Effect of an Entrepreneurship Program on GPA and Retention*, The

Journal of Engineering Education, Oct 2004 by Ohland, Matthew W, Frillman, Sharron A, Zhang, Guili, Brawner, Catherine E, Miller, Thomas K III

Of the 107 lower division students who had taken the EEP in one or more of the semesters listed in Table 3, university data was available for 103 of them. Twelve of these students were transfer students and were deleted from further analysis since their transfer status would confound both of the outcomes under study, leaving 91 students in the sample under study.

Longitudinal study techniques used: Using the SUCCEED longitudinal database [31], each of the EEP participants was matched with a student who was not in the EEP but who had the same gender, ethnicity, cohort, engineering major, and similar SAT scores. This method helps control for a number of possible biases associated with a course for which students self-select. Student Grade Point Averages, cohort, graduation, and retention data for the experimental group (labeled EEP) and the control group (labeled Non-EEP) were also obtained from the SUCCEED longitudinal database. Retention figures shown in Table 6 include students graduated or still enrolled in an engineering discipline as of fall 2000. For each matched pair, there are four possible outcomes-both are retained, both leave engineering, the EEP participant is retained while the non-EEP leaves engineering, or vice versa.

Results of longitudinal study of retention: Table 6 cross-classifies NCSU engineering students in the study according to whether they participated in the EEP program and whether they were retained. A total of 91 pairs of students were studied, each pair consisting of an EEP participant and a non-participant matched as described above. Of those 91 pairs, 34 pairs showed both EEP participants and non-participants were retained; in 15 pairs, neither participants nor non-participants were retained; in 30 pairs, the participant was retained while the non-participant was not retained; and in 12 pairs, the non-participant was retained while the participant was not retained. Overall, out of the same number of students, 64 of the EEP group were retained while 46 from the Non-EEP group were retained.

For such data, we want to study whether a statistically significant association exists between EEP and retention. We seek to answer the question, "are students who participate in the EEP programs more likely than those who do not participate in the program to graduate or to stay in engineering, or is graduation independent of whether the students are in the EEP program?"

The McNemar Test for matched pairs was used to test for marginal homogeneity for matched binary responses [32]. The null hypothesis was: π^sub 1 ^ = π^sub 1 ,^ or the probability of retention is the same for the EEP group and the Non-EEP group. The alternative hypothesis was: π^sub 1 ^ > π^sub 1 ^, or the probability of retention for EEP group was higher than that for Non-EEP group. The McNemar Test statistic provides very strong evidence that the EEP group's retention rate was significantly higher than for the non-EEP group (x^sup 2^ = 7.71, df=1, p = .005).[32]

While the McNemar Test indicates that the probability of retention was higher for the EEP group than for the non-EEP group, the test does not tell us by how much the probability of retention was higher. To answer this question, we estimated the population difference in proportion by using sample difference in proportion and constructing a 95 percent confidence interval on the difference, because a confidence interval for the difference of proportions is more informative than a hypothesis test. Based on the work of Agresti, the 20 percent improvement in retention observed for EEP participants leaves us 95 percent confident that EEP participants are at least 6.5 percent more likely to be retained in engineering.