Reliability generalization meta-analysis of nine mathematics anxiety scales

Abstract The purpose of this study is to estimate pooled reliability coefficients for nine commonly used math anxiety scales (Mathematics Anxiety Rating Scale [MARS], Revised Mathematics Anxiety Rating Scale [RMARS], Abbreviated Version of Mathematics Anxiety Rating Scale [A-MARS], Mathematics Anxiety Rating Scale for Adolescents [MARS-A], Suinn Mathematics Anxiety Rating Scale for Elementary School Students [MARS-E], Abbreviated Version of Mathematics Anxiety Scale [AMAS], Revised Mathematics Anxiety Scale [RMAS], Suinn and Winston Mathematics Anxiety Rating Scale Short Version [SWMARS], Mathematics Anxiety Scale [MAS]). The number of studies reporting alpha coefficients for these scales ranged from 11 to 120. Separate reliability generalization meta-analyses were conducted using Bonett-transformed values to estimate the pooled reliability coefficient for each scale. The pooled reliability coefficients were estimated as follows: MARS = 0.964, RMARS = 0.933, A-MARS = 0.937, MARS-A = 0.918, MARS-E = 0.891, AMAS = 0.872, RMAS = 0.907, SWMARS = 0.931, and MAS = 0.898. The study found that different scales were affected by a number of statistically significant factors. These included the school level (MARS, RMARS, A-MARS, AMAS, and MAS), the country (MARS, RMARS, A-MARS, MARS-A, SWMARS, and MAS), the number of items (MARS, A-MARS, MARS-A, MARS-E, and MAS), the female percentage (AMAS, RMAS, and SWMARS), the sample size (RMARS and MAS), and the publication year (MARS). The findings led to several suggestions for the researchers.