Poster #70 - A Meta-analysis Investigating the Relation Between Math Anxiety and Math Achievement

• In Event: 2-163 - Poster Session 08
  In Poster Session: PS 08 Section - Education, Schooling

Integrative Statement

Introduction
Mathematics knowledge is vital to academic success and everyday math-related functioning (Robinson, 2000). Previous meta-analyses have found a significant and negative relation between math achievement and math anxiety, or the fear of math (Hembree, 1990; Ma, 1999); however, no meta-analysis has been done to summarize the collection of research in this area from the last 20 years. Thus, the present meta-analysis provides an updated synthesis on 1) the overall average correlation between math anxiety and math achievement, and 2) whether this relation is moderated by grade level, gender, continent, race/ethnicity, math anxiety and achievement measures, or math content area.
Method
We searched for studies with relevant data available published from 1992 to 2018 through online journal databases and a developmental science listserv. We included 224 studies representing 752 correlation coefficients from 334 independent samples (N = 386,548).
Results
Average correlations for the overall sample and subgroups within each moderator are shown in Table 1. The overall average correlation between math anxiety and math achievement was small-to-moderate, negative, and statistically significant (r=-0.28). A significant amount of variance was unaccounted for in this overall effect size (I2=90.39). Gender and race/ethnicity were not found to moderate the relation, but several other significant moderators of the relation were found (see Table 2 for selected comparisons). Students in grades 3-5 (r=-.21) had a significantly lower average correlation compared to all other grade categories, except for postsecondary students. Postsecondary school students (r=-.24) had a significant weaker relation than students in grades 6-8 (r=-.33) and 9-12 (r=-.33). Samples from North America (r=-.27) had significantly weaker relations than samples from Asia (r=-.32).
We also found significant differences in correlations between several math anxiety and achievement measures. The Math Anxiety Scale for Young Children (r=.12) and the Child Math Anxiety Questionnaire (r=.16) had significantly stronger correlations than studies that used the Math Anxiety Questionnaire. The Fennema Sherman Math Anxiety Scale (r=.37), Achievement Emotion Questionnaire Math Anxiety scale (r=.42), and Math Anxiety Scale by Betz (1978; r=.35) were also found to have significantly stronger relations with math achievement than the other postsecondary/adult measures. In addition, the correlation between approximate number system skills and math anxiety (r=-.12) was significantly weaker than the relations with all other math content areas. Basic number knowledge (r=-.18) and whole number calculation (r=-.23) correlations were also found to have significantly weaker relations than relations with word problem solving (r=-.33) and fractions, decimals, and percentage knowledge (r=-.37). A funnel plot suggested the potential for publication bias in this literature, however the Fail Safe N indicated that 4,903,263 articles with null results would need to exist to make the results non-significant.
Discussion
Our results indicate that a significant relation exists between math anxiety and math achievement starting as early as childhood, remains significant through adulthood, and differs depending on the sample or measure characteristic. These findings provide support for interventions to improve math skills and reduce math anxiety at all ages, although targeting anxiety and achievement for more complex types of math content may be prudent.

Authors

• Connie Barroso, Florida State University
  Presenting Author

• Colleen M Ganley, Florida State University
  Non-Presenting Author

• Amanda L McGraw, Florida State University
  Non-Presenting Author

• Elyssa A Geer, Florida State University
  Non-Presenting Author

• Mia Daucourt, Florida State University
  Non-Presenting Author

• Sara Hart, Florida State University
  Non-Presenting Author