Multiple Linear Regression: A Return to Basics in Educational Research

Abstract

A content analysis of the American Educational Research Journal and the Educational Evaluation and Policy Analysis journal for the use of multiple linear regression (MLR) was conducted. Two hundred articles were randomly sampled and coded to determine if basic reporting guidelines were followed. Results showed that standard reporting methods for MLR were not followed and the use of stepwise MLR was on the rise. Manuscripts using MLR did not apply necessary corrections for inflated Type I error rates. The majority of the sampled articles did not include key summary statistics, which violated the American Educational Research Association’s principle of transparency. The lack of consistency in reporting hindered critique of work, meta-analysis, and theory development.