Teacher Certification Exams: Predicting Failure on the TExES History (8-12) Content Exam (A Nonparametric Approach using Classification Trees)

Abstract

Previous research efforts concerning teacher certification in Texas focused primarily on the Pedagogy and Professional Responsibilities exam; an exam that all teacher candidates must pass regardless of their specific content area. Few studies have attempted to explore which variables are useful for predicting the outcome of the TExES content-area certification exams, which represents a major gap in the literature. Because of its high failure rate, this study focused on identifying factors that were influential in predicting failure on the TExES History (8-12) certification exam. A convenience sample was used and only those who had taken the TExES History (8-12) exam from 2002 – 2008 were selected (n = 181).

The study is an exploratory data design using classification trees—a nonparametric statistical technique often associated with data mining. The study was different from previous studies in two important aspects: a) the study included a much wider range of variables, and b) nonparametric, classification tree methodology was used to build predictive models.

Using the proportional chance criterion and Press’ Q to assess significance, the models were statistically significant (p < .05), indicating that the models were capable of predicting outcomes well beyond what would be expected based on chance. Because classification trees produce a set of decision rules that can be graphically depicted, a model based on a decision tree paradigm is more intuitive, and more easily interpreted and implemented compared to regression methods.

Although classification trees are not widely used in social science research, the success of the technique in the current study suggests that classification trees can be an effective, nonparametric alternative to the more traditional multiple regression and logistic regression methods and provides researchers a glimpse of the capabilities of classification trees.