A number of studies have quantified the problem of high attrition in engineering programs, especially after the first year of study. For example, one large scale study by Astin (1993) involving approximately 25,000 students in over 300 institutions found that less than half of the students who began in engineering graduated with an engineering degree. Another study showed first-year attrition from engineering increased from about 12% in 1975 to 25% in 1990. A report issued by the The National Academies shows undergraduate programs in science and engineering to have some of the lowest retention rates among academic disciplines.

Studies examining students leaving engineering have shown that those who persist and those who leave engineering are quite similar in cognitive ability and scholastic achievement, but tend to show differences in affect characteristics. Models designed to predict retention using multiple measures of academic performance and affect attributes show the greatest promise of predicting student retention.

Therefore, the goal of this study was to develop a hybrid artificial neural network (NN) model capable of predicting student success after the first year. Such a model has the potential to help faculty and advisors provide informed course selection advice to beginning engineering students and/or could facilitate the development of intervention programs to improve retention. The NN model used a combination of a student’s prior academic performance as well as his/her self-reported affect characteristics as input from which student success after the first-year was output. For the purpose of this presentation, "student success" is operationalized as retained in engineering. Student data was collected at a large Midwestern University prior to the start of the students first year during 2004 (N=1595), 2005 (N=1814) and 2006 (N=1838). These data were used as input to artificial neural network (NN) models developed to predict students’ retention.

P.K. Imbrie, the Director of the College of Engineering Honors Program and an Associate Professor in the School of Engineering Education. He holds B.S., M.S. and Ph.D. degrees in Aerospace Engineering from Texas A&M University.

An advocate for research-based approaches to engineering education, curricular reform, and student retention, Dr. Imbrie conducts research in epistemologies, assessment, and modeling of student learning, student success, student team effectiveness, and global competencies. He helped establish the scholarly foundation for engineering education as an academic discipline through lead-authorship of the landmark 2006 JEE special reports "The National Engineering Education Research Colloquies" and "The Research Agenda for the New Discipline of Engineering Education." He co-led the creation of the First-Year Engineering Program’s Ideas to Innovation (i2i) Learning Laboratory, a design-oriented facility that engages students in team-based, socially relevant projects. Directly building on the enhancements to the first-year experience, he led the First-Year Honors Program toward an enabling environment that attracts and retains the highly motivated and academically successful student with a broader and more enriched educational experience promoting scholastic achievement, breadth of knowledge, global awareness, and leadership development.

He has been involved with various research projects sponsored by NSF, NASA, and AFOSR, ranging from education-related issues to traditional research topics in 3D stress measurements using photo-stimulated luminescence spectroscopy (PSLS) and environmental effects on titanium-based metal matrix composites. Prior to joining Purdue, Dr. Imbrie was involved with NSF’s Foundation Coalition, during which time he established himself as one of the nation’s experts on collaborative learning and teaming. He is co-author of a text on teaming called Teamwork and Project Management.