Self-Concept and Subjective Value as Predictors of Math Anxiety: Latent Interaction Modeling Approach

• In Event: Personal and Contextual Factors Impacting the Development and Expression of Math Anxiety

Abstract

Objectives & Theoretical Framework

The objective of this study was three fold: (1) to investigate the role of self-concept and subjective value in predicting math anxiety (2) to examine the additional benefit of the latent interaction between self-concept and subjective value in predicting math anxiety, and (3) to evaluate whether the relationships between self-concept, subjective value and math anxiety generalize across gender.
Despite its significant negative impact on math achievement and career choices of individuals and its negative impact on a nation’s resource base in STEM, far too little attention has been paid to the causal antecedents of math anxiety. The current study used Control-Value Theory (CVT) of academic emotions (Pekrun,2006) as a theoretical framework. Briefly, CVT proposes that perceived ability to do well in math and perceived value of math as well as the interaction between the two variables are most proximal causal antecedents of math anxiety.

Method

The current research employed data from the Program for International Student Assessment (PISA) 2003 U.S. sample (N= 5456, 49.8 girls, diverse racial/ethnic groups), The math self-concept scale consisted of five items measuring students perceived competence in mathematics. Subjective value of mathematics was measured using four items assessing the usefulness of math for the students’ future. Math anxiety was measured using five items that assessed the students’ anxiety towards mathematics.
All items in the measures were rated on a 4-point scale ranging from ‘strongly agree’ to ‘strongly disagree’. The three measures have very good psychometric properties (OECD,2004).

Results and Discussion

To examine the influence of self-concept, subjective value and the interaction between the two variables, we used the unconstrained approach to latent variables interaction modeling within SEM framework(Marsh etal.,2006). We used multigroup SEM to examine whether the relationships between self-concept, subjective value and math anxiety generalize across gender. However, before we tested the equivalence of the structural relations between the variables, we conducted multi-group confirmatory factor analysis to test for measurement invariance across gender. The overall sample SEM in which self-concept, value and the interaction between the were posited to predict math anxiety fitted the data well (χ2(113, N=5456)= 2591.37; CFI=. 97; TLI=. 98; RMSEA=. 06). Whereas self-concept predicted math anxiety negatively (β= -.77; t = -51.26), subjective value predicted math anxiety positively (β= .10; t = 6.65.). More importantly the effect of latent interaction was also significant but negative (β= -.05; t = -3.44). A closer examination of the slopes showed that, the higher one's value, the more negative was the slope between self-concept and anxiety. The multi-group confirmatory factor analysis established the invariance of the measurement model across gender. The multi-group SEM analysis showed that the effects of self-concept, value and the interaction between the two were all significant in both girls and boy samples. However, the magnitudes of the effects were a little higher for boys than for girls. The findings of the current study lend support to CVT and suggest although both self-concept and subjective value are important predictors of math anxiety, having the right balance of the two seems to be more adaptive.

Author

• Wondimu Ahmed, The University of Akron