Logit models, the area under receiver characteristic curves, sensitivity, and specificity for Co-enrollment density in college networks dataset

Eric Leonardo Huerta-Manzanilla; Matthew W Ohland; Manuel Toledano-Ayala; Juan Carlos Jáuregui-Correa

doi:10.1016/j.dib.2021.107509

Logit models, the area under receiver characteristic curves, sensitivity, and specificity for Co-enrollment density in college networks dataset

Data Brief. 2021 Oct 26:39:107509. doi: 10.1016/j.dib.2021.107509. eCollection 2021 Dec.

Authors

Eric Leonardo Huerta-Manzanilla¹, Matthew W Ohland², Manuel Toledano-Ayala¹, Juan Carlos Jáuregui-Correa¹

Affiliations

¹ Universidad Autónoma de Querétaro, México.
² Purdue University, United States.

Abstract

This article describes the data related to co-enrollment density (CD), a new network clustering index, that can predict persistence and graduation. The data hold the raw results and charts obtained with the algorithm for CD introduced in ``Co-Enrollment Density Predicts Engineering Students' Persistence and Graduation: College Networks and Logistic Regression Analysis.'' There are data for eight institutions that show CD as a predictor for graduation at four years, graduation at six years, and ever graduated. The files were processed using R to estimate CD at one, two, three, and four years. Logistic regression models, receiver operating characteristic curves, specificity, sensitivity, and cut-off points were estimated for each model. The R code to reproduce the metanalysis for the summary data is included. The displays for the logistic regression models, receiver operating characteristic curves, density curves for classes, models, and parameters are included.

Keywords: Logistic regression in education; Receiver operating characteristics roc curves; Sensitivity and specificity; Social network relations strength; Student retention in college.