.

Monday, September 30, 2019

Quality of Emba Program

Problem Definition Background to the problem Dhaka University's Evening MBA program started in 2002 as an effort to bring the Faculty of Business Studies up to the standard with other private, public and international academic institutions. The program is currently on its 18th batch. Although the University authority started the program almost 8 years ago, there are still doubts among people about the quality of the Evening MBA program offered by the Dhaka University Faculty of Business Studies and many are confused about where this program stands against the MBA program offered by Dhaka University Institute of Business Administration (IBA). So a study was due in this field to determine the quality of the DU evening MBA program. And no one knows about the program better than those who are studying in the program already. Problem Statement The problem statement for our research is â€Å"The quality of the Evening MBA program in Dhaka University is not very high†. We will be using several statistical theories and tools to test our problem statement and create a model from it and also test the significance of the model. Approach to the Problem As our problem statement suggests, our objective of the study is to determine the quality of the Evening MBA program in Dhaka University. Now a product's quality can be easily determined through the use of different quality checks but what determines the quality of an academic program? After looking into some secondary data and taking initial opinion from a small group we have come to conclusion that quality of an academic program is closely related to quality of the learners, the environment the learning process takes place, the teacher, content of the learning program, the process and the outcome of it. Based on these analyses we have come to the conclusion that we will need to develop a survey questionnaire which will be used as our primary data and will include questions relating the above mentioned criteria with the topic in question- quality of the evening MBA program. Based on the data collected we will run regression analysis to develop the model and then use Discriminant analysis to classify the results. Research Design A research design should include all the important information about the research process such as type of research done, information needs, data collection, scaling techniques, questionnaire development, sampling techniques and fieldwork. Type of research To determine what type of research is needed to be done to achieve our objectives, we have explored different possibilities. First let us revisit our objective- we would like to determine the criteria that form quality of an educational program and then create a model out of it to identify significance of each of the criteria in the model. In other words, we will be determining the effect of some independent variables and determine how they affect the dependant variable. Based on this initial assumption we could say that we are looking into a cause and effect relationship and hence we will be doing a Causal research. However, further probing into the matter made us realize that the most important criteria of a causal relationship is to be able to manipulate the variables and observe their effect on the model, which in our case is not at all possible, neither we will be doing much experimentation. So ours is anything but a causal research. What our research is rather capable of is determine some characteristics relating to the problem in hand and based on primary data and observation develop a model to show an overall relationship. Based on such analysis we have come to the conclusion that a Descriptive research is more appropriate in our case. Information Needs For our analysis, at first we needed to know the perception of students about the quality of the evening MBA program. After that we needed their opinion on quality of the students enrolled in the program, students’ perception of the brand value of the program, quality of the teachers, grading system and admission test. So we have determined to create a questionnaire that will include questions about both the dependent variable (quality of the program) and independent variables (quality of students, brand value perception, quality of teachers, grading system, admission tests etc. ) and use it to run a survey to gather all the necessary information we will need in developing the model. Scaling Techniques Now all of our required data are about opinions or perceptions of students. For such situations a Likert scale is the most appropriate scaling technique to use because the main feature of Likert scale is a list of opinions ranging from extreme positive to extreme negative about a statement. Each opinion is assigned a score and based on the respondent’s response the score is considered in further data analysis. While using Likert scale we had to careful to maintain a consistency in the statements so that the positivity of an opinion always gets the highest score and the negative one the lowest. Based on this rule we needed to reverse the scoring on questions that leaned to negativity. Questionnaire Development and Pre-testing We have already defined our dependent and independent variables for the research and what we needed was to collect the sample population’s opinion about each of the statements made using a Likert scale. So setting up the questionnaire was quite simple for us. All we had to do is form a statement relating to each of the variables and attach a Likert scale table to let the respondents choose to what level they agreed or disagreed to the statements. Based on our initial analysis, we have determined 1 dependent and 12 independent variable and so we created a questionnaire with 13 statements and some apparently-relevant open-ended questions. To avoid confusion, the statements were worded as simple as possible. However, after pre-testing the questionnaire among a very small group, we observed that people still got confused about a question (question 12) related to grading. While our objective was to correlate fairness in grading with the quality of education, students related it to the grading system. So this statement didn’t serve its intended purpose and hence in our final analysis we have decided to leave it out. Due to this change, from now on every instance of Q12 will refer to Q13 in real. A sample copy of the questionnaire used has been included in the appendix section of this report. Sampling technique Sampling is another important part of a research design. The first and foremost job in sampling is to define the target population. Now we have already discussed that as the problem is to determine whether the quality of the EMBA program is up to the standard or not, no one but the students would know the most about it. So our target population is defined as all the students in the EMBA program; that includes not only the students from marketing department but all other departments too. After the target population was defined, the next job was to determine a sampling technique. Now there are a lot of sampling technique available but due to several limitations not all of them were appropriate. Considering the fact that it would be a lot difficult to contact and convince students from other departments to participate in the survey, we have decided to leave them out. And considering the fact that we started working on the research project right at the end of the semester made it difficult to communicate with all the students in marketing department too. So to run our survey we had to rely on the places of convenience where we would be present along with more students from different batches. In other words, the sampling technique we used was more of a convenience sampling. however we were careful not to select one respondent more than once. After the sampling technique was determined, our next job was to determine the sample size. Now determining the sample size is a very complicated process even on situations where relevant data like total population size etc. are available. When no such data exists, the sample size determination becomes just that much harder. However, we were lucky not to have gone through any hardship at all because we were instructed by our respectable course teacher to keep the sample size to somewhere around 30 and we followed his instructions to the book. Fieldwork As this is a very small-scale academic research, there was no need for additional fieldworker to run the survey. Instead we, the researchers took matters into our own hands and did the fieldwork ourselves. Now the positive side of it was that we didn’t have to train anyone to run the survey most effectively. As we were the ones setting up the questionnaire, we had a clear idea about what to do and how to do it. We used our available classes as place of convenience and used the class breaks to collect our data. Data Preparation After the fieldwork was done, we were left with 36 responses, out of which 4 were found to be incomplete. On such situations it is suggested to assign missing values to the incomplete survey papers. However, as we still had a margin from the required minimum of 30, we decided to leave the incomplete ones out. Once we agreed on that, the 32 valid survey papers were coded and transcribed into the computer to be used with SPSS, the statistical tool we ought to use. Data Analysis Once the data were transferred into SPSS, we were ready to start data analysis. Methodology The data were analyzed by conducting multiple regression analysis and Discriminant analysis. Regression was conducted to see whether a relationship exists between quality of EMBA program (dependant variable) and factors we have determined to indicate the quality (independent variables). Discrminant analysis was used to give us further insights. Plan for Data Analysis For regression we have considered Q1 as dependant variable and Q2-Q12 as independent variables. Then we have used SPSS to get the output. For Discriminant analysis, independent variables were converted from nine-point Likert scale into two-group categorical variable. For conversion we followed the following rule: -4= 1 (Low); that is EMBA program is perceived to be of low quality. 6-9= 2 (High), that is EMBA program is perceived to be of high quality. Note that we have disregarded the neutral value 5. From our analysis we have found that none of the respondents have chosen this. So we have taken only two variables and planned for Two-Group Discriminant analysis. So in this case the newly converted group variable is the dependent and Q2–Q12 are considered as predictors. Results Regression Strength of Association In the SPSS output Table below we can see the value of R2 is . 878. The R-square value is an indicator of how well the model fits the data e. . , an R2 close to 1. 0 indicates that we have accounted for almost all of the variability with the variables specified in the model. Our R2 is close to 1. ModelRR2Adjusted R SquareStd. Error of the EstimateChange Statistics R2 ChangeF Changedf1df2Sig. F Change 1. 937(a). 878. 811. 989. 87813. 0651120. 000 (a) Predictors: (Constant), Q12, Q7, Q3, Q2, Q4, Q9, Q6, Q11, Q10, Q8, Q5 Significance Testing The following formula is used to test whether an R2 calculated is significantly different than Zero. The Null Hypothesis is that the population R2 is Zero. where N is the number of subjects, k is the number of predictor variables and R? s the squared multiple correlation coefficient. The F is based on k and N – k – 1 degrees of freedom. In our case, N = 32, k = 12, and R? = . 878. In the SPSS output Table below we can see that F = 13. 065 which is significant at ? =0. 05. We can also see that significance is . 000; as it is smaller than . 05 we can say that it is highly significant. ANOVA (b) ModelSum of SquaresdfMean SquareFSig. 1Regression140. 6461112. 78613. 065. 000(a) Residual19. 57320. 979 Total160. 21931 (a) Predictors: (Constant), Q12, Q7, Q3, Q2, Q4, Q9, Q6, Q11, Q10, Q8, Q5 (b) Dependent Variable: Q1 In addition to testing R? or significance, it is possible to test the individual regression coefficients (Beta) for significance and it is shown in the SPSS output in the following table. Coefficients (a) ModelUnstandardized CoefficientsStandardized CoefficientstSig. 95% Confidence Interval for BCorrelations BStd. ErrorBetaLower BoundUpper BoundZero-orderPartialPart 1(Constant). 4491. 171. 383. 705-1. 9942. 892 Q2-. 046. 188-. 035-. 246. 808-. 437. 345. 395-. 055-. 019 Q3-. 091. 131-. 074-. 694. 496 -. 364. 182. 236-. 153-. 054 Q4. 102. 190. 084. 539. 596-. 294. 499. 407. 120. 042 Q5-. 188. 232-. 129-. 809. 428-. 672. 296. 464-. 78-. 063 Q6. 507. 196. 4372. 582. 018. 097. 916. 856. 500. 202 Q7. 015. 141. 011. 103. 919-. 279. 308. 273. 023. 008 Q8. 508. 170. 4652. 982. 007. 153. 864. 878. 555. 233 Q9. 035. 151. 029. 231. 819-. 280. 350. 464. 052. 018 Q10-. 132. 167-. 111-. 791. 438-. 482. 217. 524-. 174-. 062 Q11. 188. 145. 1711. 292. 211-. 115. 491. 600. 277. 101 Q12. 165. 119. 1461. 386. 181-. 083. 414. 621. 296. 108 (a) Dependent Variable: Q1 In the above table, we can see that all of significant levels corresponding to individual Beta are greater than . 05 except two. The significant for coefficient for Q6 and Q8 is less than . 5. So these are found to be significant. Therefore teacher’s delivery and student’s seriousness are important in explaining quality of education program. Regression Model From the whole regression analysis, we can finally generate a mode l that shows the total relationship between the independent variables selected and the dependent variable. Assigning each of the independent variables with Xn starting with Q2 as X1, Q3 as X2, Q4 as X3 and so on and assigning the dependent variable Q1 as Y, we form a generic regression model- Y= C + B1X1+ B2X2+ B3X3+ B4X4+ B5X5+ B6X6+ B7X7+ B8X8+ B9X9+ B10X10+ B11X11 Now putting the relevant Bs in the equation, we get- Y=0. 449 – 0. 046X1 – 0. 091X2 + 0. 102X3 – 0. 188X4 + 0. 507X5 + 0. 015X6 + 0. 508X7 + 0. 035X8 – 0. 132X9 + 0. 188X10 + 0. 165X11 This is our regression model to determine the quality of education in the EMBA program. Discriminant Analysis The significance of univariate F ratios shown in table below indicates that when the predictors are considered individually Q8, Q6 and Q12 are highly significant (significant level . 000) in differentiating between those who perceive EMBA program to be of high quality and those who perceive it to be low quality. That is teacher’s delivery (Q8), student’s seriousness (Q6) and seriousness of administration in enforcing quality (Q12) are important differentiating factors toward high or low quality perception of EMBA program. Tests of Equality of Group Means Wilks' LambdaFdf1df2Sig. Q2. 8007. 501130. 010 Q3. 9531. 480130. 233 Q4. 8067. 240130. 012 Q5. 72311. 518130. 002 Q6. 29073. 336130. 000 Q7. 9551. 410130. 244 Q8. 26881. 874130. 000 Q9. 8584. 949130. 034 Q10. 8037. 350130. 011 Q11. 8087. 123130. 012 Q12. 62517. 996130. 000 Because there are only two groups, only one discriminant function is estimated. The Eigenvalue associated with the function is 5. 37; as shown in table below and it accounts for 100 percent of the explained variance. The canonical correlation associated with this function is 0. 924. The square of this correlation, (0. 924)2 = 0. 85, indicates that 85% of the variance in the dependent variable (High/low quality perception) is explained or accounted for by the model. Eigenvalues FunctionEigenvalue% of VarianceCumulative %Canonical Correlation 15. 837(a)100. 0100. 0. 924 (a) First 1 canonical discriminant function was used in the analysis. It can be noted from table below Wilks' Lambda associated with the function is 0. 146 which transforms to a chi-square of 47. 98 with 11 degree of freedom. This is significant beyond the . 05 level. Wilks' Lambda Test of Function(s)Wilks' LambdaChi-squaredfSig. 1. 14647. 09811. 000 The table below shows the inter-correlation between the predictors and we can assume a low correlation. Q2Q3Q4Q5Q6Q7Q8Q9Q10Q11Q12 CorrelationQ21. 000. 324. 512. 340. 215-. 209. 064. 441. 317. 131. 125 Q3. 3241. 000. 088. 243. 218. 094. 072. 066. 044. 398. 261 Q4. 512. 0881. 000. 667. 096-. 236. 170. 197. 497. 266. 055 Q5. 340. 243. 6671. 000. 290-. 438. 053. 080. 336. 364. 149 Q6. 215. 218. 096. 2901. 000. 032. 186. 095. 323. 529. 015 Q7-. 209. 094-. 236-. 438. 0321. 000. 29. 100. 220. 163. 113 Q8. 064. 072. 170. 053. 186. 129 1. 000. 450. 390. 260. 186 Q9. 441. 066. 197. 080. 095. 100. 4501. 000. 531. 296. 206 Q10. 317. 044. 497. 336. 323. 220. 390. 5311. 000. 517. 153 Q11. 131. 398. 266. 364. 529. 163. 260. 296. 5171. 000. 135 Q12. 125. 261. 055. 149. 015. 113. 186. 206. 153. 1351. 000 An examination of standardized discriminant function coefficient shown in the following table given the low inter-correlation between predictors, it is revealed that Q8 (teacher’s adequate lecture delivery) and Q6 (seriousness of the students to learn) is the most important predictors (having highest value of . 13 and . 704 respectively) in discriminating between groups, followed by Q12 ( administration’s seriousness in enforcing quality) and Q5 (competitive value of achieving degree in the industry). Standardized Canonical Discriminant Function Coefficients Function 1 Q20. 164 Q3-0. 199 Q40. 122 Q50. 245 Q60. 704 Q70. 276 Q80. 713 Q9-0. 118 Q10-0. 387 Q11-0. 264 Q120. 254 It is interesting to note that the same observation is obtained from examination of the structure correlations (structure matrix shown in table below). In this table these simple correlation between predictors and discrminant function are listed in order of magnitude. Structure Matrix Function 1 Q80. 684 Q60. 647 Q120. 321 Q50. 256 Q20. 207 Q100. 205 Q40. 203 Q110. 202 Q90. 168 Q30. 092 Q70. 090 SPSS offer a leave-one-out cross validation option. In this option, the discriminant model is re-estimated as many times as there are respondents in the sample. Each re-estimated model leaves out one respondent and the model is used to predict for that respondent. The output for this is shown in the table on the following page. From the table hit ratio or the percentage of cases correctly classified can be estimated as (18+13)/32*100 =96. % [considering correct number of predictions of 18 and 13 for two groups] Classification Results (b,c) GroupPredicted Group MembershipTotal 1. 002. 00 OriginalCount%1. 0018018 2. 0011314 1. 00100. 0. 0100. 0 2. 007. 192. 9100. 0 Cross-validated(a)Count%1. 0017118 2. 0011314 1. 0094. 45. 6100. 0 2. 007. 192. 9100. 0 (a)Cross validation is done only for those cases in the analysis. In cross validation each case is classifie d by the functions derived from all cases other than that case. (b) 96. 9% of original grouped cases correctly classified. (c) 93. 8% of cross-validated grouped cases correctly classified. Thus we can say most important factors are Q8 (teacher’s adequate lecture delivery) and Q6 (seriousness of the students to learn). This result is consistently found both by regression and correlation. Limitations No research project is free from limitations of some form- be it time or resources. Same is true for our research project. Although time given was adequate for a small-scale project like this, but considering the topic of our discussion such small-scale research hardly means anything. Considering the number of students currently enrolled in the EMBA program as a whole, a sample size of 30 is hardly representative of the population. Added to that is our inability to communicate and select respondents from other departments. So considering all these, this project, although best efforts were given to complete, doesn’t completely satisfies its main purpose of determining the quality level of the EMBA program. Conclusion In conclusion, we can say that even though the research project didn’t serve its purpose completely, it at the least gives some idea about students’ perception of quality education and overall quality of the EMBA program. From the research, based on multiple data analysis, we have found out that people put great emphasis on students’ eagerness to learn and teachers’ delivery of knowledge to determine the quality of education program. So it is imperative that students are encouraged in different ways so that they feel inspired to learn new things on their own. And teachers also should keep in mind the duty they have sworn to fulfill and give their best efforts in teaching the students properly while staying above all influences. –X–

No comments:

Post a Comment