Predicting Time to Graduation in Engineering by Student Behavior and Gender

Our state, like many, is currently under pressure to reduce time to graduation of college students to satisfy various local, city and state stakeholders. As a consequence, we seek to develop a systematic method to take into account both good reasons for delaying graduation (such as spending a semester in a co-op or internship) and negative reasons (such as failing classes). We also want to quantify an estimate for the delay based on each experience. The hope is that this model helps inform the discussions in our institution’s upper administration and state legislature regarding time to graduation. This study looks at retention and graduation patterns in engineering by gender, seeking specifically to understand why students tend to take longer than the advertised 4 years to graduate, and focusing on one institution only. While much attention has been given in the past to the issues of retention of women and under-represented minorities in engineering, most analyses use data collected at multiple institutions (thus blending together results coming from vastly different campus cultures) and tend to focus on the experiences of those who graduate versus those who don’t. In this work, we focus on why the students who do graduate usually take much longer than the four (4) years that are routinely advertised. This study follows in the footsteps of previous work. Introduction Retention of engineering undergraduate students is not a new research topic. As a persistent source of worry in industry, government and academe, low graduation rates of engineering students could damage the supremacy of the US in its technical prowess. If current trends continue, the country could potentially lose to other countries (such as China and India) in the numbers of qualified engineering graduates overall. Another well-known issue with engineering is the field’s persistent lack of diversity. Since engineering pervades every aspect of modern life, it is vital that engineering products and processes reflect the diversity of the population they aim to serve. Engineering products (such as, say, voice recognition devices) used by a highly diverse population should not be designed overwhelmingly by Caucasian males. In addition, the changing demographics of the US (the fact that the Caucasian population will be a minority in the next few decades) spell doom for engineering programs if the engineering community doesn’t do a better job of attracting and retaining a more representative and diverse segment of the overall American population1. In this work we define retention the customary way, that is, by the number of first-time, full-time students that graduate from the institution within 6 years. Much past research on retention has focused on students who leave engineering (so-called “non-persisters”) and what caused their departure. For example, Marra et al.2 shows students of both genders tend to drop out of engineering primarily for two reasons: 1) the curriculum is too challenging and the quality of teaching too poor, and 2) students don’t believe they belong. P ge 24997.2 Ohland et al.3 present an extensive analysis of retention measures and student educational experiences at the undergraduate level, and for the first time uses a semester-level measure of retention (rather than year-level retention which is the overwhelming approach). His group uses the large, multi-institution dataset MIDFIELD (Multiple-Institution Database for Investigating Engineering Longitudinal Development) which contains records of over 75,000 students in engineering during the years of 1988 through 1998. Ohland and his colleagues3,4 found that in general, it is shown that paths of persistence are nonlinear, gendered and racialized, so that it’s important to use multiple measures to assess retention when dealing with diverse populations of engineering students. A frequent concern in most retention studies is that researchers must lump disparate populations together to achieve adequate statistical meaning – for example, looking at STEM (Science, Technology, Engineering and Mathematics) majors as a whole without differentiating between the disciplines, and/or lumping multiple institutions together. Even then, it has long been known that women studying engineering often suffer from a lack of peer support, role models, and adequate academic preparation, and these problems can cause these students to transfer out of engineering5-8. Research also shows that women are more likely than men to report that teaching styles, subject matter relevance, and the culture of the discipline affect their retention and eventual completion of the degree9. They’re also more likely to report that tutoring services are important to their academic success and ultimately, their retention (even controlling for academic preparation and race/ethnicity). In general, women find the classroom climate in engineering to be chillier than White men, and are less likely to work as a practicing engineer in industry, government, or nonprofit organization after graduation – though both effects are lessened significantly if the women engage in engineering clubs or programs supporting women engineering students such as SWE (Society of Women Engineers) or NSBE (National Society of Black Engineers). Multiple studies (by Marra and others) show that unlike men, women studying engineering tend to be better retained if the institution offers support such as strong ties to faculty and other students, tutoring, availability of numerous student clubs, and living/learning communities10,11,12. Despite these complaints, multiple studies, such as Consentino et al.13 and Lord et al.14 found that retention is not the primary reason for the low percentage of women in engineering, but rather, recruitment. That is, when women enter college intending to study engineering, they usually do eventually graduate with an engineering degree and don’t transfer to a nonengineering field. However, very few female high school seniors do in fact choose engineering as a field of study in college. Our study is different in that rather than focusing on persisters and non-persisters, we focus on persisters who take longer than 4 years to graduate. We are specifically interested in finding the causes for students’ delay in their graduation, and whether these causes are good (gaining work experience through co-op and internships, for instance) or bad (failing classes and having to repeat them). We’re also interested in quantifying the delay these causes produce, in semesters, per cause. Does co-oping delay graduation more than, say, getting a F? P ge 24997.3 This study is motivated in large part by our state’s legislature, which wants to tie our share of state financial support to our graduation times. If we can document that the longer times to graduation that we see in our students are caused by good reasons (better readiness for the workforce) rather than bad ones, we can make a strong case to the legislature stakeholders that our funding needs to be maintained. At our institution, the Georgia Institute of Technology (GT), the graduation rate is significantly above the national average. For students of both genders, the 6-year graduation rate is above 85%. Therefore, the issue is not persisters versus non-persisters for us. Another unique feature of our study is that it’s done at the institutional level. Our numbers of engineering students and graduates are large enough that they have statistical meaning without pooling them with other institutions. Therefore, we are truly comparing like behaviors, done in the same educational environment. This also provides a tremendous advantage in that the data is consistent, and mistakes are minimized. Multiple institutions often define things such as transfer credit or AP credit differently, which necessarily muddies the results of the analysis. To gain this level of depth and accuracy in the data across institutions normally requires Deanlevel contacts to be made. We are very interested in expanding our work to other universities, should there be interest, now that the “proof of concept” has been established. This paper is a follow-up on our prior work15 that presented demographics for the cohort of study (students who enrolled in the College of Engineering between 2000 and 2005 and have since overwhelmingly graduated, representing 6 years of complete data) and analyzed time to graduation in the following situations: Citizenship and residency status, Whether they were a student-athlete at any time during their studies, Whether they received a poor grade (D, F, or Withdrew), AP credit or transfer credit. In this paper, we seek to expand the study to look at the impact of work experience (such as coop and internships), and changing majors or Colleges (not institutions as a whole – we’re only considering persisters – but instead, transferring from, say, the College of Engineering to the College of Sciences). Ultimately, our goal is to present a statistical formula assessing and quantifying the delay these various behaviors cause on the student’s time to graduation. This formula will be presented in future work. Before we start presenting our analysis and results, we want to stress that we in no way claim that our conclusions can extend to other institutions. Our institution is pretty unique in its size, limited focus of degree offerings (we are not a comprehensive university) and selectivity. But we instead hope that our work will inspire others to conduct similar investigations at their institutions. Most of them (at least public, state-supported institutions) face the same accountability issues we have from our state, and it would be fascinating to see how the same student behavior (say, getting a failing grade and having to repeat a class) affects time to graduation across different institutions. P ge 24997.4