sample size of
the number of deliveries conducted in the last two years.
CHAPTER FOUR
DEVELOPING A PERFORMANCE
MONITORING PLAN
To ensure that a development project achieves its objective, it is very

important to monitor its performance simultaneously. For this, a
systematic collection and analysis of the project performance data is done,
and based on this, feedback is provided to the project management staff. In
this chapter the practitioner will understand what performance monitoring
is and why it is important. Subsequently, the practitioner will learn what a
Performance Monitoring Plan (PMP) is and about the tools used by
development practitioners for project performance monitoring. The
practitioner is taken step by step through the process of developing a PMP
while understanding its key elements. Finally, the practitioner will learn
about some key aspects which are taken care of to ensure the quality of the
monitoring data.
At the completion of the chapter the practitioner will be able to:
" Understand the fundamentals and key concepts of a PMP

" Define the key components of a PMP
" Develop a PMP
" Understand the key levers for ensuring the quality of the
monitoring data
4.1 What is Performance Monitoring Plan?

Project management is about managing 'what we are doing' in the domain
of activities. The results of these activities are outputs, outcomes and
impact which depict the performance of the project. Performance
management is, therefore, geared towards 'what has been done' in terms of
the results achieved in a project. It is to ensure that project objectives and
goals are being met consistently in an effective and efficient manner.
Unlike project management which is limited to process management or

managing project activities, time and cost, performance management is
38 Chapter Four
about managing project results. In project management, the starting point

is activities and how to better manage or organise them to achieve the
required results. However, in the case of performance-based results
management, results become the basis on which all project planning is
done, wherein the result that is to be achieved determines the activities that
are to be carried out.
By developing the results chain, a conceptual map of the process of

change is developed within the project. To track the performance of the
project, it is necessary to track the project results i.e., project outputs,
outcomes and impact. Thus, a PMP is needed to capture and monitor
project performance. PMP, therefore, is a tool by which the outcomes or
objectives are monitored and evaluated concurrently.
Once the results chain is developed, it is essential to translate it into a PMP

that can monitor and assess the achievement of project results. This helps
in managing the results by continuously providing evidence on the
performance of the project.
PMP is also called Performance Measurement Framework (PMF) or the

Results Framework. PMP serves as the reference document for concurrent
performance monitoring of the project. A typical PMP format is shown in
the matrix below.
Key components of the PMP are mentioned below:
1. Statement of expected results in their hierarchy-outputs, outcome

and impact - or the conditions that are to be achieved
2. Indicators informing what information to seek so that the
practitioner knows that the condition is achieved
3. Periodicity describing when the measurement of the indicator is
made
4. Source from where the information is received at the defined
periodicity
5. Baseline or the starting value of the indicator before the
intervention situation or condition
6. Milestones that are planned for the changed condition, as the
project moves forward
7. Target of the condition to be achieved at the completion of the
project
8. Responsibility of who will fetch the information at the desired
periodicity from the defined source.
Developing a Performance Monitoring Plan 39
Table 1.1: Project Monitoring Plan Format
Expected results Indicator Periodicity Source Base Milestones Target Responsibility

line Y1 Y2 ....
Value
Impact
Outcome
Outputs
1.............
2.............
40 Chapter Four
4.2 How to develop a Performance Monitoring Plan

The practitioner has understood by now that PMP is a tool for developing
an overarching plan for managing the performance of a project, which
includes activities for monitoring and evaluating the project. Using this
tool a plan is formalised for collecting and monitoring the project
performance related data (USAID Centre for Development Information
and Evaluation, 1996). PMP, at the bare minimum, includes the following
details:
1. Identifying performance indicators at various levels and

subsequently defining them.
2. Indicator-wise details like data sources, method of collection,
frequency, timing etc.
3. Building teams and delineating roles and responsibilities for
supervision and collection of performance data.
4. Plan for analysis of performance data.
5. Developing processes to ensure that it is reported and further used
to provide feedback to the programme for informed decision
making.
A step by step process is given below for developing a PMP:
Step 1: Populate the Expected Results Column
In the Results Chain developed in Chapter 2, all levels of results viz.,

impact, outcome and outputs were defined. As the first step of developing
a PMP, the outputs and outcome from the Results Chain or the PPM
(Project Planning Matrix) are put in the respective Expected Results
column at the assigned level. Outputs and outcomes are focused on as the
project is accountable for their achievement. Inclusion of impact in the
Expected Results column is kept optional as a project always contributes
towards the impact but does not achieve it by itself.
An example of a Reproductive, Maternal, Newborn and Child Health

(RMNCH) focused project is used to develop its PMP. The project aims to
improve MCH in its target area by increasing community awareness and
sensitising them on maternal health issues as well as by increasing the
accountability of government institutions in implementation of their
programmes. The project will create village level monitoring committees
for spreading awareness about the critical MCH care practices. It will also
hold the government health system accountable through advocacy to
ensure that beneficiaries have access to their entitlements under the

national flagship health programmes. The project also aims to encourage
participatory formulation of the Annual Implementation Plan (AIP) for the
government plan at the district level.
The first step of populating the expected results column is given below:
Expected Results Indicators …...................... Responsibility

Impact
Ensuring universal
access to maternal
health services for
achievement of MDG-5
Outcome
Improved status of
women’s health in the
project target area
Output
1. Women have
increased access to
obstetric care
2.Increased capacity of
community to demand
for the rights and
entitlements under
government health
programmes
3. Increased awareness
and knowledge about
important maternal and
child practices and
contraception methods
4.Participatory
formulation of AIP in
consultation with
village health
committees (VHCs)
(Practice Level Change)
42 Chapter Four
Step 2: Develop Indicators for Results at Each Level
As mentioned above, during its implementation, change in result level

indicators of the project is tracked and measured through the PMP.
Indicators provide information regarding change in the condition and its
magnitude of change. In Chapter 3, the practitioner learnt about the
essential characteristics of a good indicator and how to design and define
them.
Developing or selecting suitable indicators remains a challenging task in

any performance measurement framework. For example, poverty-
reduction can be measured by using multiple indicators ranging from per
capita income to calorie consumption. Depending upon the indicator
selected, information that is collected is either income or kilocalories
consumed at time t1 (start of the project) and time t2 (end of project),
which depicts whether the condition of poverty had increased or decreased
or remained the same over a period. It needs to be reiterated here that it is
only the condition that is changing, with the indicator simply pointing out
whether it has increased or decreased or stayed the same over a period of
time. Being a unit of information, an indicator, by its very nature, is value-
neutral and can at best be an approximation. Based on the SMART criteria
of developing an indicator, a suitable indicator for each expected result is
selected. It is also important that there are complete definitions of each
indicator in order to avoid confusion and create consistency. For each level
of result, it is suggested that there should be a minimum of one indicator
and a maximum of three. More than three indicators imply that it is not
clear what the project is trying to achieve as its result.
In the example used, indicators are defined in the next step for developing
the PMP and at each level of the results, they are populated with impact,
outcome and output.
An example of the PMP after identification of suitable indicators is given

below:
Table 1.3: Developing PMP-Example
Expected Indicators ......................... Responsibility

Results
Impact Maternal Mortality
Ensuring Ratio (MMR)
universal
access to
maternal
health services
for
achievement of
MDG-5
Outcome Number of women
Improved reporting serious
status of health problems
women’s health related to child
in the target birth
project area Number of women
consuming iron
rich/ fortified
foods or
supplements
Output Number of births
1. Women with attended by skilled
increased health
access to professionals
obstetric care Number of women
benefitted under
the Janani
Surkasha Yojana
(JSY) scheme
Number of
primary health
clinics (PHCs)
with referral
services for
complicated
pregnancies
44 Chapter Four
Number of women
with access to
contraceptives and
safe abortion
services
2.Increased Number of village
capacity of level communities
community to formed which are
demand for the capable of
rights and demanding
entitlements entitlement under
under the national
government government health
health programme
programmes
Number of VHCs
which are capable
of registering
complaints with
the health
department in case
any entitlement
under the
government health
programme is not
received
3. Increased Number of women
awareness and who are able to
knowledge articulate the key
about danger signs
important during pregnancy
MCH practices Number of women
and who have
contraception knowledge of
methods temporary method
of contraception
Number of women
who are aware
about the
availability of
contraceptive and
abortion services at
various levels of
the public health
system
4.Participatory
formulation of
annual
implementation
plan (AIP) in
consultation
with VHCs)
(Practice Level
Change)
Step 3: Assign Periodicity, Identify Sources and Assign Responsibility
The next step is to decide the period or frequency at which each indicator
in the PMP is to be measured. Periodicity is decided based on factors like
the minimum time that is required for change to be visible in the condition
under observation, the cost of data collection of a specific indicator,
requirement of indicator etc. For e.g., if the change is expected within a
short duration, the periodicity can be low such as quarterly or half-yearly.
If the project intervention is expected to take a longer duration to have an
effect on the indicator value, then the periodicity would be higher, say
annual or biannual. In the case of indicators for which data collection is a
complex process or where special investigating skills are required which
make the data collection costly, e.g., in agriculture production, testing the
haemoglobin (Hb) level in blood etc., it is better to collect this data over a
longer periodicity or for a smaller sample.
Apart from assigning the periodicity, it is important to identify the source

from where the indicator information is drawn. The exact source of this
information is reported so that anyone can access and understand this
information. Also, if there are multiple sources for availing the indicator
information, then at least two of those sources, which seem to be the most
credible and accessible, are mentioned. The primary sources of indicator
value can be census surveys, sample surveys or concurrently collected
project MIS data. The secondary sources of information can be past
sample surveys or government census surveys or some secondary MIS
data. While using secondary data, it is important to consider the gap
between the time of the actual data collection of the secondary survey and
46 Chapter Four
the time when it is being used. In case the time gap is substantial, it is not
advisable to use this secondary data source. Data sources are decided
while considering factors like availability, reliability, cost etc.
It is important to assign the responsibility of gathering information specific

to each indicator to someone from the project team, to a stakeholder or to
an external M&E team if deployed.
The PMP of the sample project with its periodicity, source and the
person/agency responsible for gathering information is populated in the
table below:
Expected Results Indicators Periodicity Source ...... Responsibility

.......
Impact MMR Annual Sample Monitoring,
Ensuring universal Registration Evaluation and
access to maternal Survey Learning (MEL)
health services for Manager
achievement of MDG-5
Outcome Number of women Annual HH Structured Designated
Improved status of reporting serious health Interview, Project Research Agency
women’s health in the problems related to Annual Survey
project target area childbirth (PAS)
Number of women Annual HH Structured Designated

consuming iron rich/ Interview, PAS Research Agency
fortified foods or
supplements
Output Number of births attended Annual HH Structured Designated
1. Women with by skilled health Interview, PAS Research Agency
increased access to professionals
obstetric care Number of women who Annual HH Structured Designated
benefitted under the JSY Interview, PAS Research Agency
scheme
48 Chapter Four
Number of primary health Annual CHC/PHC Designated

clinics (PHCs) with Register Audit, Research Agency
referral services for PAS
complicated pregnancy
Number of women with Annual HH Structured Designated
access to contraceptives Interview, PAS Research Agency
and safe abortion services
2.Increased capacity of Number of VHCs formed Annual Project MIS MEL Manager
the community to which are capable of
demand for the rights demanding entitlement
and entitlements under under the national
government health government health
programmes programme
Number of VHCs capable Annual VHC Designated

of registering complaints Representative in- Research Agency
with the health department depth interviews
in case any entitlement (IDIs), PAS
under government health
programme is not received
3. Increased awareness Number of women who Annual HH Structured Designated
and knowledge about can articulate the key Interview, PAS Research Agency
important MCH danger signs during
practices and pregnancy
contraception methods Number of women with Annual HH Structured Designated

knowledge of temporary Interview, PAS Research Agency
methods of contraception
Number of women who Annual HH Structured Designated
are aware about the Interview, PAS Research Agency
availability of
contraceptives and
abortion services at
various levels of the public
health system
4.Participatory
formulation of AIP in
consultation with VHCs
(Practice Level Change)
50 Chapter Four
Step 4: Put Milestone Values for each Indicator from the Baseline to
the Endline of the Project
For monitoring the performance of a project and ensuring that it is result

oriented, it is very important to define a starting point for gauging or
calculating the change and tracking it intermittently. The baseline is the
value of the indicator at or just before the start of the project. This acts as a
reference point for assessing change in the condition after the project has
been initiated.
After putting in the baseline value, key milestones and target values are
assigned at specific time gaps for the indicators. The target is a posteriori
condition that is envisaged at or after completion of the project. In other
words, it is the concrete result that is sought at the end of the project
efforts. Milestones are the benchmark values that are hoped to be achieved
at specific points of time within the project implementation years. The
milestone values are to be assigned based on the way the project
implementation has been planned. For instance, for a five-year project,
formative work is undertaken or the foundation is built during the first
year of project implementation. The implementation intensity picks up
from the second year and peaks during the third or fourth year. The fifth
year is usually reserved for the handing over of the project after ensuring
its sustainability. Therefore, the project milestones are set according to the
planned pace or intensity of work.
The PMP of the sample project after the key milestones values are
assigned is given below:
Expected Indicators Periodic Source Baseline Milestone Milestone Milestone 3 Respons-

Results ity 1 (Year2) 2 (Year 3) (Year 5) ibility
Impact MMR Annual Sample 200 MMR to be MMR to MMR to be MEL
Ensuring Registration brought be brought Manager
universal Survey down by brought down by
access to 5% from down by 25% from the
maternal the baseline 10% from baseline
health value in the the value in the
services for target area baseline target area
achievement value in
of MDG-5 the target
area
Outcome Number of women Annual HH Structured 30% 27% 20% 10% Designated
Improved who are reporting Interview, PAS Research
status of serious health Agency
women’s problems related to
health in the childbirth
target project Number of women Annual HH Structured 5% 8% 15% 30% Designated
area who are consuming Interview, PAS Research
iron rich/fortified Agency
foods or supplements
Output Number of births Annual HH Structured 70% 75% 80% 90% Designated
1. Women which are attended by Interview, PAS Research
with skilled health Agency
increased professionals
access to
obstetric care Number of women Annual HH Structured 40% 45% 50% 60% Designated
who benefited under Interview, PAS Research
the JSY scheme Agency
52 Chapter Four
Number of PHCs with Annual CHC/PHC 2 PHCs 5 PHCs out 10 PHCs 20 PHCs out Designated
referral services for Register Audit, out of 30 of 30 PHCs out of 30 of 30 PHCs Research
complicated PAS PHCs have PHCs have referral Agency
pregnancy have referral have transport
referral transport referral support
transport support transport
support support
Number of women Annual HH Structured 5% 10% 15% 25% Designated
with access to safe Interview, PAS Research
abortion services Agency
2.Increased Number of VHCs Annual Project MIS Out of Out of 50 Out of Out of 250 MEL
capacity of formed which are 500 intervention 100 intervention Manager
community to capable of demanding interventi villages, interventi villages,
demand for entitlement under the on none have on none have
the rights and national government villages, VHCs villages, VHCs
entitlements health programme none have none have
under VHCs VHCs
government
health Number of Village Annual VHC Zero 50 100 250 Designated
programmes Health Committees Representative, Research
that are capable of IDIs, PAS Agency
registering complaints
with the health
department in case any
entitlement under
government health
programme is not
received
3.Increased Number of women Annual HH Structured 15% 20% 25% 50% Designated
awareness who can articulate the Interview, PAS Research
and key danger signs Agency
knowledge during pregnancy
about Number of women Annual HH Structured 30% 35% 45% 70% Designated
important with knowledge of at Interview, PAS Research
MCH least two temporary Agency
practices and methods of
contraception contraception
methods
4.Participator
y formulation
of AIP in
consultation
with VHCs
(Practice
Level
Change)
54 Chapter Four
Step 5: Develop Measurement Protocols for Behavioural Change
By following Step 2 to Step 4, physical and financial changes sought

through the project are taken care of and tracked via their respective
indicators. However, different measurement protocols or systems for
behavioural outputs are needed, which are essentially practice-level
changes that are sought through the sample project. At the same time, it is
to be noted that measurement protocols are developed for practice level
indicators only in cases where the project aims to achieve institutional
changes or behavioural changes.
In the case of projects where the perceived behavioural change falls in the
realm of outputs or outcomes which are not physically observable or
verifiable, outcome mapping serves as a vital tool to map and track
process-level changes along the pathway of change. Using a graded
measure of change known as 'progress markers', it seeks to unpack the
multiple layers of change.
Progress markers are statements describing the practice or process- level

changes that lead to the envisioned outcome. For example, the key
progress markers for a project aiming to improve participative governance
and budgeting at the panchayati level are: transparency in public budgets,
public participation in the budget formulation process, simplification of
budget documents, social audit of expenditure by the public etc. All these
constitute key practice-level demands that require institutional change.
These are best mapped through progress markers and ideally a project
should not have more than five progress markers.
These progress markers define and describe the milestone changes that are
required to achieve the behavioural change from the current situation to
the envisaged situation. These are progressive milestones, ranging from
the low-hanging fruit to the higher-order or difficult to attain changes. The
first milestone is an 'expect to see' yardstick, which refers to the minimally
acceptable level of change or first-level change expected from the
intervention. The second milestone is what is called 'like to see' or the
change that could realistically be achieved, while the final milestone is
known as ‘love to see’ or the most desired or ambitious change that is
foreseen. To use a real-life example, passing an exam is the minimally
acceptable or what we 'expect to see', whereas getting a first division is
what we would 'like to see' and finally, to get a first-class distinction is
what we would ‘love to see’.
These progress markers are analogous to milestones and targets for

indicators and their periodicity, source and the person/agency responsible
for gathering information is assigned accordingly.
The PPM of the example project, is presented below after adding the its
behavioural outcomes and related progress markers
56 Chapter Four
Table 1.6: Developing PMP Example

Expected
Results Indicators Periodicity Source Baseline Milestone Milestone Milestone Responsibility
1(Year2) 2(Year 3) 3(Year 5)
Impact MMR Annual Sample 200 MMR to be MMR to be MMR to be MEL
Ensuring Registration brought brought down brought down Manager
universal access Survey down by by 10% from by 25% from
to maternal 5% from the baseline the baseline
health services the baseline value in the value in the
for achievement value in the target area target area
of MDG-5 target area
Outcome Number of Annual HH Structured 30% 27% 20% 10% Designated
Improved women who Interview, Research
status of are reporting PAS Agency
women’s health serious
in the target health
project area problems
related to
childbirth
Number of Annual HH Structured 5% 8% 15% 30% Designated
women who Interview, Research
are PAS agency
consuming
iron
rich/fortified
foods or
supplements
Output Number of Annual HH Structured 70% 75% 80% 90% Designated
1. Women with births which Interview, Research
increased access are attended PAS agency
to obstetric care by skilled
health
professionals
women who Interview, Research
benefitted PAS Agency
under the
JSY scheme
Number of Annual CHC/PHC 2 out of 30 5 out of 30 10 out of 30 20 out of 30 Designated
PHCs with Register PHCs have PHCs have PHCs have PHCs have Research
referral Audit, PAS referral referral referral referral Agency
services for transport transport transport transport
complicated support support support support
pregnancy
women with Interview, Research
access to safe PAS Agency
abortion
services
2.Increased Number of Annual Project MIS Out of 500 Out of 50 Out of 100 Out of 250 MEL
capacity of the VHCs intervention intervention intervention intervention Manager
community to formed villages, none villages, villages, none villages, none
demand for which are have VHCs none have have VHCs have VHCs
rights and capable of VHCs
entitlements demanding
under entitlement
government under the
health national
programmes government
58 Chapter Four
health
programme
Number of Annual VHC Zero 50 100 250 Designated

VHCs which Representative Research
are capable IDIs, PAS Agency
of registering
complaints
with the
health
department
in case any
entitlement
under the
government
health
programme
is not
received
3.Increased Number of Annual HH Structured 15% 20% 25% 50% Designated
awareness and women who Interview, Research
knowledge can articulate PAS Agency
about the key
important danger signs
MCH practices during
and pregnancy
contraception Number of Annual HH Structured 30% 35% 45% 70% Designated
methods women with Interview, Research
knowledge PAS Agency
of at least
two
temporary
methods of
contraception
Output(Practice Progress Periodicity Source Expect to See Like to See Love to See
Level Change) Marker
4.Participatory District wise State Inputs of VHC Representative
formulation of AIP government representatives of VHC is
AIP in planning officials are taken included in the
consultation including while executive body
with VHCs VHC developing an which
representative, AIP develops the
AIP district level
development AIP
meetings
60 Chapter Four
To sum up, in this sub chapter, the practitioner has learnt to develop a
PMP, which is a plan to monitor and manage results, by following a step-
by-step process.
4.3 Ensuring Quality of Monitoring Data

It is said that the quality of project monitoring is as good as the quality of
data collected. It is very important to collect good quality data so that the
programme management team and other important decision makers can
trust it and make use of it for tracking and improving the programme.
Good quality data should be accurate, complete, consistent (across
different sources), timely, useful, precise, and accessible.
Robust data quality assurance mechanisms and regular data auditing

ensures the veracity of the monitoring data.
Data Quality Assurance
For data quality assurance, systems, document protocols and guidelines for
ensuring monitoring data quality are developed throughout the various
project stages. Checks and procedures are also defined right across the
various stages viz., designing of monitoring formats, translation of
monitoring formats, during data collection, data entry or digitisation etc.
Key aspects that should be included while developing a data quality
assurance plan are listed below:
1. Listing quality assurance mechanisms to be followed at all project

stages.
2. Assigning roles and responsibilities for data collection, data
auditing and deciding their frequency.
3. Conducting concurrent data auditing and investigating the reasons
for data variance if any.
4. Calibration of the monitoring and reporting formats to check their
precision, and translation, while ensuring they are not biased in any
way.
5. Validation of data and data cleaning systems.
6. Conducting training to build the capacities of the data collection
team and to maintain consistency in the way the tools are
administered.
7. Defining roles and responsibilities for data collection and data
auditing.
8. Defining data quality benchmarks and action to be taken in case the

collected data falls below these benchmarks.
Data Auditing
As part of quality assurance, it is also necessary to devise the plan for

conducting data auditing. Data auditing is an exercise by means of which
the veracity of data is checked. Data auditing is done primarily using two
techniques-one, by conducting spot-checks and two, by conducting back-
checks. In spot-checks, the data quality auditor is present at the time when
the data collector is collecting data. In this way, real time feedback can be
provided to the data collector on tool administration or other related issues.
Spot-checks done during the starting phase of data collection help to assess
how survey tools are being administered by the data collectors. In back-
checks, the data auditor goes to a randomly selected sample of respondents
from whom data has been collected by the main survey team. It is
suggested that there should be a separate team for conducting back-checks,
and these should be conducted immediately, i.e., on the very next day of
the data collection. Both data sets are then matched to find if there is any
variance between the two sets. This helps in gauging the quality of the data
being collected, and based on this, improvement or rectification measures
are taken up as and when required. At the same time, it is important to
keep in mind that mostly factual questions should be used as part the back-
check tool so that the possibility of variance due to different responses by
the respondent or any other external factor is at a minimum. Triangulation
with other data sources is also useful for auditing the quality of the data.
CHAPTER FIVE
APPROACHES TO MONITORING
In the previous chapter the practitioner learnt how to develop a project

PMP. In this chapter, the practitioner will learn about the different
approaches that are adopted for monitoring a project, about progress and
process monitoring, how monitoring can be done in a participative way
while involving all project stakeholders, and how to conduct monitoring
without using indicators.
At the completion of the chapter, the practitioner will be able to:
" Understand and explain the definition and key concept of progress
and performance monitoring
" Understand the definition and concept of Participatory Monitoring
" Explain the concept of Most Significant Change (MSC) and
Outcome Mapping
" Understand and explain the key levers for ensuring quality of
monitoring data
5.1 Progress and Process Monitoring

In this section, the practitioner will learn why progress and process
monitoring are important, what the key differences between them are and
how they can be implemented for monitoring a project.
Progress Monitoring
Progress monitoring, as the name suggests, aims to assess the progress of

a project towards its objectives and target milestones. It is advisable to do
progress monitoring concurrently or intermittently along with the project
implementation to ensure that the project is on track. Progress monitoring
is an ongoing process which involves collecting and analysing the output
level or even outcome level indicators related to the project to see if the
project is on track to deliver its planned results.
Approaches to Monitoring 63
Taking the example of a project which aims to eradicate OD and improve

sanitation by building individual household latrines (IHHL)in the project
area, one lakh HHs are identified without toilets using a baseline study.
The project aims to cover this deficit over a period of five years. During
the project planning stage, it was decided that the project would build ten
thousand toilets in the first year, twenty thousand in the second year,
twenty-five thousand in the third and the fourth year respectively and
twenty thousand in the fifth year.
In this case, through progress monitoring, a tab can be kept concurrently

on the number of toilets that are constructed per year. Through progress
monitoring, the number of toilets constructed is assessed and then it needs
to be seen if the project can achieve the numbers it has planned to achieve.
Based on this, feedback is provided to the project implementers to check
whether the project is on track or whether it is lagging. In progress
monitoring, the focus is more on the quantity of target achievement in
comparison to its quality.
Project MIS (Management Information System) data is generally the

source of data required for progress monitoring. This should ideally be
done by the project management team itself.
Process Monitoring
Process monitoring, as the name suggests, includes monitoring of the

processes and the activities done as part of the project implementation. Its
objective is to focus on the quality of the implementation rather than
focusing only on the targets or the milestones achieved by the project.
As an initial step, the processes that need to be adopted for implementation

of each project activity are delineated and listed. Activity-wise, this is an
ideal process and sequence in which each activity that is to be
implemented is first envisaged so that the desired results are achieved. A
process intensive approach is required for ensuring quality and
sustainability of project outcomes.
A process check-list is developed by making a list of all the steps that are
followed as part of the ideal process implementation. Process monitoring
is usually done using these process checklists. The activities and the
processes are observed and recorded on the checklist. In case any
deviation is observed from the ideal required process, it is recorded.
64 Chapter Five
For example, to monitor the process of subsidy transfer to beneficiaries for

construction of IHHL, at first, all the steps are listed in a sequential order.
1. The beneficiary should take the required photographs of the toilet

site before, during and after its construction.
2. Get the toilet constructed.
3. Go to the project district office and intimate them about the toilet
construction and get the form for availing subsidy.
4. Fill the form requesting for the subsidy amount and attach the
following documents:
a. Three photographs of the various stages of toilet construction as
mentioned above.
b. Identification card of the beneficiary to prove that the
beneficiary is a resident of the mentioned area.
c. A verification letter from the sarpanch that the beneficiary's HH
does not have a toilet.
d. Bank account number in the name of the beneficiary.
5. The form is submitted to the district project office after being duly
filled.
6. The subsidy should be transferred to the beneficiary's account
within 15 days of the receipt of the application form at the district
office.
Based on the above-mentioned steps, a checklist for process monitoring is

prepared. Using this checklist as a reference, the processes that need to
be followed as part of the project activity are monitored. Deviation from
the ideal process is also recorded. As part of process monitoring, the
reason for this deviation is assessed and the feedback is given to the
project management for improvement of the programme.
5.2 Participatory Monitoring and Evaluation

Participatory M&E, as the name suggests, is a process through which all
the project stakeholders at various levels are engaged in monitoring and
evaluating a project or a programme. Unlike conventional M&E, in
participatory M&E, all stakeholders of the project including donors,
implementation agencies, primary stakeholders and other stakeholders
share control over the process, content and results of the M&E activity.
Also, all stakeholders are involved in identifying or taking corrective
action in case any issues or deviation is observed (Forster, 2002).
Participatory M&E’s core principle is that it considers the project

beneficiaries as active participants and not just information providers. It
believes that their capacities should be built so that they can be actively
engaged in analysing and reflecting on the project performance so that the
performance can be improved based on the inputs from the ground. During
participatory M&E, all stakeholders are engaged throughout all stages of
the M&E, which include formulating the M&E framework, information
collection, collation, analysis, interpretation and finally, decision making.
Why is Participatory M&E necessary?
Why is Participatory M&E (PM&E) necessary and what are its

advantages? Engaging all stakeholders has increasingly been considered as
important for the M&E process as it offers new ways of assessment and
deriving learning from the project. PM&E makes the M&E process more
inclusive i.e., internal and contextual to the programme and also more
responsive to the needs and expectations of the primary stakeholders of the
project.
PM&E aims not only to measure the effectiveness of a project but also
ensures accountability to its beneficiaries and increases transparency in the
M&E process of the project. Making the process participatory empowers
the project beneficiaries and inculcates a sense of ownership amongst the
stakeholders of the project. Also, making the process participatory
buttresses the process of formulating corrective action for improving the
performance and outcomes of the project.
PM&E also helps to build the beneficiaries’ commitment and increases

their understanding of the design, planning and implementation of the
project.
Steps in conducting Participatory M&E
Key steps involved in conducting PM&E are:
Step I: Identify the Key Stake holders
As the first step, all the stakeholders in the project, right from the project
funders, project implementers, government agencies, other key boundary
partners and its target beneficiaries are identified to be included in the
M&E process.
66 Chapter Five
Step II: Developing a framework for PM&E
After identifying the key stakeholders, the next step is to develop a

framework for PM&E. This is one of the most important and challenging
stages as it involves getting all the stakeholders together on board for the
first time to provide their inputs, raise their concerns and expectations
from the process (Gaventa). At this stage, the objective of the PM&E is
established. Usually there are lot of deliberations and negotiations between
the stakeholders at this stage. This process can answer key fundamental
questions like:
x What are the resources and limitations at hand for the PM&E?
x What is the role and the responsibility of each stake holder?
x What information is required?
x From whom is this information to be collected?
x How is it to be incorporated to improve the programme
implementation?
Step III: Developing the Indicators and Performance Monitoring Plan
As in the case of conventional M&E, after formulating the objective, the

next step is to develop the indicators at each level of the Results Chain and
define them fully. While developing the indicators in PM&E, special
thought needs to be given to who the end users of this information are, and
how it can be used to further improve the programme. The indicators
should be developed keeping in mind the SMART criteria which was
elucidated in Chapter 3.
Subsequently, the project PMP and the matrix will be developed as it has
been done in Chapter 4. All the stakeholders are jointly involved in
developing the PMP. Also, the responsibilities are distributed among all
stakeholders including the beneficiaries, unlike in the case of the
conventional M&E.
Step IV: Collecting Data
After the objective and framework are finalised, the next critical step is to
determine how to go ahead with the data collection. For this, the tools and
techniques used for data collection are identified. A wide range of tools is
available for this and a tool based on the context and requirements of the
project is selected. After the tool is selected, the actual process of data
collection is started. In the PM&E framework, it is already detailed who

will do the data collection.
Step V: Data Analysis
The next step after the data collection is to analyse this collected data.
Unlike in conventional M&E where analysis is usually done by a third
party or by funders or implementers of the project, in PM&E, all
stakeholders, including the project beneficiaries, are involved in
conducting the data analysis. Workshops or joint meetings are conducted
in which all stakeholders analyse the data collectively, mull over the
problems and constraints in achieving the project objective, look for
solutions and examine the results of the project activities.
Step VI: Programme Reporting and Feedback
After the data is analysed and insights derived from it, the last step is to
document and report the process and the information generated from the
PM&E process. It is also very important to share and disseminate this
information to all stakeholders and relevant external audience. Based on
the insights derived from this process, feedback and recommendations are
given to the project implementation team for further improvement and
course correction.
Limitations
Though there are various advantages in conducting PM&E, there are also
some practical limitations while applying the same. Conducting a
participatory evaluation which fulfils the core objectives requires a lot of
commitment and focus from the donors and the implementers. Conducting
such an evaluation is very resource intensive, both in terms of time and
finance. Moreover, adept professionals are required to facilitate this kind
of process.
5.3 Monitoring without Indicators using the Most

Significant Change technique
Most Significant Change
The Most Significant Change (MSC) technique is a participatory

technique that is used in both monitoring and evaluation. It was invented
by Rick Davis in an attempt to meet some of the challenges associated
with monitoring and evaluating a complex participatory rural development
68 Chapter Five
programme (Davis, 2005). This programme, run by the Christian

Commission for Development in Bangladesh (CCDB), had diversity in
both implementation and outcomes. MSC is a participatory technique in
which all stake holders are first involved in deciding the criteria of change
to be recorded, then in collecting these stories of change and lastly, in
analysing them. It is essentially a type of monitoring tool as this activity is
done through the project cycle to provide feedback to programme
managers about the direction of the programme. MSC also contributes to
the evaluation as it provides information about the project i.e., whether it
is moving towards achieving its intended long term results while
commenting on the effectiveness of the programme.
The process involves collection of significant change (SC) stories from the
grassroots level. The MSC stories are then shortlisted by the panel of
designated stakeholders or the project team. Subsequently, various
representatives sit together and read out these SC stories and further
discuss the value of these stories.
When should MSC be used?
While learning about a technique, it is of utmost importance to know when

the technique should be used and when it should not be used.
Effectiveness or suitability of MSC for a programme depends upon the
programme context, its objective and its implementation strategy. MSC is
suitable for programmes that:
x Are complex and have multiple outcomes

x Aim at creating behaviour and social change
x Struggle with conventional monitoring systems.
MSC is more suitable for monitoring because it is focused on deriving

learning for the improvement of the programme rather than just its
accountability.
The MSC technique is implemented in the programme using the ten steps
formulated by Rick Davis in 2005. They are:
1. Getting started: Establishing champions and getting familiar with

the approach. This step includes introducing this technique to the
key people and groups and convincing them how useful it can be
for the programme while being easy to implement. After all
stakeholders have agreed to adopt this technique then the
champions, who will act as facilitators and catalysts to involve

others and further facilitate the process, need to be identified.
2. Establishing the domains of change: The second step is to identify
the domains of change in the programme, which are broad
categories in which SC stories are identified. Unlike in
conventional M&E where the indicators need to be SMART,
domains of change are left open ended and fuzzy so that different
stakeholders can interpret them in their own way.
3. Defining the reporting period: The next step is to define the period
or frequency of collecting and reporting these stories. This period
could vary from a fortnight to a year depending on the rigour
required by the programme team and donors, although quarterly
reporting is considered the most common.
4. Collecting stories of change: The next critical step is collecting
these stories of SC by asking the respondents what according to
them has been the MSC that the project has brought about in the
quality of life of its intended beneficiaries. These stories are
captured in different ways by fieldworkers during work, through
interviews conducted by people designated for the purpose, through
group discussions or by encouraging beneficiaries to write their
stories directly etc. The key information documented includes the
background information of the person who collected the story, the
context of the story, the narrative, and why the storyteller thinks
this story describes a SC.
5. Reviewing the stories within the organisational hierarchy: The next
step is to review and shortlist the most significant stories from the
pool of stories collected. The stories are reviewed by a group of
people who select the most significant stories amongst them and
forward them to their seniors in the hierarchy. It is very important
to decide and document the criteria, the process, and the people
involved in the process.
6. Providing stakeholders with regular feedback about the review
process: As an objective that is imperative in any MLE tool, MSC
also aims at providing useful feedback to the project team.
Information regarding the MSC observed on the field is fed back
into the system. This helps the team in handling the programme to
qualitatively understand the perceived significant changes that have
been brought into the people’s lives. This feedback is provided
using different methods like formal reports, newsletters, seminars
or through public meetings etc.
70 Chapter Five
7. Verifying the authenticity of stories if necessary: As a lot of stories

are collected, there is merit in verifying some of them. Getting
details would help to ensure that the reported stories are authentic.
Also, if a process of verification is put in place, people who are
collecting these stories will be more careful while collecting them.
Before verifications are made, it is decided, who will do the
verification, what content will be verified, how many verifications
will be done etc.
8. Quantification: Though MSC focuses on reporting qualitative
change, there is also scope for reporting quantification of change
using three methods. The first is at the level of the individual story,
in which the number of people involved in that activity, the amount
of spent, the amount of money that was saved etc. is considered.
The second way of quantification is by quantifying the frequency
of instances where this change got selected and the MSC was
observed. The third way of quantification is to collect all the SC
stories, and amongst them, to keep a count of the number of times a
specific type of change was observed.
9. Conducting secondary analysis of the stories: Besides doing the
participatory selection and analysis of the stories, additional
analysis is done which includes secondary analysis and meta-
monitoring of data. Though this is not a mandatory step under
MSC, the information generated could be useful in making the
process more rigorous. Both these techniques involve analysing the
complete set of SC stories. Secondary analysis involves going
through the SC stories and then classifying and analysing them.
Meta-monitoring involves analysing all the SC stories based on the
attributes of the study. This macro level monitoring involves
finding out about the number of SC stories written during each
reporting period, identifying the people who had written these
stories, recognising whether there was any pattern in the stories that
got selected and if any action had been taken based on these stories
etc.
10. Revising the MSC process: The last step is to revise the MSC
process for the next rounds based on the feedback and learnings,
while or after the first round has been conducted. The change could
be related to any of the steps from 1 to 9 that have been explained
above. The change could be related to the stories collected, the
frequency at which the stories were reported, it could be associated
with the people involved in the story collection, the method of
analysis and shortlisting of stories, providing feedback to the
programme based on the MSC process and so on. The aim of

reviewing is to improve the MSC post after every round of
implementation.
CHAPTER SIX
CHOOSING EVALUATION DESIGNS
This chapter deals with evaluation and its fundamental concepts, and
experimental and quasi experimental evaluation designs. At the
completion of the chapter, the practitioner will be able to:
" Describe evaluation and its fundamental concepts

" Describe experimental evaluation design, its advantages and
disadvantages
" Describe various quasi experimental evaluation designs, their
advantages and disadvantages
" Choose the most suitable evaluation design depending upon the
project situation.
Evaluation is defined as a type of research, which measures an effect and

attributes the measured effect to a project or intervention. Impact
evaluation determines the impact of the programme on the project
outcome levels and assesses whether these effects are intended or
unintended. Thus, there are two key questions that are to be answered in
an evaluation:
1. Is there a change? What is the magnitude of this change?

2. Can this change be attributed to the programme?
Evaluation designs help to answer these questions with some degree of

certainty. Evaluations are important to understand and assess whether the
programme has achieved its intended goal. Understanding the efficacy of
the programme helps policy makers, the government, development
funding agencies and other organisations to take decisions regarding
expansion, modification or elimination of the programme.
Evaluation designs are determined by the choice of methods used to

identify a counterfactual, or in other words, a group of comparable non-
participants in a project. This counterfactual should be as similar to the
target group as possible, apart from the fact that its members do not
participate in the project or receive the intervention. The more similar the
Choosing Evaluation Designs 73
counterfactual built through the evaluation design is to the target group,

the more robust the evaluation becomes in attributing the impact of the
intervention to the programme. An estimate of impact can then be derived
by comparing the outcome of interest, e.g., HH income, MMR across the
counterfactual group and the treatment/intervention group.
Evaluation designs can be broadly classified into experimental, quasi-

experimental and non-experimental designs. In the case of experimental
evaluations, the counterfactual is known as the control group while in the
case of quasi experimental evaluations, the counterfactual is known as the
comparison group.
These evaluation designs vary in feasibility, cost, generalisability of

results and the degree of selection bias in selecting the treatment and
control groups. Only after understanding the situation at hand and
considering the resources at disposal for conducting the evaluation, an
appropriate evaluation design should be selected.
Different types of evaluation designs are explained as below:
6.1 Experimental Design

The Experimental design involves gathering a set of individuals or other
units of analysis who are equally eligible and willing to participate in the
project and then randomly dividing them into two groups: those who will
receive the intervention, i.e., the treatment group and those who will not
receive the intervention, i.e., the control group.
Experimental or randomised designs are the gold standard of evaluation

designs and are the most robust of evaluation methodologies. By randomly
allocating the intervention among eligible beneficiaries, the assignment
process itself creates comparable treatment and control groups that are
statistically equivalent to one another, given appropriate sample sizes
(Paul J. Gertler S. M., 2011). This is a very powerful outcome because in
theory, the control groups generated through the random assignment
process serve as a perfect counterfactual, free from the troublesome
selection bias issues that exist in all evaluations. Experimental evaluation
design needs to be embedded in the programme design phase as the
randomisation has to be in place while deciding on the programme
intervention area.
74 Chapter Six
The main benefit of this technique is that it interprets results in a simple

way, which means that the project impact on the outcome which is being
evaluated can be measured by the difference between the mean of the
samples of the treatment group and the control group.
While experimental design is considered the optimum approach towards

estimating the project impact, in practice, there are several problems while
implementing this design.
1. Randomisation may be unethical owing to the denial of benefits or

services to otherwise eligible members of the population for the
purposes of the study.
2. It can be politically difficult to provide an intervention to one group
and not to the other.
3. The scope of the intervention may rule out the possibility of
selecting a control group, as with a nationwide project or policy
change.
4. Individuals in treatment or control groups may change certain
identifying characteristics during the experiment that could
invalidate or contaminate the results. For example, people may
move in and out of the treatment or control group. Alternatively,
people who were denied a project benefit may seek it through
alternative sources, or even those being offered a project may not
take up the intervention.
5. It may be difficult to ensure that the assignment is truly random. An
example of this might be administrators who exclude high-risk
applicants to achieve better results.
6. Experimental designs can be expensive and time consuming in
certain situations, particularly in the collection of new data.
The above mentioned challenges exist while conducting evaluations using

the Randomised Control Trial (RCT) design. However, as it is the gold
standard of evaluation and the most robust of all designs, evaluators prefer
to use this design whenever it is feasible.
Example of Evaluation using Cluster RCT:
To illustrate an example of an evaluation design using Cluster RCT, one

may take into consideration a programme, in which separate toilets are
built for girl students. This programme is implemented in schools where
the attendance rate of girls is less than 50 per cent. Based on this criterion,
300 schools are identified in the programme area. Out of these 300 schools
which are comparable and eligible for the programme, intervention is

provided to 150 randomly selected schools, while the balance 150 schools
act as the control group. This evaluation design is called Cluster RCT as
the programme is implemented at the cluster level, i.e., at the school. For
evaluating this intervention, information regarding the relevant indicators
at baseline and endline in both project and control clusters is collected.
The impact of the programme is calculated as the double difference across
baseline and endline in both project and control clusters.
Pipeline Design
A practitioner can use the experimental Pipeline evaluation design

approach for programmes which are supposed to have nationwide or
universal coverage, but are rolled out in a phased manner. Pipeline
evaluation design helps to exploit the delay in the programme
implementation as the areas/groups where the programme is implemented
at a later phase can be used as the comparison area. However, here too it is
important to ensure that the treatment and control areas are similar and
comparable.
Also, it is essential to ensure that there is appropriate gap between the

rollout phases so that the project can achieve its expected outcome.
Example of Evaluation using Pipeline design:
Taking the example of a state flagship programme which is to be

implemented in 20 low performing similar districts of State X. Due to the
scarcity of funds, the programme was initially rolled out in 10 districts and
is to be further expanded in the next 10 districts after three years. In such a
case, the second phase districts which are like the first phase districts can
be used as a control group.
6.2 Quasi-experimental Design

The Quasi-experimental or semi-experimental designs are those in which
the allocation to treatment and comparison groups is not randomised. As
discussed above, due to various constraints, it is not always feasible to
conduct a pure experiment in the form of RCT. Quasi-experimental
designs are the best available option for evaluations, in cases where it is
not feasible to apply experimental designs. Unlike experimental designs,
quasi-experimental designs are used when the programme intervention
area is pre-decided. This design consists of various techniques which can
76 Chapter Six
be used for constructing a comparison group. The quasi- experimental

design used depends upon the type of programme, the stage of programme
implementation, data availability and resources available for evaluation.
Various quasi-experimental techniques are explained in detail below.
6.2.1 Difference in Difference

The Difference in Differences (DID) design, as its name suggests,
compares the change in outcome over time between the treatment group
and the comparison group. In a situation where the programme
implementation area is pre-decided but the programme has not started,
DID can be used as a powerful statistical tool to measure and attribute the
impact of the programme to the intervention. The DID design requires two
cross sections of data, viz., the baseline or the pre-programme data for
both the treatment and the control group and the post-programme data for
the treatment and the control group.
Figure 1.1: Difference in Difference
The figure above illustrates the DID method where A is the value of the
output indicator of the treatment group at time t1 i.e., before the
intervention and B is the value of the output indicator of the treatment
group at time t2, i.e., after the intervention. Similarly, C is the value of the
output indicator of the comparison group at time t1 and D is its value at
time t2. DID helps to separate the impact on the output indicator due to
time and estimates the change in output variable which can be attributed to
the project. The computation of DID is estimated to be (B-A)-(D-C) as
shown in the table given below.
Table 1.7: Computation of DID
After Before Difference

Treatment Area B A B-A
Comparison Area D C D-C
Difference B-D A-C DD=(B-A)-(D-C)
Example of Evaluation using the DID design:
Using the example of a programme which aims to improve MCH, the

project management, based on some predefined criteria, has already
selected the project intervention districts. In such a case, DID is the most
suitable technique to measure and attribute the change in outcome
indicators of a project. In this case, based on some predefined
characteristics like education level, SC/ST population ratio, sex ratio etc.,
comparison districts are selected which are similar to the intervention
districts. Data regarding useful indicators are collected at both the baseline
(before the project) and at the endline (after the project), based on which
the impact evaluation of the programme is done.
6.2.2 Propensity Score Matching

Matching involves using statistical techniques to construct an artificial
comparison group which is comparable in essential characteristics to the
treatment group. Both groups should be matched on the basis of observed
characteristics that are believed to influence project outcomes. It can be
rightly said that matching is as good as the characteristics used for
matching (Paul J. Gertler S. M., 2011). This method is applied in the
context of almost any programme assignment rule, as long as there is a
group that has not participated in the programme. Matched comparison
groups can be selected before project implementation (prospective studies)
or afterwards (retrospective studies).
The key advantage of evaluations using matching methods is that they

draw on existing data sources and are thus often quicker and cheaper to
implement. The principal disadvantage is that the reliability of the results
is often lower, as the methodology may not completely solve the problem
78 Chapter Six
of selection bias; moreover, as the matching methods can be statistically

complex, considerable expertise is required in the design of the evaluation
and in analysis and interpretation of results.
The most widely used type of matching is propensity score matching

(PSM), in which the comparison group is matched to the treatment group
based on the propensity score. The propensity score is the predicted
probability score of each unit for participating/enrolling in the programme.
This score varies between zero to one, where zero depicts zero per cent
and one depicts one hundred per cent probability of enrolling in a
programme. Using this method, a comparison group is found from a
sample of non-participants who are closest in terms of observable
characteristics to a sample of project participants.
PSM is a very useful method when there are many potential characteristics
to match between a sample of project participants and a sample of non-
participants. Instead of aiming to ensure that the matched control for each
participant has the same value of the control variable X, the same result
can be achieved by matching the predicted probability of project
participation P, given X, which is the propensity score of X. The range of
propensity scores estimated for the treatment group should correspond
closely to that of the retained sample of non-participants. The closer the
propensity score, the better the match. A good comparison group comes
from the same economic environment and is administered the same
questionnaire as the treatment group by the same group of trained
interviewers.
Example of Evaluation using the PSM design:
To illustrate evaluation using the PSM design, one may take the example
of a programme which was started four years ago, and is currently on the
verge of completion. The objective of the programme was to create SHGs
and work towards giving its members greater access to financial resources.
In such cases, where the baseline data is not available, the matching
propensity score is used to identify the comparison units for the treatment
units. The propensity score is calculated based on variables which are
expected to affect the probability of the person becoming a member of the
SHG. Variables like the education level, above poverty line (APL)/below
poverty line (BPL) status, caste, engagement in agricultural activities, and
participation in gram sabha activities can be used to calculate the
propensity score.
6.2.3 Regression Discontinuity

The Regression Discontinuity (RD) design is used to evaluate programmes
in which an index or a continuous variable is used to decide the eligibility
of participation in a programme. In such programmes, all potential
participants are ranked based on a continuous index or variable and the
eligible participants are selected based on a cut-off point.
The RD design exploits the discontinuity around the cut-off score in the
regression line to estimate the counterfactual. The cases just above and
below the cut-off point are similar in their characteristics and the
intervention is the main difference between them. Development
interventions like subsidy programmes based on poverty score, pension
programmes based on age, scholarship programmes based on marks etc.,
can be evaluated by using the RD design.
Although this design is useful in evaluating programmes where eligibility

is decided based on the cut off score, there are a few limitations to this
design. This design is less robust and has weak external validity i.e., less
generalisability to the population, as only a few units across the cut off line
are compared to estimate the impact on the population. Also, in many
cases, assignment rules are not strict, thus, a few ineligible beneficiaries
can get the intervention while a few beneficiaries who are eligible may not
get the intervention. Evaluation of these programmes using the RD design
becomes challenging in such cases.
Example of Evaluation using Regression Discontinuity design:

80 Chapter Six

The practitioner may consider the example of a food subsidy programme

which aims to improve the food security of its target families having its
eligibility criterion based on a poverty score. For this programme, the
impact can be evaluated by comparing HHs which are just below and
above the poverty line. The figures given above show the distribution of
the population with respect to the poverty score and the food security
index. All families with a poverty score below 40 are eligible for the
programme. The cases just below and above the cut off score (circled in
the diagram) can be compared for evaluating the impact of the programme.
During the endline survey, the difference in the food security index
between the treatment group and the comparison group is assessed. This
difference is said to be the attributable impact of the programme.
6.2.4 Instrumental Variable

The Instrumental Variable evaluation design is used to evaluate programmes
which have universal coverage and voluntary or open enrolment. For such
programmes, the programme administrators do not have control over who
will and who will not participate in the programme. Comparing people
who have enrolled in the programme and those who did not enrol in the
programme is not appropriate for attributing the impact, as both these sets
of people are not similar. There are some endogenous factors like
motivation, talent, access to information, opportunity cost etc., because of
which participants who enrolled in the programme and those who did not
enrol in the programme are different.
The Instrumental Variable (IV) design is one of the econometric

techniques that can be used to compare project participants and non-
participants correcting for selection bias (Paul J. Gertler S. M., 2011). It
consists of using one or more variables (instruments) that matter to
participation but not to outcomes given participation. This identifies the
exogenous variation in outcomes attributable to the project, recognising
that its placement may not be random but purposive. The instrumental
variables are first used to predict project participation, then the project
impact is estimated using the predicted values from the first equation.
Randomised Promotion can be used as an instrument variable that allows

the practitioner to create variation between units and further use these
units to create a valid comparison group. The target population of the
programme is classified into three groups. The first is the ‘Never Enrol’
group, i.e., people who will never enrol for the programme whether the
programme is promoted amongst them or not. The second is the ‘Always
82 Chapter Six
Enrol’ group, which constitutes of people who will enrol in the programme
even if the programme is not promoted. The third group is ‘Enrol if
Promoted’, and these are the people who will enrol in the training
programme if the training programme is promoted amongst them. The
lattermost are the set of participants on which the impact of the
programme can be evaluated (Paul J. Gertler S. M., 2011).
The fundamental assumptions in using randomised promotion as the

instrument variable is that the promoted and the non-promoted group are
similar. The instrument variable used in this case viz., randomised
promotion, should be effective enough to increase the enrolment rate
substantially amongst the ‘Enrol if Promoted’ group. Also, it is to be made
sure that randomised promotion does not have an effect on the outcome of
the programme.
Example of evaluation using the Instrument Variable design:
The practitioner may take the example of evaluation of a voluntary skill

development training programme which aims to improve the monthly
income of its participants. In this example randomised promotion can be
used as an instrument variable to measure and attribute the impact of this
programme. A set of similar target participants is chosen to constitute the
evaluation sample. The evaluation sample is then randomly divided into
two clusters. The programme is promoted in one cluster by an outreach
worker who encourages the target participants to enrol in the programme
but no promotional activity is done in the other cluster. It is expected that
there will be a considerable increase in enrolment to the training
programme in the cluster where the programme was promoted.
The impact of the vocational skill training programme is calculated by

dividing the change in the monthly income in the promoted and non-
promoted cluster divided by the increase in the enrolment rate to the
programme. One may assume that the monthly income in the promoted
and the non-promoted group is found to be INR 6,000 and INR 5,000
respectively, and that there is an increase in the enrolment rate from 40 per
cent to 80 per cent due to the promotion. In this case, the estimated impact
will be change in the monthly income divided by the increase in the
programme enrolment i.e., 1000 divided by 40 per cent which is equal to
INR 2,500. Therefore, there is an increase of INR 2,500 in the monthly
income because of the training programme.
6.2.5 Reflexive Comparison

The Reflexive Comparison design is another type of a quasi-experimental
design. In a reflexive comparison, the counterfactual is constructed on the
basis of the situation of project participants before the project. Thus,
project participants are compared to themselves before and after the
intervention i.e., they function as both treatment and comparison groups.
This type of design is particularly useful in evaluations of full-coverage

interventions such as nationwide policies and projects in which the entire
population participates and there is no scope for a control group.
There is, however, a major drawback in using the reflexive comparison

design. The situation of project participants before and after the
intervention may change owing to myriad reasons independent of the
project. For example, participants in a training project may have improved
their employment prospects after the project. While this improvement may
be due to the project, it may also be due to the fact that the economy is
recovering from a past crisis and employment is growing again. Unless
they are carefully done, reflexive comparisons may not be able to distinguish
between the project and other external effects, thus compromising the
reliability of results.
6.3 Non-Experimental Designs

Non-Experimental designs are impact evaluation designs that do not
include a matched comparison group. These days, an increasing proportion
of development interventions fall into categories where it is difficult to
apply conventional i.e., experimental and quasi-experimental designs. In
situations when the development programmes are complex and it is not
possible to define a comparison group, non-experimental designs seem to
be potentially the best possible and feasible designs that can be adopted
(Richard Blundell, 2000).
In these designs, outcomes and impact are assessed without having a

conventional counterfactual to address the attribution question, i.e., “What
would have been the situation of the target population if the project had
not taken place?”
Creating a counterfactual for a longitudinal design is an example of a non-

experimental design. The Longitudinal design is a design in which more
than one cross-section of data is collected. A counterfactual can be created
84 Chapter Six
in case of longitudinal research design to attribute impact before and after

the intervention. In order to analyse the longitudinal/panel data, the fixed
effect and random effect model is used to create the counterfactual.
Fixed Effect Model
The Fixed Effect (FE) model explores the relationship between predictor
and outcome variables within an entity. It assumes that something within
the individual may impact or bias the predictor or outcome variables and
during the study, the practitioner needs to control for the same.
The FE model removes the effect of those time-invariant characteristics

from the predictor variables so the predictor’s net effect can be assessed.
The equation for the FE model is:
Yit= ȕ1 Xit+ Įi+ uit
Random Effect Model
In the case of the Random Effect model, the key rationale is that unlike in
the FE model, the variation is assumed to be random and uncorrelated with
the independent variables included in the model or to quote “…the crucial
distinction between fixed and random effects is whether the unobserved
individual effect embodies elements that are correlated with the regressors
in the model, not whether these effects are stochastic or not” (Greene,
2008).
One of the advantages of the random effect method over the FE method is
that one can include time invariant variables. The equation is written as:
Yit= ȕXit+ Į + uit+ İ it
Fixed Effect or Random Effect Model
To decide whether to use the Fixed or the Random Effect model, the
Hausman test can be carried out wherein the null hypothesis is that the
preferred model is the random effect vs the alternative, the fixed effect
(Greene, 2008).
Example of Non-Experimental Design:
The practitioner may consider the example of a programme aiming to

increase awareness about health and hygiene being spread to eliminate OD
at the village level. All the villages are covered as part of the intervention
and it is not possible to select a control or comparison group. In this case,

a cohort design can be used to evaluate the performance of the intervention
in which the same HHs are surveyed as part of the baseline and the
endline. This is the best possible option to evaluate the project impact
considering the current situation.
6.4 Comparison Matrix

As discussed above in detail, each impact evaluation design has its own
merits and demerits. It is important for an evaluator to understand the
project and the present condition and then consider the availability of
critical resources like time, data, and financial resources. Based on these
factors, an evaluation expert can choose the best possible design. The table
given below gives a detailed comparison between the various impact
evaluation designs, when an evaluator should use them and what their key
advantages and disadvantages are.
Table 1.8: Comparison Matrix of Evaluation Designs
Design When to use Advantages Disadvantages
Randomised x Whenever feasible x Gold standard x Not always

Control Trial x Most powerful feasible
x Not always
ethical
x Costly
Instrument x When an x Provides x Only looks at
Variable- intervention is external the sub-group
Randomised universally variation for a of the sample
Promotion implemented sub-set of x Power of
beneficiaries encouragement
design only
known ex post
Regression x If an intervention x Project x IE can be
Discontinuity has a clear, sharp beneficiaries generalised only
assignment rule often must to a sub-group
qualify through i.e., cases near
established the cut-off line
criteria x Assignment rule
in practice often
not
implemented
strictly
86 Chapter Six
Difference-in- x If two groups are x Eliminates x Can be biased if

Differences growing at similar fixed trends change
rates differences not x Ideally have
x Baseline and related to two pre-
follow-up data are treatment intervention
available periods of data
Matching x When other x Overcomes x Assumes no
methods are not observed unobserved
possible differences differences
between (often
treatment and implausible)
comparison x Large sample
required
Non- x When it is possible x The only x Does not have a
Experimental to select a different possible option robust design
Design control/comparison in the case of x Does not
group many real control for
world complex external factors
development causing
projects observable
changes
6.5 Practice Exercise- Choosing the appropriate Impact

Evaluation Design
Project: - Beti Padhao Yojna – Improving Educational Outcomes
Background:
According to the 2010 education survey in YY state in XXX country,

primary school attendance in the country was estimated at 45 per cent
nationally, while it was only 40 per cent for rural areas and 36 per cent for
girl children. To help address this problem, the corporate social
responsibility (CSR) foundation of XXXX organisation funded a four
year, twenty-million-dollar project focused on increasing education
outcomes and hygiene conditions in schools that were especially for girls.
The Beti PadhaoYojana (BPY) programme was implemented in 200
villages across the 10 districts which had the lowest levels of enrolment
rates for girls.
It focused on building a school (with separate toilets for girls, and access
to drinking water) in each selected village. Also, some additional
interventions were introduced in these schools which included:
x Mid-day Meals: Free mid-day meals provided to all students daily

x Menstrual Hygiene Kits: Free menstrual hygiene kits are provided
to adolescent girls
x Bridge Classes: Bridge classes in the evening for children who had
dropped out of school
x Take-home Rations: Girls with high attendance are given food to
take home to their families
x School Kits and Textbooks: Free material provided to all students
An Impact Evaluation was commissioned to address three key questions:
1. What was the impact of the programme on school enrolment?

2. What was the output and outcome level impacts of the programme?
3. Was the impact on girl students different to the impact on boy
students?
Team Tasks for Practitioners
You are an employee of XYZ consulting services which has been

contracted as the Impact Evaluation agency for this evaluation. You will
be provided a few different scenarios. Please respond to the questions for
each scenario given below.
Scenario 1:
It has been decided by the programme team to make inclusive education

(IE) an integral part of the programme design. Your team has been called
during the design phase to conduct an impact evaluation to answer the
evaluation questions. Based on eligibility criteria, 450 villages were
identified initially for this programme, though due to a resource constraint,
only 200 out of these 450 villages which are to be chosen will be covered
under this phase of BPY. Please provide your response to the following
questions:
1. What IE design will you recommend?

2. What will be the level of assignment of the design?
3. What could be the practical challenges and disadvantages in
implementing this design?
4. What are the strengths of this design?
88 Chapter Six
Scenario 2:
A new CSR head joined the foundation of XXXX organisation six months
after the commencement of the BPY project. He appreciates the
importance of having a robust IE and wants IE to be conducted for this
project. A baseline was not conducted initially, so the programme team is
uncertain about the IE design that should be adopted. The CSR head, who
has had a positive experience of working with your organisation, requests
you to conduct an IE. In your background research, you find that the YY
state Education Ministry had conducted a village level survey in the 20
worst performing districts which includes the 10 districts covered by BPY.
This data includes characteristics like the number of girls in the village,
the enrolment rate of girl students, the distance to the nearest school from
the village, the number of students in the nearest school etc. This
application data had been used by the implementing agency to rank each
village of 450 initially shortlisted villages in terms of need, with the 200
villages ranked as the neediest being selected to receive the programme
interventions. Consider the following questions and prepare your
responses.
1. Which IE design will you recommend?

3. What could be the practical challenges in implementing this
design?
4. What are the strengths of this design?
5. Can this design be adopted in case the programme team has the
option to collect the baseline data? What are the advantages and the
disadvantages of applying this design in that case?
Scenario 3:
At the time when the five year BPY programme is about to end, the CSR
board realises that they should assess how effective the programme has
been. The programme team is left with a limited budget, making it difficult
to do large scale primary data collection. However, it is found that the
state Ministry of Education has annually been collecting very good village
level data in 20 low performing districts, including in the 10 project
districts. The data includes important indicators like the enrolment rate of
girl and boy students, the number of toilets in the school, and the learning
test scores of both girl and boy students. Consider the following questions
and prepare your responses.

3. What could be the practical challenges in implementing this design
and also, what are the weaknesses of this design?
4. What are the strengths and advantages of this design?
Scenario 4:
The BPY programme is about to start. The board of the CSR foundations
is considering conducting an IE for the project. Due to strategic reasons
and political pressure, the 200 villages where the programme will be
implemented have already been decided. Your organisation is being called
to conduct an IE for this project. Please present your responses to the
questions mentioned below:

2. What would be the practical challenges in implementing this
design? What are the weaknesses of this design?
3. What are the strengths and advantages of this design?
4. Would you also be able to use this design if you were called in to
evaluate after the programme had already started?
CHAPTER SEVEN
APPROACHES TO EVALUATION
In this chapter, the different approaches used in evaluations are briefly

touched upon. It is very important to understand that there are different
approaches to evaluation which are adopted while being integrated with
the conventional evaluation methods and designs, as per their suitability to
a particular project context. At the end of the chapter various approaches
to evaluation will be introduced and described, including:
" Utilisation Focused Evaluation

" Feminist Evaluation
" Transformative Evaluation
" Real World Evaluation
" Using Appreciative Enquiry in Evaluation
" Realistic Evaluation
" Equity Focused Evaluation
" Developmental Approach to Evaluation
7.1 Utilisation Focused Evaluation

Utilisation focused evaluation (UFE) is an evaluation approach where the
focus is on its intended users, the ‘utilisation’ of the evaluation findings
and recommendations. This approach, developed by Michael Quinn
Patton, is based on the premise that the evaluation should be judged for its
worth, based on its usefulness to its intended users. UFE aims to facilitate
a learning process through which the users of the evaluation can use it to
improve and buttress the programme (Ramirez, 2013). It is just a guiding
framework for conducting evaluations and does not prescribe any
particular methodology or theory. It includes various evaluation methods,
but the key is that they should be implemented in a participatory approach
while including the users of the evaluation. The evaluation is planned and
implemented in such a way that the users are involved in the designing of
the evaluations and it is made sure that the evaluation findings are utilised
for the improvement of the project or the programme.
Approaches to Evaluation 91
This evaluation approach has two key components. Firstly, the specific
and real life primary intended users of the evaluation are identified and
engaged from the beginning to design the evaluation and decide the key
outputs of the evaluations based on their requirement. Secondly, the
decisions about the evaluation process should be made keeping in mind
the intended use of the evaluation by its intended users.
The seventeen-step framework listed below, was outlined by M.Q. Patton

in the year 2012.It is used as a guide while adopting UFE.
1. Assess and build programme and organisational readiness for UFE

2. Assess and enhance evaluator readiness and competence to
undertake a UFE
3. Identify, organise, and engage primary intended users: the personal
factor
4. Situation analysis is conducted jointly with primary intended users
5. Identify and prioritize primary intended uses by determining
priority purposes
6. Consider and build in-process uses if and as appropriate
7. Focus on priority evaluation questions
8. Check that fundamental areas for evaluation inquiry: implementation,
outcomes, and attribution questions are being adequately addressed
9. Determine what intervention model or theory of change is being
evaluated
10. Negotiate appropriate methods to generate credible findings that
support intended use by intended users
11. Make sure intended users understand potential methods of
controversies and their implications
12. Simulate use of findings; this is the evaluation's equivalent of a
dress rehearsal
13. Gather data with ongoing attention to use
14. Organise and present the data for interpretation and use by primary
intended users, viz., analysis, interpretation, judgment, and
recommendations
15. Prepare an evaluation report to facilitate use and disseminate
significant findings to expand influence
16. Follow up with primary intended users to facilitate and enhance use
17. Meta-evaluation of use: be accountable, learn, and improve.
92 Chapter Seven
7.2 Feminist Evaluation

Feminist evaluation and gender approaches offer researchers methods of
evaluation to be applied to particular circumstances. Feminist evaluation is
based on feminist theory. According to feminist researchers, a story based
on only a man’s experience is equivalent to missing half the picture, while
adding a woman’s perspective enables a researcher to fully construct the
picture of reality (M. Connelly, 2000). Feminist evaluation emphasises
participatory, empowering, and social justice agendas (Patton M. , 2008).
Denise Seigart (2005) suggests that feminist theory and feminist research,
influenced by the women’s movement, encourages researchers and
evaluators ‘to question what it means to do research, to question authority,
to examine gender issues, to examine the lives of women, and to promote
social change’. She further explains that over the years, feminist research
has moved from feminist empiricism to standpoint theory, and finally to
postmodern feminism. Feminist evaluation is often described as ‘fluid,
dynamic, and evolving’ (Brisolara, 2002). Feminist evaluation theorists
tend to describe feminist evaluation as flexible and do not recommend a
strict approach or provide a framework; rather, they describe it as a way of
thinking about evaluation. Although feminist evaluation follows a
dynamic structure, it stands on six basic tenets. Sielbeck-Bowen et al.,
(2002) defined these six tenets as follows:
1. Feminist evaluation has, as a central focus, the gender inequities

that lead to social injustice.
2. Discrimination or inequality based on gender is systemic and
structural.
3. Evaluation is a political activity; the contexts in which evaluation
operates are politicised and the personal experiences, perspectives,
and characteristics that evaluators bring to evaluations (and with
which we interact) lead to a particular political stance.
4. Knowledge is a powerful resource that serves an explicit or implicit
purpose.
5. Knowledge should be a resource of and for the people who create,
hold, and share it. Consequently, the evaluation or the research
process can lead to significant negative or positive effects on the
people involved in the evaluation/research. Knowledge and values
are culturally, socially, and temporally contingent. Knowledge is
also filtered through the knower.
6. There are multiple ways of knowing; some ways are more
privileged than others.
Difference between Feminist Evaluation and Gender Approaches
Table 1.9: Gender Approach vs Feminist Evaluation
Gender Approaches Feminist Evaluation

Gender approaches do not Feminist evaluation challenges and
challenge women’s position in attempts to strategically improve
society (Gender and women’s lives. Feminist evaluation is
Development (GAD) approach) used to guide the evaluation
methodology if the evaluation
questions seek to understand why
differences exist between men and
women
A technical fix that leaves the A feminist evaluation would take a
prevailing and unequal power more activist approach and through its
relations intact evaluation design, explore the
possibility that perhaps all women do
not want what men have
Gender approaches provide Feminist evaluation offers broad
more concrete guidelines and guidance that encourages an evaluator
prescriptive methods how to think about an evaluation and
how to use that reflection
7.3 Transformative Evaluation

Transformative evaluation (TE) is an evaluation approach which was
proposed by Donna Martens in the year 2009. Transformative evaluation
believes that equity and social justice should be the central values of the
evaluation approach. This approach places importance on the lives and
experiences of the socially marginalised groups such as women, ethnic
minorities, people with disability etc., and advocates that they should be
given special focus and priority in this type of evaluation approach
(Kosheleva, 2016). This approach also believes that the impact evaluation
should be methodologically inclusive. By having a methodology which is
inclusive, evaluators can create opportunities for all social groups to
participate in the evaluation. This helps in making the evaluation
contribute towards increasing social inclusion. This approach is congruent
with the International Organization for Cooperation in Evaluation's
(IOCE) philosophy on evaluation.
94 Chapter Seven
The transformative paradigm believes that evaluators should establish an

interactive and trusting relationship with the communities involved in the
evaluation so that they understand the realities and the actual condition of
these communities and social groups. This approach believes that all social
groups and communities have their own strengths and assets and people
from socially excluded groups are experts of their own situation and
context. Hence, this approach treats people as investors who can provide
valuable knowledge about their community and their life experiences, and
in this way contribute significantly to the evaluation. Therefore, TE
requires evaluators to develop a network with these socially excluded
groups which the project or programme evaluates as targets. For its
implementation, the TE approach requires the involvement of all stake
holders including programme donors, implementers and the intended and
unintended beneficiaries of the programme.
There are a few critical points that need to be taken care of while
implementing TE. It is suggested that the researchers should have a
discussion at the beginning of the planning phase in order to understand
and ascertain the background and cultural context of the programme. Also,
it is good to use mixed-method designs, as quantitative and qualitative data
facilitates the responsiveness of different participants to different issues.
Also, the methodology adopted should be able to capture the cultural
complexity and it should be appropriate for the cultural groups in the
research study area.
7.4 Real World Evaluation

Real world evaluation (RWE) was developed to address the conundrum of
conducting a methodological rigorous evaluation under real-world
scenarios which face a lot of practical constraints. In real world scenarios,
there are constraints like time constraints, budget constraints, and data
constraints like non-availability of data or poor quality data. Evaluators are
also often required to modify the evaluation design as per political
pressures and influences. Also, sometimes evaluations need to be tailored
according to organisational systems involving different agencies and
actors, as well as administrative procedures that are not well suited to
conduct a rigorous evaluation. Very often, evaluators are brought in to
conduct the evaluation only towards the closure of the programme. Such
situations lead to a poor quality baseline data or no comparison group
during the baseline. The RWE approach helps evaluators to adapt to these
constraints and come up with the best possible methodology in the context
of the programme.
RWE envisages that the highest level of methodological rigour is adopted

with respect to the circumstances and the constraints under which the
project evaluation is conducted. Michael Bamberger, Jim Rugh and Linda
Mabry first authored a book on this evaluation approach which was
published in 2006 (Real World Evaluation, 2016). The figure below
summarises the seven steps of the RWE approach (Michael Bamberger,
2006) (Real World Evaluation, 2016).
RWE tries to address the quality challenge to ensure that the study has the
minimum acceptable methodological rigour and adapts the evaluation
design to the level of rigour required by decision makers. One of the key
advantages of RWE is that it can lead the increase in the uptake of
evidence into policy making as project stakeholders are involved across all
phases including design, implementation, analysis and dissemination. This
would also ensure that the evaluation focuses on the key issues highlighted
by the stakeholders and that it uses the client’s preferred style of
communication.
96 Chapter Seven
Figure 1.4: Steps in Real World Evaluation

7.5 Evaluation Approach Using Appreciative Inquiry

Appreciative Evaluation is an evaluation approach which is influenced by
Appreciative Inquiry (AI), which is originally from the field of
organisational development. It is a change management tool developed in
the 1980s. To explain briefly AI is a process that inquiries into, identifies
and further develops the best of what is present in organisations in order to
improve them further. This approach aims to focus on the strengths,
positive experiences, best practices etc., of the organisation and is
implemented using the 4-D model which includes four stages:
x Discovery: Discover what is the best in the present system and

identify the system strengths
x Dream: Dream about what can be the best case scenario and
envision the probable results in such a case
x Design: Prepare a detailed action plan at this stage to convert
dreams or imagination into action
x Delivery: Implement the overall vision and specific activities of the
design phase.
Using AI as part of the evaluation approach is a recent development. In

terms of evaluation, AI can use the 4-D model to measure change, develop
the programme logic model, clarify the evaluation purpose, determine the
key evaluation questions, develop indicators and evaluation plans etc. The
AI approach can also be used to modify the existing survey instruments or
developing new ones to include the AI questions. An evaluation adopting
the AI framework would focus on stories of best practices, positive
learnings, successful processes etc. This enables the system to look in the
direction of success and to look forward to building a future on these
positive experiences. An assessment of the programme using the method
of AI evaluation seeks answers to what is working well in the programme
and how to enhance or replicate those aspects. The AI approach to
evaluation is a participatory approach which involves all stakeholders in
the evaluation process.
This approach is apposite in cases where previous evaluation methods

have failed, the relationship among individuals or various stakeholders
have deteriorated, where there is a desire to build the evaluation capacity
to help others learn from evaluation practice, and to develop specific data
collection methodologies etc.
98 Chapter Seven
The advantages of using this approach is that it focuses on developing the

positive aspects of the programme that is being evaluated rather than by
simply focusing on the problems. Being a participatory technique, it
involves various stakeholders and maximises their ownership to the
evaluation process and its results. It helps to make the evaluation mode
democratic, participatory, deliberative, engaging and enlightening.
7.6 Realist Evaluation

Realist Evaluation is a theory driven evaluation approach. This approach
believes that the most durable and practical programme recommendations
that evaluators can offer come from research that commences with
programme theory and ends with refinement of the same. Therefore, apart
from determining whether and how the programme worked in a particular
context, realist evaluation also works to test and further refine the
programme theory.
This approach was proposed by Ray Pawson and Nick Tilley in 1997.
While conventional evaluations aim to answer the question, ‘What
works?’ or ‘Does the programme work?’ realist evaluation aims to answer
the question ‘What works for whom in what circumstances and in what
respect, and how?’ (Tilley, 2004). To understand this, realist evaluation
aims to identify the mechanisms and processes and explain how the output
was achieved under the influence of the programme milieu.
Realist evaluation believes that the fundamentals of a programme is a

theory which can be practically implemented. Every programme is based
on a theory and the efficacy of the programme depends upon the efficacy
of the theory. In this evaluation approach, various programme theory
elements such as mechanisms/processes, programme context, outcomes
etc., are made explicit at the evaluation design stage as it enables design of
data collection tools for testing of the programme theory.
Realist evaluation is method neutral i.e., it does not impose any particular
method. As with other evaluation approaches, realist evaluation too
believes that the choice of data collection and analysis methods and tools
depends upon the type of data that is needed to answer the evaluation
questions and to test the programme theory. Quantitative data is collected
to test the various levels of results and the programme context, whereas
qualitative data is collected on the processes and generative mechanisms.
This evaluation approach is useful in cases where complex situations need

to be assessed. Thus, this approach is more useful to policy makers as it
helps them to understand how the intervention works and the conditions or
the catalysts that are required to make it work.
7.7 Equity-Focused Evaluation

Before discussing equity-focused evaluations, it is important to first
understand what equity is. As defined by the World Health Organization
(WHO), equity is the absence of avoidable or remediable differences
among groups of people, whether these groups are defined socially,
economically, demographically, or geographically. Often, the term equity
is used as a synonym for equality. However, it is important to recognise
that these terms are not synonymous. Equality means sameness or no
differences at all, whereas, the aim of equity is not to eliminate all
differences so that everyone has the same level of income, health, and
education. Rather, the goal is to eliminate the unfair and avoidable
circumstances that deprive people of their rights. Therefore, inequities
generally arise when certain population groups are unfairly deprived of
basic resources that are available to other groups. A disparity is ‘unfair’ or
‘unjust’ when its cause is due to the social context, rather than biological
factors.
In recent times, there has been increased demand and stronger focus on
equity in human and social development. A greater number of
organisations working in the development sector including national
governments are focusing on achieving equitable development results for
the vulnerable or disadvantaged groups amongst the target population.
While the focus is on equity focused interventions, there is also the
challenge of how to evaluate the effect of interventions on equitable
outcomes.
Equity-focused evaluation, as the name suggests, focuses on evaluating

projects, policies or programmes which aim to achieve equity focused
results. As a standard definition, equity-focused evaluation is judgement
made on the relevance, effectiveness, efficiency, impact and sustainability
of policies, programmes and projects concerned with achieving equitable
developmental results (UNICEF Evaluation Office, 2006). This type of
evaluation involves going through a systematic and rigorous process of
designing, analysing and interpreting what works and what does not work
in reducing inequality. It highlights intended and unintended results of the
project/programme on the vulnerable or disadvantaged group. It also
100 Chapter Seven
highlights the gaps between the various strata of beneficiaries viz., 'the
best-off group', 'the average group' and 'the worst-off group'.
As the name suggests, the key difference in equity-focused evaluations is

that it places a special focus on evaluating the impact of the programme
from the perspective of equity. It aims to assess if the programme has had
any impact to decrease the inequities, as well as to evaluate the change in
the outcome indicators for 'the worst-off group' against the stand-alone
index and in comparison, to 'the average group' and 'the best-off group'. In
case of equity evaluation, special focus is given to sampling to ensure the
representation of disadvantaged or vulnerable groups in the evaluation
sample. Also, evaluation questions are explicitly included to assess the
impact of the programme on 'the worst-off group'.
Equity-focused evaluation can be used to evaluate the sanitation programme

to assess if the programme has been able to ensure participation of the
poorest and the most vulnerable groups. The central goal of this evaluation
then is to assess the equity outcomes of the project.
7.8 Developmental Approach to Evaluation

Not all projects and the milieu in which they work are simple, controlled
or easily predictable at the onset. Sometimes the situation at hand is
complex and the project team cannot say with any degree of certainty what
they will achieve through the project, although they are confident that by
doing their planned activities they will find a way forward. Sometimes
these programmes have to be very innovative and dynamic in order to
respond to situations, i.e., they can be modified while being implemented
or have emerging outcomes. The developmental evaluation approach has
been developed to respond to such situations. The focus of developmental
evaluation is to assist social innovators to develop innovative social
change programmes to respond to complex or uncertain situations. In
developmental evaluation, the evaluator is part of a team whose members
collaborate to conceptualise and design new approaches in the long-term,
on-going process of continuous improvement, adaption and intentional
change (Gamble, 2008). In this approach, the evaluator’s primary role is to
embed evaluative questions, data and logic models in the team discussions
and also to facilitate data based decision making in the development
process.
Such situations are suitable for developmental evaluation if:

x The project does not have an intervention model yet i.e. it needs to
be developed
x The model is not fool proof and needs to be developed or modified
x The programme situation or context is complex i.e., there is no
simple cause and effect situation. Most of the important relationships
between cause and effect are not clear
x When a rapid response needs to be developed in case of a sudden
major change in the project or situation.
Instead of single loop learning where it is first assessed whether the

programme is working out or not and then giving feedback to the system,
developmental evaluation aims to focus on double loop learning. Double
loop learning aims to question the assumptions in the theory model, while
it does a deeper assessment of the situation and the strategy. More than
problem solving, double loop learning re-evaluates and re-frames the
goals, values or assumptions about the strategies that will be successful in
a given situation.
The key characteristics of developmental evaluation is that it envisages a

tight integration between evaluators and the programme staff. It also aims
to use the data concurrently for continuous programme improvement. In
this evaluation approach, measures and tracking mechanisms are
developed quickly as outcomes emerge. Also, as the process unfolds, these
measures can be changed during the evaluation.
Thus, developmental evaluation is suitable in innovative settings where

programme goals are evolving and emergent rather than pre-determined
and fixed, time periods are fluid and forward looking and not imposed by
external deadlines and the objective of evaluation is to facilitate innovation
and change and derive learning rather than external accountability.
CHAPTER EIGHT
ANALYSING M&E INFORMATION
The M&E system generates a substantial amount of programmatic data

which needs to be analysed and interpreted for programmatic
improvement and effective decision making. This chapter explores ways in
which M&E data is analysed and interpreted meaningfully for
programmatic decision-making. The chapter starts with revisiting data
basics and goes on to explore the fundamentals of qualitative and
quantitative analysis conceptually and it also provides an overview of
doing analysis using statistical software. At the end of this chapter the
practitioner will be able to:
" Understand and explain the fundamentals of data basics

" Understand and explain the fundamentals of univariate data
analysis
" Understand and explain the fundamentals of bivariate data analysis
" Use MS-Excel for data analysis
" Use SPSS for basic and advanced data analysis
" Understand the fundamental concepts of Qualitative Data Analysis
8.1 Data Basics

At the outset, some of the basics of M&E and data collection and
collation, which is done as part of the M&E effort, are revised. The M&E
process is an attempt to collect information about variables of interest and
assessing change in those variables as a function of the internal and
external environment. It is the process of grouping observations about
variables of interest in a systematic and coherent way to provide data,
which could be qualitative or quantitative in nature depending on the
nature and type of variables. Qualitative data is segregated by words,
pictures or images, while quantitative data is segregated by numbers on
which are based basic mathematical operations.
A variable is defined as an attribute or a characteristic of a case, which is

different for different cases. Its variability is usually captured on the
Analysing M&E Information 103
measurement scale, varying between two scale values to potentially an

infinite number of scale values for the binary scale or continuous metric
scale.
Variables can be classified into three categories depending on typology:
1. Independent Variable: A variable, which at a particular point of

time, can take values on its own. It can also be described as a factor
that is selected and manipulated or controlled by the experimenter.
2. Dependent variable: A variable whose value is dependent upon
one or more independent variable or on other dependent variables.
This measurable behaviour is exhibited by the participant in the
research.
3. Extraneous variables: Extraneous variables are variables that may
influence the outcome of a study but are not directly related to the
study.
There are four types of measurement scales that are used in the
measurement of variables: nominal, ordinal, interval, and ratio scales,
which follow a hierarchy, with nominal being at the lowest rung of the
hierarchy. These four types of measurement scales are briefly defined
below:
1. Nominal variables: The values of the nominal variable data have

no numeric meaning as no mathematical operation except counting
can be done on this type of data. They are in fact used to classify
whether the individual items belong to distinctively different
categories. Typical examples of nominal variables are gender, race,
colour, city, etc.
2. Ordinal variables: Ordinal variables, unlike nominal variables,
allow the measured items to be ranked in terms of order, where the
higher order item that is specified represents more of the quality
represented by the variable, though it may not be possible to
specify exactly how much more than the other items. A typical
example of an ordinal variable is the rating assigned to the impact
of the programme, i.e., whether it is excellent, average or poor.
Even if the x% rated programme is considered to be excellent and
the y% rated programme is average while the one rated z% is poor,
the researcher is not able to estimate accurately whether the
difference between excellent and average is the same as that
between average and poor. In the case of ordinal variables, only
certain mathematical variables such as 'greater than' or 'less than'
104 Chapter Eight
are feasible and only measures such as median and range can be
calculated on this type of data.
3. Interval variables: Interval variables provide more flexibility in
terms of measurement as they not only allow measured items to be
ranked but also help to quantify the sizes of differences between
them. For example, temperature that is measured in degrees
Fahrenheit or Celsius constitutes an interval scale. Thus, although
the temperature of 80 degrees is higher than the temperature of 40
degrees, the temperature at 80 degrees is not twice as hot as 40
degrees. Another example of the interval variable is that of the time
system such as B.C. or A.D., wherein time is measured taking B.C.
or A.D. as the reference point (initial point of reference is assumed
as zero). However, this does not mean that time did not exist before
B.C. or A.D. A reference scale to measure time has been
constructed, but it does not have a true or rational zero.
4. Ratio variables: Ratio variables are measured by scale and they
have an equidistant point as well as a rational zero. Thus, in
addition to all the properties of interval variables, ratio variables
feature an identifiable absolute zero point. A typical example of the
ratio scale is the Kelvin temperature scale, where not only is the
temperature of 60 degrees higher than the temperature of 20
degrees, but it can be specified that the temperature of 60 degrees is
thrice as high as the temperature at 20 degrees. Similarly, an object
weighing 20 kg is twice as heavy than an object weighing 10 kg.
Most of the variables used for measuring in field situations
conform to ratio scale properties, though most statistical data
analysis procedures do not distinguish between the interval and
ratio properties of the measurement scales.
Data Classification
Data, as explained earlier, is a set of observations of variables,

characteristics or indicators. Based on the nature, source, values and scale
of measurement, data has a different typology. The typology and basis of
classification of data are mentioned in the following table.
A broad snapshot of Data Types
Figure 1.5: Data Types

106 Chapter Eight
Table 1.10: Data Classification
Basis Classification
Nature Qualitative
Quantitative
Source Primary
Secondary
Value Discrete
Continuous
Qualitative Data
Data which describes features or attributes, such as the eye colour of a

group of individuals, that cannot be measured or computed by arithmetic
relations is termed as qualitative data. They are categorical variables that
serve as labels which advise in which category or class an individual,
object, or process falls.
Quantitative data
Numerical data which results from a process of measurement is termed

quantitative data. Measurement is commonly defined as the assignment of
numbers to represent the properties of objects as per rule. Quantitative data
can be further classified into metric and non-metric data based on the
metric scale that is associated with the measurement of data.
1. Non-Metric Data: The measurement scale used by such data does

not possesses a metric scale with which the distance between the
scale values can be measured. Nominal and ordinal data falls under
this classification.
2. Metric Data: Unlike non-metric data, this data can be measured on
a metric scale, thus, the researcher can define distances between the
scale values while measuring the data. Metric data can be further
divided into two groups:
x Discrete data: This is countable data which can take values
designated by natural numbers, e.g., population, age, etc.
x Continuous data: Continuous data can take values that can be
expressed in rational numbers. This data is measured on a
continuous scale, e.g., height, weight etc.
Primary Data
Information collected directly by researchers through first hand

investigation for the specific purpose of the research project is termed
primary data. Primary data is original data obtained from a primary source
which provides direct evidence related to the issue, e.g., – data collected
using questionnaires, interviews, focus groups etc., during baseline survey.
Secondary Data
Secondary data is data that has already been collected and is available
from other sources. When primary data derived from primary sources is
made available to the researcher, it is called secondary data, e.g., data
collected from the Census of India, AHS (Annual Health Survey), data
that is already published in literature etc. Secondary data is readily
available so it saves time, as the researcher is provided with a larger
database. It also helps the researcher to identify the gaps and deficiencies
in the available data and the additional information that is required.
8.2 Univariate Analysis

When only one variable is being analysed, it is called univariate analysis.
The two main types of analyses that are undertaken for a single variable
are called measures of central tendency and measures of variability.
8.2.1 Measures of central tendency

Measures of central tendency are descriptive measures that indicate where
the centre of the data lies. The goal of descriptive statistics is to describe
the data with one single measurement or statistic. This should be an
average or typical score if it is to accurately describe or represent the
group. There are three measures of central tendency used in different
occasions, the mean, the median and the mode. These can be easily
calculated and help to simplify the understanding of raw data in a variety
of situations.
Mode
The mode is the simplest measure of central tendency; it is the frequency

in a data set that occurs for the maximum number of times. In the case of
nominal scale data, it is the only analysis that can be done along with
percentages.
108 Chapter Eight
Median
The median is the middle value in a data set that divides the ordered data
set exactly in half, with one half of the data being greater than the median
and one half being less than the median. The goal here is to determine the
precise midpoint of the data set. When the number of observations (N) is
an odd number, the median is the middlemost number in the ordered data
set. When N is an even number, the median is the average of the two
middlemost scores in an ordered data set. One very important property of
the median is that it is less sensitive than the mean to extreme scores.
Mean
The mean is the arithmetic average, i.e., the sum of all data divided by the
number of cases.
Mean = (sum of observations)/ (total number of observations)
The mean is the ‘balancing point’ of a distribution. If one score is

changed, then the balance point shifts (unlike the median, the mean is
affected by extreme scores). Since this involves mathematical calculations,
it is only used for interval/ratio data.
Taking an example of the monthly savings of seven SHGs in a village, and

assuming that their monthly saving data is to the tune of 120, 130, 150,
160, 120, 140 and 120. Here the mode is 120 (most occurring frequency as
it appears three times), the median is the mid value when the data set is
arranged in ascending or descending order (this being 120, 120, 120, 130,
140, 150, 160), therefore, 130, while the mean is 134.3, which is the sum
of all values, i.e., 940 divided by the number of observations, i.e., 7.
How is it decided which measure of central tendency should be used?
Often it is best to calculate more than one measure since each measure
tells us something different, although all refer to the centre of the data.
Generally, the mean is the most preferred measure of central tendency
because it uses every score in the distribution. The practitioner needs to
remember that the goal is to use one score to best represent the entire data
set though it might not always be the best.
When the Mode is the best: Since the mean has to use the interval or the
ratio data, the mode is often the best in nominal or ordinal data though it
can be used for all types. Moreover, it is very easy to use, since no
calculations are necessary.
When the Median is preferred over the Mean: There are several
instances when the median is the preferred statistic:
1. When there are extreme scores: An extreme score pulls the mean
towards that score, which means that an extremely high score pulls
the mean higher, while an extremely low score pulls the mean
lower. If, for example, one of the groups in the data given above
saved INR 1000, the mean would not be the best way by which to
represent the central tendency.
2. If some scores are missing.
3. If one has ordinal data; since this data is discrete, the mean is
generally not used.
8.2.2 Measures of variability

A measure of central tendency alone does not suffice to tell the complete
story if one is to understand data distribution completely. Unlike the
measures of central tendency which capture similarities, a measure of
variability is a measure of the dispersion of the scores around the mean,
i.e., how spread out is the data, or how different is the data? In the data set
given above, measures of variability are required to know the minimum or
the maximum savings, the range of the group's savings, and the average
deviation from the mean savings etc. The goal here is to determine how
much variability is there in the data set or the distribution.
Measures of variability have three main purposes:
1. To describe the distributions (how disperse or variable they are)

2. It gives an idea of how accurately the mean describes the
distribution
3. It gives an indication of how well a sample represents the entire
population (distribution).
The key measures of variability are:
Range
The range is the difference between the highest and lowest score in the
distribution. Expressed as a single number, the range is the result of the
following formula:
110 Chapter Eight
Range = Highest score - Lowest score
For the above example, Range = 160 – 120 = 40
Range checks are a very straight-forward procedure used on numeric fields

and can be a good initial check on the data. In practice, however, people
usually list the minimum and the maximum rather than the single number
range.
Variance
The variance, s2, is the mean of the squared deviations:
s2 = Ȉ (x - x )2 or SS
N N
where SS = the sum of squares, which refers to the sum of squared

deviations from the mean Ȉ(x - x )2.
The variance detects the differences that the range does not detect because
it uses all the data points in the calculation and not just the minimum and
the maximum.
The key to variance is the SS since if the SS is large, the variance is large,
etc.
The variance is very useful in more sophisticated inferential statistics, but

its use in descriptive statistics has one major flaw, as the calculated value
is expressed in squared units of measurement. Thus, it is seldom used for
interpreting descriptive variance.
Standard Deviation
To work around this problem of using the variance, the researcher should
go back to the original units, and take the square root
ı= Ȉ(X- X ) 2 or SS
n n
This is the standard deviation, which is the most commonly used measure
of variability. It is frequently cited with the mean as the main piece of
information about a distribution.
It considers every piece of data in the distribution (as does the mean) as it
measures, on an average, how much each piece of data deviates from the
mean.
This is done in several steps:
1. Calculate the mean (134.3 in the example)

2. Measure the deviation of each score (X) from the mean ( X )
3. Square the deviations to get rid of the (-) signs
4. Divide the sum of the squares by the number of observations
5. Take the square root of the total value
The standard deviation has been calculated below taking the SHG example
discussed above.
Table 1.11: Calculating the Standard Deviation - Example
S. no. Savings (X- X ) (X- X )2

1 120 -14.3 204.49
2 130 -4.3 18.49
3 150 15.7 246.49
4 160 25.7 660.49
5 120 -14.3 204.49
6 140 5.7 32.49
7 120 -14.3 204.49
Total 0 1571.43
If the sum of the square of deviations is divided by the number of

observations (1571.43/7), the figure 224.49 is obtained. The square root of
this figure is 14.98. This is the standard deviation. Even though the
variance provides some idea of deviation, there is still another problem to
deal with. The square of the numbers magnifies the numbers considerably
and the units are NOT in their original form. Therefore, the square root of
the variance is taken, which takes one back to the original units; it makes
the correction for having squared the distances from the mean (the
deviations) and gives the standard deviation.
112 Chapter Eight
Standard deviation = variance
The standard deviation (and the variance) is very sensitive to extreme

scores (as is the mean). Therefore, one needs to be very careful in
interpreting the standard deviation or the variance for a distribution that
contains even a few very extreme scores.
This should give the practitioner an idea of how much dispersion around
the mean there is in a particular distribution. It provides the overall notion
of dispersion in a distribution needed to complete the description of the
distribution.
8.3 Bivariate analysis: Correlation and Regression

Correlation
Correlation is one of the most widely used measures of association

between two or more variables. In its simplest form, it signifies the
relationship between two variables, i.e., whether an increase in one
variable results in increase in the other variable. The measure of
correlation is employed to explore presence or absence of correlation, i.e.,
whether there is correlation between the variables in an equation.
Correlation coefficient also describes the direction of correlation, i.e.,
whether it is positive or negative. At the same time, it also describes the
strength of the correlation, i.e., whether an existing correlation is strong or
weak.
There are various measures of correlation between nominal and ordinal

data. Pearson product-moment correlation coefficient is a measure of
linear association between two interval-ratio variables. The measure,
represented by the letter ‘r’ varies from -1 to +1. A zero correlation
indicates that there is no correlation between the variables.
A correlation coefficient indicates both types of correlations as well as the

strength of the relationship. The coefficient value determines the strength,
whereas the sign indicates whether variables change in the same direction
or in the opposite direction. A positive correlation indicates that as one
variable increases, the other variable also increases in a similar way. A
negative correlation signified by a negative sign indicates that there is an
inverse relationship between two variables, i.e., an increase in one variable
is associated with decrease in the other variable. A zero correlation
suggests that there is no systematic relationship between the two variables

and any change in one variable is not associated with change in the other
variable.
Regression
Regression is one of the most frequently used techniques in social

research. It is used in estimating the value of one variable based on the
value of the other variable. It does so by finding a line of best-fit using the
ordinary least square method. The relation between the variables could be
linear or non-linear, correspondingly, the regression equation could also be
linear or non-linear.
Regression analysis extends correlation by asserting a direction of

causality (from explanatory or independent variable to dependent or
endogenous or explained variable) and measuring the influence of the
explanatory variable(s) on the explained variable.
By convention, Y is used for the dependent variable, and X for the

independent variable. The aim is to find the ‘line of best-fit’ to the data.
Initially, the relationship between Y and X is defined as a linear one.
Y = a+bX
In regression equation, ‘a’ is defined as the intercept and ‘b’ is known as

the regression coefficient. The value of b indicates the change in the
dependent variable for every unit change in the independent variable.
Regression coefficient is another widely used measure of association

between two interval-ratio variables. Regression coefficient is an
asymmetric measure of association, which is why the regression coefficient
of the dependent variable on the independent variable is different from the
regression coefficient of the independent variable on the dependent
variable. Further, depending on the application of the regression method,
the practitioner can use an asymmetric measure of association or a
symmetric measure of association. However, when the practitioner needs
to predict one variable by another variable, then an asymmetric measure is
preferred.
114 Chapter Eight
8.4 Quantitative Data Analysis using Computer-aided

Analytical Software
8.4.1 Performing analysis using MS-Excel
MS-Excel provides a vast range of functions for data management and
analysis. If the practitioner knows the functions needed and how to use
them, they can enter the functions into a cell and start analysing the data.
However, in case they are not sure of which function to use for a particular
task or analysis or even how to use any particular function, MS-Excel can
assist them with that too.
Finding the Right MS-Excel Function
To identify the functions available in MS-Excel, the user needs to click the
on Formulas tab on the Ribbon and then click the Insert Function button
in the Function Library group. Alternatively, the user clicks the button on
the formula bar. Either approach provides the Insert Function dialog box
shown in the figure below:
Figure 1.6: Insert Function

The Search for a function box at the top of the Insert Function dialog box
enables the user to type a brief description of what they want to do. After
that, when the Go button is clicked, MS-Excel searches for and displays
the functions that may accomplish the task in the Select a function box. In
many situations, however, the user may want to browse through an entire
category of functions to see what is available. For this task, the user has
the choice of Or select a category box which contains a drop-down list of
several categories of functions provided by MS-Excel. The figure below
shows that the Statistical category is selected. As a result of this, MS-
Excel’s statistical functions appear in alphabetic order in the Select a
function box. The user will find the AVEDEV function listed first,
followed by the AVERAGE function, and so on.
Figure 1.7: Insert Function in Statistical Category
For data analysis, the user can choose out of the key MS-Excel functions
displayed. For this purpose, the sample dataset Health_Data_2013-14.xlsx
is used.
116 Chapter Eight
The figure below provides a screenshot of the dataset:
Figure 1.8: Dataset - Example
The first step in any data analysis is to understand the dataset. In the
example used, it can be observed that the dataset consists of the following
variables:
x District Name
x Month of reporting
x Total number of pregnant women registered for ANC
x Deliveries conducted at facility (including C-sections)
x The number of newborns breastfed within one hour of birth
The data is culled from three districts viz., District-1, District-2 and
District-3 and contains 36 cases in all. Each case contains the monthly data
for a district and based on these observations, the user analyses the data.
The AVERAGE function is used on the variables listed above to compute

the average number of deliveries in a month.
The first step is to select the cell in which the output is to be displayed.
The cell next to the last entered value at the end of column D is selected.
In this case, the cell number is D38. After this cell has been selected, the
user needs to:
1. Go to the Formulas tab in the Ribbon and select the Insert Function
2. Select All from the drop-down menu of Or select a category
3. Select AVERAGE from the list of functions and press OK.
The dialog box given above will appear.
The AVERAGE function requires the user to give an input of the list of
numbers that are used to compute the average. To estimate the average
number of deliveries in facilities in a month, the entire range of the
variable for institutional deliveries i.e., D2 to D37 is selected. The user
needs to click OK to get the average number of deliveries conducted at the
facility per month across the three districts, which is 2684.7.
Using the AVERAGE function, the average number of deliveries in a

month is obtained. However, this average was computed for all the three
districts. If, for instance, District-1 had a higher delivery load than the
other districts, the estimation of the average number of deliveries is
inaccurate.
One way to tackle this issue is to compute the average number of

deliveries separately for each of the districts. As in the figure above, the
dialog box for the AVERAGE function asks the user to give an input of
the range of numbers to be used to calculate the average. To calculate the
118 Chapter Eight
averages of institutional deliveries separately for the districts, the process

is repeated three times, and each time the range of cells belonging to a
particular district is specified to calculate the average.
The following are the functions to calculate the average number of

deliveries in a month for each of the districts:
Average number of deliveries in a month for District 1 = AVERAGE

(D2:D13)

(D14:D25)

(D26:D37)
Though the required average figures may be obtained using this process,
several steps are involved in it. An easier way to do this is by using MS-
Excel Pivot Tables.
For the process of using MS-Excel functions, the user needs to:
1. Click on the cell from which the output is required

2. Insert the appropriate function using the formula list and specify
the range
3. Click on OK
The desired output can be obtained using this process and subsequently the
results can be interpreted. Given here is the list of functions that are most
frequently used:
Table 1.12: Frequently Used Functions
Function Use
MAX Returns the largest value from a list of supplied numbers
MIN Returns the smallest value from a list of supplied numbers
SUM Returns the sum of a list of supplied numbers
IF Tests a user-defined condition and returns one result if the
condition is TRUE, and another result if the condition is
FALSE
COUNT Returns the number of numerical values in a supplied set of
cells or values
COUNTIF Returns the number of cells (of a supplied range) that satisfy
a given criterion
CORREL Returns the correlation coefficient between array 1 and
array 2 cell ranges
AVERAGE Returns the Average of a list of supplied numbers
MEDIAN Returns the Median (the middle value) of a list of supplied
numbers
MODE Returns the Mode (the most frequently occurring value) of a
list of supplied numbers (replaced by MODE.SNGL
function in MS-Excel 2010)
STDEV.S Returns the standard deviation of a supplied set of values
(which represents a sample of a population) (New in MS-
Excel 2010 - replaces STDEV function)
STDEV.P Returns the standard deviation of a supplied set of values
(which represents an entire population) (New in MS-Excel
2010 - replaces STDEVP function)
120 Chapter Eight
Using Pivot Tables in MS-Excel
Pivot Tables are one of the most powerful features of MS-Excel. Pivot
tables provide an easy way to summarise, analyse, explore and present
data. A Pivot Table is a versatile and user friendly reporting tool that
makes it easy to extract information from large tables of data without the
use of formulae. By moving or pivoting fields of data from one location to
another or by using drag and drop, the same data can be seen in several
different ways.
To analyse the data using Pivot Tables, the first step for the user is to
insert a Pivot Table for the dataset and click on any single cell in the
dataset.
The user goes to the insert tab and clicks Pivot Table.
Figure 1.11: Insert Pivot Table
The following dialog box appears. MS-Excel automatically selects the data
and the default location for a new pivot table is a New Worksheet.
Figure 1.12: Creating a Pivot Table
The Pivot Table Fields will appear on the right side of the screen. For
example, to get the total number of women registered for ANC for each
district, the user drags the following fields to different areas on the Pivot
Table Field:
1. District Name to Rows area.

2. Number of women registered for ANC to Values area.
122 Chapter Eight
Figure 1.13: Selecting Pivot Table Fields
With these settings, an output showing the total number of pregnant

women registered for ANC for each of these three districts is obtained.
Figure 1.14: Pivot Table Output
Pivot Tables also provide options to summarise the data using functions
such as count, average, standard deviation etc. These options can be
chosen from the value field settings in the values area.
To summarise data with the help of the Pivot Tables tool, the user needs
to:
1. Insert Pivot Tables for the data set

2. Drag columns in the appropriate fields based on the type of output
required
3. Select the appropriate function from value field settings as per the
summary required.
Installing the MS-Excel Analysis ToolPak add-in
The Analysis ToolPak is an MS-Excel add-in programme that provides

data analysis tools for a variety of statistical analyses. The steps that need
to be followed to load the Analysis ToolPak are mentioned below:
Step 1: Click the Office Button

Step 2: Click Excel Options (on the taskbar at the bottom)
Step 3: When the Excel Options dialog box appears:
x Select Add-Ins from the list of options (on the pane on the left)
x In the Manage box, select Excel Add-Ins (in the second last
row)
x Click Go
Step 4: When the Add-Ins dialog box appears:
x Select Analysis ToolPak
x Click OK
The data analysis functions in MS-Excel are ready to be used now.
Calculating descriptive statistics using ToolPak
Suppose the descriptive statistics for the number of children breastfed

within one hour of birth from the dataset are to be examined, the user
needs to:
1. Go to the Data tab

2. Select Data Analysis
3. Wait for the dialog box to appear
124 Chapter Eight
Figure 1.15: Analysis using the ToolPak-Data Analysis
4. Select Descriptive Statistics from the list of analysis tools and click
OK. The following dialog box appears.
Figure 1.16: Selecting Descriptive Statistics

The user gives the input about the data to be analysed in the Input Range
box. To analyse the number of children breastfed within one hour of birth,
E1:E37 is selected as the Input Range. Each column in the dataset
represents a variable, the user clicks on the Columns in the Grouped By
checkbox. Also, because Row 1, which includes the variable name is
included, the user will click on the Labels in First Row checkbox.
Next, under Output Options, the user decides where the results are to be
written and gives the command for a new sheet by clicking on the New
Worksheet Ply radio button so that accidentally some of the data is not
overwritten.
In the example given above, the Summary statistics checkbox is selected.
The user needs to click OK for the new results to appear in a New
Worksheet.
Calculating correlation using ToolPak
Using the ToolPak add-in programme, the relationship between two

variables can be calculated. To see whether there is any relationship
between the number of institutional deliveries and the number of children
given breast milk within one hour of birth, the user: needs to:
1. Go to Data and select Data Analysis.

2. Select Correlation from the list of analysis tools and tick the
checkbox OK.
The following dialog box will appear.

126 Chapter Eight
Figure 1.17: Correlation using the Analysis ToolPak programme
The first box is the Input Range, where the user gives the input about the
variables for which the correlation is to be analysed.
To check the relationship between the number of institutional deliveries

and the number of children breastfed within one hour of birth, D1:E37 is
selected as the Input Range. In the dataset, each column represents a
variable, so, the Columns radio button in the Grouped By option is clicked.
Also, because Row 1 which includes the variable name is included, the
Labels in First Row checkbox is ticked. Finally, the user clicks OK to view
the output.
Performing Regression using ToolPak
To check how a single dependent variable is affected by one or more

independent variables, the ToolPak is used to perform a linear regression
analysis.
In the dataset, to see how early initiation of breastfeeding is affected by

institutional deliveries, the user needs to:
1. Go to Data and select Data Analysis

2. Select Regression from the list of analysis tools and click on OK
The following dialog box appears:

Figure 1.18: Regression using the Analysis ToolPak programme
For the Input Y Range, all the values of the dependent variable (the
number of children breastfed within one hour of birth) is selected.
In the next step the independent variable (the number of deliveries in the
facility) is entered in the Input X Range box.
The Labels checkbox is ticked since the first row of data is the variable
label. Once again, the output is entered on a new spreadsheet.
When the user selects OK, the result appears in the New Worksheet.
128 Chapter Eight
Figure 1.19: Regression Output

Interpretation of the Output
As explained above, the dependent variable in this case is the number of

children breastfed within one hour of birth, while the independent variable
is the number of deliveries in the facility. The coefficient of the
independent variable denotes the size of the effect that this variable has on
the dependent variable and the sign on this coefficient denotes the
direction of the effect. In this case, it is observed that with the change in
each unit of the independent variable, the dependent variable changes by
1.33 times in the positive direction. Also, as the p- value is 1.709E-19, less
than this proves that this value is significant. The value of intercept in this
case is 422.18 but as its p value .07 is greater than .05, it is not significant.
Therefore, the equation for our model for this example would be:
Y (The number of children breastfed within one hour of birth) = 1.33 X

(the number of deliveries in the facility)
Another important result that is derived from this model is the R squared
value. The R squared value denotes the fraction of the variance in the
dependent variable that is accounted for by the independent variable. In
this case, the R square value is 0.91, which suggests that 91 per cent of the
variation in the dependent variable is accounted for the independent
variable.
The process of using the Analysis ToolPak programme is as follows:
1. Go to Options Æ Add-Ins and select the Analysis ToolPak

programme to be added to MS-Excel
2. Go to Data and select Data Analysis to open the ToolPak
programme
3. Select the appropriate analysis option mentioned in the list
4. Select the input range and other parameters to get the output.
8.4.2 Overview of data analysis using Statistical Package

for Social Science
Statistical Package for Social Science (SPSS) is the most popular quantitative
analysis software used today in social research. SPSS is a comprehensive
and flexible statistical analysis and data management software, which can
utilise data from almost every type of dataset format to generate tabulated
reports, distribution charts, trends, descriptive statistics and complex
statistical analyses. Besides this, SPSS is compatible with almost every
130 Chapter Eight
operating system such as Windows, Macintosh and open sources systems

like UNIX or LINUX.
SPSS, unlike other statistical analysis software, provides a window based

user interface which makes it very user friendly. Researchers can use
simple menus, pop up boxes or dialog boxes to perform complex analyses
without writing even a single line of syntax. SPSS is an integrated
software which also provides a spreadsheet-like utility function for
entering data and browsing the working data file. SPSS and spreadsheet
SPSS have an advantage over other prevalent statistical software such as
SAS and STATA because of its window driven options. Its operation is
pivoted around three windows as mentioned below:
1. The Data Window, with a blank data sheet ready for analysis, is the
first window the user encounters. It is used to define and enter data
and to perform statistical procedures. Data files have the suffix ‘.
sav’ at the end of the file.
2. The Syntax Window is used to keep a record of all commands
issued in a session. The researcher does not necessarily have to
know the language for writing syntax, instead they can just select
the appropriate options from the menu and the dialog box, after
which they can select the Paste function option. This command
pastes the equivalent syntax of the selected operation in the Syntax
window. Beyond serving as a log for operations, it is possible to
run commands directly from the syntax window. It is recommended
that a syntax file is maintained with the syntax of all data analyses.
This ensures regeneration of the required tables and analysis
quickly. The syntax files have the suffix ‘.sps’ at the end of the file.
3. Output window: Whenever a procedure is run, the output is
directed to a separate window called the Output Window. The
tables can be directly copied from the Output Window and be saved
in MS-Excel or MS-Word files. Output files have the suffix ‘. spo’
at the end of the file.
Entering Data
The researcher can create a data file simply by entering the data. The
present section describes the step-by-step procedure for creating a data file
by entering variable information about subjects or cases. The SPSS main
data window provides two options on the bottom left hand corner of the
screen and the researcher can access either the Data View or the Variable
View window as shown in the figure given below.
Figure 1.20: Data Editor Window in SPSS

132 Chapter Eight
Figure 1.21: Variable View in SPSS

The Variable View window depicts the characteristic of the variable in

terms of name, type and nature as mentioned below:
x Name: The variable names in SPSS are not case sensitive but they
must begin with a letter. Further, the variable name should not
exceed eight characters.
x Type: SPSS also provides the facility to specify the data type, i.e.,
whether the data is in numeric, string or date format etc., though the
default format is numeric.
x Labels: SPSS also provides the facility of attaching labels to a
variable name. A variable name is limited to a length of eight
characters but by using a variable label, a researcher can use up to
256 characters to attach a label to a variable name. This provides
the ability to have very descriptive labels that appear as the output.
The researcher can enhance the readability of the output by using
the Labels option.
x Values: The researcher can assign values, e.g., the male and the
female can be coded as 1 and 2, respectively.
Researchers should familiarise themselves with the data characteristic at

the initial stage. Though SPSS also provides the facility by which the
researcher, at later stages of data analysis, can list information about data
nature, type and characteristic by just selecting the data dictionary option
which is explained below.
Data Dictionary
A data dictionary provides information about the nature, type and

characteristics of data. It provides all information i.e., variable name, type,
variable label, value labels, missing value definition, and display format
for each variable in the data set. In other words, it documents how each
variable in the data set is defined.
A data dictionary is easily produced by selecting the File Info option from
the Utilities menu. The data dictionary first provides a list of variables on
the working file, wherein the variable name appears on the left-hand side
and the column number of the variable appears on the right side of the
output window. The data dictionary also provides the print and write
format after the variable name, followed by special characteristics of the
variable such as value labels.
134 Chapter Eight
Opening Data Files
Data can be exported in several ways, though the simplest is the Open
option. In this case, the requisite database file can be directly opened by
simply selecting the ‘Open’ option as displayed in the figure given below.
Secondly, data files of other formats (dBase, Access) can be imported to
SPSS through database capture. It can also be used to read ASCII data as it
is the third option to read files that are saved in ASCII format, which
further provides two options - Freefield and Fixed Columns.
Figure 1.22: Opening Data File in SPSS

136 Chapter Eight
a) Opening an SPSS file
An SPSS file can be opened quite easily by clicking on the menu item and
selecting the option File -> Open -> Data. The SPSS file type (having a '.
sav' extension) can then be selected and the checkbox ticked to open it.
Similarly, an SPSS output file can be opened by selecting File -> Open ->
Output document files type (having a '.spo' extension).
b) Opening an Excel file
An Excel workbook can be opened by selecting the Read Text Data option
by navigating through File->Read Text Data. Excel file types (.xls) can be
selected in file type and any excel file can be selected to be opened as
shown in the figure given below.
Figure 1.23: Option of Opening an Excel File
After this, the Excel workbook can be selected which can be opened by
selecting the option 'Open and Continue'. This opens a screen like the one
shown in the figure below.
Figure 1.24: MS Excel File Opened in SPSS

138 Chapter Eight
It is important to point out that in case the first row of the Excel workbook
contains the variable names, then it should be ensured that the checkbox
for this option is ticked. In case the data is not on the first worksheet, the
worksheet can be changed using the down arrow to select the worksheet
which has the data. SPSS also provides the facility to select the range from
the worksheet which the researcher wants to use, in case the researcher is
not interested in using all the data in the worksheet.
c) Importing Text File
SPSS provides the facility to import the text file through ‘Text Wizard’.
Text data files can be read in a variety of formats mentioned below:
• Tab-delimited files
• Space-delimited files
• Comma-delimited files
• Fixed-format files.
At the first stage, either a predefined format that can be applied is

specified, or the remaining steps in the Text Wizard are followed.
Text Import Wizard at Step 2 requests information about how variables are
arranged, i.e., whether the variables are defined by a specific character
(spaces, tabs or commas) or whether the variables have a fixed width,
where each variable is recorded in the same column. In the next step, it
further needs to be seen if the variable names are included at the top of the
file as the file may or may not contain variable names. In case the file
contains variable names of more than eight characters, then the variable
names are truncated. Further, in case the files do not contain variable
names, SPSS can allocate default names.
In the third step, the Text Import Wizard requests information about how
cases are represented and the number of cases that are to be imported.
Usually the first case begins with line 1 if no variable name is supplied and
line 2 if the variable name is supplied. Further, each line normally
represents a case. In this step, it is also necessary to specify the number of
cases the researcher wants to import, i.e., whether the researcher wants to
import all the cases, the first n cases, or a random sample.
Step 4 of the Text Import Wizard requests information about the file, i.e.,
whether it is comma delimited or space delimited. This step, in the case of
delimited files, allows selection of the character or symbol used as a
delimiter, whereas in case of fixed width files, the step displays vertical
lines on the file which can be moved if required.
Step 5 of the Text Import Wizard requests information about variable

specifications. This step allows the researcher to control variable names
and the data format to read each variable which is included in the data file.
Values which contain invalid characters for the selected format are treated
as missing.
In the last step, i.e., step 6 of the Text Import Wizard, SPSS provides a
facility to save the file specification for later use, or to paste the syntax.
Basic Data Management Function
a) Recoding Variables
Sometimes, while analysing and reporting, fewer categories are required

by the researcher than as envisaged in the survey.
In such a situation, the option of recoding variables is used as a way of

combining the values of a variable into fewer categories.
For this, there are two options provided by SPSS for recoding of the
variable, viz., Recode into Different Variables and Recode into Same
Variables. It is strongly recommended that initially the researcher uses
only the Recoding into Different Variables option because even if the
researcher makes any error, the original variable would still be in the file
and the variables can be recoded.
The data can be recoded using several options. A particular value can be
changed into a new value by entering the value to be changed into the Old
Value box, while the new value is entered into the New Value box.
b) Creating New Variables using Compute
In SPPS, new variables can be created by using the Compute command as

demonstrated in the figure below. This is done by typing the name of the
variable that is to be created in the Target Variable field.
After typing the target variable, the computation that involves the
variables in the list is specified to create the target variable. The
researcher can use all the operations listed at the bottom of the screen,
though operations within parentheses are performed first.
140 Chapter Eight
Figure 1.25: Using Compute to Create New Variables

c) Creating New Variables Using If
The If command can also be used to create new variables based on the old
variables. This can be done by selecting Transform and then selecting the
Compute function. Further, the Include if Case Satisfies Condition dialog
box should be selected to select subsets of cases using conditional
expressions.
There are several operators which can be used, though in the majority of
cases, one or more of the six relational operators (<, >, <=, >=, =, and ~=)
are used.
d) Spilt File option
SPSS provides the facility of splitting a data file into separate groups for
analysis based on the values of one or more grouping variables. In case
multiple grouping variables are to be selected, then cases are grouped by
each variable within categories of the first grouping variable, based on the
groups selected. For example, if Occupation is selected as the first
grouping variable and Education as the second grouping variable, then
cases will be grouped by Education classification within each Occupation
category.
Split a Data File for Analysis
The Data-> Split File option can be very easily chosen from the menu as
shown in the figure below. Further, at the next stage, groups are selected to
organise output by groups. Before splitting the file, it is important to select
the Split File option from the data editor window.
142 Chapter Eight
Figure 1.26: Splitting a Data File for Analysis

The Split File option can be turned off by selecting Data and then the Split
File option from the Data Editor window by clicking on the Analyze all
cases option.
e) Using Select Cases
SPSS also provides the facility to select subsets of cases for further
analysis. To arrive at this option, first Data is selected and then the option
Select Cases. This opens the Select Cases box and if the researcher wants
to select a subset of these cases, they can select the option If condition is
satisfied to select the subset of cases.
Further, at the bottom of the window, there is a check box to specify that
the unselected cases are filtered. By selecting this option, the unselected
cases can be used later by clicking on All Cases. However, if the Delete
option is selected, the unselected cases are permanently deleted.
f) Weighting Cases: In particular situations, it may be necessary to weight

some cases in the data more heavily than others. For instance, there is a
HH in a survey which has an equal probability of selection as other HHs,
and there is more than one person eligible in the HH out of which one
individual is randomly selected.
To weight the data using this variable just created, the researcher selects
Data and then Weight Cases. In the next step, the researcher selects the
circle to the left of Weight Cases.
g) Missing Values
Missing data can be due to various factors, such as the way in which the
interviewer has administered the question, or left certain questions blank,
the respondent has declined to respond to certain questions, or due to some
human error in data coding and data entry etc. There are a few techniques
to handle data with missing values: (1) complete case analysis (list wise
deletion), (2) available case methods (pair-wise deletion), and (3) filling in
the missing values with estimated scores (imputation).
In case a response is missing, it is recommended that the researcher fixes a

pre-determined code for the same, for instance, 999 so that it is understood
that the data is missing rather than it being a data entry error. Missing
values can be coded in the data by recoding the missing values. This can
be done by navigating through the following tabs Transform->Recode Into
Same Variable. The recoding can be done into the same variable or
144 Chapter Eight
different variables as required. As shown in the figure below, the System –

missing radio button can be checked under the old value category and the
value can be assigned, which is 999 in this case, after checking the Value
radio button under the New Value category. After this the Add button can
be checked and the system Missing values can continue to be recoded, as
demonstrated in the figure below.
Besides this, the researcher may use various functions and simple
arithmetic expressions to extract missing values in different ways.
Thus, the researcher can use the expression AVERAGE in the example
(var1+var2+var3)/3 where the result is treated as missing, only if any of
the three variables i.e., var1, var2 or var3 has a missing value.
The researcher can use the expression MEAN (var1, var2, var3) where the
result is treated as missing only if a case has missing values for all three
variables.
SPSS also provides the facility of specifying a statistical function, where

the researcher can specify the minimum number of variables that have no
missing values, e.g., the function MEAN.2 (var1, var2, var3).
Figure 1.27: Coding Missing Values

146 Chapter Eight
h) Reliability Analysis
Reliability signifies the issue of consistency of measures i.e., the ability of

a measurement instrument to measure the same thing with consistency
each time the instrument is used. Reliability inherently depends on three
key elements, such as stability, internal reliability and inter-observer
consistency.
SPSS provides the option of doing reliability analysis to assess the

additive nature of individual items. This option can be accessed by going
to Analyze->Scale->Reliability analysis as demonstrated in the figure
below.

148 Chapter Eight
As the next step, the selected indicators should be entered as items to

assess the reliability of scale. To do so, it is necessary to have an idea
about the scale mean, scale variance and cronbach alpha for each item, in
case that particular item is to be deleted from the scale. These statistics can
be easily computed by selecting the Descriptives for Scale option from the
Reliability Analysis window as demonstrated in the figure below.

150 Chapter Eight
The figure above details the reliability coefficients for an awareness scale,
which involves eleven awareness variables about the leprosy symptom,
having a high value of alpha, i.e., .7672. A rule of thumb for alpha value
states that 0.60 is the lower level of acceptability for the alpha, though
alphas in the 0.70s and 0.80s range are preferable.
Further, the Inter-Item Correlations and Scale if item deleted are also very
important indicators to assess the reliability of scale. The Inter-Item
Correlation option allows the researcher to see if any of the items are
negatively correlated with the other items in the scale, while the Scale if
item deleted option reveals the alpha if each item is deleted from the scale.
Data Analysis
a) Univariate Statistics
Univariate analysis, as its name suggests and explained above in this

chapter, provides analytical information about one variable, which could
be metric or non-metric in nature. SPSS provides the facility for univariate
analysis such as Frequencies, Descriptives and Explore, all of which are
located under the Analyze menu as shown in the figure below.
Figure 1.30: Analysing Frequencies

152 Chapter Eight
a) If the Statistics option at the bottom of the window is selected, this

option offers several summary statistics. Based on the options selected, the
summarised information gets displayed in the output window. The
Frequencies option generates the frequency information in addition to the
measures of central tendencies and dispersion. Each time when statistical
procedures like Frequencies and Descriptives are selected, the results are
immediately displayed in an Output window.
b) Frequencies
The Frequencies option can be used to list detailed information of the

selected variables. The option is extremely useful for nominal and
categorical data such as in the case of gender, wherein data is coded in two
categories. The Frequencies option provides a table, which shows counts,
percentages and statistics including percentile values, central tendency,
dispersion and distribution.
c) Descriptives
The SPSS package provides the facility to obtain summary information

about the distribution, variability, and central tendency of continuous
variables by using the Descriptives option. The researcher can select
various measures such as mean, sum, standard deviation, variance, range,
minimum, maximum, standard error (SE), mean, kurtosis and skewness by
using the Descriptives option.
d) Explore
The SPSS package provides the option to examine the central tendency
and distributional characteristics of continuous variables by using the
Explore option. The researcher can select statistics such as M estimators,
outliers, and percentiles by using the Explore option. It also provides
options like grouped frequency tables, displays, as well as stem and leaf
and box plots.
e) Cross Tabulations
Cross tabulation is one of the easiest ways of summarising the data which
can be of any size in terms of rows and columns. It generally allows the
researcher to identify relationships between the cross tabulated variables
based on their cell values.
The SPSS package provides the facility for generating bivariate cross
tabulations. A cross tabulation helps in analysing the association of one
variable with another variable and is extremely useful in cases where each
variable contains only a few categories.
The crosstab option is found by navigating through Analyze->Descriptive

Statistics->Crosstab. Subsequently, the dependent and the independent
variables are selected to generate the crosstab option. After selecting the
crosstab option, the Statistics button is selected for the Chi-square test to
obtain a measure of statistical significance, i.e., Phi and Cramer’s V as
shown in the figure below. If both variables are dichotomous, the Phi
Correlation Coefficient is used. Cramer’s V is preferred over the Phi
Square Coefficient in the case of larger tables.
154 Chapter Eight
Figure 1.31: Crosstab Statistics

f) The Mean
The mean is a very important measure of the central location of distribution,

especially in the case of interval and numerical data. Researchers can use two
methods from the SPSS package to produce the mean. One option is to use
the Frequencies sub option from the Descriptive Statistics option.
Alternatively, the Compare Means option is selected from the Analyze

menu option. Subsequently, the variable which needs to be put in the
Independent List box and the one which needs to be put in the Dependent
List box can be selected accordingly.
g) The Chi-square Test
The chi-square test is only used with measures, which place cases into
categories. This test indicates whether the results from the two measures
are about what one would expect if the two were not related. This test can
be conducted by selecting Analyze from the top menu and Descriptive
Statistics and Crosstabs to open the Crosstabs dialog box. Further, in the
Crosstabs dialog box, the variables which the researcher wishes to crosstab
can be selected. Further, the researcher can select the Crosstabs: Statistics
box option, and then select the Chi-square box option to continue.
h) Independent-Samples t-Test
The Independent-Samples t-test procedure compares the means for two groups
of cases. In fact, there are two variants of unpaired t-tests based on the
assumption of equal and unequal variances between the two groups of cases.
In the case of an unpaired t-test, subjects are randomly assigned to two groups,
so that after employing the Significance-test, it can conclude if the difference
in response is due to the treatment and not due to other factors.
The Independent-Samples t-Test can be accessed by selecting Analyze and

then pointing the mouse at Compare Means and clicking on the
Independent-Samples t-Test option. To explain the test further with an
example, TV viewership is compared across the gender variable. This test
can be done by putting the TV viewership frequency variable into the Test
Variable Box and the gender variable in the Grouping Variable Box.
At the next stage, the groups are defined by selecting the Define Groups
button, which opens the Define Groups box. Since males are coded as 1
and females are coded as 2, the researcher types 1 in the Group 1 box and
2 in the Group 2 box.
156 Chapter Eight
Table 1.13:Independent Samples t-test

The table above shows the results of the two t tests. The table also gives
the values for the degrees of freedom and the observed significance level.
It is important to point out that in this test the null hypothesis is that men
and women have the same TV viewership frequency. The null hypothesis
can be easily tested by comparing the p value with the specified
significance level. This significance level value is the probability that the t
value would be this big or bigger simply by chance if the null hypothesis
were true. Since this probability is less than .05 (the significance level
researchers use by convention in social studies), the researcher can reject
the null hypothesis and conclude that probably there is a difference
between men and women in terms of TV viewership.
i) Paired-Samples t-Test
The paired-samples t-test is very similar to the unpaired-samples t-test,

except for the difference that the paired-samples t-test is related to
matched samples. It tests the difference between the raw scores and
assumes that the data is measured in the interval or the ratio scale. This
test assumes that the observed data from the matched samples is drawn
from a population with a normal distribution.
The test statistic is t with n-1 degrees of freedom. If the p-value associated
with t is low (< 0.05), there is evidence to reject the null hypothesis. This
proves that there is evidence that there is a difference in the means across
paired observations.
The researcher can access the Independent-Samples t-test by selecting

Analyze on the menu option and then pointing the mouse at Compare
Means and finally selecting the Paired -Samples t-test option.
j) Correlation
Correlation, as described earlier in this chapter, is one of the most widely

used measures of association between two or more variables. In its
simplest form, it signifies the relationship between two variables i.e., to
show whether increase in one variable results in increase in the other
variable.
For example, one could hypothesise that as education increases, the level
of prestige of one’s occupation also increases. The Pearson Correlation
Coefficient could tell us the correlation between these two variables. This
can be assessed by selecting the Analyze menu option, then Correlate and
Bivariate sub-options.
158 Chapter Eight
Though there are various measures of correlation between nominal or

ordinal data, Pearson product-moment correlation coefficient is a measure
of the linear association between two interval-ratio variables. The
measure, represented by the letter r, varies from -1 to +1. A zero
correlation indicates that there is no correlation between the variables. The
SPSS package includes another correlation test, Spearman’s rho, besides
Pearson's Correlation Coefficient to analyse variables that are not normally
distributed, or ranked. The researcher can opt for both Pearson correlation,
and Spearman’s rho. The output screen shows two tables: one for
Pearson's Correlation Coefficient, and one for Spearman’s rho.
A correlation coefficient indicates both types of correlation as well as the

strength of the relationship. The coefficient value determines the strength,
whereas the sign indicates whether the variables change in the same
direction or in the opposite direction.
If the coefficient is away from 0, regardless of whether it is positive or

negative, it signifies a stronger relationship between the two variables.
Thus, a coefficient of .685 is exactly as strong as a coefficient of -.685.
The only difference lies in the fact that positive coefficients tell us there is
a direct relationship and as one variable increases, the other variable also
increases. Negative coefficients indicate that there is an inverse
relationship and so when one variable increases, the other one decreases.
k) Regression
Regression is widely used in estimating the value of one variable based on

the value of the other variable. This is done by finding a line of best fit
using the ordinary least square method. The relation between variables
could be linear or non-linear, which would make the regression equation
also linear or non-linear.
Further, depending on the number of variables, the regression technique is

classified into simple regression and multiple regression. In simple
regression, there is one dependent variable and one independent variable,
whereas in multiple regression, there is one dependent variable and many
independent variables. Beta values (partial regression coefficients) can be
computed in case of multiple regression, which gives an idea of the
relative impact of each independent variable on the dependent variable.
The output generates the R squared value, which is a summary statistic of

the impacts of all the independent variables taken together. It is important
to point out that regression assumes that the dependent variable is
measured on an interval, continuous scale, though an ordinal scale can also

be used. Another condition for multiple regression is to ensure normality
i.e., the distribution of all the variables should be normal.
Regression analysis in SPPS can be accessed by selecting the menu option

Analyze-> Regression-> Linear. Subsequently, the dependent variable can
be selected from the variable list. After the dependent variable is selected
from the list, the researcher selects Continue and then Options as shown in
the figure below, which shows the default options.
160 Chapter Eight
Figure 1.32: Regression

At the next stage, the method of data analysis is selected by clicking on the
Method button which is right below the Independent(s): box. The SPSS
statistics package provides several choices for doing regression analysis
though this is the one which is used most frequently.
Graphs
Graphical representation of data is a more effective medium for

representing data, not only because of its visual appeal but also for its ease
of interpretation by users. There are various ways in which data can be
represented, viz., by using bar graphs, line graphs and pie graphs. SPSS
also, like other statistical software, offers graphical presentations in
various formats like bar graphs, stacked bar graphs, pie charts and line
graphs etc. However, not all graphic formats are appropriate for all types
of data. Bar graphs and pie charts are suitable for all nominal, ordinal and
ratio data types, whereas line graphs are appropriate to show time series
data, i.e., how variables vary over a continuous period.
a) Bar graphs
To select the bar graph option, first Graphs is selected from the menu,
after which the Bar option is selected, which includes the options of
Simple, Clustered or Stacked bar graphs. After this, the independent
variable is placed on the vertical axis into the category box, while the
dependent variable is placed on the horizontal axis into the Y box.
b) Pie charts
The pie chart sub-option can be selected by going to the ‘Graph’ tab
through the menu bar. After selecting pie chart as the graphical
representation option, the variable for which the pie chart needs to be
created can be selected.
c) Line graphs
In the same way, the line graph chart option can also be selected from the
Graph tab in the menu bar option. SPSS provides the option of creating
different types of line graphs like simple line graph, multiple line graph or
drop line graph.
In the data view in SPSS, if the grey column heading is clicked on, the
whole column gets selected or highlighted. Similarly, if the grey row
number is clicked on, the whole row gets selected.
162 Chapter Eight
Writing Syntax
One of the key advantages of SPSS is that it provides both options of

performing analysis using Windows menu driven selection and by using
the programming option to write syntax codes. SPSS saves the syntax file
with the extension .sps and is the source of the commands it runs. Any file
can be opened with the extension .sps by going to File->Open Menu path,
or new syntax files can be created with File->New. Commands typed or
pasted into a syntax file.
Directly Editing Syntax Files
Syntax can also be directly edited into syntax files in much simpler terms
than menu-generated syntax. Users usually find it more efficient to
program by directly typing in the commands.
Like any other language, programming in SPSS is subject to a specific set

of rules. First and foremost, programming in SPSS is not case sensitive
and a command begins with the Command name keyword. Each command
ends with ‘.’, which is the default terminator. SPSS subcommands have
two delimiters: the ‘=’ sign distinguishes additional specifications which
follow the subcommand and the ‘/’ slash distinguishes one subcommand
from another. If the researcher wishes to type in a comment, the researcher
can use the asterisk ‘*’.
It is suggested that the syntax should be saved in syntax files for analysis
and tabulations done on a data set. This is very helpful as the analysis and
tables can be regenerated instantly multiple times using this syntax. It is
very easy to run the syntax commands. To run any syntax command, the
appropriate syntax which needs to be run in the Syntax Editor Window is
highlighted or selected. After the Syntax Editor Window is selected, the
Run button on the Syntax window toolbar is clicked or the same choice is
selected from the Run menu and the shortcut Control+R is used.
Interpreting SPSS Output
The output of the analysis or the tabulations done can be viewed in the
SPSS Output Viewer window. The SPSS Viewer window displays output,
including the warning and error messages. The left-hand side of the
window contains the log history highlighting navigation of all the outputs
run at a single output format in the opened output file.
It is important to explore both warning and error messages to debug syntax

errors. Warning messages should be carefully examined to prevent earlier
programming problems from confounding later analyses in ways that can
be frustratingly difficult to debug.
SPSS Help and Documentation
The functionality of Help and Documentation is very elaborate and helpful

in SPSS. The SPSS Help Index can be opened by selecting Help->Topics
on the menu bar. The researcher can then go through the several aspects of
Help such as the basic steps of Syntax, Command Syntax and SPSS.
To get Help in the SPSS Viewer for interpreting any output table, the
researcher right-clicks an output table and selects Results Coach in the
popup menu. Results Coach is a well-organised, context specific
introduction to interpreting output items.
8.5 Qualitative Data Analysis

Like quantitative data analysis, qualitative data analysis also aims to
describe and summarise the qualitative data generated in the form of text,
graphics or audio through interviews and observations. It envisages
revealing hidden theories and phenomena by relating them to concepts,
behaviour and ideas.
The initial preparation is like that of quantitative analysis regardless of the

formal analysis procedure and methodology used for data analysis.
Qualitative analysis pays a lot of emphasis on observation during data
collection, procedures and the initial process of analysis which starts from
the first day of data collection.
Qualitative data is usually in the form of written texts, though nowadays, a

lot of graphic and audio data is also collected to complement the textual
data. Textual data is in form of transcripts of field notes, observations, or
findings of interviews and in-depth discussions. Non-textual data, which
nowadays is analysed in volumes, could be in the form of pictures, musical
scores, tape recordings or video films.
In qualitative research, it is believed that data analysis is as good as the

transcription of the raw data. Transcription is an important stage in
qualitative data analysis and some degree of transcription is involved in
almost all qualitative research studies. Transcription is not about jotting
down or summing up what a researcher, interviewer or transcriber feels,
164 Chapter Eight
it’s all about noting down what the respondent has reported about what he
or she feels. Thus, it is important to point out that there is the potential
problem of a researcher's bias during transcription, due to which the
researcher notes only those sections that seem relevant or interesting to
them. During qualitative data analysis, the researcher should take special
care to avoid this issue. Transcription is not an act which is done in
isolation; it involves lot of processes, which ultimately contribute to
collection of quality raw data that is ready for analysis. Transcription
coupled with content analysis, takes care of qualitative data analysis and
usually includes the following steps:
x At the first stage, all relevant documents that need to be analysed in

the project are collected.
x The documents and even the segments of text within the document
should be indexed.
x Interesting passages or segments within the passages should be
coded or noted.
x Based on codes and notes in the document, the researcher should
look out for frequently occurring text and phrases which are of
analytical importance.
x The researcher should make notes and memos about the emerging
concepts, ideas and theories as part of relationship and network
building process.
x The researcher should revisit previously collected data after getting
some lead into theory building to further support the emerging
theory.
Analysis Methods
There are various distinct approaches to undertake qualitative analysis and

none of these can be defined as the right way or the wrong way of
analysing qualitative data. Some of the commonly used data analysis
techniques are listed below:
a) Typology: Typology, as the name suggests, is a classification system

which classifies patterns, themes, or other kinds of groups of data (Patton
M. , 1990). As these categories are often not mutually exclusive and
exhaustive, data analysis or the classification system may not be very
accurate.
b) Grounded Theory: Grounded theory was developed in the late 1960s. It

uses an inductive method of qualitative research and today it has become
by far the most widely used framework for analysing qualitative data.
Grounded theory has been defined as ‘theory that was derived from data,
systematically gathered and analysed through the research process. In this
method, data collection, analysis and eventual theory stand in close
relationship with one another’ (Corbin, 1998). It consists of ‘plausible
relationships’ among sets of concepts, which are directly developed from
data analysis. Thus, two central features of grounded theory are that it is
concerned with the development of theory out of data, and while the
approach is iterative, it means that data collection and analysis proceed in
tandem.
Grounded theory starts with a clear research question and it owes a lot of
its analytical skill to its constant comparative method. In this method,
concepts or categories emerging from one stage of the data analysis are
compared with concepts emerging from the next. In a way, by constantly
comparing the concepts and categories, the researcher looks for the
relationship between them, while looking out for the emerging theory.
This continues until the stage of theoretical saturation has been reached,
i.e., when there are no new significant categories or concepts which are of
interest. At the initial stage of data analysis, after familiarisation with field
notes, the researcher looks for indicators of categories in events and
behaviour, which are subsequently coded.
The coding of the notes of text passages which are of interest is done and
codes are assigned (mentally or by scribbling) to text passages, which are
then compared to find similarities and differences between the codes to
reveal concepts and categories. At the next stage, memos are developed on
the comparisons and emerging concepts and categories till the point of
saturation. Thus, eventually the emerging concepts and theories can be
combined in a logical way to find the relationship between them.
c) Framework Analysis:
Framework relationship, in some way, is like grounded theory as it also

entails coding and indexing for the preparation of charts and maps, while
looking out for the relationship. It is also different in many ways, for
unlike grounded theory, framework analysis was developed explicitly for
applied policy research and is used widely nowadays in health related
research.
The process of framework analysis starts by getting familiar with the raw
data. After this, the thematic framework needs to be developed by
166 Chapter Eight
reviewing and analysing the emerging issues. At the next stage, numeral
and text codes are assigned to specific pieces of data based on the
developed thematic framework. The data is represented pictorially under
the broad heads of the thematic framework by using charts and maps,
which later pave the way for identifying patterns, associations and
relationships. Framework analysis thus involves a few key stages, which
are mentioned below:
x Familiarisation with the data

x Identifying and developing a thematic or conceptual framework
x Reading, indexing and coding
x Charting, mapping and interpretation
d) Analytic Induction: Analytic induction is a data analysis approach in

which universal explanations of a phenomenon are sought through data
collection. This is done until no cases that are inconsistent with the
hypothetical explanation are found. At the next stage, the hypothesis is
revised to encompass all observed examples to develop a hypothesis that
accounts for all the observed cases.
It begins with a rough definition of a research question and continues to

formulate a hypothetical explanation of the research question and then
proceeds to look out for deviant cases to reformulate or confirm the
hypothesis.
e) Logical Analysis/Matrix Analysis: Logical analysis, as the name

suggests, is based on the logical reasoning process and uses flow charts,
diagrams, etc., to pictorially represent ideas, theories and concepts.
f) Event Analysis/Microanalysis: Event analysis focuses on finding precise

beginnings and endings of events by finding specific boundaries and
things that mark boundaries or events. It has found its application mostly
in the case of video data.
g) Metaphorical Analysis: Metaphorical analysis takes its cue from

various metaphors and analyses how well they fit in an observed scenario.
The process entails asking participants for metaphors and to jot down their
spontaneous answers, which can be linked with the observed phenomenon.
h) Domain Analysis: Domain analysis describes a social situation and the

cultural patterns within a domain. It emphasises describing the meanings
of the social situation to participants and interrelates the social situation
with the cultural meanings.
Domain analysis focuses on different kinds of domains, viz., folk domains,

mixed domains and analytic domains. The various steps integrating
domain analysis are mentioned below:
x Select semantic relationships

x Prepare domain analysis worksheet
x Suggest broad and narrow terms to describe semantic relationships
x Formulate questions about semantic relationships
x Repeat the process to look out for a different semantic relationship
x Make a list of all domains covered
i) Hermeneutical Analysis: Hermeneutical analysis not only looks for the

objective meaning of text, but looks for the meaning of the text in the
context of the people involved. Hermeneutical analysis looks at different
layers of interpretation of text.
j) Semiotics: Semiotics is the study of signs. It is defined as an approach

which seeks to analyse and document phenomenon to reveal deeper
meanings of that phenomenon. It explores the meaning of signs and
symbols encapsulated in a relationship or phenomenon and further
analyses how signs and symbols have an impact on the prospective users
of those signs.
k) Meta-analysis: Meta-analysis is the process of combining results of

several different studies on the same subject or question to look for
answers to the current research problem. Though it is quantitative in
nature, it involves sorting and coding techniques found in qualitative
research.
i) Content Analysis: Content analysis is an approach for the analysis of

documents and texts, which seeks to quantify content into meaningful and
predetermined (in line with the objectives of the study) categories in a
coherent way. It involves manual or automated coding of the text in the
form of documents, transcripts or even of the audio of the video media to
obtain phrases and theories for analysis.
Qualitative Research using Computers
In contrast to quantitative research, qualitative research uses a logical

structure to make an inference about the context and meaning of the
qualitative data generated. It tries to reveal the underlying theme
encapsulated in a plethora of words or images generated through data
collection. It does so by establishing theories and the relationship between
168 Chapter Eight
various identified themes to look for relationships and networking. The

strength of the question lies in their effectiveness in ensuring depth and
detail and the ability to generate new theories and reveal subtle
phenomena. Thus, it is imperative to avoid pre-judgements and
preconceived notions to capture people's perspectives and opinions and
avoid bias.
Computer-Assisted Qualitative Data Analysis
In practical situations, and even in qualitative studies nowadays, a high

volume of data needs to be analysed and that too quickly. It’s this quest for
data analysis which has led to the development of a range of specialist
qualitative packages with text retrieval and data handling capacities. In
qualitative data analysis, whether it is beneficial to use a software package
depends upon various factors like - the objectives of the study, the kind of
data that needs to be analysed and an analytic framework that the
researcher wants to adhere to. For instance, if the researcher wants to build
theory and relationships, they should use ATLAS/TI and N Vivo series
software programs, which use advanced search and query functions to
retrieve data.
Software packages offering qualitative analysis facility depend upon

procedures for coding the text. Indexing and coding (marking or attaching
tags to relevant segments of text) form an integral part of any qualitative
research process. These coded texts are then analysed to reveal the hidden
or the emerging theory. Qualitative computer software programs facilitate
the attachment of these codes to the strips of data and allow the researcher
to retrieve all those instances in the data that share a code by formulating a
query or search function.
Popular Software for Data Analysis
Qualitative data analysis using statistical software packages is gaining

ground and is now used widely by researchers, not only because of the
ease with which it can be used but also because it saves time. Now there is
a whole range of statistical softwares that is available for data analysis.
Some of the popular qualitative softwares available in the market are
described below:
a) Ethnography
Ethnography uses segment as the basic unit to retrieve the text from
documents and can be described as a text retrieval statistical software.
Segments are then coded in the form of twelve code words, which can be
nested or overlapped upto seven levels deep as a sort of query function to
retrieve text. It is important to point out that the search results depend on
the query function in the form of nesting using codes.
b) HyperQual
HyperQual is an integrated software, which encompasses all features for

data entry, storage and retrieval. It thus helps in entering data collected
from interviews, observations and documents for further analysis.
c) Hypersoft
Hypersoft provides an integrated facility for data entry and data retrieval
for theory building. This software also provides the options of indexing,
searching and analysing textual data. Hypersoft does so by providing
facilities for annotating, categorising, and coding of raw data.
e) Textbase Alpha
Textbase Alpha provides the facility for coding of both narrative text as
well as data which has internal structure. Textbase Alpha also provides the
facility for assembling coded segments, which can be represented by
frequency counts and data matrix output. Textbase Beta, the new and
modified version of Textbase Alpha provides the facility to code longer
portions of lines.
f) SONAR
SONAR is basically a text retrieval software which also provides the

facility for coding. While using SONAR, it is not required to extensively
prepare the text files, while the data can be retrieved instantly based on the
coded segments. Data can be edited by putting comments or an index of
comments. SONAR also provides the facility for conditional searches
based on Boolean operators.
g) ATLAS/ti
ATLAS/ti is perhaps the most powerful software that is available today

and has an impressive user interface. ATLAS/ti works on the same logic
for searching, coding and indexing as NUDIST, another qualitative data
analysis system. However, in the case of ATLAS/ti, the concept of node
and hierarchical logical system which is the key to the analysis process in
NUDIST is not followed. Rather, it focuses on the inter-code relationships
170 Chapter Eight
and theory building, rather than straight code and retrieve. Codes and
super codes which represent a series of statements expressing textual
relationships are used to build theory, concept and relationship through
operators.
There are various qualitative data analysis softwares available, however,

as stated above, it should be cautiously decided on which software should
be used in a project depending on the data that is to be analysed and the
analysis theory that is to be followed.
ATLAS/ti is probably the most powerful qualitative analysis tool for text,
video and audio data. It can work with text, video and audio files of
various formats. Though it follows the analysis paradigm of Grounded
Theory and the qualitative content analysis, it is used universally by
qualitative researchers from all realms of analysis theory. It helps to reveal
the complex phenomena hidden in qualitative data in an exploratory way.
Qualitative data analysis using ATLAS/ti usually involves the following

sequence of steps for systematic data analysis:
Step I: At first, a Hermeneutic Unit (HU), which signifies a project or an

idea container, is created to contain all qualitative data including findings,
codes, memos, and structures under a single name. ATLAS/ti is started by
clicking on the software icon. It opens the main window known as
Hermeneutic Unit (HU) Editor, which is the main work space area. It is
the main editing tool which provides access to all other tools like the title
bar, main menu, tool bar, main tool bar and the primary document toolbar.
HU contains all the primary documents, which can be texts, graphics, or

sound files for a given project. Further, within each primary document, the
analyst selects specific quotations and assigns codes. Codes can be
grouped into families and quotations and coded objects can be further
documented in memos or graphically displayed as networks of
relationships.
Step II: After creating a HU, all the data files relevant to the project,
including the primary documents in the form of text, graphics and audio is
associated with the created project or the HU.
Step III: After associating all relevant data files or Primary Documents
with the HU, the relevant text passages or pieces of special interest are
selected and further assigned codes or memos. This step transforms all
relevant documents in codes and memos, which are used in later stages of
query indexing or theory building.
Figure 1.33: ATLAS/ti Main Window
Step IV: After the initial stages of assigning codes and memos, the coded
data segments are reviewed thoroughly to ascertain whether it is necessary
to associate more files with the HU.
Step V: After assigning codes and memos to all relevant data files
associated with the project, all the primary documents, codes, and memos
are organised using ‘Families’. This is used later in developing theory.
Step VI: After organising the primary documents, codes and memos, as the
next step, the semantics, conceptual or terminological networks are built
from the codes and memos which have been created and organised. These
conceptual networks along with codes, memos, super codes and families
help in crystallisation of the emerging theory.
Step VII: As the last step, based on the emerging theory, a report is
prepared. In case any further analysis is required, the data is exported for
further analysis using the statistical software.
172 Chapter Eight
h) NVivo
The NVivo series software developed by QSR International Corporation is

also one of the qualitative analysis softwares widely used by researchers. It
is based on the concept of building a logical hierarchical system of Nodes
where each Node is logically related to other Nodes, which it does by
flagging and text search to construct a possibly large and highly structured
hierarchical database indexing into the documents to be analysed. Node
search uses seventeen operators including Boolean, proximity and other
operators to generate new nodes to build hierarchical indexing categories
encouraging generation of new ideas.
After the NVivo program is started by clicking in the software icon, one of
the three options provided can be selected viz., Run a Tutorial, Start a New
Project or Open an Existing Project (Figure 8.31). To start a new project,
the dialog box of ‘New Project’ is checked.
The NVivo menu provides the option to start a New Project, or to Open an
Existing Project.

Starting a New Project
A project in NVivo can be said to be like a big container which contains

all unstructured data, analysis and reporting tools relevant to a specific
project.
A project in NVivo encapsulates the following broad themes:
x All Documents: 1) Internal documents such as transcripts of focus

group discussions (FGDs) and field/project notes 2) External
documents such as books, audios and videos
x Relevant Nodes which represent the emerging themes, cases or
relationships
x Attributes (such as age or gender) assigned to cases and Memos
which represent the researcher's notes
x Text Queries to ascertain patterns and pursuing ideas
x Models which can visually demonstrate and explore themes
The NVivo workspace is categorised into three different spaces viz.,

Navigation View, List View and Details View. The first upper side pane of
Navigation View provides information about the all project related
documents.
174 Chapter Eight
Reports
Using the NVivo program, it is possible to produce a range of different

reports such as:
Project summary – As the name suggests, this is a summary of the entire

project, e.g., the number of documents included, the number of attributes,
cases, memos etc.
Node summary –This provides a summary of each node, e.g., the number
of words coded for this node, the number of sources and cases.
Coding summary –This summarises the allocation of words to each node

as a percentage of the overall word count of the document.
To summarise, the following activities can be done using the NVivo

software:
x Associating all documents, viz., 1) Internal such as transcripts of

FGDs and field/project notes 2) External documents such as books,
audios and videos
x Creating relevant Nodes which represent the emerging themes,
cases or relationships
x Assigning Attributes (such as age or gender) to cases and Memos
which represents the researcher’s notes
x Text Queries to ascertain patterns and pursuing ideas
x Models which can visually demonstrate and explore themes
x Preparation of reports based on emerging findings.
CHAPTER NINE
EXECUTING MONITORING EVALUATION

AND LEARNING
The final chapter of this manual presents the operational dimensions of

executing the Monitoring, Evaluation and Learning (MLE) component of
developmental projects and programmes. The MLE function of a
project/programme involves its planning, management, execution and
finally using its results. Executing MLE involves several logical and
sequential steps starting from finalising the theory of change, aligning it to
project goals and objectives, defining clear hypotheses, selecting the M&E
design, selecting a sampling scheme, implementing the M&E framework
and analysis, and finally communicating the results through a well-defined
learning plan. The chapter also discusses the quality assurance plan along
with analysis and representation of the M&E result.
At the end of this chapter the researcher will be able to:
" List the steps involved in executing MLE for development projects
or programmes
" Understand the operational dimensions involved in executing MLE
for developmental projects or programmes
9.1 Finalising the Theory of Change

The first and foremost answer that an evaluator seeks is to the question,
'What needs to be evaluated?' Different stakeholders, including funders
and implementers, have different views on what needs to be evaluated.
Moreover, sometimes even within a project, different people have
different evaluation priorities. This holds true in real world situations also,
primarily because different stakeholders have different priorities and look
at evaluation from different vantage points. To have a common view point
on evaluation that all stakeholders are agreeable with, it is essential to
have a theory of change which presents the evaluation objective and its
reasoning equally to all stakeholders.
176 Chapter Nine
The theory of change (ToC) provides the blueprint or pathway by mapping

out long-term goals and linking them to existing preconditions, besides
specifying the causal link for each precondition. It also lists the basic
assumptions about achieving a specific set of outcomes underpinning the
importance of the context, which in turn helps in agreeing with the logical
narrative for the intervention. In some cases, the programme may not have
a ToC or even if the ToC exists, there is no agreement on the ToC among
all stakeholders. Thus, in both cases, it is important for an evaluator to
have detailed discussions with all stakeholders to develop/align the ToC
and to attempt to articulate the reasons for the sequence of results to occur.
It is also important to point out that in real world situations, considering

the dynamic nature of the interventions, the ToC is built-in for the
underlying assumptions, risks, mechanisms, and external factors that may
affect the ability of the interventions to achieve its intended results. Hence,
the ToC developed at the start of the intervention should be flexible in
nature with scope of constant evolution with the changing environment.
An assessment of the ToC during project implementation helps not only in

the course correction of the project implementation which may be
necessary due to change in ground realities and assumptions, but it also
helps in redefining the evaluation objective and design thereafter.
9.2 Finalising the Logic Model

Different evaluators have different viewpoints on whether to first finalise
the ToC or the Logic Model. The ToC provides a roadmap or a blue print
of achieving an outcome, whereas the Logic Model provides intervention
inputs and activities, the outputs they will produce, and the connections
between these outputs and the desired outcomes. Thus, while the ToC
summarises work at the strategic level defining causality, the Logic Model
unpacks the implementation paradigm of the ToC by defining a specific
set of interventions.
After building on an agreed ToC, the evaluator works on the Logic Model
with the programme team and in case the programme does not have a clear
Logic Model, it is essential that the evaluator develops one in
collaboration with the relevant stakeholders. While working on the logic
model, it is important to have an agreement about its specific components,
viz., the inputs or the resources required to execute the intervention,
activities that need to be carried out, outputs, outcomes and the impact to
be achieved.
Executing Monitoring Evaluation and Learning 177
Further, as explained in the previous chapters, the Logic Model must also
outline the specific indicators for measuring the aspects of each
component.
9.3 Finalising the Purpose of Evaluation

At the next stage, building on the agreed ToC and the Logic Model,
evaluators work in collaboration with stakeholders to define the purpose of
evaluation. To do so, evaluators re-confirm about the need of evaluation to
define the evaluation’s purpose and scope. It is important to probe whether
the focus is on evaluating the project comprehensively or whether the
focus is on a specific component? If the focus is on a specific component,
the reason for the same is investigated and information regarding the
components of the programme that should be evaluated first is solicited.
Based on the purpose of the evaluation, the different terminologies that are
used to describe evaluations are as follows:
x Comprehensive Evaluation
x Formative Evaluation
x Process Evaluation
x Outcome Evaluation/ Impact Evaluation
In an ideal situation, one should go in for comprehensive evaluation of the

project/programme. Comprehensive Evaluation captures the change in the
intermediate output, process and outcome level indicators to assess the
attribution questions and links the outcome level change to the change in
the process/output level to demystify the story of contribution. Comprehensive
evaluation further ascertains the cost-effectiveness, scalability and
sustainability of the intervention.
The Cost-Effectiveness Analysis (CEA) compares the costs in monetary

units with the outcomes measured in quantitative terms. The outcomes are
measured through rigorous evaluation and at the same time, the study also
captures both financial and non-financial relevant costs incurred for the
implementation. This is also extended to include the benefits to compute
the Cost Benefit Analysis (CBA). Kalpan-Meier (KM) or product limit
method is used to assess changes in outcome using STATA or specialised
software such as TreeAge in mutual consultation with the key
stakeholders.
178 Chapter Nine
Comprehensive evaluation is not complete without commenting on the

sustainability and scalability aspect of the intervention. The evaluation
should also focus on finding out whether the programme is sustainable and
scalable. The CBA also helps in providing critical information required for
answering questions regarding sustainability and scalability.
While comprehensive evaluation summarises the intervention result,

formative evaluation is an assessment of efforts usually done at the start of
the evaluation and prior to their completion for improving the efforts. Its
core objective is to provide feedback regarding the efficacy of the
intervention and its progress.
Integrating Mix Method Design
The objective of comprehensive evaluation is not only to answer the

attribution question but also the contribution aspect of the programme. It is
essential that as part of evaluation, mix method approaches are used. Mix
method approaches help in understanding the storyline of the effect of the
intervention. Mix method approaches can help us in understanding the
change in a specific context and culture.
One can use participative approaches or traditional qualitative methods

such as FGDs and IDIs. Additionally, to map the process of change, one
can build on qualitative approaches such as the success case method and
MSC (Dart, 2003) stories method to collect stories of change to understand
and analyse the process of change. As explained in the previous chapters,
these stories are of changes both positive and negative, and attempt to
document the outliers. Analyses of these stories across domains help the
researcher to understand what and why some things work while others do
not.
Process monitoring and observation data can also be very useful in

mapping the process of change and in understanding the context in which
they are operating. This provides answers to how and why the programme
impacts were or were not achieved. Additionally, through this process, the
practitioner aims to identify the facilitating and restraining factors and
synthesise the lessons learnt for successful replication. This also ensures
that feed-forward is provided for implementation across different
interventions for improving systems and processes.
Outcome Evaluation and Impact Evaluation usually follow a similar

trajectory of change. While outcome evaluation attributes the change in
the outcome because of project intervention, impact evaluation looks at the
long-term outcomes or impact that might have resulted from that

programme. For example, in the case of specific health interventions,
while outcome evaluation primarily captures changes at the ANC level,
impact evaluation ascertains change in mortality.
Process Evaluation ascertains whether the programme activities have been

implemented as intended or planned. Process evaluation plays a critical
role in evaluating how a specific set of processes as enshrined in the
project document have been implemented at the field level and whether it
has further resulted in change in the outcome. Process evaluation is usually
done at the end of the project but a concurrent process monitoring exercise
can also be carried out to provide specific feedback about the course
correction. Process evaluation helps in identifying processes and how or
whether the processes have resulted in the desired outcome.
Besides ascertaining the process of change, it is also critical to identify

success stories and delineate the lessons learnt for adoption, adaptation
and replication. This is done using documentation and exploration
techniques. Documentation is focused towards a practice or the process. In
the implementation context, process innovations and institutional
innovations are documented. Apropos the content of documentation, the
context or background of the issue is explained first as it is important to
explain the context of the problem before discussing how the problem is to
be addressed. Documentation should also include detailing the thought
process behind the development of the practice and then the actual
process. Also, the favourable factors that have helped in the establishment
of the practice and the restraining factors that have been overcome are
detailed.
Based on the above tasks, the lessons learnt and best practices are
delineated so that they can further be adopted, adapted and replicated
across other projects for facilitating their efficient and effective
implementation.
In addition to finalising the purpose of evaluation, it is also important to

finalise the scope of the evaluation, which means, it is necessary to
determine what in terms of resources, i.e., budget, staff, time etc., are
available for the evaluation. It is also necessary to look at the long term
learning and dissemination plan and whether the same has been factored in
as part of the evaluation.
180 Chapter Nine
9.4 Finalising the Evaluation Question

After finalising the purpose of the evaluation, the next step is to finalise
the evaluation questions that are highly relevant to the aspect(s) being
evaluated. The evaluation questions are:
Result Questions
The result questions focus on assessing the impact of interventions at

different levels. At each level, the result questions try to assess if there has
been any improvement in the coverage and quality of intervention.
Further, depending on the hierarchy of the result, they can be further
defined as impact or outcome level questions.
Process Questions
The process questions are also related to the inter-linkage and sustainability
questions. While the process questions at each level try to draw causal
inferences about the intervention, the inter-linkage questions attempt to
analyse the outcome of interventions at one level as well as at the next
level of implementation. The questions framed under sustainability are
cross-cutting in nature and seek answers to issues like the viability of
interventions, the possibility of taking the interventions to scale and the
scope of internalisation.
9.5 Finalising the Key Areas of Enquiry

Building on the broad question of evaluation, it is important to list the
broader areas of inquiry across different components for evaluation.
Though only the key questions are listed as part of the evaluation
questions, it is important to ensure that all areas of enquiry are captured
comprehensively including the intervention and confounding variables.
9.6 Finalising Indicators

After finalising the research questions, at the next stage, the evaluator is
enabled with a better understanding and appreciation of the indicators
mentioned in the Logic Model. During this process, each of the indicators
are critically reviewed with respect to its harmonisation with the Logic
Model, practical considerations of collecting data on the indicators and the
nature of the indicators, i.e., whether they are Specific, Measurable,
Achievable, Relevant and Time bound as explored in detail in Chapter 3.
This process leads to a better presentation of the logframe for stakeholder

consultations. This set of indicators is presented to the stakeholders and
through a consultative process, the final list of indicators is established.
Table 1.14: Review of the Indicators with Respect to the Logic Model
and Theory of Change
As part of the MEL planning process, it is essential for the MEL team to
review the indicators regarding the various stages of Logic Model and the
ToC. It should be assessed if they are appropriate to each stage and
whether they fulfil the SMART characteristics which each indicator
should ideally possess.
In addition to the tracking of the impact trajectory, it is also essential to

have an estimated timeline for interventions to have a measurable impact.
Absence of a timeline makes it difficult to have realistic expectations of
the outcomes and makes it hard to differentiate between a failed outcome
and an outcome that has not occurred as yet. Hence, an evaluation
framework and impact timeline should be developed to assess the impact
of policy interventions.
Definitions of Key Indicators
As a part of the evaluation exercise, another step is to finalise the

definitions of the key indicators for analysis and reporting. Indicators have
been defined in Chapter 3 and it is crucial that the definitions of the
indicators is specified beforehand to avoid different interpretations of the
meaning of an indicator or a sub-indicator. For instance, an indicator that
calculates the percentage of women who have received complete ANC is
defined by the number of ANC check-ups, components of the ANC check-
up like height, weight, abdomen check-up etc., the number of Tetanus
Toxoid (TT) injections and the number of Iron Folic Acid (IFA) tablets
that have been consumed to compute the percentage of women who have
182 Chapter Nine
received complete ANC. It is important to reach a concurrence on the

definitions of such key indicators for computing the coverage level
estimates from the data and then validating the data.
The method of analysis of key indicators is also decided keeping in mind

the definitions of key indicators.
9.7 Finalising Research Instruments

After finalising the indicators, the next task is to develop and finalise the
research instruments. It is important that the research instrument, tool or
questionnaire is developed in a systematic way. Not following appropriate
and systematic procedures in questionnaire development, testing, and
evaluation may undermine the quality and utilisation of data (Esposito,
2002).
Translation of Study Tools
After the research tools are developed, they are translated into local
languages focusing on three basic aspects: semantic equivalence, conceptual
equivalence, and normative equivalence of items. For verification of the
translation quality, the translation/back translation method is used, having
independent translators translate from one language to another, and then
back again, to see if the original and re-translated item remains the same.
9.8 Pilot Testing

After developing the research tools, it is very important to thoroughly test
them in order to ascertain their suitability in actual field conditions. Field
testing of the developed tools is conducted in the non-sampled areas. Pilot
testing is not only critical for identifying the issues in the questionnaire but
also helpful in removing ambiguities and other sources of bias and error.
Both pre-field and field type of pre-testing should be done. In field type of
pre-testing, the study tools are tested under field conditions by
professionals and investigators. The team members should be involved in
the pre-testing exercise for at least two days. After the field type of pre-
testing sessions, a debriefing session is conducted with the field team
members and necessary suggestions are incorporated in the tools. It is
important that not only the questionnaire but the entire survey protocol is
pilot tested.
One of the key objectives of pilot testing is to ascertain the validity to

analyse systematic error in measurement (Norland, 1990). Evaluators should
try to assess all kinds of validity errors, viz., content, construct, criterion, and
face. One of the key aspects of this assessment is to ensure that the
questionnaire comprehensively captures all indicators and information with
respect to the finalised evaluation objective. Additionally, the evaluator
should also assess whether the questionnaire is measuring what it intends
to measure. Besides this, the evaluator should assess whether the
questionnaire adequately represents the content.
In addition to validity, it is also important to establish the reliability of the

questionnaires. Practitioners can measure the accuracy of the research
instrument using the reliability assessment test (Norland, 1990). They can
use different types of measures such as reliability types (test-retest, split
half, alternate form, internal consistency) depending on the nature of the
data. Pilot testing is the key to conduct a reliability assessment test. In
terms of the number of interviews, pilot testing should be conducted on
approximately 20-30 subjects. Reliability can be measured by reviewing
the initial data and by calculating the reliability coefficient (alpha). If the
reliability coefficient is .70 or more, the research tool can be considered
reliable.
There are other important questions that need to be answered as part of the
pilot test regarding the research tool, viz.,
1. Do the questions convey the meaning they are supposed to

communicate and whether the language used is appropriate with
respect to the local context?
2. Do all respondents interpret the question in the same way?
3. Are the required instructions included as part of the question?
Additionally, it is also important to assess the time taken to complete the

questionnaire. It is important to assess this to make sure the research tool
does not become too long and does not exhaust the respondent, for if tired,
the respondent will not be able to provide quality information. Thus, it is
important to ascertain at what level respondents feel fatigued and structure
the length of the questionnaire accordingly.
Revising the Research Instruments
Based on the results of pilot testing, it is essential to revise the research

instruments. Sometimes, the pilot testing findings may call for major
184 Chapter Nine
changes. In that case, it is suggested that another round of pilot testing be

conducted to finalise the research instruments.
The field-tested tools are finalised in consultation with the client and serve
the following purposes:
x Finalisation of the areas of inquiry: To enable the team to touch

base with all the areas of inquiry enlisted in the ToR and give the
communities an opportunity to add-on new areas of inquiry to the
study.
x Finalisation of the activity schedule: This helps in getting the
timeline right for field-level processes.
9.9 Preparation for Electronic Data Collection

Parallel to finalising the research instruments, it is important to also
finalise the data collection protocol. The data collection protocol can be
broadly classified into two broad categories of Paper-Assisted Personal
Interviewing (PAPI) and Computer-Assisted Personal Interviewing
(CAPI). In addition, in specific cases where resources are limited,
Computer-Assisted Telephone Interviewing (CATI) and Computer-
Assisted Web Interviewing (CAWI)/Online Data collection approaches are
also used.
Regarding the questionnaire’s design in both CAPI and PAPI, a standard

template should be developed for entering the data. The only difference is
that in the case of CAPI, the template or data entry design is the data entry
programme. However, in the case of PAPI, one has the choice of making
the data entry template later as interviews are conducted using paper based
research instruments.
In the case of CAPI, all enumerators require a personal digital assistant

(PDA) for conducting the interviews, which turns out to be a substantial
cost. It is suggested that over time, research organisations should build an
inventory of PDAs so that they can be used for different projects. In the
case of pen and paper based interviewing, PDAs are not required. Thus,
only a few PDA are required for data entry for which PDAs with minimal
configuration can also suffice. Most institutions outsource the data entry
task for efficiency. One of the key challenges regarding the hardware is
the rapidly changing technology. As the wear and tear of PDAs while the
practitioners are working in the field is inevitable, in most cases,
mobiles/tablets used for an evaluation do not last for more than two to
three projects and hence, it is prudent to budget and plan accordingly.
One great advantage of CAPI is that the enumerators need to spend less
time thinking about what question to ask next as the skips are automated in
the programme. Thus, the CAPI interview has a more natural
conversational flow than a PAPI interview and it also tends to move faster,
reducing the total time it takes to administer the questionnaire. Further, it
is observed that respondents (and enumerators) are less likely to become
tired and/or frustrated with the interview.
Another critical benefit in the case of CAPI is the timeliness of data. One
can look and analyse the data in almost real time and there is no time lag
between data collection, cleaning and analysis. This saves lot of time,
which may be to the tune of one to three months depending on the sample
size. More importantly, timely availability of quality data helps to speed
up decision making. It also helps in building ownership, as all stakeholders
can look at the data on real time basis.
Overall, the key differentiating factor in favour of CAPI based survey

administration is the assurance of data quality, which is worth more than
the expense incurred on hardware and the time taken to build the right
kind of software for CAPI. Different kind of checks such as validity
checks, range checks and logical checks can be incorporated in the CAPI
programme. These checks function on a real-time basis and provide cues
to the enumerators to probe further and restrict the possibility of entering
incorrect or illogical data. For example, if a value is out of range, the
programme shows an error in a prompt window.
In developing countries, most of the evaluation organisations have already

shifted to CAPI and with respect to long term investment, CAPI will pay
off as it ensures good-quality data. Institutions are using state of the art
technological interfaces viz., PDAs, laptops etc., as part of Computer
Assisted Survey Interviewing (CASI). Census and Survey Processing
System (CSPro) programmes which are custom designed are used on
mobile phones or tablets or laptops. Some institutions also use survey
platforms like Survey CTO for collecting data electronically.
Once the survey is completed, the complete (survey) data is electronically

transferred back to the database server. Automated reports regarding the
status of data collection can be generated from the database on real time
basis. By reviewing the data, issues in the data quality can be identified
186 Chapter Nine
and feedback can be shared with field managers, supervisors and

enumerators on real time basis.
The schematic diagram presented below describes the processes used in

data collection:
Figure 1.36: Data Collection Process using PDAs
9.10 Recruitment of the Field Team

Recruitment of enumerators, supervisors and field managers in the survey
team is based on their educational qualifications and experience in
conducting similar surveys.
The quality of the data depends on the skill of the enumerators recruited,
trained and managed during the data collection exercise. Enumerators are
the fulcrum of the data collection exercise and, hence, they are also the
custodians of the data. The field staff consisting of field enumerators and
supervisors are primarily responsible for the data collection effort.
Thus, it is essential that the enumerators selected for the data collection
exercise possess the requisite skills. The first and foremost skillset
required is good communication or people skills as the key job of
enumerators is to communicate the purpose of the survey, solicit consent,
highlight the importance of the respondent’s participation and then solicit
responses to research questions listed in the research instrument. It is
always helpful to have enumerators who have either basic understanding
of thematic areas wherein they are supposed to collect data or have prior
experience of collecting data on similar thematic areas viz., public health,
nutrition, livelihood etc. For instance, the data collection exercises which
involve technical knowledge as in the case of collection of data on the
quality of care at the health facility, it is in the interest of the programme
to have enumerators who are trained in nursing. Similarly, in case the
survey is being done using PDAs, it needs to be assessed if the
investigators are comfortable using the same. The composition of the
survey team i.e., the number or males or females required is as per the
nature of the survey. For example, in the case of surveys related to
reproductive health and family planning to be administered to women, it is
advisable to have female investigators.
The recruitment process of enumerators should be standardised and

according to the evaluation objective. In most cases, enumerators are
freelancers and not working on a contract basis with any specific agency.
Depending on the availability of work and established work-relations, the
majority of the enumerators prefer working with certain organisations.
Thus, recruitment for any particular survey becomes easy as the
established organisations usually have a pool of enumerators to choose
from, based on the nature and the location of the survey. In the case of
large surveys or surveys of a different kind, recruitment is also facilitated
through advertisements. Subsequently, interviews or written tests or both
are conducted to check the suitability for a specific evaluation exercise. In
both situations, whether the enumerator is recruited from an existing pool
of enumerators or a new set of enumerators is recruited, it is important that
recruitments are done based on competence i.e., skills required to conduct
the survey effectively.
188 Chapter Nine
Another key skill set required for enumerators is the ability to grasp and
follow detailed instructions in addition to literacy, basic numeracy, and the
ability to absorb training. Language becomes one of the key criteria for the
selection of an enumerator and enumerators should be fluent in the local
language(s) which is native to respondents. In specific cases, an additional
language may come handy, and more so in the case of multi-lingual
surveys.
A few personal attributes are also important in the selection of enumerators.

Enumerators should be able to approach strangers and build rapport with
them, which is a key factor in soliciting consent for the interview.
Additionally, enumerators should also be diligent in adhering to the
protocol and following instructions. They should be responsible and
should act as custodians of data by showing ownership in ensuring quality
data.
As most of the enumerators are freelancers, it is important to ascertain that

they are available for the entire duration of the training and the data
collection effort before their recruitment is finalised.
Generally, as a contingency plan to cater to drop outs and rejection of

enumerators who are not found to be adept enough to be part of the data
collection, it is advisable to have an additional 15 to 20 per cent of
enumerators and supervisors in the training programme so that enough
team members are available to conduct the survey at the end of the
training.
9.11 Preparation and Implementation of Data Collection

Careful and thoughtful field work enable teams to collect data
systematically. The following steps ensure quality assurance of the data.
9.11.1 Development of the field manual

Development of the field guide or manual for data collection is essential to
ensure smooth functioning and standardisation of field operations. The
training manual serves as a guide for facilitators, enumerators and
supervisors for managing the data collection effectively. The manual
should be in the local language to serve as a ready reference during the
survey for all research and field staff. It broadly consists of:
x Background and the research objective

x Research design and methodology
x Role and task of research and field staff
x Quality control mechanism and protocols which need to be
followed
x Brief description of the key questions and explanation of the key
concepts which are required for administering the questionnaire.
9.11.2 Training of field survey teams

Different enumerators have different skill sets and they may also have a
different understanding of the study. Thus, to minimise interviewer bias
and to ensure consistency and accuracy in the data collection exercise, it is
important to train the data enumerators well. The objectives of the training
exercise are to train enumerators on how to use the survey instrument(s),
how to conduct interviews, and how to perform other data collection
activities in a coherent and consistent manner according to the evaluation
objective. This will help in minimising non-sampling errors and response
errors substantially. Training of the survey team to deliberate on survey
objectives, survey tools, sampling design, as well as on expected data
quality are some of the critical steps that will ensure data quality. It is
suggested that all the researchers and field staff go through the field
manual carefully before the training session starts.
All members of the data collection team are continuously monitored

during the training and based on the internal assessment of their
performance, the teams are finalised. Those who are not up to the standard
or are not performing well should not be included in the team in order to
ensure data quality.
The training should be facilitated by a team of professionals. The field

team should also be included in the exercise of identifying the errors in the
questionnaire and the plausible questions and responses on which
automated checks are applied to minimise data entry errors.
Enumerator training should take place in a systematic way and in a

structured learning environment. A training agenda for the training
exercise should be made and shared with all stakeholders. All stakeholders
should be in agreement regarding the duration, design and method
employed for the training.
190 Chapter Nine
The training should have a proper mix of classroom sessions and field
practice. Classroom training should focus on training the field team on the
structured interview schedule and related fundamental concepts. Besides
lectures, the training should also include mock exercises and role play
demonstration. Field practice sessions should be scheduled in non-sampled
areas, once enumerators are trained adequately on research instruments.
Debriefing sessions to reiterate and assess whether investigators have
understood the questions correctly should be scheduled thereafter.
The training agenda should have sufficient time for classroom training,
simulation and mock exercise, field training and then feedback. The
training methods selected should include active learning opportunities for
the trainee enumerators whenever possible. The training method should
provide enumerators with an opportunity to practice, apply, or respond. In
case PDAs are being used for conducting the surveys, technical training is
also important and sufficient time should be given to train enumerators on
the usage of PDAs.
The lead trainer should discuss the survey instrument or questionnaire in

detail with the data collectors to point out specific instructions. The
training should include an example of a completed instrument or an
interview transcript for data collectors. There should be clear, written
instructions and/or a script for the data collectors to follow while
interviewing respondents.
Sufficient time should be allocated on how interviewers should introduce

themselves as it has a strong influence on the respondent’s reaction and
willingness to cooperate. One of the key and critical aspects of collecting
data is the ability to introduce oneself well and to establish rapport with
the respondent as soon as possible. It is expected that enumerators should
be able to communicate the purpose of the survey, what information is
being sought, how the information will be used and to be able to assure
respondents about the confidentiality of the information provided. It is
important to have interviewers practice in role plays or mock interviews
during the training sessions to help the enumerators to prepare for difficult
or unusual situations which may arise. It is also important to ensure that
ethics and protocols are followed at the field level. Following ethical
guidelines such as, providing study related information, soliciting consent
and ensuring confidentiality of data not only ensures transparency but it
also ensures that the data collected is of good quality and the non-response
rate is minimum.
Training using Visual Aids
Training of enumerators is a critical aspect of data collection and the

effective use of training aids can enhance the quality of training. Though it
is important to start the training on a lecture mode, the monotony of this
mode of training can set in quite quickly and after a point, enumerators
may not pay sufficient attention. In such situations, audio visual aids along
with group or simulation exercises can be a more effective way of
imparting training. Even during lectures, it is suggested that presentations
are made with interesting graphics and videos. Flip charts can be used to
make the training sessions more participatory. Simulation exercises along
with mock interviews and role plays are also very helpful in making the
training sessions participatory.
Testing and Scoring One way of ensuring the quality of the field team
and to ensure that they are ready to go to the field for data collection is to
rate and score them. One can use a mock survey exercise where one
enumerator chooses another enumerator as a respondent and completes a
mock questionnaire. Each investigator can be scored based on their
performance.
Finalising the Team Size
The team size, i.e., the number of enumerators required, depends on the
productivity of the enumerators, which in turn depends on the length and
complexity of the research instruments which determines the time allowed
for the survey to be conducted. In addition to this, geographical location or
terrain plays a part in determining the team size. Thus, the number of
enumerators required may vary from survey to survey.
Dry Run and Initial Feedback
The dry run is another activity which is the key to ensure data quality.
During the dry run, the field team is informed that their field work has
started, though in reality, it is like a practice session. The dry run helps the
research team to understand how the field team will perform in real world
situations. The research team and field managers can take stock of the
situation based on the dry run.
192 Chapter Nine
Logistic Plan including Transportation
Transportation is a key aspect of logistics and can be a critical factor in

ensuring data quality. Thus, the team size should be planned in a way that
vehicles may be hired to build efficiency.
Generating a Sample Frame
In many cases, to generate a robust sampling frame at the primary

sampling unit level, HH listing or mapping exercises are conducted.
Enumerators or address listers can begin at the northwest corner of the
segment and record all the residential addresses they encounter while
travelling in a clockwise direction around the segment. They can proceed
around a segment, always turning right when the opportunity presents
itself, and in this way, add the address of each structure to the list. To
ensure that house-listing is completed just before the survey begins (to
maintain integrity of the reference period of cases), it is proposed that
house-listing should be conducted at least one day before the actual
survey.
9.11.3 Roles and responsibilities of the data collection team

It is very important to make a list of the roles and responsibilities of
enumerators, supervisors and field managers or executives. These roles
and responsibilities should be explained to the field team in detail and
included in the training manual for their reference. The roles and
responsibilities of the field team members at various levels are listed
below:
Role and Responsibilities of the Enumerator
Enumerators have a central role in the study since they are the ones who
collect information from respondents. Thus, the success of the survey
depends on the quality of each interviewer’s work. Successful
interviewing is an art that should not be treated as a mechanical process.
The skill of interviewing develops with practice but there are certain basic
principles which should be followed by every successful interviewer. One
of the key tasks of an enumerator is building rapport with the respondent.
The enumerator should keep some of these points in mind while
conducting an interview:
Make a good first impression: When approaching the respondent for the
first time, the investigator should do their best to make the respondent feel
at ease. With a few well-chosen words, the respondent can be put in the
right frame of mind for the interview. The interview should be started with
a smile and salutation such as ‘namaste’ followed by the introduction of
the investigator.
The approach should always be positive: An apologetic approach should

not be adopted by using opening words such as, ‘Are you too busy?’,
‘Could you spare a few minutes?’ or ‘Would you mind answering some
questions?’ Such questions invite refusal even before the process starts.
Rather, the investigator should engage with the respondent by saying, ‘I'd
like to ask you a few questions’ or ‘I'd like to talk with you for a few
moments.’
Stress the confidentiality of responses: The investigator should explain to

the respondent that the information that will be collected will remain
confidential, and that no individual names will be divulged at any stage. It
is important to reassure the respondent about the confidentiality of the
survey, especially if the respondent is hesitant about responding to the
interview. Also, the interviewer should never mention other interviews or
show completed questionnaires to other interviewers or supervisors in
front of a respondent or any other person.
Answer the respondent’s questions frankly: Before agreeing to be

interviewed, the respondent may ask the interviewer some questions about
the survey or how the respondent got selected to be interviewed. The
interviewer should be honest and answer any queries the respondent may
have in a pleasant manner and put the respondent at ease.
Interview the respondent alone: The presence of a third person during an

interview is a hindrance in getting frank answers from a respondent.
Therefore, it is very important that individual interviews are conducted
privately and that all questions are answered by the respondents
themselves. The investigator should keep in mind the points mentioned
below while conducting the interview:
x Be neutral throughout the interview

x Never suggest answers to the respondent
x Not change the wording or sequence of the questions
x Handle hesitant respondents tactfully
x Not raise the respondent's expectations to believe that they would
be getting any direct benefits by participating in the interview
x Not rush through the interview
194 Chapter Nine
The interviewer's responsibilities include:
x Locating the HHs in the sample as allocated by the team

supervisor.
x Conducting interviews with the identified HH member in each of
the HHs assigned to the investigator by their supervisor.
x Administering the complete questionnaire in the required logical
order.
x Giving the respondent information about the survey and taking
their consent before starting the interview.
x Checking completed interviews to ensure that all questions were
asked and the responses neatly and legibly entered.
Roles and Responsibilities of the Supervisor
Controlling and ensuring the quality of data collection is the most

important responsibility of the field executive or supervisor. Throughout
the fieldwork, they are responsible for observing interviews and carrying
out field-editing. It should be noted that it is necessary to observe the
interviewers more frequently at the beginning of the survey and again
towards the end of the survey. In the beginning, interviewers may make
mistakes due to lack of experience or lack of familiarity with the
questionnaire. These can be corrected by discussions with the team
members in the evening, getting solutions for their problems and getting
answers for their queries. Towards the end of the survey, interviewers may
become bored or lazy in anticipation of the end of the fieldwork, and their
lack of attention to detail may result in carelessness with the data. To
maintain the quality of data, the supervisor should check the performance
of the interviewers thoroughly.
Collection of high-quality data is crucial for the success of the survey. It is

important that the quality of work is monitored through editing, spot
check, back check and re-interviews. This is especially important during
the initial phases of the fieldwork, when it is possible to eliminate
interviewer error patterns before they become habits. Thus, the supervisor
should:
x Observe some of the interviews to ensure that the interviewers are

conducting themselves well, asking the questions in the right
manner, interpreting and recording the answers correctly
x Spot check some of the HHs selected for the interview to be sure
that the right HHs are being interviewed
x Review a few of the filled questionnaires to ensure that they are

complete and consistent
x Meet each member of the team daily to discuss their performance
and point out areas where they can improve
x Help to solve problems that the interviewer might face in finding
the assigned HHs, understanding the concepts in the questionnaire
or with difficult respondents.
During the fieldwork, the supervisor should assign work to the interviewers
after considering the competence of individual interviewers and ensuring
that there is an equitable distribution of the workload, allocate the HHs in
such a way that all the investigators are working close to each other. This
will help the supervisor in proper supervision and monitoring and
communicating any problems to the field coordinator/project director.
Assigning work to interviewers
While assigning work to the interviewer/investigator, the supervisor

should:
x Assign a maximum of two calls or interviews at a time to an

investigator in usual circumstances so that it is easy to back-check
the investigator's work.
x Assign more interviews than an interviewer can do in one complete
day. This is necessary because some HHs and/or women may not
be available for the interview at the time of the interviewer’s visit.
Sometimes, there may be as many as three or four such cases in a
day for a particular interviewer. Fewer HHs should be assigned at
the beginning of the survey to allow time for discussion of
problems and for close supervision.
x Provide sample details such as structure number, HH number, name
of the head of the HH, the respondent's name etc., to the
investigators so that it is easy to locate the sample HH.
x Assign calls or interviews by following the sequence of the list so
that no sample is missed.
x Make daily work assignments. The supervisor should make sure
that each interviewer has enough work to do for the day,
considering the duration of an interview as well as the working
conditions in the area.
x Distribute the work fairly among the interviewers, assign work
considering the capabilities and strengths of each interviewer but
196 Chapter Nine
also make sure that a larger work load is not assigned to a certain
set of investigators consistently. The supervisor should also ensure
that each interviewer has all the required information and materials
for completing the work assignment.
x Make sure that all the selected HHs have been interviewed before
the team leaves that area.
x Maintain a checklist of all completed records for each day, and
carefully monitor all assignments and work completed by each
interviewer and for each work area for completeness and accuracy.
x Make sure that all the mobiles, laptops etc., are charged and are in
proper functioning condition in case the survey is being done using
PDAs.
x Make sure that the interviewers fully understand the instructions
given to them and that they adhere to the work schedule.
Roles and Responsibilities of Field Executive/Field Manager
The responsibility of the field manager is to manage the field work starting
with the recruitment of the investigators and supervisors, their training,
ensuring data collection is carried out in adherence with the required
quality standards and finally providing data to the field team. The field
manager should concurrently monitor the data collection, make spot
checks and back checks and have a look at the data collected regularly to
provide feedback to supervisors and enumerators on data quality. The field
manager should also conduct debriefing sessions initially and intermittently to
discuss any issues which the field team might be facing and also provide
feedback to the investigators based on his and the supervisor's observations.
9.12 Quality Assurance Protocols

The lead researcher and the core team members are directly responsible
for ensuring that the data collection norms are adhered to. There should be
regular visits from the field managers during data collection to ensure that
the quality protocols are followed. The field executives and the supervisor
should observe each interviewer multiple times throughout the course of
the fieldwork. At first the investigators should be observed during the
training sessions and then screened based on their understanding and skill
and by observing them during mock sessions, field practice and dry runs.
Each interviewer should also be observed during the first two days of
fieldwork so that any error made consistently is brought to their notice
immediately. Additionally, observation of each interviewer’s performance

is done during the rest of the fieldwork.
The field manager and executives, in tandem with the supervisors, should
ensure best practices are followed at the data collection stage. Along with
this, there should be regular visits by the research team. Handholding
support should be provided to ensure data quality through spot checks and
back checks.
Back checks
The field manager and executives should ensure that back checks are done
by them or by the supervisors in all sampled areas wherein the completion
rate is found to be low or it seems to be a problem. Back checks are a
powerful tool for checking the quality of the data in which the information
for a few respondents is systematically checked. This is done by
conducting a short re-interview with some respondents and matching the
results with what had been collected by the interviewer. Back checks or re-
interviews help reduce the types of problems that affect the accuracy of the
survey data.
Spot checks and observation
Some of the interviews conducted should be observed by the field

managers, executives and supervisors to ensure that the interviewers are
administering the survey tool properly. Rigorous spot checks and
observations need to be done at the start of the survey so that immediate
feedback can be provided to the investigators and the quality of the field
work can also be monitored. All observations during spot checks should be
noted and discussed with the complete team during the de-briefing
sessions.
Validation Checks and Rules
To ensure the data quality and to minimise logical and data entry errors,
basic checks should be included in the programme or the survey tool.
Validity checks should be run on one question or field at a time to ensure

that the fields do not include invalid characters and that all the essential
fields have been completed (e.g., no field is left blank where a number is
required). A sample check is presented as follows: the example specifies
that the value of the indicator for the question on outcome of delivery can
either be 'live birth' or 'still birth'. The value of the indicator for this value
198 Chapter Nine
cannot be missing or cannot have any other value than the two options of
'live birth' or 'still birth'. A log of cases with such kinds of errors should be
generated for each survey.
Example:
Table 1.15: Example Validation Check
Indicator Variable value check Variable value

What was the Live birth-1
outcome of this Still birth-2
delivery?
CHECK Value=1 or 2
Range Checks
For data fields containing information about a continuous variable e.g.,

height, weight, month etc., observations should fall within a specified
range. Thus, if the value of a variable falls outside the normal range, it
should be checked. Any value outside this range is erroneous and needs to
be checked.
Example:
Table 1.16: Example Range Check
Indicator Variable Value

At which month of your
pregnancy did the ASHA visit
you for the first time?
CHECK Value 1 should be from 0-9
Consistency Checks
Often, certain combinations of values of different variables that are within

range are logically impossible or very unlikely. The validation rules are
defined to have some checks to ensure data consistency. As mentioned in
the table given below, the number of hours after which the woman
delivered after reaching the facility should be less than the number of
hours that the woman stayed in the facility in total. Such consistency
checks are documented for different indicator values for the three types of
data.
Table 1.17: Example Consistency Check
Value Indicator
Value 1 How many hours did you stay in this facility?
Value 2 After how many hours of reaching the facility did you
deliver?
CHECK Value 1 > Value 2
Data Cleaning
After the completion of data collection, when the field team hands over the
raw data to the research team, data cleaning is done by the research team.
Data cleaning consists of checking the data thoroughly and checking to see
if the data has all the variables, including the required number of cases.
Open ended questions should be post coded as part of data cleaning. It is
also necessary to ensure that the data is properly labelled. The data analyst
looks for areas of data inconsistency and clarifies with the field team if
required.
9.13 Data Analysis

The next step after data cleaning is to prepare a data analysis framework
and go ahead with the data analysis. Though the data analysis paradigm is
usually finalised at the start of the project, it is important to revisit the data
analysis protocol and paradigm. In an ideal situation, an evaluation should
have a mixed method approach, where the quantitative method provides
the attribution story and the qualitative method contributes to the story
line.
Qualitative Data Analysis
In contrast to the quantitative data collection, qualitative data collection

through FGDs and IDIs are mostly conducted using paper-based
qualitative in-depth research tools. There is a process sheet to document
the information which includes the time and place of the interview, the
subjects covered, the person providing the information, the field team
member conducting and documenting that interview, and other relevant
information.
The content analysis format is usually developed beforehand, which is not

only helpful in the documentation process but it also helps in supporting
data analysis and synthesis.
200 Chapter Nine
The qualitative data collected is transcribed and translated into English.

Detailed field notes also supplement the transcription and the translation.
After the transcription, the detailed coding framework is developed. This
serves as a reference to the qualitative analysis software as per the pre-
defined set of assigned criteria.
The data/content is appropriately coded and validated. The coding of

content is carried out in parallel by two coders and processed to find out
the inter-coder reliability. Evaluators use NVivo or ATLAS.ti to code
textual data, wherein they attach annotation, indexing and memo at the
query stage to build theory. The analysis not only looks for the major
themes that cut across groups but also at the key insights. During analysis,
different aspects and contextual themes are considered to build theories
with respect to the broad areas of a) Context b) Internal Consistency c)
Frequency d) Extensiveness e) Intensity and f) Specificity.
In terms of presentation of qualitative data, evaluators base the findings of

the research around the categories or themes that have emerged. These are
presented as sections with relevant sub-sections. Quotes are used to
demonstrate and or to inform or support findings building on reliability
and validity of each quote. Further, and if required, qualitative data is
represented in a quantitative form.
Data Analysis Plan: Quantitative Protocol
The core team members, under the guidance of the project coordinator and
the data analyst, prepare the analysis or tabulation plan. The required
tables are generated using the latest versions of analysis software like
SPSS or STATA.
The choice and nature of data analysis depends on several factors such as
the type of the variable, the nature of the variable and the mode of analysis
performed. Evaluators can choose either SPSS 17.0 or STATA 10.0 for
analysis. SPSS is the most popular quantitative analysis software used
today in social research. SPSS 17.0 can be used for basic analysis and for
generating tabulated reports, descriptive statistics, and complex statistical
analyses, in addition to which the SPSS package provides easy to use
interface for analytics. STATA is another strong statistical software
package which provides strong modules for evaluation methodologies
such as DID, RD and IV. Nowadays, evaluators are also using R for its
strong analytical power. The data analysis is usually categorised into two
broad categories:
Basic Analysis
x Simple percentage of all variables of interest

x Descriptive statistics of key variables in socio-economic surveys
x Parametric test of association to ascertain the statistical significance
of existing associations between variables of interest
Advanced Analysis
x Significance testing
x Calculating correlation coefficient
x Analysing internal consistency reliability of the scale using Alpha
Coefficient
x Inequality Analysis
x Multivariate Analysis
9.14 Ethical Approvals

Ethics plays an important role in any research, and more so in the case of
evaluations involving human subjects. Evaluation guidelines prescribed by
various international and national organisations place central emphasis on
ethics as the key tenet of evaluation. A snapshot of the evaluation
guidelines presented in the table below highlights the key tenets of
evaluation.
202 Chapter Nine
Table 1.18: Snapshot of Evaluation Guidelines
UNDP Norms UNEG American African

for Evaluation Evaluation
Evaluation Association’s five Guidelines
guiding
principles are
Independent Intentionality of Systematic Utility
Evaluation: Inquiry
x Utility
x Necessity
Intentional - x Obligation of Competence Feasibility
Clarity in Evaluators
Rationale x Independence
x Impartiality
x Credibility
x Conflict of
interest
x Honesty and
integrity
x Accountability
Transparent Obligation to Integrity/Honesty Propriety
Participants
x Respect for
Dignity and
Diversity
x Rights
x Confidentiality
x Avoidance of
Harm
Ethics - Evaluation Process Respect for Accuracy
Devoid of and Product People
personal or x Accuracy
sectoral x Transparency
interests x Reporting
x Omission and
wrong doing
Impartial Responsibilities
for General and
Public Welfare
High Quality
Timely
Used
In terms of the operational aspects and in case it is required, it is important

that approvals are sought from the local Institutional Review Board (IRB)
prior to the study. In addition, it is critical that the key tenets of ethics such
as taking proper consent, ensuring privacy and confidentiality are adhered
to during the evaluation.
9.15 Learning Framework and Structure

One of the critical objectives of M&E is to derive learnings for course
correction and to suggest changes with respect to programme replication
and scale up. Learning as part of MEL becomes as critical as conducting
M&E itself. More so, in practice where different partners and stakeholders
are involved in contributing towards the project goal, it is imperative that
the learning is shared not only among the M&E team members and the
project implementation team but also with the donor and the external
audience, including policy makers and evaluators, so that the lessons learnt
can be built upon and replicated in other settings. It is in this context that it
is imperative that a clear-cut learning structure is spelt out in which the
modalities of knowledge sharing are also clearly specified. The learning
structure should be in accordance with the reflective and integrated nature
of the external MLE support and constitutes both ‘informal’ and ‘formal’
spaces.
There are four aspects to developing learning structures for intervention:
x Develop a learning framework

x Build dialogue/feedback spaces to support the ongoing learning
x Develop structures to support the ongoing feedback
x Pay attention to mechanisms of policy influence
Developing a Learning Framework
The first task after having an initial agreement on an M&E framework is

to develop a Learning Framework.
One of the areas on which there is key focus in M&E is to explore

questions on the configurations of contexts, mechanisms and outcomes
204 Chapter Nine
which inherently result in a broad range of learning that includes

knowledge development, policy, organisational process and impact
learning (Sridharan, 2008). Thus, to formalise the learning, a key action
step should be taken to develop a learning framework which will: (i)
categorise the type of learning that is needed over the course of the
evaluation; (ii) the timing when such learning is needed; (iii) the
dissemination vehicle for such learning.
In this context, examples of learning include:
x Policy learning
Policy makers typically want to know more than just if a programme

works or not (Sanderson, 2003). Based on the evaluation, policy makers
also want to know how future versions of similar interventions can better
translate the underlying policy theory into more clearly developed
programmes.
x Organisational learning
‘Social inventions are complex systems thrust amidst complex systems’

(Pawson, 2004). Thus, the key questions related to organisational learning
that need to be answered are:
x What kind of organisational structures are needed to successfully

implement complex programmes?
x Which organisational structures support coordination across the
different teams implementing the multiple interventions?
x Process Learning
It is important that the learnings related to processes are also derived from
the programme evaluation. Key areas of non-compliance and bottle necks
need to be identified and communicated to the programme team as part of
process learning.
x Knowledge Development
Programmes often do not have sufficient knowledge about the lives of the
programme recipients or beneficiaries they are trying to impact.
Evaluations provide an opportunity to build such knowledge. Evaluations
also provide opportunities to learn about the barriers that need to be
overcome. This knowledge can be fed back into the system with the aim of
further improving the programme or improving the future versions of the

support programme.
x Programme Impacts
One of the most important focus areas of evaluations is impact.

Evaluations aim to answer the programme related learnings, viz.,
x Does the programme model work?

x Are there demonstrable impacts of the interventions on programme
recipients?
x Is there empirical support for programme impacts on interim and
proximal individual-level outcomes based on the programme
theory?
Dialogue Spaces for Ongoing Learning
Building on the learning framework, dialogue spaces are created for

interaction between multiple stakeholders. Dynamic interaction is required
among programme planners, implementers, and evaluators when contexts
and mechanisms are being worked out. This means moving away from a
purely product focused view of knowledge and encouraging spaces in
which dialogue and ongoing learning is encouraged. Learning needs to be
supported by well thought out processes and products. Learning processes
need active and intentionally planned spaces for the ongoing dialogue.
Develop Clear Structures to Support Feedback and Learning
One key goal is to develop clear structures within the MLE team to
support the learning and feedback. This implies:
x Having at least one person whose responsibility is to ensure that

feedback is occurring periodically and that the same person is
responsible for the ongoing alignment of learning from the
evaluation
x Collaboratively developing the learning framework discussed
above
x Working collaboratively with the implementing organisation and
funders to develop both clarity of the evidence needs of the variety
of stakeholders and a policy influence plan to connect the learning
with the stakeholders
206 Chapter Nine
x Developing a clear responsibility for both a framework of influence

and a dissemination strategy for all stakeholders and developing
metrics for accountability to learning collaboratively
x Identifying events (both policy meetings, conferences etc.) that can
support the dissemination strategy.
Developing a Policy Influence Plan
As one of the key objectives of monitoring is to influence decision

making, similarly at the last mile, evaluation aims to provide evidence to
replicate and scale-up intervention and to influence policy (Carden,
Knowledge to Policy: Making the Most of Development Research, 2009)
(Mark, 2004). For policy recommendations that are given based on the
evaluation, there should be compliance with the points mentioned below:
x Fit with the latest ‘political zeitgeist’ (The defining spirit or mood
of a period of history as shown by the ideas and beliefs of the time).
For example, the political zeitgeist may be conducive to
implementing health equity policies. Working closely with policy
makers and other policy and programme stakeholders can provide
evaluators an avenue by which to understand the policy landscape
and in some cases, impact the same through a broadening of policy
horizons and enhancing policy capacities (Carden, Knowledge to
Policy: Making the Most of Development Research, 2009).
x Raise the salience of the key policy considerations emerging from
the evaluations to those with the power to decide (Mark, 2004)
x Be simple enough to be communicable
x Be simple enough for those delivering to be able to apply it
x Be cogent enough to persuade the various agents in the delivery
chain that it is worth doing
x Be cogent enough to win over the required leadership - from the
start, and from the top, to the end
x Be embedded within the intervention teams, e.g., explore
perspectives with action research approaches which are helpful but
not captured by the intervention team - with people who can remind
the intervention team about the gap between their measures and the
intended outcome. Such embedded evaluators can mediate/present
the experience of others in similar contexts and advise on what is
likely to work
x Educate policy makers and practitioners that an evaluation is not
simply an instrument to determine if a programme is working
x Explore loss in translation from policy aspirations to programme

design,as translating the policy aspiration into a demonstration
project can be a challenge. Thinking about contexts, mechanisms,
and outcomes in a focused manner can help raise questions about
the match between the programme design and the longer-term
policy aspiration. As described by Sanderson, functions of
evaluations can include ‘influencing the conceptualization of
issues, the range of options considered and challenging taken-for-
granted assumptions about appropriate goals and activities’
(Sanderson, 2003).
The challenge of ‘making the evaluation matter’ is not just to generate

learning(s) but to align such learning with policy priorities. A key focus of
the dialogues or feedback is aligning the evaluation learnings to the
changing policy landscape. However, there is also a growing body of
literature on external validity (Deaton, 2009), which establishes how
learning from evaluations that can be spread to impact populations has
only a limited evidence base (Rodrik, 2008). Questions of alignment
between the evaluations and policy priorities are especially important
given dynamic policy landscapes, changes in the government and changes
in policy priorities.
To give an example, a policy influence framework that focuses on

inequities needs to go considerably beyond the details of a specific
programme (Sridharan, 2008). A programme theory usually informs the
practitioner why a programme may be effective, though it rarely describes
why and how a programme can impact inequities at the population level.
More formal, explicit and pragmatic thinking is needed on how and why
an intervention can impact the population at large, and how a policy
influence plan can help. Questions about contexts and mechanisms should
not be confined to the programme level, raising questions about contexts,
mechanisms and their relationship to equity outcomes at the population
level can help enhance wide scale policy implementation, mainstreaming
and sustainability. The dialogue spaces discussed below will help in
developing such a policy influence plan.
Operational Mechanism of Learning Structures
In order to have coherent strategy, it is important for every agency and

expert to have role clarity about their specific tasks and responsibilities. A
reporting management structure is prepared and operationalised to address
these issues and ensure clarity among partners.
208 Chapter Nine
A simple operational aspect such as initiating a system of conference calls

fortnightly between all the key personnel from different partners for
updating each other on the progress made so far also adds to the efficiency
of the process.
A dropbox or samepage can been created to act as a depository of all

relevant documents produced, and all documents and updates made
available there. All partners have online access to the created document
repository which ensures ease of document retrieval.
For sharing learnings with the implementation organisation, informal

sharing of lessons should take place as and when required and at different
levels and forums. However, for a systematic sharing of learning, it is
important to have monthly meetings where it is expected that routine data
issues and MLE learning is shared and discussed.
As the project starts being implemented at the ground level, it is

imperative that the progress is shared with the funders and other key
stakeholders frequently to ideate, operationalise the design and outcome,
besides apprising the team there of the challenges faced by the project as
well as the measures being taken to address them. A routine reporting in
the form of an activity reporting template as part of reporting requirement
is used.
In addition to the above, meetings to share the learnings with the funders
should continue to take place as and when required.
To share the project learnings with the outside world, i.e., the policy
makers and other evaluators, the project should generate working papers
or methodological notes for presentation in local and international
conferences as well as in peer reviewed journals. These documents are
aimed primarily at policy makers/evaluators and highlight the new
approaches attempted in the project.
9.16 Evaluation Report

Evaluation reports present findings, conclusions, and recommendations
from an evaluation, including recommendations regarding how evaluation
results can be used to guide programme improvement and decision
making. The evaluation report can have the following chapterisation plan:
x Title Page
x Executive Summary
x Intended Use and Users
x Programme Description
x Evaluation Design Scope and Focus
x Results and Conclusion
x Programmatic Implication
x Learning
x Annexure
Executive Summary: An Executive Summary provides a summary of the

complete evaluation starting from its design, intended users, evaluation
focus, key results and its implication on programme application.
Intended Use and Users: In line with the spirit of utilisation focused
evaluation, the evaluation report clearly specifies its intended use and the
intended users for transparency and better use of evaluation findings. This
helps in building the demand for evaluation use and also helps the reader
or the user to interpret findings in a better way.
Project Description: Project description forms an important component of

the evaluation report. It provides a background section on the project along
with a ToC and logic model. In addition, it also provides a literature
review to set the context for evaluation findings.
Evaluation Focus, Scope and Design: This chapter presents data sources
and methods along with evaluation indicators, performance measures, data
sources, and methods used in the evaluation which are described in this
section.
Results and Conclusion: The Results and Conclusion section provides the
findings derived from the evaluation study. This section also aims to
provide a meaningful interpretation of the data, which is more than mere
presentation.
Programmatic Implication: Based on the evaluation result, it is critical to

provide insights for programmatic improvement or replication.
Use, Dissemination, and Sharing of Research Results: Ensuring the use of

evaluation as its key tenet. Building on this key tenet, this section
describes plans for the use of evaluation result findings, along with a
specific dissemination plan clearly outlining user specific strategies. It is
210 Chapter Nine
also important to list the ways in which different stakeholders should be

involved taking cognisance of the complex and dynamic environment in
which the intervention is executed.
BIBLIOGRAPHY
Aquaknow. (2016, 01 06). Retrieved from

http://www.aquaknow.net/en/water-toolkit-trainings/monitoring-
evaluation-and-audit-water-development-projects/monitoring-
evaluation-and-audit-water-dev/15689
Brisolara, D. S. (2002). Feminist Evaluation:Explorations and
Experiences .New Directions for Evaluation.
Carden, F. (2009). Knowledge to Policy: Making the Most of Development
Research. New Delhi: Sage.
—. (2009). Knowledge to Policy: Making the Most of Development
Research. New Delhi: Sage Publications.
Chemonics. (2016, 05 08). Beyond Indicators: Why We Need Qualitative
Monitoring ? Retrieved from The Compass:
http://blog.chemonics.com/beyond-indicators:-why-we-need-
qualitative-monitoring
Connelly, L. M. (2000). Feminism and Developmenet:Theoretical
Perspectives.
Corbin, A. S. (1998). Basics of Qualitative Research: Techniques and
Procedures for Developing Grounded Theory. Sage Publications.
Dart, R. D. (2003). A Dialogical, Story-Based Evaluation Tool: The Most
Significant Change Technique. American Journal of Evaluation.
Davis, J. D. (2005). The 'Most Significant Change(MSC)' Technique- A
Guide to Its Use.
Deaton, A. (2009). Instruments of development: Randomization in the
tropics, and the search for the elusive keys to economic development.
Dharmendra Chandurkar, N. S. (2014). Developing Monitoring and
Evaluation Framework for Budget Work Projects. A How-To Guide for
Practitioners.
FAO. (n.d.). The logical framework. Retrieved from FAO Corporate
Document Repository:
http://www.fao.org/WAIRdocs/x5405e/x5405e0p.htm
Forster, R. (2002, April 19). Participatory Monitoring and Evaluation:
Principles, Action Steps, Challenges : Introduction.
Foundations of Success. (2007). Using Results Chain to Improve Strategy
Effectiveness: An FOS How-To Guide. Foundations of Success,
Bethesda, Maryland, USA.
212 Bibliography
Gamble, J. A. (2008). A Developmental Evaluation Primer. The J.W.

McConnell Family Foundation.
Gaventa, M. E. (n.d.). Who Counts Reality ? Participatory Monitoring and
Evaluation: A Literature Review. Institute of Development Studies-
Working Paper 70.
Greene, W. H. (2008). Econometric Analysis. Upper Saddle River, N.J.
07458: Prentice Hall.
Kari, O. (2004). The Logical Framework Approach. SIDA .
KEPA. (2015, 01 06). Retrieved from https://www.kepa.fi/:
https://www.kepa.fi/tiedostot/differences-between-monitoring-
evaluation.pdf
Kosheleva, N. (2016, 06 17). Methodologically Inclusive Transformative
Evaluation as an Enabler of Social Inclusion. Retrieved from
http://www.unesco.org/:
http://www.unesco.org/new/fileadmin/MULTIMEDIA/HQ/SHS/pdf/W
orkshop-Social-Inclusion_IOCE.pdf
Mark, M. &.-5. (2004). The Mechanisms and Outcomes of Evaluation
Influence. Evaluation, 35-37.
MDF Training & Consultancy (2016, 01 07). Retrieved from Urban
Reproductive Health:
https://www.urbanreproductivehealth.org/sites/mle/files/10%20Indicat
ors.pdf
Michael Bamberger, J. R. (2006). Real World Evaluation : Working under
Budget, Time, Data and Political Constraints. Sage Publications.
NORAD. (1999). The Logical Framework Approach: handbook for
objectives-oriented planning.
Patton, M. (1990). Qualitative evaluation and research methods. Sage
Publications.
—. (2008). Utilization-Focused Evaluation(4th edition). Sage Publishing.
Patton, M. Q. (2012). Essentials of Utilization-Focussed Evaluation. Sage
Publishing.
Paul J. Gertler, S. M. (2011). Impact Evaluation in Practice. The World
Bank.
—. (2011). Impact Evaluation in Practice. World Bank.
Pawson, R. G. (2004). Realist synthesis: an introduction. ESRC Research
Methods Programme. Retrieved from
http://www.ccsr.ac.uk/methods/publications/documents/RMPmethods2
.pdf
Ramirez, D. B. (2013). Utilzation Focused Evaluation-A primer for
evaluators. Southbound Sdn. Bhd.
A Practitioners’ Manual on Monitoring and Evaluation 213
Real World Evaluation. (2016, July 2). Retrieved from Real World
Evaluation: http://www.realworldevaluation.org/Home_Page.html
Republic of Serbia. (2011). Guide to Logical Framework Approach: A Key
Tool to Project Cycle Management .
Richard Blundell, M. C. (2000). Evaluation Methods for Non-
Experimental Data. Fiscal Studies , 427-468.
Rodrik, D. (2008, October). The New Development Economics: We Shall
Experiment, but How Shall We Learn? Faculty Research Working
Papers Series-John F. Kennedy School of Government - Harvard
University .
Sanderson, I. (2003). Is it ‘What Works’ that Matters? Evaluation and
Evidence-Based Policy Making. Journal of Research Papers in
Education, 329-343.
Sielbeck-Bowen, S. B. (2002). Exploring femenist evaluation:The ground
from which we rise.New Directions for Evaluation .
SMART, Characteristics of Good Objectives. (2016, 01 04). Retrieved
from Community Empowermnet Collective:
http://cec.vcn.bc.ca/cmp/modules/pd-smar.htm
Sridharan, S. G. (2008). Evaluation of Primary Prevention Intervention:
Have a Heart Paisley Phase 2. . Glasgow: NHS Health Scotland.
Tilley, R. P. (2004). Realist Evaluation. Sage Publications.
UNAIDS. (2010). An Introduction to Indicators .
UNDP Evaluation Office. (2002). Handbook on Monitoring and
Evaluation for Results.
UNICEF Evaluation Office. (2006). How to design and manage Equity-
focused evaluations. UNICEF.
USAID Centre for Development Information and Evaluation. (1996).
Preparing a Performance Monitoring Plan.

A Practitioners Manual On Monitoring and Evaluation

Uploaded by

Copyright:

Available Formats

A Practitioners Manual On Monitoring and Evaluation

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Practitioners Manual On Monitoring and Evaluation

Uploaded by

Copyright:

Available Formats

A Practitioners’

By Kultar Singh, Dharmendra Chandurkar and Varun Dutt

This book first published 2017

Cambridge Scholars Publishing

Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK

British Library Cataloguing in Publication Data

Copyright © 2017 by Kultar Singh, Dharmendra Chandurkar

ISBN (10): 1-4438-8992-X

List of Figures........................................................................................... viii

List of Tables ............................................................................................... x

List of Abbreviations .................................................................................. xi

Preface ...................................................................................................... xiv

Chapter One ................................................................................................. 1

Chapter Two ................................................................................................ 9

Chapter Three ............................................................................................ 27

Chapter Four .............................................................................................. 37

Chapter Five .............................................................................................. 62

Chapter Six ................................................................................................ 72

Chapter Eight ........................................................................................... 102

Chapter Nine............................................................................................ 175

Bibliography ............................................................................................ 211

Figure 1.1: Project Cycle

Figure 8.20: MS Excel File Opened in SPSS

Table 3.1: Monitoring vs Evaluation

AHS Annual Health Survey

IHHL Individual Household Latrines

VHC Village Health Committee

This book is an effort to provide handholding support to professionals

A Practitioner's Manual on Monitoring and Evaluation of Development

Chapter 1 of the manual aims at developing a fundamental understanding

Chapter 2, which focuses on understanding how to design projects, begins

Chapter 4 takes the readers through the process of developing a Performance

intended results. Subsequently, the key activities that should be done to

Chapter 5 describes different approaches to monitoring including progress,

Chapter 6 delineates in detail the various evaluation designs by explaining

Chapter 7 explains the fundamental concepts of the different approaches

In Chapter 9, practitioners are taken through a step by step process of

collection and quality assurance protocols, getting ethical approvals,

To summarise, this book will act as a hands-on manual for professionals

The mainstay of work in the development sector is in the form of projects,

On completion of the chapter the reader will be able to:

" Describe what is a project

1.1 Defining a Project

Every project has a specific objective and it is envisioned that through

Hence, a project may be defined as:

1.2 Project Cycle

Figure 1.1: Project Cycle

Stage I: Situation Analysis

Stage II: Gap Analysis

Stage III: Project Planning

staff) based on the defined roles and responsibilities. Physical and

Stage IV: Implementation and Monitoring

Taking the current example of the ODF project, an evaluation should be

To consider another example of a project which aims to increase the rate

1.3 Project Goal and Objective

The SMART criteria is widely used to judge project objectives. A project

x Specific: The objective is said to be specific as it specifically

objective is likely to lose focus, while it renders the objective open to

Therefore, a well-designed project should ideally have one objective