International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013), pp.221-236 http://dx.doi.org/10.14257/ijseia.2013.7.6.19

Factors Influencing Software Development Process: A statistical Outlook Madhumita Singha Neogi1*, Soubhik Chakraborty2 and Vandana Bhattacherjee3 1

Departmen. of Information Technology, Xavier Institute of Social Service, Ranchi-834001, India 2 Department. of Applied Mathematics, Birla Institute of Technology, Mesra, Ranchi-835215, India 3 Department of Computer Science & Engineering, Birla Institute of Technology, Lalpur, Ranchi-834001, India *corresponding author’s email: [email protected] Abstract While dealing with students in solving their assignments and projects of software development we have experienced that students are ready to adopt agile process model because it is very close to their nature. It is assumed that certain concepts were already prevalent in student fraternity. The factors considered here are ‘quality of code’, ‘preference of group work’, ‘regular interaction with teammate’, ‘responding to changes’, ‘early testing’, and ‘fast delivery of working software’. Presence of linear relationship, if any, was tested by testing the significance of correlation coefficient r using t-test while non-linearity was tested by testing the significance of correlation ratio (η) using F-test. The overall results show that there is strong non-linear relationship among the factors. Keywords: Agile, relationship, correlation-ratio, t-test, f-test

1. Introduction Agility, swiftness only can cope up with this fast changing scenario of software industry. Since 2001 when Kent Beck et al. signed the “Manifesto for Agile Software Development”, a lot has been talked about agile process models where we have been experiencing the change from traditional process model to agile process model. While dealing with students in solving their assignments and projects of software development we have experienced that students are ready to adopt agile process model because it is very close to their nature. There are certain factors which were preferred by the students and agile manifesto also suggests giving emphasis to these factors. These factors are drawn from our previous study [11] and Abrahamson’s report [8] and [9]. We concluded that certain concepts were already prevalent in student fraternity. In this paper we have shown that there are few factors which influence software development pattern of agile type. Our objective was to know whether or not there are any relationships--linear or non-linear--existing among the factors and to measure the degree of such relationships, if any, among those factors. The factors considered here are quality of code, preference of group work, regular interaction with teammate, responding to changes, early testing, and fast delivery of working software. Earlier we have investigated the co relationship among factors, namely, ‘quality of code’, ‘preference of group work’, ‘regular interaction with teammate’, ‘responding to changes’, ‘early testing’, and ‘fast delivery of working software’ in small data set [12, 15]. In continuation with our previous work, here we

ISSN: 1738-9984 IJSEIA Copyright ⓒ 2013 SERSC

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

have tested the presence of both linear and non linear relationships using appropriate statistical measures.

2. Related Work This section presents the papers that have discussed about various studies related to factors which affect software development techniques. Similar to our work another study was conducted during senior student software development projects in which they had applied agile principles working with industry clients under minimal supervision. Those results indicated a strong relationship between student perceptions of value delivered to the client and their perception of the value of utilizing agile principles to achieve their goal. [13]. Germain and Robillard suggested that in-process monitoring is required to control the process activities. In-process monitoring is likely to provide causal information between the actual process activities and the quality of the implemented components. They have mentioned that traditional software engineering processes are composed of practices defined by roles, activities and artifacts. Software developers have their own understanding of practices and their own ways of implementing them, which could result in variations in software development practices [3]. Agile software development promotes feedback, discipline and close collaboration between all members of the development team. In one of the research papers the focus was on functionality of two key physical artifacts’ the story card and the Wall - which, individually and in combination, underpin the team's activity. These artifacts’ have two main roles - one which enables a shared understanding of requirements and one which facilitates the development process itself. In one of the paper it shows how the two perspectives - the notational and the social - intertwine and are mutually supportive. Any attempt to replace these physical artifacts’ with alternative support for an agile team needs to take account of both perspectives, and the complex relationships between them [6]. Melnik and Maurer conducted a research on using agile methods in software engineering education. They explored the perceptions of students from five different academic levels of agile practices. Information was gathered through the collection of quantitative and qualitative data over three academic years, and analysis revealed student experiences, mainly positive but also some negative. Student opinions indicated the preference to continue to use agile practices at the workplace if allowed [14]. Abbas et al., have conducted a study on a set of data to see the effectiveness of 58 different agile practices using factor analysis. The analysis extracted 15 factors; each was associated with a list of practices. These factors with the associated practices can be used as a guide for agile process improvement. They have also calculated correlations between the extracted factors and the significant correlation findings suggested that people who applied iterative and incremental development and quality assurance practices had a high success rate [1]. In an industry setting it has been found that there are strong correlations between the factors of open source software components reuse and software development economics. Software organizations can achieve some economic gains in terms of software development productivity and product quality if they implement open source software components reuse adoption in a systematic way [2]. In another study which was conducted using an unprecedentedly large-scale survey-based methodology, consisting of respondents who practiced ASD and who had the experience of practicing plan-driven software development in the past. This study indicated that nine of the 14 hypothesized factors have statistically significant relationship with “Success”. The important success related factors that were found are: customer satisfaction, customer

222

Copyright ⓒ 2013 SERSC

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

collaboration, customer commitment, decision time, corporate culture, control, personal characteristics, societal culture, and training and learning [4]. In an industrial setting a study was conducted in which the empirical comparison of accuracy of fault prediction offered by statistical prediction models with the accuracy of expert estimations [7]. There are other papers which studied on factors that impact implementing a system development methodology. They present the findings of empirical research from 61 companies, mostly from the United States, to identify the factors that may impact implementation of a system development methodology (SDM) [5]. But these papers only give some ideas of conducting empirical research in this area and have explored or tested statistically the presence of linear or non-linear relationship in a systematic manner. So our paper makes amends for this lapse.

3. Methodology A structured interview questionnaire was prepared in English. We have considered nine questions for this study (Appendix-1). Each question was measured in a Likert type 3 time scale – (i) Yes, (ii) Neutral, (iii) No. We have measured them as 1, 0, -1 respectively. Our objective was to measure the degree of relationship, if any, among these variables among the few factors that influence software development process. Earlier we have investigated the co relationship among factors quality of code, preference of group work, regular interaction with teammate, responding to changes, early testing, and fast delivery of working software in small data set [12, 15]. In continuation with this, here we have shown the significance of correlation coefficient depicting linear relationship using t-test in some cases while in others we have shown that the relationships are non-linear using correlation ratio test i.e., by calculating eta and proving its significance of using F-test. We framed a set of research questions which were investigated by obtaining relevant measures as defined in the next section. Research questions led to the framing of a set of hypotheses to be evaluated. Following are the research questions and relevant hypotheses for this experiment: RQ1

Did the students prefer working with a partner?

RQ2

Did they feel that team work helped create better (error free) programs?

RQ3

Did they place emphasis on fixing up roles and responsibility?

RQ4

Did they go in for early testing?

RQ5

Did they entertain changes in requirements during development?

RQ6

Did they put emphasis on incremental development and fast release?

An overview of the study’s hypotheses is presented in Table 1.

Copyright ⓒ 2013 SERSC

223

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

Table 1. Description of Null Hypothesis H01

Group work does not improve quality of code

H02

Working in groups is not amenable to entertain changes in requirement during development

H03

Regular interaction with teammates does not improve quality of code

H04

Early testing does not cause delay in completion of the project

H05

Changes in the requirement do not affect the following of a predefined plan

The above set of hypothesis has been derived from previous study along with the paper of [14]. 3.1. Experiment Description 3.1.1 Metrics used: To find out the relationship among the factors such as quality of code (error free), preference of group work, regular interaction with teammate, responding to changes, early testing, and fast delivery of working software. In bivariate distribution one is interested to find out if there is any correlation or covariation between the variables under study. If the change in one variable affects a change in the other variable, the variables are said to be correlated. If the two variables deviate in the same direction, i.e., if the increase (or decrease) in one results in a corresponding increase (or decrease) in the other, the correlation is said to be direct or positive. But if they constantly deviate in the opposite directions, i.e., if increase (or decrease) in one result in corresponding decrease (or increase) in the other, correlation is said to be diverse or negative. As a measure of intensity or degree of linear relationship between two variables, Karl Pearson, developed a formula called Correlation Coefficient [10]. The coefficient of correlation is defined as the covariance divided by the standard deviations of the variables. Correlation coefficient between two random variables X and Y, usually denoted by r(X, Y) or simply rxy, is a numerical measure of linear relationship between them, and is defined as

r( X ,Y ) =

Cov( X ,Y )

σσ x

(i)

y

If (xi, yi); i=1, 2, 3, …,n is the bivariate distribution, then

Cov ( x, y ) = Ε[{ X − Ε( X )}{(Y − Ε(Y )}] =

σ X 2 = Ε{ X − Ε( X )}2 =

224

(

1 ∑ xi − x n

)

(

)(

1 ∑ xi − x y i − y n

)

(ii)

2

(iii)

Copyright ⓒ 2013 SERSC

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

σ Y 2 = Ε{Y − Ε(Y )}2 =

(

1 ∑ yi − y n

)

2

(iv)

The summation extending from 1 to n over i Measure of t-test: In next step we tested significance of correlation. Significance of a correlation coefficient can be tested using t test. If r is the observed correlation coefficient in a sample of n pairs of observations from a bivariate normal population, then Prof. Fisher proved that under the null hypothesis: population correlation coefficient is zero, the statistic:

(v) [10] If the absolute value of t, i.e., |t| exceeds the table value of t at p% level of significance (say) and (n-2) degrees of freedom, then the value of r is significant at p% level otherwise insignificant. Here it is assumed that the n pairs are coming from a bivariate normal distribution. Sometimes due to a large value of n in the numerator, an apparently small value of r can also become significant. One should be careful in making interpretation in such cases. Measure of Correlation Ratio: As we know, when variables are linearly related, we have the regression lines of one variable on another variable and correlation coefficient can be computed to tell us about the extent of association between them. However, if the variables are not linearly related but some sort of curvilinear relationship exists between them, the use of r which is a measure of the degree to which the relation approaches a straight line “law” will be misleading. We might come across bivariate distributions where r may be very low or even zero but the regression may be strong, or even perfect. Correlation ratio ‘η’ is the appropriate measure of curvilinear relationship between two variables. Just as r measures the concentration of points about the straight line of best fit, curve of best fit, η measures the concentration of points about the curve of best fit. If regression is linear η=r, otherwise η>r. Suppose each observation is where indicates the category that observation is in and is the level of the particular observation. Let be the number of observations in category and

(vi)

(vii) Where is the mean of the category x and is the mean of the whole population. The correlation ratio η (eta) is defined as to satisfy

Copyright ⓒ 2013 SERSC

225

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

which can be written as

i.e., weighted variance of the category means divided by the variance of all samples [10]. Measure of F-test: After calculating η (eta) in third step we go for F-test for testing significance of an observed sample correlation ratio ηyx. Under the null hypothesis that population correlation ratio is zero, the test statistic is

(xii) where N is the size of the sample (from a bivariate normal population) arranged in h arrays [10]. 3.1.2. Test Programs In this study we selected the assignments to be solved for various programming languages within a specified time period. The test programs were the sheets of assignments from C and C++ programming languages. Assignments sheet consisted of 16 questions and 22 questions of C and C++ programming languages respectively. The questions were of various level of complexity. The assignment sheets started with simple questions to gradual increase in the complexity of the questions. The assignments were to be solved in weekly basis. The assignments were evaluated every week. The lab sessions were of 1hr. 15 minutes’ duration. Towards the end few assignments were given to solve in groups. Towards the end of the experiment duration the subjects were surveyed through a questionnaire. In the last lab session the questionnaire was distributed and responses were collected. 3.1.3. Subjects The subjects were chosen from two different colleges. A group of computer science students was considered for the experiment from an Engineering college of West Bengal and the other group of post graduate students undergoing a course of MBA systems specialization at Ranchi. Class strength varied from 50 to 60. All the students had the knowledge of programming language. The questionnaire (1) was distributed among 160 students and response was obtained from 155. Few respondents were absent. The questionnaire (Appendix 1) consisted of 9 questions.

226

Copyright ⓒ 2013 SERSC

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

4. Analysis and Findings 4.1. Discussion on Correlation Coefficient In all five hypotheses were proposed as given in Table 1. To test the main effect of group work and regular interaction on quality of code we propose hypothesis H01 and H03. In hypothesis H02 and H05 we tried to gauge how far group accepts changes in requirement during development stage and whether it really has an effect on following a predefined plan. Whether early testing affects project completion time is studied in H04. Correlation coefficient r was calculated using for (i), (ii), (iii) and (iv) (Section 3.1.1) for various pairs of factors (Table 2) considered in hypothesis (Table 1). This test shows that there is a significant relationship el) between group work and quality of code. They have a strong belief that working in groups definitely enhances the quality of code, that is, it exhibits less error because it allows cross checking and verification by the teammates. The study also reveals that students’ approach is always ready to adopt changes in the requirement. There is a significant relation el) between group work and entertaining changes in the requirement. Furthermore, analysis revealed significant el) relationship between the quality of code and regular interaction with teammates. However, there is an exception from the above cases that there is a negative el) between early testing and completion of the project correlation ( on time. Early testing seems to cause delay in completion of the project. Further our study also reveals that students’ approach towards software development is very flexible. Whenever there is a need to make some changes in the requirement then they are ready to do so instead of following a predefined plan because it gives better quality product. Table 2. Results of Pearson Correlation Factors considered for correlation

Pearson correlation

Group work and quality of code

Significant at the .01 level

r = 0.447, p = .000, Ν = 155 Group work and changes in the requirement

Significant at the .05 level

r = 0.176, p = .028, Ν = 155 Regular interaction with teammate and quality of code

Significant at the .01 level

Early testing and completion of the project in time

Significant at the .01 level

r = 0.409, p = .000, Ν = 155

r = -0.406, p = .000, Ν = 155 Changes in the requirement and following a predefined plan

Copyright ⓒ 2013 SERSC

Significant at the .01 level

r = 0.465, p = .000, Ν = 155

227

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

Table 3. Results of Hypothesis Tests H0

Reject?

Result

H01

Yes

Correlation between students’ preference to work in groups and quality of code is significant at the 0.01 level ( r = 0.447, p = .000 ) It has a perfect correlation.

H02

Yes

Relationship between group work and changes in the requirement is significant with r = 0.176, p = .028 at .05 levels.

H03

Yes

There is a significant (with r = 0.409, p = .000 at .01 level) relationship between regular interaction with teammate and quality of code. Regular interaction with teammate improves quality.

H04

Yes

Relationship between early testing and completion of project on time is negatively significant at .01 levels ( r = -0.406, p = .000 ).Students was of the view early testing caused delays.

H05

The null hypothesis is rejected at .01 levels ( r = 0.465, p = .000 ). To meet the user requirements students approach is to deviate from the pre defined plan. They give more emphasis on quality product.

Yes

Result of the study shows a positive correlation el) between ‘changes in the requirement’ and ‘following a predefined plan’. It is also found that factor ‘working in the same room with teammate’ shows significant results with more factors such as for ‘quality of code’ ( el), ‘entertain changes in the requirement el), and ‘working software first than during development’ el). Overall, the results of correlation coefficients documentation’ explored and revealed that student approach of software development has an inclination towards agile type of style to the extent that they do not prefer the strict formal approach of the traditional methods. Analysis of the results of Pearson correlation test reveals that student category is very positive towards the agile view of software development approach. Table 2 contains the list of factors considered for Pearson correlation test and respective values. The overall result of hypothesis testing is given in Table 3. 4.2. Discussion on t-test for Significance of Correlation Coefficient Further to verify the significance of correlation coefficient r, t-test was carried out using formula (v) in Section 3.1.1. Results of t-test are shown in Table 4.4. Table 4.4. Results of t-test

228

R

Level of significance

N

0.447

.01

155

6.18096675

0.176 0.409 -0.406 0.465

.05 .01 .01 .01

155 155 155 155

2.211521249 5.543956148 -5.495229897 6.496851871

0.347

.01

155

4.576514068

0.207

.01

155

2.617133232

-0.239

.01

155

-3.044497591

Copyright ⓒ 2013 SERSC

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

For all the significant value of r the absolute value of t as shown in Table 4.4 is greater than the t-Table values which are as follows

(10). Hence, results show that the above mentioned factors are linearly related. 4.3. Correlation Ratio Test for non-linearity Once it is proved that the factors are linearly related in the collected data set we then check for non-linearity using correlation ratio. In the third step to test non-linearity we calculated correlation ratio (eta) using formula (viii) defined in previous Section 3.1.1. In our case we have nine questions i.e., the value of 9 and 155 respondents. Group mean for each question was calculated first i.e., , then overall mean was calculated i.e., . The group means for the nine questions are as follows: 0.922580645, 0.922580645, 0.8, 0.922581, 0.554839, 0.845161, 0.129032, 0.8, -0.27742 using formula (vi) of Section 3.1.1 and the overall mean is 0.62437276 using formula (vii) of Section 3.1.1. Then the value of η2 is calculated as 0.29106974 using formula (viii) of Section 3.1.1 and hence η is 0.539508795. 4.4. F-Test for Significance of Correlation Ratio In order to test the significance of correlation ratio, F-test was carried out using formula (xii) under the null hypothesis that population correlation ratio is zero. Here the value of 9 ,

Since Cal F > Table F so η is significant at 5% level of significance indicating strong non-linear relationship. So overall the data set contains linear relationship among few factors and it also shows non-linear relationship among other factors. The factors such ‘group work’ and ‘fast delivery of working software’ are related to each other but in a non-linear fashion. Similarly there is a non-linear relationship between ‘quality of code’ and ‘responding to changes’ and ‘quality of code’ and ‘fast delivery of working software’. The data set shows all the variables or factors considered for the study are related to each other because the result of correlation coefficient through SPSS (Appendix 3) shows all the values of r are either greater or less than zero. No value of correlation coefficient is zero which indicates that the data set is not containing any uncorrelated factors or rather factors are not independent of each other.

5. Summary This paper shows that there is significant relationship between factors identified from the previous studies [11, 12, 15]. The variables considered for this study such as ‘quality of code (error free)’, ‘preference of group work’, ‘regular interaction with teammate’, ‘responding to changes’, ‘early testing’, and ‘fast delivery of working software’ are correlated with each other. Four hypotheses viz. ‘quality of code’ and ‘group works’, ‘regular interaction with teammate’ and ‘quality of code’, ‘changes in the requirement’ and ‘group work’, and

Copyright ⓒ 2013 SERSC

229

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

‘changes in the requirement’ and ‘following a predefined plan’ are positively correlated whereas one hypothesis i.e. ‘early testing’ and ‘completion time’ are negatively correlated. In order to test the significance of correlation coefficient r, t-test was carried out and results show that the factors are significantly related. In the data set, both linear and non-linear relationship among factors has been noticed. To test non-linearity we calculated correlation ratio (eta). Further to test the significance of correlation ratio, F-test was carried out. The result shows that there is strong non-linear relationship among the factors. We have found that the data set does not have any independent factor as all the values of correlation coefficients are either greater than or less than zero but not equal to zero. So data set contains factors in search are related to each other, and not uncorrelated. Overall all the factors identified in this study are playing an important role in software development. The results of correlation coefficients, correlation ratio explored and revealed that student approach of software development has an inclination towards agile type of style to the extent that they do not prefer the strict formal approach of the traditional methods.

References [1] [2] [3] [4] [5]

[6] [7]

[8] [9] [10] [11]

[12]

[13] [14]

[15]

230

N. Abbas, A. M. Gravell and G. B. Wills, “Using Factor Analysis to Generate Clusters of Agile Practices”, IEEE Computer Society, (2010), pp. 11-20. S. A. Ajila and D. Wu, ”Empirical study of the effects of open source adoption on software development economics”, The Journal of Systems and Software, vol. 80, (2007), pp. 1517-1529. E. Germain and P. N. Robillard, “Towards software process patterns: An empirical analysis of the behavior of student teams”, Information and Software Technology, vol. 50, (2008), pp. 1088-1097. S. C. Misra, V. Kumar, U. Kumar, “Identifying some important factors in adopting agile software development practices”, Journal of Systems and Software, vol. 82, no. 11, (2009), pp. 1869-1890. T. L. Roberts, Jr., M. L. Gibson, K. T. Fields and R. K. Rainer, Jr., “Factors that Impact Implementing a System Development Methodology”, IEEE Transactions On Software Engineering, vol. 24, no. 8, (1998) August, pp. 640-649. H. Sharp, H. Robinson and M. Petre, “The role of physical artifacts in agile software development: two complementary perspectives”, Interacting with Computers, vol. 21, no. 1-2, (2009), pp. 108-16. P. Tomaszewsk, J. Hakansson, H. Grahn and L. Lundberg, “Statistical models vs. expert estimation for fault prediction in modified code – an industrial case study”, The Journal of Systems and Software, vol. 80, (2007), pp. 1227-1238. P. Abrahamsson, O. Salo, J. Ronkainen and J. Warsta, ”Agile Software Development Methods Review and Analysis”, VTT Publications, (2002). D. Bell, “Software Engineering for Students, A programming Approach”, Fourth Edition, Pearson Education, (2007). S. C. Gupta and V. K. Kapoor, “Fundamentals of mathematical statistics”, Sultan Chand & Sons, Tenth Revised Edition, (2001). V. Bhattacherjee, M. Neogi and R. Mahanti, “Software Development Patterns in University Setting: A Case Study”, Proceedings of the national seminar on Recent Advances on Information Technology, ISM Dhanbad, (2007) February, pp. 40-43. V. Bhattacherjee, M. S. Neogi and R. Mahanti, “Software Development Approach of Students: An Evaluation”, Proceedings of the national conference on Methods and Models in Computing (NCM2C), JNU, New Delhi, (2008), pp 23-30. C. Coupal and K. Boechler, “Introducing Agile into a Software Development Capstone Project”, Proceedings of the Agile Development Conference (ADC’05), (2005). G. Melnik and F. Maurer, “A Cross-Program Investigation of Students’ Perceptions of Agile Methods”, Proceedings of International Conference on Software Engineering archive Proceedings of the 27th international conference on Software engineering, ICSE ’05, (2005) May 15-21, pp. 481-488. M. S. Neogi, V. Bhattacherjee and R. Mahanti, “An Evaluation of Student Preferences during Software Development”, Proceedings of national seminar on Recent Advances Information Technology (RAIT), ISM Dhanbad, (2009) February 6-7, pp. 239-245.

Copyright ⓒ 2013 SERSC

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

Appendix 1: 1. 2. 3. 4. 5.

Questionnaire - 1 Do you prefer to work in groups? Yes Neutral No Do you think working in groups improves quality? Yes Neutral No Do you think working in the same room with the teammate is beneficial? Yes Neutral No Regular interaction with teammates improves quality? Yes Neutral No Do you go for testing at the early stage? Yes Neutral No

6. Do you entertain any changes in the requirement during its development stage? Yes Neutral No 7. Do you prefer to go for working software first than documentation? Yes Neutral No 8. Do you respond to changes instead of following a plan? Yes Neutral No 9. Do you complete your assignments on time always Yes Neutral No Appendix 2: Data with Response: Q1 Q2 R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11

1 1 1 1 1 1 1 1 1 1 1

Copyright ⓒ 2013 SERSC

Q3 1 1 1 1 1 1 1 1 1 1 1

-1 1 1 1 1 0 1 1 1 1 1

Q4

Q5 1 1 1 1 1 1 1 1 1 1 1

1 1 1 1 1 1 1 1 -1 1 -1

Q6

Q7 1 1 1 1 1 1 1 1 1 1 1

Q8 1 -1 -1 0 -1 1 -1 -1 1 1 -1

Q9 1 1 1 1 1 1 1 1 1 1 1

1 -1 -1 -1 1 -1 -1 -1 -1 1 1

231

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

R12 R13 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23 R24 R25 R26 R27 R28 R29 R30 R31 R32 R33 R34 R35 R36 R37 R38 R39 R40 R41 R42 R43 R44 R45 R46 R47 R48 R49 R50 R51 R52

232

1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 1 -1 1 1 1 1 1 1 1 1 1 1 1 0

1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 -1 1 1 1 1 1 1

1 1 1 1 1 -1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 1 1 1 1 0 1 1 -1 -1 1 -1 1 0 0 1

1 1 1 1 1 -1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 1 1 0 1 1 1 1 1 1 1 1 1 1

1 1 1 -1 -1 1 1 1 1 1 -1 -1 -1 1 1 1 1 1 1 1 -1 1 1 1 1 0 -1 1 1 0 1 0 -1 1 0 1 1 1 1 1 1

1 1 1 1 1 -1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 1 1 1 1 0 1 1 1 1 1 1 1 1 1

0 1 -1 1 1 1 -1 -1 -1 -1 -1 -1 -1 -1 1 -1 -1 -1 1 1 1 1 1 1 0 1 1 -1 1 1 1 1 0 1 1 -1 -1 1 0 1 1

1 1 1 1 1 1 -1 -1 -1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 -1 1 1 1 1 1 1 1 1

-1 -1 0 1 1 -1 1 1 1 -1 1 1 1 -1 1 -1 -1 -1 -1 -1 -1 -1 -1 1 0 -1 -1 0 1 0 0 -1 0 -1 1 -1 -1 -1 -1 -1 -1

Copyright ⓒ 2013 SERSC

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

R53 R54 R55 R56 R57 R58 R59 R60 R61 R62 R63 R64 R65 R66 R67 R68 R69 R70 R71 R72 R73 R74 R75 R76 R77 R78 R79 R80 R81 R82 R83 R84 R85 R86 R87 R88 R89 R90 R91 R92 R93

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 1 1 1 0 1 1 1 1 1

Copyright ⓒ 2013 SERSC

1 1 1 1 1 1 1 1 1 1 1 1 -1 1 -1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 1 1 1 1 1 1 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 -1 1 1 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

1 1 1 1 1 0 1 0 -1 1 1 1 1 -1 1 -1 0 1 1 1 1 1 1 1 1 0 0 0 1 1 1 0 1 1 1 1 1 0 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1

1 0 1 1 1 0 1 1 1 1 1 1 0 0 1 1 1 1 1 1 1 1 0 0 -1 0 0 1 1 1 1 0 1 1 0 0 0 0 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 0 1 1 1 1 1 1 1 -1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1

-1 -1 -1 -1 -1 1 -1 1 1 -1 1 -1 -1 1 -1 -1 0 -1 -1 0 -1 -1 -1 -1 -1 1 1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0 -1 -1 -1

233

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

R94 R95 R96 R97 R98 R99 R100 R101 R102 R103 R104 R105 R106 R107 R108 R109 R110 R111 R112 R113 R114 R115 R116 R117 R118 R119 R120 R121 R122 R123 R124 R125 R126 R127 R128 R129 R130 R131 R132 R133 R134

234

1 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1

1 1 -1 1 1 1 1 -1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 -1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1

-1 1 1 1 1 0 1 1 1 1 1 1 1 1 1 -1 1 -1 1 1 1 -1 -1 1 1 1 1 1 -1 -1 -1 1 1 1 1 1 1 1 1 1 1

1 1 1 1 -1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1

1 1 1 -1 1 0 1 1 -1 -1 0 -1 1 -1 -1 1 1 -1 0 1 -1 1 1 1 -1 -1 -1 -1 -1 -1 -1 -1 1 -1 -1 -1 1 1 -1 -1 0

-1 1 1 1 1 1 -1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 -1 -1 -1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

-1 0 -1 1 -1 -1 1 1 -1 -1 -1 1 -1 -1 -1 -1 1 1 -1 -1 0 1 1 -1 1 1 1 -1 1 1 1 -1 1 -1 -1 -1 -1 1 -1 -1 -1

Copyright ⓒ 2013 SERSC

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

R135 R136 R137 R138 R139 R140 R141 R142 R143 R144 R145 R146 R147 R148 R149 R150 R151 R152 R153 R154 R155

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1

Copyright ⓒ 2013 SERSC

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1

1 0 1 1 1 1 1 1 1 1 1 1 -1 1 1 1 1 1 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1 -1 1 1 1 1 1 1 0 1

1 1 1 1 -1 1 -1 1 1 1 -1 -1 1 1 1 1 1 -1 -1 -1 1

1 1 1 1 1 1 1 1 1 1 1 1 -1 0 0 0 1 1 1 1 1

-1 1 -1 -1 1 1 -1 0 1 -1 1 1 1 -1 -1 -1 -1 -1 -1 -1 -1

1 1 1 1 1 1 1 1 1 1 1 1 1 -1 -1 -1 1 1 1 1 1

1 -1 -1 -1 -1 1 1 -1 -1 0 1 1 -1 1 1 1 -1 1 1 1 -1

235

International Journal of Software Engineering and Its Applications Vol.7, No.6 (2013)

Appendix 3:Results of correlations: Correlations Q1

Q2

Q3

Q4

Q5

Q6

Q7

Q8

Q9

Q1 Pearson Correlation 1.000 Sig. (2-tailed) . N 155 Pearson Correlation .447** Sig. (2-tailed) .000 N 155 Pearson Correlation .055 Sig. (2-tailed) .493 N 155 Pearson Correlation .115 Sig. (2-tailed) .155 N 155 Pearson Correlation .166* Sig. (2-tailed) .039 N 155 Pearson Correlation .176* Sig. (2-tailed) .028 N 155 Pearson Correlation -.031 Sig. (2-tailed) .703 N 155 Pearson Correlation -.013 Sig. (2-tailed) .869 N 155 Pearson Correlation -.049 Sig. (2-tailed) .543 N 155

Q2 .447** .000 155 1.000 . 155 .118 .144 155 .109 .179 155 .110 .174 155 .006 .944 155 -.009 .911 155 -.013 .876 155 -.067 .410 155

Q3 .055 .493 155 .118 .144 155 1.000 . 155 .347** .000 155 -.130 .108 155 .207** .010 155 -.239** .003 155 -.102 .205 155 .068 .404 155

Q4 .115 .155 155 .109 .179 155 .347** .000 155 1.000 . 155 .01 6 .84 8155 .40 ** 9 .00 0155 -.009 .91 1155 -.076 .350 155 -.007 .935 155

Q5 .166* .039 155 .110 .174 155 -.130 .108 155 .016 .848 155 1.000 . 155 -.012 .878 155 -.019 .815 155 .017 .834 155 -.406** .000 155

Q6 .176* .028 155 .006 .944 155 .207** .010 155 .409** .000 155 -.012 .878 155 1.000 . 155 -.029 .716 155 .465** .000 155 -.010 .901 155

Q7 -.031 .703 155 -.009 .911 155 -.239** .003 155 -.009 .911 155 -.019 .815 155 -.029 .716 155 1.000 . 155 .194* .016 155 -.165* .041 155

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

236

Copyright ⓒ 2013 SERSC

Q8 -.013 .869 155 -.013 .876 155 -.102 .205 155 -.076 .350 155 .017 .834 155 .465** .000 155 .194* .016 155 1.000 . 155 -.248** .002 155

Q9 -.049 .543 155 -.067 .410 155 .068 .404 155 -.007 .935 155 -.406** .000 155 -.010 .901 155 -.165* .041 155 -.248** .002 155 1.000 . 155