Development of a longitudinal data base for tracking and analysis of attrition data
This study examined factors related to student persistence in college, with particular attention to engineering students. All direct from high school and transfer students who entered a large midwestern university in the engineering college were tracked through fourteen semesters and summer sessions between fall 1990 and spring 1995. The study was conducted for the following goals and purposes: (1) To assemble and design a data base that would allow for the characterization and tracking of all undergraduate engineering students. (2) To examine the relationships between significant variables in a model that could identify students potentially at risk of attrition. (3) To provide baseline retention data that could be compared to data from other cohort years and after retention initiatives are implemented;In order to accomplish these objectives it was necessary to obtain and reformat a very large data base. A dependent variable called "category of risk" was established to classify students according to their progress toward the goal of completing their engineering degrees;Statistical analyses were performed to provide college-wide, descriptive information. Methodologies were developed to predict which students were likely to be lost due to attrition. It was believed that if students who were at higher risk for attrition could be identified early in their college careers, appropriate interventions could be developed to reduce the likelihood of attrition;The approach taken in this study was supported by the information found in the review of the literature. The articles reinforced the necessity for individual colleges or institutions to monitor, analyze, and to the greatest extent possible predict student retention. The study will add to the current body of knowledge about retention and contribute to the spectrum of ideas that feed into the various theories of retention. It could also serve as a model for other universities interested in developing similar data bases or descriptive, predictive, or comparative measures. The statistical methods used in this research could be adapted to other sets of independent and dependent variables.