Assessing Licensure Tests as an Indicator of Teachers' Science and Mathematics Content Knowledge for Teaching

• In Event: Developing and Validating Indicators for Teachers' Science and Mathematics Content Knowledge for Teaching

Abstract

It makes a great deal of sense given growing evidence about the importance of teacher quality for student achievement that policymakers are focusing on the human capital of the nation’s STEM teachers (e.g., National Research Council 2013; PCAST 2010) as a lever for improving STEM outcomes in K-12 schools. Unfortunately, the current options for tracking the progress that the nation is (or is not) making toward having a high-capacity STEM teacher workforce are somewhat limited. In particular, while a number of potential indicators of STEM teachers’ content knowledge for teaching—e.g., licensure status and degree level—have face validity, these indicators are not highly predictive of student achievement at the secondary level (Wilson, Floden, & Ferrini-Mundy, 2001).
We argue here that there is a potential indicator of teachers’ content knowledge for teaching that is already being collected in nearly every state in the country: teacher licensure test scores. A small body of empirical research at the elementary level finds evidence of the predictive validity—i.e., ability to predict the achievement of students assigned to teachers with varying licensure scores—of some licensure tests (Clotfelter, Ladd, Vigdor, & Wheeler, 2006; Author, 2013a; Author, 2007). There is also preliminary evidence, based on cross-subject comparisons, of a relationship between teacher licensure test scores and student achievement in high school (Clotfelter, Ladd, & Vigdor, 2010), but no evidence that relies on within-subject comparisons (i.e., comparing STEM teachers to other STEM teachers).
To fill this gap in the literature and assess the predictive validity of teacher credential exam scores on student performance in secondary math and science classes, we use data on 2,405 teachers in Washington State who took at least one of seven different credential exams and who are matched to 145,428 student – year observations between the 2009-10 and 2012-13 school years. We focus on middle school math, high school Algebra, and high school Biology classrooms because it is possible to apply standard value-added methodologies (i.e., measures of both prior-year and end-of-year assessments of student learning are available at these levels) and generate a within-subject estimate of the predictive validity of teacher credential test scores.
Preliminary estimates from these models suggest that, all else equal, students who are assigned to teachers with higher credential exam scores do modestly better on math and biology tests in grades 7 – 9. Specifically, students assigned to a teacher scoring in the top quartile of the state’s WEST-B math credential test are predicted to score between 0.05 and 0.12 standard deviations higher than similar students assigned to a teacher who scored in the lowest quartile. These findings are robust to the inclusion of school and school-by-year fixed effects, and persist in models that are identified solely by teachers who teach in multiple grades and schools during our years of data (Hendricks, 2015). These results suggest that teacher credential exams scores could be used as a pre-service indicator of math and science teachers’ content knowledge for teaching.

Authors

• Dan Goldhaber, American Institutes for Research

• Roddy Theobald, American Institutes for Research

• Trevor Gratz, Center for Education Data and Research