Premium
An Engineering Student Retention Study
Author(s) -
MollerWong Cheryl,
Eide Arvid
Publication year - 1997
Publication title -
journal of engineering education
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 3.896
H-Index - 108
eISSN - 2168-9830
pISSN - 1069-4730
DOI - 10.1002/j.2168-9830.1997.tb00259.x
Subject(s) - attrition , logistic regression , descriptive statistics , psychology , set (abstract data type) , multitude , population , regression analysis , data collection , tracking (education) , computer science , mathematics education , medical education , statistics , mathematics , demography , medicine , pedagogy , sociology , philosophy , dentistry , epistemology , programming language
National engineering enrollment peaked in the early 1980's and, with rare exception, has declined or remained flat for the past fifteen years. Historically, engineering enrollment has focused on new student recruitment, but recently much more attention has been directed toward the issue of student retention. Our efforts at Iowa State University to examine retention issues were divided into two parts. Phase I of the study, which is presented in this paper, was targeted to accomplish several objectives. First, we had to design and assemble a data base that would allow for individual tracking of students. Once a complete profile of our students population had been assembled, it was possible to identify accurately a range of descriptive variables. Next, using the established data base we developed a retention analysis tool that would statistically suggest and identify students who are potentially at risk of attrition. To do this we examined a range of independent variables against a set of dependent risk categories. Statistical analyses and logistic regression methods were performed to provide a predictive model. Phase II of this study is currently in process and involves a multitude of attrition variables which were not considered, quantified, or integrated into Phase I. When factors such as background, social integration, attitude, etc. can be successfully measured and integrated into the data base the accuracy of our current model is expected to improve.