UNIT 6 MEASUREMENT OF VARIABLES

Structure
6.0 Objectives
6.1 Introduction
6.2 Types of Variables
6.3 Measurement of Qualitative Data
6.4 Census versus Sample Survey
6.5 Sampling Procedure
6.6 Types of Sampling
6.7 Summary
6.8 Answers to Self Check Exercises
6.9 Keywords
6.10 References and Further Reading

6.0 OBJECTIVES
After going through this Unit, you will be able to:
• define a variable;
• distinguish between various types of variables;
• measure ordinal variables through scaling techniques;
• distinguish between census and sample survey;
• explain the steps involved in carrying out a sample survey; and
• distinguish between various types of sampling.

6.1 INTRODUCTION
During the course of research you come across situations where you have to measure
various characteristics. These characteristics could be of various types, viz., age, height
or income level of visitors to a library; educational qualifications, social status or reading
habits of a person; gender, religion, or area of interest of a library user. Note that all
these characteristics are not similar from the point of measurement. While age, height
or income can be measured in quantitative terms (in number of years, in centimetres, in
rupees) religion or gender can be put to certain categories only. In this Unit we discuss
the issue of measurement of qualitative variables, particularly the scaling techniques.
Another issue discussed in the Unit pertains to collection of data on the basis of sample
survey. Very often it is not possible to survey all the units bearing the characteristic
under study. The constraints could be inadequate funds, time limit, and manpower. In
such situations we survey only a subset of the population, called sample. We discuss
various concepts associated with sampling procedure.
We begin with the types of variables and their measurement. 91
 Measurement of
6.2 TYPES OF VARIABLES Variables

Let us begin with the concept of a variable. It is a characteristic of the sample or the
population that we intend to measure. Thus age of the reader is a variable, so is gender,
educational level or mother tongue. As we have mentioned earlier all variables are not
similar.
Variables can be of two types: qualitative and quantitative. Qualitative variable is one
that cannot be expressed in numerical terms. For example, marital status is a qualitative
variable. Here we can have two categories: married and single. Of course, if you want
a more detailed categorisation you can further divide single in to widow/widower, divorcee
and never married. Similarly, gender (male or female), mother tongue (Hindi, Bengali,
Oriya, Tamil, Urdu, etc.), subject categories (economics, history, physics, medicine,
etc.), religion (Buddhism, Christianity, Hinduism, Islam, etc.) are examples of qualitative
variables. Here we study an attribute or quality that cannot be quantified, but can be
divided into various categories. Moreover, we cannot say that one category is higher or
greater than another category. Such variables are also called nominal variables.
There is another type of qualitative variable where we can divide the observations into
various categories and also say that one category is higher or greater than another
category. An example could be the educational qualification of a visitor to a library.
Here we can divide the visitors on the basis of their educational qualification into
categories such as ‘secondary’, ‘senior secondary’, ‘graduate’ and ‘post-graduate’. In
this case, obviously, the category ‘Senior Secondary’ is higher than the category
‘Secondary’ in terms of number of years of schooling and expected mental maturity. In
this case we arrange the categories in an ascending or descending order. This sort of
variables are called ordinal variables.
In the case of nominal variables we cannot perform any mathematical operations (such
as addition, subtraction, multiplication, division,) or logical operations (greater than,
less than) across categories. We can simply count the number of observations in each
category. In the case of ordinal variables we can say that one category is greater than
another category. But we cannot quantify the difference between categories. For
example, we cannot express numerically the difference between two categories (say
secondary and senior secondary). Also we cannot say that the difference between two
categories (say secondary and senior secondary) is the same as the difference between
two other categories (say graduate and post-graduate).
A quantitative variable can be expressed in numerical terms. Hence it is also called
numerical variable. Examples of numerical variable could be age, income, weight, height,
distance travelled, etc. This category of variables can be subjected to various mathematical
and logical operations. Thus we can express the monthly income of a librarian in Rupees
and also say by what percentage it exceeds the salary of a library assistant.
Numerical variables can be of two types: discrete and continuous. Discrete variable is
one where the observations assume values in complete numbers. For example, the
number of children in family can only be whole numbers; it cannot be fractions. On the
other hand, continuous variables can assume any value in an interval. For example,
weight of a person can be measured to any precision and thus can take any value in
between two points.
Let us distinguish between variable and data. We obtain data by measuring a variable
(qualitative or quantitative) on certain individuals or units. For example, suppose we
measure the height of 50 employees in a library. Here height is the variable and we 89
obtainfor
Tools 50 Research
observations. These 50 numerical values that we obtain are our data. Thus
we have discrete data or continuous data depending upon whether the variable is discrete
or continuous.
Similarly there are primary data and secondary data. Primary data refers to data collected
by the researcher by undertaking a field survey. On the other hand, secondary data
refers to collection of data from published sources, e.g., census, budget, handbooks,
etc. Thus when you undertake a field survey, collect data, analyse the results and present
it in some forum, it is primary data. But when I use that data for further analysis it
becomes secondary data for me.

6.3 MEASUREMENT OF QUALITATIVE DATA

The measurement of quantitative data does not pose problems as these are expressed
in numerical terms. Measurement of qualitative data, however, is a complex issue and
needs to be discussed further. There are two types of variables that are usually measured:
i) social behaviour and personality, and ii) cultural and social environment. The purpose
is to study socioeconomic status and its impact on various issues, attitude towards a
particular event or behaviour, participation in certain activity, etc.
In order to measure these types of issues no readymade scale of measurement is available.
Secondly, there is a lot of ambiguity and inter-personal variability in the definition of
these qualities. However, attempts have been made to construct ‘attitude scales’ to
measure these qualities. These scales are in the form of equal-appearing intervals. Such
scales are widely used in sociology, psychology, education and administration apart
from other branches of study.
In designing such scales we should keep one thing in mind that it should be ‘valid’ and
‘reliable’. In other words, i) it should measure what we intend to measure, and ii) it
should yield the same consistent results when applied under the same conditions.
In designing attitude scales we form a number of statements and ask the respondents to
react to these statements. The statements should be brief, unambiguous and relevant. It
should be expressed in such a form that it could be endorsed or rejected in terms of
definitely expressed attitude.
A widely used scale is ‘Likert Scale’ which is also referred to as ‘technique of summated
ratings’. The basic steps in construction of Likert scale are given below:
A series of propositions representing attitudes are compiled. The attitudes of persons
towards these propositions could range from extremely negative to extremely positive.
The statements or propositions express values rather than facts. Each statement indicates
the position of a person towards the issue concerned. For example, a statement could
be ‘there should be separate reading rooms for boys and girls in a library’. The response
to such a statement could range from ‘strongly agree’ to ‘strongly disagree’ depending
upon the attitude of the respondent towards the issue.
Each statement is so formulated that the response of persons can be given in any of the
five terms such as i) strongly approve, ii) approve, iii) undecided, iv) disapprove, v)
strongly disapprove. Many times expressions such as i) strongly agree, ii) agree, iii)
cannot say, iv) disagree, v) strongly disagree are also used. Although a five-point
continuum is common, there can be three-, four-, or six-point continuum also. For
example, a seven-point continuum could be i) always, ii) almost always, iii) frequently,
iv) occasionally, v) rarely, vi) almost never, vii) never.
90
Weights are assigned to each response for a statement. For example,Measurement
in a five-point
of
Variables
continuum the weights that we attach to each response could be 1, 2, 3, 4, and 5. These
weights indicate the intensity of the attitudes of a respondent.
When we have a number of statements, we can arrange the statements according to
their relative intensity. Also groups of individuals can be arranged in a rank order on the
basis of scale scores. Moreover, the scale scores can be used to predict any outside
variable.
Self Check Exercise
1) State the type of variable (nominal, ordinal and numerical) for the following:
a) tribes in India
b) height of children
c) number of books
d) accession number of books
e) subject codes
2) With an example explain the use of Likert scale.
Note: i) Write your answers in the space given below.
ii) Check your answers with the answers given at the end of the Unit.
........................................................................................................
........................................................................................................
........................................................................................................
........................................................................................................

6.4 CENSUS VERSUS SAMPLE SURVEY

There are quite a few methods for collection of data that we will discuss in detail in
Block 3 (questionnaire method, personal interview, participatory observation, etc.). An
important issue is whether we should collect data on all units present or only a part of it.
For example, if we wanted to know the reading habits of economics students of
Sambalpur University and for that purpose we designed a questionnaire. We found that
there are 800 economics students and it would not be possible to survey all the students
due to time and money constraints. Thus, we decided to administer the questionnaire to
100 students only.
In formal statistical language we distinguish between ‘population’ and ‘sample’. Since
our objective is to study the reading habits of economics students in Sambalpur University,
all 800 economics students qualify to be studied by us and thus constitute our population.
If we decide not to collect data on all students but study only 100 students, these
selected 100 students constitute our sample.
Thus, population is a collection of individuals or objects having the desired characteristics
we are interested in a sample which is a part of the population. Obviously we can draw
more than one sample from the same population. If you go to Sambalpur University on
a different day and select 100 students your sample may be different from what another
person had selected.
91
Collection
Tools of information on all units of a population is called census. On the other
for Research
hand, collection of information on a sample is called sample survey. The process of
selection of a sample is called ‘sampling’. The advantages of sample survey are:
• Sample survey is less expensive than a census.
• Sample survey requires lesser time and manpower than a census.
• Sample survey can be monitored closely and more accurate information can be
colleted than in census.
• Many times we do not require information on all units. For finding out the
effectiveness of a new medicine we do not have to test it on all patients; only a
representative sample will do.
In sampling we are surveying only a part of the population. Remember that we do not
know the exact population characteristics, as we do not survey the population. But
there are scientific methods by which we can estimate a population characteristic (called
parameter) on the basis of sample characteristic (called statistic). This topic will be
covered in Unit 17 while dealing with estimation and hypothesis testing.
There are two types of errors in a survey: sampling error and non-sampling error.
Sampling error is due to the fact that only a part of the population is being surveyed. It
is the difference between the parameter and the statistic. It can be reduced by adoption
of scientific sampling procedure, specifically probability sampling so that a random
sample is selected. Non-sampling error is due to errors in measurement, non-response
by the selected units, wrong recording of data, and personal bias of the investigators.
Thus, non-sampling error is present in both census and sample survey. In a census,
however, many investigators are involved and large number of units are to be surveyed.
So non-sampling error could be very high. On the other hand, in a sample survey
greater care can be taken in collection and recording of data since lesser number of
units are surveyed. Thus non-sampling error can be minimised in a sample survey. The
total error is the sum of sampling and non-sampling errors. It may so happen that the
non-sampling error in a census is more than the sampling and non-sampling errors in a
sample survey! In such cases sample survey would give better results than census.
Self-Check Exercises
3) What are the advantages of sampling over census?
4) Distinguish between sampling error and non-sampling error.
Note: i) Write your answers in the space given below.
ii) Check your answers with the answers given at the end of the Unit.
........................................................................................................
........................................................................................................
........................................................................................................
........................................................................................................

6.5 SAMPLING PROCEDURE

While carrying out a sample survey there are certain steps that we should follow. These
are as follows:
92
Specification of objectives: The objectives of carrying out samplingMeasurement
is the first and
of
Variables
foremost step in sampling procedure. Because all other steps will follow from this
objective.
Definition of population: In this step we define the units that should be included in the
population. Many times there are certain border cases where proper definition is
important. For example, if you want to survey the library personnel, it has to be decided
whether you should include part-time employees in the population or not.
Preparation of ‘sampling frame’: Once you have defined the units to be included in
the population, the next step is to prepare a list of the units from which the sample is to
be drawn. Many times problems come up because the sources from which you want to
prepare the sampling frame may be incomplete or obsolete.
Identification of sampling procedure: Sampling procedure refers to the method of
selecting the sample. There are quite a few sampling procedures available. We should
select a method that: i) gives us a representative sample, ii) is feasible to carry out
keeping in view our constraints, and iii) is cost effective. There are broadly two types of
sampling: probability sampling and non-probability sampling. We will discuss these types
in the next Section.
Determination of sample size: The next step is the determination of sample size. The
factors that influence sample size are i) population size, ii) variance of units, iii) desired
precision level, iv) response rate, and v) availability of resources. The sample size should
be relatively larger if population size is larger, variability among units is higher, more
precise results are required, and lesser is the response rate. However, we should settle
for smaller sample size if there are constraints such as availability of funds or manpower
or time. Sample is considered to be large if its size is greater than 30. If sample size is
less than or equal to 30 it is considered as small sample. The procedures of drawing
inferences (See Unit 17) for large and small samples are different.
Selection of sampling units: Once you have decided on the sampling procedure and
the size of the sample, the next step is draw the sampling units (that is, units included in
the sample) from the sampling frame.

6.6 TYPES OF SAMPLING

As mentioned earlier there are two types of sampling: probability and non-probability.
In probability sampling some element of randomness is involved in selection of units, so
that personal judgement or bias is not there. Here we have simple random sampling,
systematic random sampling, stratified random sampling and multistage sampling.
Remember that the word ‘random’ here does not mean haphazard or unsystematic,
rather it implies lack of bias or control in selection. In non-probability sampling we have
judgement sampling, quota sampling, and snowball sampling.
Simple Random Sampling (SRS): It is the basic sampling procedure where each unit
in the population gets an equal chance of being included in the sample. There are two
commonly used methods to draw a simple random sample, viz., i) lottery method, and
ii) random numbers selection method. In both the methods we assign a unique number
to each unit in the sampling frame. In lottery method we mix up the numbers very well
and draw the numbers one by one. In random number selection method we refer to
‘random number tables (RNT)’ available from various sources (including the Internet)
and select the units which are there in the RNT. Simple random sampling should be
used with a homogeneous population. That is, all the units in a population possess the
93
same attributes that the researcher is interested in measuring. The characteristics of
homogeneity
Tools may include age, sex, income, social status, geographical region, etc. If
for Research
the population is not homogeneous then we need a larger sample size to obtain a
representative sample. But we can divide the population into various homogeneous
strata and obtain a stratified random sample. A limitation of the simple random sampling
is that if the population size is too large, selection of the sampling units is time consuming.
These days, of course, computer has been very useful in selection of random numbers
and reduced time spent in selection process.
Systematic Random Sampling: In this case we select the units in a fixed interval.
For example, in a library you have to check the quality of maintenance of books. For
this purpose you decide take physical verification of 100 books out of the 5000 books
available in the library. Here population size is 5000 and sample size is 100. In systematic
random sampling procedure you take a random starting point and then select the 50th
subsequent unit. Thus if 12th book is the starting point, then you go on checking books
located at the positions: 62nd, 112th, 162nd, …… and so on. This procedure is useful
when preparation of sampling frame is difficult or not possible.
Stratified Random Sampling: This procedure is practiced when the population is
not homogeneous but can be divided into various homogeneous groups (called ‘strata’).
Here we select sub-samples from each stratum and add them together to obtain our
desired sample. Therefore, the stratified random sampling procedure is a better
representative of the population and sampled units reflect the population characteristics
more accurately. The steps we should follow are:
• Divide the population into strata based on some characteristic chosen by you
(example, Rural/Urban, Male/Female, etc.)
• The number of units taken from each stratum is proportional of the relative size of
the stratum and standard deviation of the characteristic within the stratum. If stratum
size is then a larger sub-sample should be taken. Similarly if you find that variability
among units is more in a stratum than other strata, then you should take a larger
sub-sample from that.
• Choose the sub-sample from each stratum using simple random sampling.
Multistage Sampling: As the name suggests, sampling procedure in this case is divided
into two or more stages. Let us take an example of a two-stage sampling. Suppose
you have to survey the aptitude of college students, who can be categorised into three
streams: arts, science and commerce. Here in the first stage you divide colleges according
to geographical regions and you select specified number of colleges from each region
on the basis of random sampling. In the second stage you select students by a random
sampling method from the colleges selected at the first stage only, not from all colleges.
Let us try to explain the difference between two-stage sampling and stratified sampling.
In stratified random sampling you divided the population (all college students) into
there strata: arts, science and commerce. From each stratum you selected a sub-
sample by simple random sampling procedure. The selected students can be from any
college and you cannot rule out a visit to any college. In two-stage sampling you are
excluding certain colleges at the first stage itself, which may prove to be economical.
Judgement Sampling: It is a non-probability sampling procedure. It is also called
purposive sampling, where the researcher selects the sample based on his/her judgment.
The researcher believes that the selected sample elements are representative of the
population. For example, the calculation of consumer price index is based on judgment
sampling. Here the sample consists of a basket of consumer items and other goods
94
and services which are expected to reflect a representative sample. The prices of
these items are collected from selected cities that are viewed as typical cities with
demographic profiles matching the national profile. The advantage of judgment sampling
Measurement of
Variables
is that it is low cost, convenient and quick. The disadvantage is that it does not allow
direct generalisations to population. The quality of the sample depends upon the
judgment of the researcher.
Quota Sampling: In this procedure the population is divided into groups based on
some characteristics such as gender, age, education, religion, income group, etc. A
quota of units from each group is determined. The quota may be either proportional or
non-proportional to the size of the group in the population. Do not confuse the quota
sampling with stratified random sampling discussed earlier. In stratified random sampling
you select random samples from each stratum or group whereas in quota sampling the
researcher/interviewer has a fixed quota and selection of units is based on judgement.
Quota sampling has the advantage that cost and time involved in selection of units is
reduced considerably. However, it has many disadvantages also. In quota sampling,
the samples are selected according to the convenience of the investigator. Therefore,
the selected sample may be biased.
Snowball Sampling: In this procedure you begin by identifying someone who meets
the criteria for inclusion in your study. You then ask them to recommend others who
they may know who also meets the criteria. Although this method would hardly lead to
representative samples, there are times when it may be the best method available.
Snowball sampling is especially useful when you are trying to reach population that are
inaccessible or hard to find. For example, if you are studying the homeless or people
suffering from a particular disease you are may not be able to find good lists respondents.
However, if you identify one or two, you may find that they know others in a similar
condition.
Self Check Exercise
5) What are the important steps in carrying out a sample survey?
6) Explain the procedure of drawing a stratified random sample.
7) Explain the following concepts:
a) Systematic random sampling
b) Parameter and statistic
c) Multistage sampling
Note: i) Write your answers in the space given below.
ii) Check your answers with the answers given at the end of the Unit.
........................................................................................................
........................................................................................................
........................................................................................................
........................................................................................................

6.7 SUMMARY
A variable is a characteristic that we are interested in analysing. It can be nominal,
ordinal or numerical. Numerical variable can be discrete or continuous. While numerical
variable is amenable to mathematical and logical operations nominal and ordinal variables
95
are not. In certain cases it is difficult to measure ordinal variables because we do not
have standard
Tools of measurement. In such cases we need to construct a measurement
for Research
scale. In this we discussed one such scaling techniques, that is summated rating.
Another issue that we covered in this Unit relates to sampling techniques. Because of
inadequate resources or infeasibility we often resort to sampling instead of census of all
the units in the population. There are two broad types of sampling procedures: probability
sampling and non-probability sampling. In probability sampling we have simple random
sampling, systematic random sampling, stratified random sampling and multistage
sampling. On the other hand, non-probability sampling procedures are judgment
sampling, quota sampling and snowball sampling.

6.8 ANSWERS TO SELF CHECK EXERCISES

1) a) Nominal, b) Numerical, c) Numerical, d) Nominal, and e) Nominal
2) Example of the use of Likert Scale: Show your agreement with the following:
Libraries will ever remain important for the advancement of the society
Strongly Agree; Agree; Undecided; Disagree; Strongly Disagree
3) Sampling is advantageous over statistic in that it saves cost and time to be
devoted to the survey. It involves fewer personnel to be deployed and also results
in more precision in the results.
4) Sampling error is due to the faulty sample selected. It may be due to non- probability
sampling techniques adopted. Non- sampling errors are due to the errors in data
measurement or analysis methods.
5) The steps to be followed in carrying out a sample survey are:
specification of objective, preparation of sampling frame, identification of sampling
procedure, determination of sample size, and selection of sampling units.
6) The procedure of drawing a stratified random sample is:
l Divide the population into strata based on some characteristic chosen by you
(example, Post graduate/ Under graduate, Male/Female, etc.)
l Decide the number of units to be taken from each stratum proportional to the
relative size of the stratum and standard deviation of the characteristic within
the stratum. If stratum size is then a large a sub-sample should be taken.
Similarly if you find that variability among units is more in a stratum than other
strata, then you should take a larger sub-sample from that.
l Choose the sub-sample from each stratum using simple random sampling.
7) a) Systematic random sampling is a type of random sampling where the bias is
minimised. Here the random sample is selected in a systematic way, e.g., in
case a sample of 100 is to be selected from a population of 1000, and then
we may select the first member of the population and every subsequent 20th
member.
b) Parameter is a summary value of the population while statistic is that of the
sample.
c) Multistage sampling is sampling done in two or more stages. In case we have
to survey reading habits of users in public libraries, we may first take a random
sample on the basis of geographical regions and then further take a sample
96 on the basis of age groups. This is an example of multistage sampling.
 Measurement of
6.9 KEYWORDS Variables

Convenience Sampling : It refers to the method of obtaining a sample that is

most conveniently available to the researcher.
Judgement Sampling : In this sampling procedure the selection of sample is
based on the researcher’s judgement about some
appropriate characteristics required of the sample units.
Multistage Sampling : The sample selection is done in a number of stages.
Parameter : It is a measure of some characteristic of the population.
Population : It is the entire collection of units of a specified type in a
given place and at a particular point of time.
Quota Sampling : In this sampling procedure the samples are selected on
the basis of some parameters such as age, gender,
geographical region, education, income, religion, etc.
Sample : It is a sub-set of the population. Therefore, it is a
collection of some units from the population.
Simple Random Sampling: This is the basic sampling procedure where all units in
the population have an equal chance of being included
in the sample.
Snowball Sampling : It relies on referrals from initial sampling units to generate
additional sampling units.
Statistic : It is a function of the values of the units that are included
in the sample. The basic purpose of a statistic is to
estimate some parameter.
Stratified Random : In this sampling procedure the population is divided
Sampling into groups called strata. Subsequently sub-samples
are selected from each stratum using a random
sampling method.
Systematic Sampling : In this procedure the units are selected from the
population at uniform interval (in time, order or space).

6.10 REFERENCES AND FURTHER READING

Bhardwaj, R. S. (1999). Business Statistics. New Delhi: Excel Books.
IGNOU Study Material (2005). EEC-13: Elementary Statistical Methods and Survey
Techniques, Block 6.
Kothari, C.R. (1985). Research Methodology: Methods and Techniques. New Delhi:
Wiley Eastern.
Young, P. V. (1988). Scientific Social Surveys and Research, Prentice Hall of India:
New Delhi.

97