I Can Do It! Expectancy as a Mediator of the ST Math Effect on Math Achievement

• In Event: Facets of Large-Scale Evaluation Illustrated With ST Math: Examining Outcomes, Mediators, and Moderators

Abstract

This paper aims to identify how STMath influences students’ math learning as measured by the CST. We propose that one mechanism is increased expectancy for math—students’ self-prediction of how well they will do on an upcoming task (Eccles, 1983).

We study math motivation from the expectancy-value framework perspective. In colloquial terms, one’s expectancy beliefs answer the question, “Can I do it?” Research has consistently found that even after accounting for ability, higher expectancies are associated with later achievement, choice behaviors, and persistence (Wigfield & Eccles, 2000). Students’ progression through STMath’s engaging games and curriculum can reinforce and build expectancy beliefs. STMath allows individually-tailored instruction and provides immediate feedback, allowing students to train their appraisals of “Can I do it?” or judgments of their ability to accomplish math tasks. When students master the material, then they can move onto more advanced concepts. It is thought that as students play STMath over the school year, they experience an increase in their beliefs about their ability to succeed on upcoming math tasks, and this boost in math expectancy explains some of the positive change in achievement.

Characteristics of the learning environment matter for students’ expectancy for success. Previous studies have found that the relationship between the classroom climate and students’ grades (Pajares & Miller, 1994) and standardized test scores (Fast et al., 2010) were mediated by increased expectancies. This current study examines math expectancy within a newly emerging learning context, educational software implemented in schools.

This study examines two years of student data (2009-10 and 2010-11) collected from the 52 schools in the STMath randomized experiment. Math expectancy was assessed at the end of both school years using the Eccles et al. (1993) expectancy-value scales. We collected demographic information such as gender, grade, ethnicity, whether an ELL, and free or reduced lunch status. To test for mediation, three ordinary least squares regression models were estimated: 1) end-of-the-year math CST scores regressed on whether student received STMath, 2) students’ reported math expectancy regressed on STMath treatment status, and 3) math CST regressed on both STMath treatment status and math expectancy (Baron & Kenny, 1986).

For the first year of STMath implementation (2009-10), we examined 321 third through fifth graders. We found that STMath increased expectancy, which in turn increased math CST (17% mediation). We attempted replication for the second year of the study (2010-11) with a sample of 362 fifth graders who had pretest (2008) CST data prior to the beginning of STMath. We were able to reproduce the positive effect on expectancy (d = 0.37, p < .01) after two years of STMath but the mediation results did not rise to the level of significance.

Our research extends previous work on elementary students by testing the expectancy-value framework in a new context, an innovative mathematics software program widely used in public U.S. schools. Given the consistent research findings demonstrating the importance of expectancy beliefs, this study highlights the value of considering math motivation--such as expectancy--when designing and evaluating educational programs.

Authors

• Arena C. Lam, University of California - Irvine

• Teomara Rutherford, North Carolina State University