Developing a model to explain IPEDS graduation rates at Minnesota public two -year colleges and four-year universities using data mining.

摘要

All postsecondary education institutions participating in Title IV financial assistance programs are required by the Student Right-to-Know Act to make available reports containing the graduation rate. The National Center for Education Statistics (NCES) collects graduation rate data with the Integrated Postsecondary Education Data System (IPEDS) Graduation Rate Survey (GRS).;The purpose of this study was to develop models using one national source of data that explain IPEDS graduation rates at both Minnesota State system public two-year colleges and four-year universities using data mining techniques. Three questions were addressed: (1) What is the relationship between IPEDS graduation rates and institutional characteristics? (2) Given these relationships, what are the predicted graduation rates? (3) How do predicted graduation rates compare to actual graduation rates at Minnesota State system institutions?;The population for this study was all postsecondary institutions responding to the IPEDS GRS in 2003. To develop the models institutions were segmented into eight sectors based on control and level of the institution. Data mining of 1,000 variables for these 5,771 institutions using the Clementine software identified 51 predictor variables. Most of the predictor variables were found in the Enrollment Survey and the Institutional Characteristics Survey. Each institution was placed in a peer group based on values of the predictor variables. A weighted predicted graduation rate was calculated for each peer group.;Based on the findings in this study, the following conclusions are drawn: (1) Data mining can be used to identify relationships between IPEDS graduation rates and institutional characteristics for all sectors of higher education using one data source and one method. The relationships and predictors differ by sector and by peer groups in each sector. (2) Using data mining, predicted graduation rates can be calculated for all sectors of higher education. (3) Predicted graduation rates can be compared to actual graduation rates at Minnesota State system institutions. The correlation between actual and predicted rates is higher using data mining techniques than using current methods. (4) Data mining can identify predictor variables not previously identified in the literature.