Non-Technical Components of Software Development and Their Influence on Success and Failure

1. Welcome Friday, 18 December 2009

2. Non-Technical components of Software Development and their influence on Success and Failure of Software Development By Dr. S.N.Geethalakshmi Reader, Department of Computer Science, Avinashilingam University for Women Coimbatore, India Friday, 18 December 2009

3. Introduction • To have a successful software project, it is essential to identify what constitutes success. • Project success and failure can rarely be described in absolute terms. If failure is no accident then success is also no accident. • Failure and success provide different perspectives on improvement: failure tells what not to do in future, where as success shows what should be done again. • Literature shows that studies on success factors have been predominantly conducted in the Western settings. • Literature reveals that decades of individual and collective efforts by project management researchers since 1960, have not led to discovery of definite set of factors leading to project success and failure. Friday, 18 December 2009

4. Introduction (Contd…) • Project success and failure is a question of perception and that the criteria could vary from project to project. • A project that has been perceived to be a failure by one stakeholder may be perceived as a success by another. • The success or failure of software development consists of two components, namely the technical and non-technical. • The technical issues of software development include those directly related to hardware and software. • Non-technical issues relate to people and process-related components of software development. • Non-technical related components of software development process tend to be under managed. Friday, 18 December 2009

5. Introduction (Contd…) • Therefore, a study was conducted in India among the industries that are into in-house software development to investigate the software development success and failure. • This study investigated the influence of the non-technical factors of software development process, on success and failure of software development. Friday, 18 December 2009

6. Need for the Study • Various studies have been conducted in the industry level and academia in Western settings on various facets of software development. • A review of extant literature shows that studies on success and failure factors have been predominantly conducted in the Western settings. Moreover, each study has yielded results that may lack generalizability and accordingly, no study has come out with a universal pattern of understanding. • In short the studies on success and failure on software projects are inconclusive. • The reasons could be attributed to methodical differences, the culture, and factors pertaining to the work role. • Studies on project success and failure factors in India are at a primitive stage. Friday, 18 December 2009

7. Need for the Study (Contd…) • Previous software engineering studies have suggested a number of non-technical components that contribute to the eventual success and failure of software development. • However, research on the joint occurrence of all the non-technical components of project management is sparse. • Over-management of technical issues and under-management of non-technical people related issues is the problem that software development is facing. • Managing technical issues tends to be more straight forward than managing non-technical, people who come with their unique personalities, strengths, weakness and opinions. • Managing non-technical, people-related components tend to be difficult. As a result non-technical issues more often plague software development than technical problems. Friday, 18 December 2009

8. Need for the Study (Contd…) • Further research on success and failure of software projects developed in-house are sparse. • In-house developed software tend to get deviated from estimates and schedule due to various reasons like; more attention for maintenance and support for already implemented projects, lack of resources, lack of required participation of stakeholders due to their day to day activities and other priorities. • Therefore, a study was conducted in India among the industries that are into in-house software development to investigate the software development success and failure. • This research investigates a list of non-technical components of software development process and, the joint effect of the chosen non-technical components that determines the success and failure of software development. Friday, 18 December 2009

9. Significance of the Study • This study will provide a greater understanding of some of the components of software development process leading to success and failure. • This research seeks to focus management attention on the importance of a number of non-technical components of the software development process. • This study is also intended to enlighten project managers with regard to the importance of practitioners’ overall perception of project success. • An understanding of the importance of software development success and failure will have significant implication for the organization and the software practitioners. • If organizations are to get better at projects, they need to learn from the project experience. They must examine the events of the project, and also the behaviors that may have led to failure or success. Friday, 18 December 2009

10. Research Focus The research focus is : • To study software development success and failure of in-house developed software projects in India. • To investigate the influence of the non-technical factors of software development process, on success and failure of software development. • To investigate the software development success and failure from the perspective of software practitioners and from the perspective of organizations (practitioners’ view). Friday, 18 December 2009

11. Study Variables The non-technical components / factors (study variables) are categorized as follows: • Sponsor/management support and participation • Customer/user support and participation • Requirement management • Estimation and schedule • Project manager and relationship with development staff • Software process management and • Software development personnel Friday, 18 December 2009

12. Objectives of the Study • To measure the percentage of software development success and failure from the perspective of software practitioners and from the organizations’ perspective (practitioners’ view). • To test whether there is any significant association on software development success and failure between the software practitioners’ perspective and the organizations’ perspective. • To predict the software development success and failure by the selected non-technical components from the software practitioners’ perspective. • To predict the software development success and failure by the selected non-technical components from the organizational perspective (practitioners’ view). Friday, 18 December 2009

13. Research Model A model was developed to predict Success and Failure. Success and failure model Customer / User Management Support Requirement Estimation and Schedule Success and failure Project manager / staff Personnel Software process management • The model treats the chosen non-technical components as predictors or independent variables; and success and failure of the software development as the dependent variable or the grouping variable. • This is a predictive model where the chosen non-technical components are tested for prediction of success and failure of software development. Friday, 18 December 2009

14. Research Methodology Instrumentation • For the purpose of studying the objectives, a questionnaire was used as an instrument to collect the data. The questionnaire has two parts; the first part measures the short profile of the respondents and the second part measures the study variables. • The variables chosen for this study are • Management support • Customer/user • Requirement • Estimation and schedule • Project manager/staff, • Software process management and • Personnel. • The items capturing each factor have been adopted from earlier research. Friday, 18 December 2009

15. Validity Test • The questionnaire was subjected to face and content validity. The face and content validity of the items were conducted with 8 experts. The experts also suggested a 5-point rating scale for all the items. The content validity ratio (CVR) was applied to each item, using the formula developed by Lawsche. • All items scored less than 0.5 on the content validity ratio have been removed from the study. Based on the face validity and content validity ratio, the final number of items in each of the factors taking part in this study was decided. • The number of items included in each of the factors is as follows: Management Support : 7 items Customer / User : 8 items Requirements : 8 items Estimation and schedule : 11 items Project manager / staff : 13 items Software process management : 20 items Personnel : 16 items Friday, 18 December 2009

16. Pilot Study • A pilot study was undertaken to assess the reliability of the research instrument. • The total number of respondents for the pilot study was 26 from 10 companies. • The discussion with the respondents during the pilot study prior to administering the questionnaire revealed that, the instrument had adequate stimulus value to gather authentic responses from the respondents. • The data collected from the pilot study was subjected to reliability test using Cronbach Alpha. It was found that the reliability coefficients for the chosen dimensions are more than 0.6, which is an acceptable value. So, the items constituting each variable under study have reasonable internal consistency. Friday, 18 December 2009

17. Sampling Frame • The geographical area of Coimbatore city was chosen as the Universe. • The main reason for choosing Coimbatore city is that the investigator is located here and is familiar with the place. • Familiarity is found to be essential for gaining accessibility to the respondents as well to solicit genuine participation by the respondents. • More significantly, Coimbatore city has the distinction of being an active commercial centre with companies such as PRICOL, LMW, Roots Industries and ELGI group of companies. Friday, 18 December 2009

18. Administration and justification of the sample • A list of companies having an in-house software development department was prepared. A total of 41 companies and 140 software practitioners were identified. The questionnaire was administered to all the 140 practitioners identified. • A thorough follow-up was done in person and over telephone to expedite the process of filling up the questionnaire. • Response rate was 71.42% (100 usable questionnaires). • The final sample size is of considerable size when compared to some relevant prior studies. • The industry sectors of respondents organizations are manufacturing sector, Textile & Sugar mills and hospitals. Friday, 18 December 2009

19. Testing the objectives (Contd…) Objective 1: To measure the percentage of software development success and failure from the perspective of software practitioners and from the organizations’ perspective (practitioners’ view). • To study this objective percentage analysis was done. • It has been found that according to the organization, 36% of the software development projects have failed and 64% of the projects are successful. • According to the individual’s perspective, 24% of the development projects have failed and 76% of the projects are successful. Friday, 18 December 2009

20. Objective 2: To test whether there is any significant association on software development success and failure between the software practitioners’ perspective and the organizations’ perspective. • McNemar test was conducted to find the association between the individual’s perspective and the organizational perspective, since the same respondents have given their opinion on success and failure from the individual and organizational perspectives. • It has been found that there is significant difference between the individual and the organizational perspective. Friday, 18 December 2009

21. Objective 3: To predict the software development success and failure by the selected non-technical components from the software practitioners’ perspective (Contd…). • Stepwise Discriminant Analysis (DA) is performed to predict the success and failure of software development by the study variables from the perspective software practitioners. • The stepwise DA resulted in a 3 – step discriminant model. • Since stepwise discriminant analysis is performed, Mahalanobis D² (Min D²) is used to evaluate the statistical significance of the discriminatory power and to determine the variable with the greatest power of discrimination. Friday, 18 December 2009

22. Objective 3: To predict the software development success and failure by the selected non-technical components from the software practitioners’ perspective (Contd…). Summary of Study variables entered the model * significant at 0.05 level • The above table gives a summary of the 3 steps involved in the discriminant analysis. The variables customer/user, software process management, and estimation and schedule entered the discriminant function. • Discrimination increased with the addition of each variable, achieving by the third step a substantial ability to discriminate between the groups. This is indicated by the Mahalanobis D² value which is significant. Friday, 18 December 2009

23. Objective 3: To predict the software development success and failure by the selected non-technical components from the software practitioners’ perspective (Contd…). • The overall results are also found to be statistically significant and continue to improve in discrimination as evidenced by the decrease in Wilks’ Lambda value (from 0.664 to 0.462). • As two groups were involved, the model produced one discriminant function. The discriminant function is significant displaying a canonical correlation of 0.75. • The function is statistically significant as measured by the Chi-Square statistic = 74.62, and that the function accounts for 100% of the variance explained. • The total amount of variance explained by the function (non-technical factors) in the dependent variable is 57.61%. Friday, 18 December 2009

24. Objective 3: To predict the software development success and failure by the selected non-technical components from the software practitioners’ perspective (Contd…). To assess the contributions of the seven predictors, the structure matrix was examined, which is indicative of each variable’s discriminating power Structure Matrix * Pooled within-groups correlations between discriminating variables and standardized canonical discriminant functions variables ordered by absolute size of correlation within function. ** This variable not used in the analysis. Friday, 18 December 2009

25. Objective 3: To predict the software development success and failure by the selected non-technical components from the software practitioners’ perspective (Contd…). The predictive accuracy is assessed using the classification matrix. The Table below shows classification. The hit ratio for the analysis is 92.0%. Classification Results 92% of original grouped cases correctly classified. • The final measure of classification accuracy is Press’ Q calculated to test the statistical significance that the classification accuracy is better than chance. • The Press’ Q statistic calculated is compared to the critical value based on Chi-square distribution. The calculated value is more than the critical value at a significance level of 0.05. Therefore, the classification results are significantly better than that which would be expected by chance. Friday, 18 December 2009

26. Objective 4: To predict the software development success and failure by the selected non-technical components from the organizational perspective (practitioners’ view). • Stepwise Discriminant Analysis is performed to predict the success and failure of software development by the study variables from the organization’s perspective. • The stepwise DA resulted in a 3 – step discriminant model. • Since stepwise discriminant analysis is performed, Mahalanobis D² is used to evaluate the statistical significance of the discriminatory power and to determine the variable with the greatest power of discrimination. Friday, 18 December 2009

27. Objective 4: To predict the software development success and failure by the selected non-technical components from the organizational perspective (practitioners’ view). Summary of Study variables Entered * significant at 0.05 level • The above table gives a summary of the 3 steps involved in the discriminant analysis. The variables customer/user, Project Manager / Staff, andsoftware process management entered the discriminant function. • Discrimination increased with the addition of each variable, achieving by the third step a substantial ability to discriminate between the groups. This is indicated by the Mahalanobis D² value which is significant. Friday, 18 December 2009

28. Objective 4: To predict the software development success and failure by the selected non-technical components from the organizational perspective (practitioners’ view). (Contd…). • The overall results are also found to be statistically significant and continue to improve in discrimination as evidenced by the decrease in Wilks’ Lambda value (from 0.55 to 0.42). • As two groups were involved, the model produced one discriminant function. The discriminant function is significant displaying a canonical correlation of 0.76. • The function is statistically significant as measured by the Chi-Square statistic = 84.62, and that the function accounts for 100% of the variance explained. • The total amount of variance explained by the function (non-technical factors) in the dependent variable is 58%. Friday, 18 December 2009

29. Objective 4: To predict the software development success and failure by the selected non-technical components from the organizational perspective (practitioners’ view). To assess the contributions of the seven predictors, the structure matrix was examined, which is indicative of each variable’s discriminating power Structure Matrix * Pooled within-groups correlations between discriminating variables and standardized canonical discriminant functions variables ordered by absolute size of correlation within function. ** This variable not used in the analysis. Friday, 18 December 2009

30. Objective 4: To predict the software development success and failure by the selected non-technical components from the organizational perspective (practitioners’ view). (Contd…). The predictive accuracy is assessed using the classification matrix. The Table below shows classification. The hit ratio for the analysis is 91% Classification Results 91% of original grouped cases correctly classified. • The final measure of classification accuracy is Press’ Q calculated to test the statistical significance that the classification accuracy is better than chance. • The Press’ Q statistic calculated is compared to the critical value based on Chi-square distribution. The calculated value is more than the critical value at a significance level of 0.05. Therefore, the classification results are significantly better than that which would be expected by chance. Friday, 18 December 2009

31. Findings from this Research • It is found that Customer/user involvement contributes most to project success and failure from both practitioners’ and organizations’ perspective. • Practitioners perceive the next important factors that contribute to project success and failure are: • Software process management • Estimation and schedule • On the other hand, organizations perceive (practitioners’ view) the next important factors that contribute to project success and failure are: • Project manager/staff • Software process management. • It is found that there is a difference between practitioners’ perception of project success and failure and their perceptions of howmanagement views project success and failure. Friday, 18 December 2009

32. Implication • The results of this study will help project managers and other project stakeholders to predict the likelihood of project success, in evaluating their on-going projects, and improve managerial decision-making, as lessons learnt are applied to other software development projects. • Findings regarding the failure will help the organizations and the software practitioners to take corrective measures and march towards successful software development. • Project managers can use the findings of this study to increase the probability that the projects they manage will be successful, both from organizational and practitioner perspective. Friday, 18 December 2009

33. Conclusion The success and failure of software project is not only a unique pattern in Western countries but also it pertains to countries like India too. Given the cultural differences in attitudes, values, and behaviors towards work, this study enables to see the pattern that is emerging in industrializing and economically progressing countries like India. The study is one among the pioneer research gleaned from several success and failure literatures, providing insight into the importance of the non-technical factors in understanding the software development success and failure. To conclude, this study gives cue to organizations and practitioners on the factors that determine the success and failure of software development. Friday, 18 December 2009

34. Thank You Friday, 18 December 2009