See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/324755571

Factors Influencing Productivity of Agile Software Development Teamwork: A

Qualitative System Dynamics Approach

Conference Paper · April 2018

DOI: 10.1109/APSEC.2017.95

CITATIONS READS

9 1,128

2 authors:

Israt Fatema Kazi Sakib

University of Dhaka University of Dhaka
3 PUBLICATIONS   11 CITATIONS    86 PUBLICATIONS   244 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Agile Teamwork Productivity View project

Mobile Application Testing Lab View project

All content following this page was uploaded by Israt Fatema on 25 April 2018.

The user has requested enhancement of the downloaded file.

 Factors Influencing Productivity of Agile Software Development Teamwork: A
Qualitative System Dynamics Approach

Israt Fatema Kazi Sakib

Institute of Information Technology Institute of Information Technology
University of Dhaka University of Dhaka
Dhaka, Bangladesh Dhaka, Bangladesh
email: [email protected] email: [email protected]

Abstract— Agile method emphasizes on the people factors and is a function of various controllable and uncontrollable
strength of teamwork that simplify the development process. A factors [2]. The relationships between some of these factors
highly productive team throughout an agile software and productivity may change under new software
development process is very instrumental in achieving project engineering practice and culture [3]. The factors change
success. Consequently, understanding of how individual over time as expectations change. In addition, actual
behaviour and productivity are affected by teamwork within productivity measurement becomes more difficult when
an agile team becomes critical. Identifying factors that impact agile software developers perform knowledge-related tasks
productivity will result in improvement of teamwork. Hence, a (e.g., creating, storing, sorting, retrieving, applying and
need emerges to recognise the significant ones. Doing so will
acquiring knowledge) where the product is usually
enable project team management to determine the areas where
to concentrate efforts in order to improve productivity. The
intangible, rarely has single way of doing it, and it is
objective of this research is to identify and analyse agile difficult to quantify [2]. Since knowledge is complex and
teamwork productivity influence factors by using system hard to evaluate, it is difficult to interpret the productivity of
dynamics (SD) approach. Identification of main factors the agile team member’s simply by source line of code
influencing productivity and how they impact agile teamwork (SLOC) or function points produced per unit of time/cost
are carried out through interviews, survey and literature [3].
review. From the perspective of agile team members, the four Despite the increasing acceptance of the agile methods,
most perceived factors impacting on their productivity are insufficient research has been empirical on the effect of
team effectiveness, team management, motivation and influence factors of software development productivity [5].
customer satisfaction. Lack of agile team management support A better knowledge of the factors and the mediators that
is found to be the most mentioned reason for failed agile influence agile teamwork productivity could help determine
project. The complex interrelated structure of different factors where to focus management efforts to improve productivity.
affecting agile teamwork productivity is modeled using Since the agile project team is the most dynamic element
influence diagram and Causal Loop Diagram (CLD) for in the software development sector, improving team
qualitative analysis. productivity has become a target for software companies in
everywhere. The aim of this study is to identify and
Keywords- agile teamwork; productivity; influence factors;
understand the complex interdependences and underlying
qualitative system dynamics.
structures at the team’s perception level, which influence
agile teamwork productivity over time. To achieve this goal,
I. INTRODUCTION this paper determines the major factors impacting teamwork
A highly productive team is the most important factor in productivity in Bangladeshi software companies through a
achieving project success at different stages of an agile survey and interviews that have been conducted with agile
software development. For efficient management and a teams and apply SD to model teamwork productivity. The
better control over the agile project team, it is important to major factors are to be modelled using a CLD based on the
understand the team dynamics and effects related to agile Authors’ earlier work in [11]. This CLD will be used to
practices that influence the development team’s examine the internal dynamics existed within the team and
productivity. the organizational resources that are used to support them.
Research has been largely carried out to identify The scope of this empirical finding considers the
productivity influence factors in traditional software Bangladeshi software companies as a case study, which can
development. There are four main factors generally in turn make the research results beneficial to these
discussed [1]: the product being developed (characterization companies.
of the specific software), people (team members, The remainder of this paper is organized as follows.
capabilities, experience, and motivation), project Section II includes a literature review, section III presents
(management and resourcing) and processes (tools and the research method and design. Section IV describes the
software methods). However, agile teamwork productivity survey results and Section V explores the structure of the
qualitative SD model. Section VI describes some limitations methodology has been applied by many researchers
of this work. Finally, Section VII describes the conclusion [19][20][24][25] for studying and managing complex
and future work. feedback system, where feedback is understood as a closed
sequence of causal relationships. The concept of a feedback
II. LITERATURE REVIEW loop reveals that any actor in a system will eventually be
There are several studies that attempted to assess the affected by its own action.
impact of some of the influencing factors on agile teamwork A number of diagramming tools are used in SD to
productivity. Only Melo et.al [2] analysed the major factors capture the structure of systems, including causal/influence
influencing agile teamwork productivity using the team’s diagrams, stock and flows. Each causal link is assigned a
perception as one potential dimension and found that team polarity, symbolized by ‘+’ indicating the two related
management is the most influencing factor on agile team variables change in the same direction, or ‘-’ showing the
productivity. There are few researches [6][7][8] that two linked variables vary in two different directions [18].
attempted to evaluate the impact of some of the influencing A. Identification of different factors affecting agile
factors on productivity separately using SD. However, the
teamwork productivity
complex inter-related structure of all the major factors
effecting the teamwork productivity was not considered by Data collection: The model developed for this work is
the previous works. Abdel-Ahmed [7] investigated the based on data collected from the software companies in
effect of various management policies on development cycle Bangladesh. Identification of the factors was initially carried
time, quality and effort. His works however adopt the out through an intensive literature review. A set of semi-
waterfall method which limits their applicability in recent structured interviews and face-to-face discussions were also
software project and more importantly, does not focus on conducted with 12 key project members from 4 software
the agile principles. companies including project managers, scrum masters,
In addition, evaluation of individual productivity may developers, project owners, and considering also different
not affect the productivity of other developers [9]. These experience profiles.
assumptions provide motivation of study team, not Using the factors identified in this first step, a
individuals. questionnaire [26] was developed and distributed to a total
Melo et al. used the ‘Input Process Output’ model to of 25 software companies in Bangladesh. Only 17
identify team productivity factors in a multiple case studies. companies (1 from each) responded to the questionnaire.
Dingsoyr et al. [16] described agile software development as The criteria for company selection of this preliminary study
a sociotechnical system comprised of human (socio) and were: companies using agile methods for at least 1 year and
technical entities. Technological interventions do not top listed companies developing software for both offshore
increase sociotechnical system effectiveness if they are not and local market.
supported by social components of the system. Such team Data were collected throughout a period of 3 months in
interactions are one of the important parts in agile software 2017 (January-March). In order to ensure the quality of data,
development. Thus, recent focus on agile software team members were all self-selected by their organization
development has increased interest in analysing self- based on their work roles as members of existing agile
managing agile teams and how to effectively make team teams.
productive [16]. Throughout the literature review, it has The returned forms were then analysed to identify major
been observed that there is a lack of well-established influence factors. Currently, more software companies are
dynamic theory about agile teamwork. This study seeks to being requested to participate in this survey, as the plan is to
fill this gap by developing an integrated model, which collect more than 100 responses from different agile teams.
represents the inter-related structure of productivity B. Selection of factors for inclusion in the model
influence factors and how they impact agile teamwork’s
productivity. In order to do so, this study applies a SD Data analysis: The important factors identified in
approach, which can study complex system by exploring literature and interviews were taken as a starting point for
underlying relationships and connections between the the system approach in this research. Initially 38 factors
components of a system that normally are not discovered by were chosen for preliminary analysis. In order to create a
the input-output-process type of models used in system model, it is required to determine the importance of
organizational studies. the individual factors, their correlation with one another and
their effects on productivity itself. The agile team members
were asked to fill in the form to indicate the strength (high,
III. RESEARCH METHODOLOGY medium or low) of the factors that they perceived influenced
their productivity [25].
The methodological approach of this research is based
The procedure followed to extract the agile team
on the SD, as a modelling and simulation methodology
member’s perception of the productivity influence factors:
enables to model complex system considering all the
1. Convert the qualitative scale to a quantitative one.
influencing factors [19] over time. SD modelling has been
The qualitative scale of high, medium or low was converted
chosen for this research because it provides a systematic
to a number scale of 3, 2, and 1, respectively.
method for description, exploration and analysis about the
dynamic behavior of complex systems [18]. SD
 2. Find the total score of each factor for frequency agile team is a difficult job for the software companies in
analysis. Then, the arithmetic mean of the total counts was Bangladesh as Table I and Fig. 4 show.
calculated as to eliminate the factors below the average
(Table. 1) mean. TABLE I. ARITHMATIC MEAN OF QUESTIONNARIE RESULTS FROM
FREQUENCY ANALYSIS
3. Cronbach’s alpha coefficient for internal
consistency reliability [21] was calculated for the selected Me- Me-
factors (Table I, highlighted factors). SL Factor an SL Factor an
1 Culture 2.23 20 What is the 1.82
4. From step 2, nineteen factors (Table I, highlighted)
staff turnover
were selected as the most influential ones. Reliability rate in the
coefficient for 19 factors was .867 (alpha), which indicates project
good internal consistency of the factors in the scale [21]. 2 Staffing 2.76 21 Reuse 2.17
3 Size of team 2.29 22 What is the 2.00
IV. SURVEY RESULTS software
reuse level in
Characteristics of the sample software companies can be the project
found in Table II. Fig. 1 presents the agile practices adopted 4 Project 2.23 23 Goals 2.29
by the participating software companies and it shows daily complexity
stand up meeting mostly used by all of them. Fig. 2 shows 5 Team 2.52 Intra group 1.94
Leadership 24 wage
that lack of management support is the main reason for
inequality
failure in agile projects. In most of the interviews, the team 6 25
Mutual 2.41 Team 2.17
members mentioned that team management have their own performance measurement
way of measuring productivity. Although at the end of the monitoring
project, the management assessed their productivity on the 7 Backup 2.41 26 Self- 2.17
basis of timeliness and quality. At the same time, 10 Behaviour management
interviewees also mentioned customer satisfaction as a 8 Team 2.52 27 Task variety 2.41
orientation and
criterion [Fig. 3]. Customer satisfaction is very important to
Innovation
software development companies in Bangladesh as a rising 9 Adaptability 2.35 28 External 2.17
market for outsourced software destination. According the Dependencies
product owner interview, dealing with cultural differences 10 Feedback 2.70 29 Tools usage 2.29
among offshore organisation influences teamwork 11 Mutual trust 2.76 30 Programming 2.05
productivity. Sometimes it becomes difficult to keep contact language
12 Coordination 2.70 31 Schedule 2.29
with the offshore client on urgent issues due to time
pressure
difference between places. Moreover, offshore client’s 13 Communication 2.82 32 Impact of 2.11
expectations are different, both in terms of their general Pair
culture and their views on life and work. programming
Five interviewees (project leads and managers) also on
mentioned that culture is a big barrier for working in an productivity
14 Staff are 1.76 33 Resource 2.41
agile team. This factor affects communication between team
appreciated for constraints
members. In addition, sometimes language barrier hinders working long
communication. Transitioning from individual work to self- hours
management team requires a reorientation not only by 15 Staff are 2.11 34 Project 2.58
developers but also by management. This changeover takes rewarded (then Management
time and resources. For this reason, these project managers or later) for
prefer fresher as a team member. Their software companies working long
hours
like to groom up with training than changing mind set up of 16 Adequate 2.41 35 Motivation 2.58
the team members. Besides that, self-management and technical
adaptability are considered key for agile development. But training for
these two factors have less influence (Table. 1) on agile team
teamwork productivity and mostly depend on well project 17 Adequate team 2.35 36 External 2.41
skills training project
management.
for team factors
Fig. 4 provides highlights of the most influencing 18 Team member 1.64 37 Dealing with 2.17
productivity factors that are perceived by the agile team turnover cultural
members. This study results show that the effectiveness of differences
an agile team lies in the interrelations of the identified among
factors such as communication, coordination, adaptability, offshore
organizations
feedback, leadership and self-management. Project manager 19 Key personnel 38 Working
2.23 2.35
is usually a technical lead and top-level management makes stayed environment
many management decisions since the majority of the throughout the
projects are outsourced projects. Getting the right person project
selected (staffing) with suitable skills and knowledge for an
TABLE II. CHARACTERISTICS OF PARTICIPATING SOFTWARE COMPANIES
Characteristic Category Number %
Main team Development project 10 58.82
assignment
Maintenance project 7 41.17
Team role Project manager 4 23.52
Developer 6 35.29
Software engineer 3 17.65
Team lead 2 11.77
Quality assurance engineer 2 11.77
Experience in 1-2 years 8 47.8
agile practice
2-5 years 7 41.2
More than 5 years 2 11.8
Development Scrum 17 100
method
Size of the 30-50 2 12
company 50-100 1 6
(person) 100-150 5 29
150-200 6 35
200-250 1 6
250-300 1 6
More than 300 1 6

Figure 4. Agile team perceived productivity influence

V. QUALITATIVE MODELING OF AGILE SOFTWARE

DEVELOPMENT TEAMWORK PRODUCTIVITY

To fulfill the research aim, an empirical qualitative

system dynamics model developed by [6] will be adapted
and used as a main reference. Fig. 6 presents the overall
conceptual model (main factors) of agile teamwork
productivity. It shows all the main influence factor’s affect
found in this study. Distinct from previous studies [7][24]
this model represents the team dynamics, which is a
Figure 1. Agile practices adopted in software companies collection of “soft factors” [23] and effects related to agile
methods that influence the teamwork’s productivity. The
soft factors that can affect the software development
teamwork productivity include motivation, team
management efficiency, customer satisfaction, skillfulness
and team effectiveness. Teams require a complex mixture of
factors that include organizational support, individual skills
and also teamwork skills [10] to work effectively.
Productivity of an organization is not only an outcome of
the skillfulness in a team, but dependent on team
effectiveness, the ways in which individuals feel and how
they perceive their team members [21]. Ignoring team in the
software development process or managing them in an
Figure 2. Main reasons for failure in agile projects inappropriate way can have a high impact on their
productivity and team effectiveness [22].
Within the model (see Fig. 5), it is shown team-learning
processes can improve team effectiveness. When a new
developer joins a development team, the better the
communication process works, the faster the new developer
can become productive and the learning curve reduced.
These learning activities are likely to create a positive
change and to influence the productivity.
Figure 3. Criterion for measuring or perceiving productivity
 The developed CLD (Fig. 5) is based on Actual team VI. LIMITATIONS OF THE STUDY
productivity = potential productivity – losses due to faulty There are a number of limitations to this study. First, this
processes [7]. Actual Team productivity (Fig. 5) is a stock study was limited to 17 survey responses and 12
representing the level of productivity the team can achieve interviewees within 17 software companies. It was difficult
over the time and it affects software development rate to gain access to more software companies due to time
directly. Actual team productivity refers to how the team constraint. Within the interviewees, they were carefully
performs (the results they get and the level of effort they put chosen from different roles on the agile team in order to get
in). Potential productivity refers to the maximum level of different perspectives of productivity in the context of
productivity the team could produce based on the individual Bangladesh software Industry.
or team skill and ability of each member in the team and the Another limitation of this study is the agile team members’
resources available. Losses due to faulty processes relate to perceptions used as a response. However, with survey, this
the issues that can get in the way of team productivity, study relies on what the respondents provided to the
preventing the team from reaching its potential productivity. researcher. The questionnaire used for this study had been
Losses are normally due to two main areas: motivational used successfully in other research [7][20] and was
faults/losses and communication and coordination overhead. developed after a detailed literature review. Some of the
There are various factors affecting agile teamwork questions were included in the survey after getting
productivity positively including skillfulness, motivation, knowledge about the working condition of software
and project management efficiency. There are some other companies in Bangladesh from the interview sessions.
factors affecting teamwork productivity negatively Finally, the CLD is not complete because it only focuses
including exhaustion, external dependency, culture and on the influence factors. The multiple feedback processes
resource constrains [Fig. 6]. and delays are not incorporated in this model.
Fig. 6 illustrates that motivation is positively related to
team effectiveness. In consequence, motivated team
involves with team learning processes over time and can be
increased team effectiveness.
CAUSAL LOOP DIAGRAM OF MAIN PRODUCTIVITY INFLUENCE FACTORS

Experienced
workforce Eft of workfore + Eft of
Communication Total workforce
experienced
& Coordination
Newly hired Eft of
+ +
workforce + Schedule pressure
Potential + Learning Communication &
Productivity + coordination overhead
Learning +
activities
+ -
Software + - Exhaustion
development rate + Actual Team Team
+ Effectivenes Motivation +
Productivity + Working
+ +
- + + Enviornment
Task Task Skillfulness
Eft of Team Reward
Backlog completed
+ + Management
+ + Need for Team +
Training Team Mgt
Rework Perceived -
- Performance+ performance Customer Satisfaction Efficiency
Goal
+ -
Work Quality

Figure 5. Causal loop diagram of Agile teamwork productivity

 [5] C. D. O. Melo, D. S. Cruzes, F. Kon, and R. Conradi, “Agile team
perceptions of productivity factors,” In Agile Conference (AGILE),
IEEE, 2011, pp. 57-66.
[6] X. Kong, L. Liu, and D. Lowe, “Modeling an agile web maintenance
process using system dynamics,” In 11th ANZSYS/Managing the
Complex V conference, ISCE Publishing, Christchurch, NZ. Dec.
2005.
[7] T.K. Abdel‐Hamid and S. Madnick, “Software productivity: potential,
actual, and perceived,”. System Dynamics Review, 5(2), pp. 93-113,
June. 1989.
[8] J. M. Lyneis and D. N. Ford, “System dynamics applied to project
management: a survey, assessment, and directions for future
Figure 6. Main factors affecting Agile teamwork productiviy research,” System Dynamics Review, vol. 23, no. 2‐3, pp. 157-189,
Jun. 2007.
[9] C.O. Melo, “Productivity of agile teams: an empirical evaluation of
VII. CONCLUSION AND FUTURE WORK factors and monitoring processes,” Ph.D. dissertation, Universidade
de São Paulo, 2015.
Teamwork productivity determines the overall project
[10] T. Dingsøyr and Y. Lindsjørn, “Team performance in agile
performance in an agile software development process. development teams: findings from 18 focus groups,” International
Therefore, it has gain more interest to study team member’s Conference on Agile Software Development, Springer Berlin
productivity. Agile team members should be taught to Heidelberg, June. 2013, pp. 46-60.
interpret and manage productivity factors regularly as they [11] I. Fatema and K. M. Sakib, “Analyse Agile Software Development
are self-managed. Productivity improvement programs Teamwork Productivity using Qualitative System Dynamics
would become effective only if all the variables are Approach,” The Twelfth International Conference on Software
Engineering Advances, 2017.
simultaneously controlled and monitored. Researchers have
tried to quantify and measure how soft factors and social [12] T. Dingsøyr and T. Dybå, “Team effectiveness in software
development: Human and cooperative aspects in team effectiveness
aspects affect teamwork productivity. Agile Software models and priorities for future studies,” Proceedings of the 5th
development process must be analyzed as a behavioral International Workshop on Co-operative and Human Aspects of
process [22]. Therefore, coordinating and managing an agile Software Engineering. IEEE Press, 2012, pp. 27-29
team is a vital activity for software companies and team [13] W. Scacchi, “Understanding and improving Software Productivity,”
dynamics have a direct influence on teamwork productivity. Advances in Software engineering and Knowledge engineering, 2005.
One effective solution to improve productivity is to look [14] F. Nasirzadeh and P. Nojedehi, “Dynamic modelling of labour
into the factors influencing productivity and also have a productivity in construction projects,” International Journal of Project
Management, vol. 31, no. 6, Aug. 2013, pp. 903-911.
dynamic strategical model that tells the project manager in
[15] A. Rodrigues and J. Bowers, “The role of system dynamics in project
advance the degree of impact that these factors will have on management,” International Journal of Project Management, vol. 14,
team productivity. In order to achieve that, the main factors no. 4, Aug. 1996, pp. 213-220.
that affect teamwork productivity are determined. [16] B. Barry, “Centre for Systems and Software Engineering,” Oct. 2012.
The validation of the conceptual model against a real- [Online] Available
world agile software development project will be presented http://sunset.usc.edu/csse/research/COCOMOII/cocomo_main.html.
in future research. Furthermore, this qualitative CLD will be [retrieved: August, 2017].
used as a basis for a stock and flow model development of [17] V. Lalsing, S. Kishnah, and P. Sameerchand, “People factors in agile
software development and project management,” International
the quantitative SD method. Journal of Software Engineering & Applications, vol. 3, pp. 117,
Jan.2012.
ACKNOWLEDGMENT
[18] L. L. R. Rodrigues, N. Dharmaraj, and B. R. Shrinivasa Rao, “System
This research is supported by ICT Division, Ministry of dynamics approach for change management in new product
Posts, Telecommunications and Information Technology, development,” Management Research News, vol. 29, no. 6, Aug.
2006, pp. 512-523.
Bangladesh. 56.00.0000.028.33.079.17-223, 20.06.2017.
[19] M. J. Mawdesley and S. Al-Jibouri, “Modelling construction project
REFERENCES productivity using systems dynamics approach,” International Journal
of Productivity and Performance Management, vol. 59, no.1, Dec.
[1] A. Trendowicz and J. Münch, “Factors Influencing Software 2009, pp. 18-36.
Development Productivity—State of the Art and Industrial
Experiences,” Advances in computers, vol. 77, pp. 185-241, Dec. 2009. [20] I. Fatema, “Agile teamwork productivity influence factors,” Jan.
2017. [Online] Available https://goo.gl/forms/I5xGdQGqFMk9he5f2.
[2] C. D. O. Melo, D. S. Cruzes, F. Kon, and R. Conradi, “Interpretative [retrieved: August, 2017].
case studies on agile team productivity and management,”
Information and Software Technology, vol. 55, pp.412-427, [21] M. Ortu, B. Adma, and G. Destefanis, P. Tourani, “Are Bullies more
Feb.2013. Productive? Empirical Study of Affectiveness vs. Issue Fixing Time,”
Proceedings of the 12th Working Conference on Mining Software
[3] K. Petersen, “Measuring and predicting software productivity: A Repositories, IEEE Press, May. 2015, pp. 303-313.
systematic map and review,” Information and Software Technology,
vol. 53, pp.317-343, Apr.2011. [22] G. Destefanis, M. Ortu, S. Counsell, S. Swift, M. Marchesi, and R.
Tonelli, “Software development: do good manners matter?,” PeerJ
[4] Y. Ramírez and D. Nembhard, “Measuring knowledge worker Computer Science, 2016.
productivity: A taxonomy,” Journal of Intellectual Capital, vol. 5, no.
4, Dec. 2004, pp. 602–628.

View publication stats