Measuring mathematics anxiety: psychometric analysis of a bidimensional affective scale

Journal of Instructional Psychology, Sept, 2009 by Haiyan Bai, LihShing Wang, Wei Pan, Mary Frey

The purpose of this study is to develop a theoretically and methodologically sound bidimensional affective scale measuring mathematics anxiety with high psychometric quality. The psychometric properties of a 14-item Mathematics Anxiety Scale-Revised (MAS-R) adapted from Betz's (1978) 10-item Mathematics Anxiety Scale were empirically analyzed on a college sample for its internal consistency reliability, parallel-item reliability, and construct validity. In addition, the bidimensionality of MAS-R was verified through the two-factor measurement model that fitted to the data significantly better than the one-factor model. Therefore, MAS-R is a psychometrically reliable and valid instrument for measuring bidimensional mathematics anxiety. Implications of the bidimensional affective scale in measuring mathematics anxiety are discussed.

Keywords: mathematics anxiety, Mathematics Anxiety Scale, mathematics education, psychometric property, factor analysis

**********

Mathematics education has been in the center stage of accountability reform since the No Child Left Behind Act went into effect in 2002 (NCLB, 2002). Mathematics literacy is regarded as a "civil issue" because of its leverage in mitigating social and economic inequity (Schoenfeld, 2002). Students and educators alike are under unprecedented pressure to achieve state proficiency standards and to close the achievement gaps (Farmer, 2005; Harris & Herrington, 2006; Jones, Jones, & Hargrove, 2003; Taylar, 2006; YeopKim, Zabel, Stiefel, & Schwartz, 2006). Among the many correlates of mathematics achievement, mathematics anxiety has been shown to be a significant factor of learning success (Cates & Rhymer, 2003; Hembree, 1990; Pajares & Miller, 1994; Ramirez & Dockweiler, 1987; Ryan & Ryan, 2005; Singh, Granville, & Dika, 2002). Individuals with high mathematics anxiety tend to perform poorly when presented with mathematics stimuli (Cates & Rhymer, 2003) with a common population correlation estimated at -.27 (Ma, 1999). Furthermore, individuals with mathematics anxiety have shown to avoid environments and careers that require the utilization of mathematics skills (Ashcraft, 2002). Therefore, it can be inferred that mathematics anxiety greatly impacts mathematics education and students' career choice.

Research on mathematics anxiety has shown that it is a multidimensional psychological construct that involves complex factors, such as feelings of pressure, performance inadequacy, and test anxiety that interfere with the manipulation of numbers and solving mathematical problems in a wide variety of ordinary life and academic situations (Kazelskis, 1998; McMorris, 2004; Rounds & Hendel, 1980). The measurement of mathematics anxiety is important for understanding the nature of this construct and the degree of its presence for intervention planning and instructional delivery (Hembree, 1990; McMorris, 2004; Newstead, 1998).

Over the past three decades, researchers have focused on developing self-report instruments to measure mathematics anxiety. A pioneering work on measuring mathematics anxiety was the Mathematics Anxiety Rating Scale (MARS) (Richardson & Suinn, 1972), a 98-item instrument with a 5-point Likert scale designed to measure college students' mathematics anxiety. MARS was once regarded as the best available measure of mathematics anxiety with the highest reliability and validity (Dew, Galassi, & Galassi, 1983; Hopko, Mahadevan, Bare, & Hunt, 2003; Richardson & Woolfolk, 1980); however, the 98-item instrument was cumbersome to administer (Pajares & Urdan, 1996), and it has been out of print since 1994.

Following the publication of MARS, several revisions were developed. For example, the Mathematics Anxiety Rating Scale for Adolescents (MARS-A) (Suinn & Edwards, 1982) and the Mathematics Anxiety Rating Scale for Elementary School Students (MARS-E) (Suinn, Taylor, & Edwards, 1988) were developed. MARS-A consists of 98 questions with a 5-point Likert scale. The Cronbach's alphas, i.e., internal consistency reliability coefficients, ranged from .89 to .96, but no validity information was clearly reported (McMorris, 2004). MARS-E has 26 items with a 5-point Likert scale, and its Cronbach's alpha was .88. The two primary factors, mathematics test anxiety and mathematics performance adequacy anxiety of MARS-E, however, accounted for less than 30% of the variance (McMorris, 2004). In his review of MARS, MARS-A, and MARS-E, McMorris (2004) suggested that additional or revised measures need to be developed to improve the measurement of mathematics anxiety.

To reduce the test length, Plake & Parker (1982) modified MARS and developed the Math Anxiety Rating Scale-Revised (MARS-R), a shorter, 24-item version of the 98-item MARS. Although it yielded a high reliability estimate of .98, MARS-R was demonstrated to be a remarkably poor fit to a large sample of undergraduates (Hopko, 2003). The latest measure of mathematics anxiety is Hopko et al.'s (2003) Abbreviated Math Anxiety Scale (AMAS), an adapted version of MARS-R with only nine items. AMAS was reported to have two factors, learning mathematics anxiety and mathematics evaluation anxiety, which accounted for 70% of the variance from an exploratory factor analysis using principal components extraction with varimax rotation. However, some factor loadings were found to be non-significant, and the varimax rotation is dubious because its orthogonality assumes no correlation between the factors, which runs counter to the concept that the two factors should be related with each other.