Testing the Ability of Two Series of Models to Predict High School Graduation Status

• In Event: Marriott Roundtable Session 7
  In Roundtable Session: Applied Methods and Measurement in Education Research

Abstract

Using secondary data, two series of predictive models aimed at projecting high school graduation status were created and tested. The first series of models tested individual academic and nonacademic variables together with school-level variables. The second series of models tested similar individual-level variables along with community-level predictors. Logistic regression and multilevel logistic regression analyses were conducted to analyze the data. Most of the individual predictors included in the models yielded findings similar to those found in previous empirical work; however, this was not true for all of the predictor variables included. These differences highlight the tension that can exist between generalizability and local specificity, particularly in challenging school contexts whose demographics differ from those included in larger studies.

Author

• David T. Marshall, Auburn University