Date of Award

Fall 2013

Degree Type


Degree Name

Doctor of Philosophy (PhD)


Engineering Education

First Advisor

P.K. Imbrie

Committee Chair

P.K. Imbrie

Committee Member 1

Sean Brophy

Committee Member 2

James D. Jones

Committee Member 3

Teri Reed


Every year a group of graduates from high schools enter the engineering programs across this country with remarkable academic record. However, as reported in numerous studies, the number of students switching out of engineering majors continues to be an important issue. Previous studies have suggested various factors as predictors for student retention in engineering. To assist the engineering students with timely advising early in their program, an effective prediction model of matriculation and retention in engineering that use available student data are highly desirable.

In the first part of this work, the author developed new prediction models of student retention based on four different modeling methods and five different sets of predictors. The four modeling methods are logistic regression, discriminant analysis, structural equation modeling and neural networks. Independent variables include high school performance indices and a collection of affective and attitudinal factors. These models are intended to help identify students at risk of leaving engineering in the early stages. The prediction performances from different methods were then compared to evaluate the strength and weakness of competing modeling methods. In the second part of this study, the knowledge and modeling skills obtained earlier were applied to develop prediction models for retention and graduation of female and male engineering students. Our findings suggest there are remarkable difference between the predictors of retention between women and men engineering students. In the third part of this work, student success models based on earlier cohorts were developed to predict student retention for cohorts in following years and promising prediction accuracy are reported.