Reliability generalization meta-analysis of statistical anxiety scales

In this study, in addition to obtaining a reliability value for the statistical anxiety scales, it was aimed to reveal a overall reliability value for each statistical anxiety scale. It was also aimed to reveal the variability of the reliability value of statistical anxiety scales. For this purpose, Web of Science, ERIC, Scopus and Google Scholar databases were searched according to certain criteria and 85 Cronbach's alpha coefficients were obtained. Reliability generalization analysis was applied to the obtained alpha coefficients within the scope of meta-analysis, which is a quantitative method. The average value of the alpha coefficients using transformation under the random effects model was .927(95% CI:.917-.937) and the findings were significant(p<.01). In addition, the average alpha value of each statistical anxiety scale was calculated as .931(95% CI:.917-.942) for STARS, .917(95% CI:.898-.933) for SAS, .918(95% CI:.896-.959) for SAS-10 and .951(95% CI:.941-.960) for WAESTA. In addition, as a result of the reliability generalization analysis for the subscales of the statistical anxiety scales consisting of more than one subscale (STARS, SAS), all subscales were found to be highly reliable(>.70). Analog-ANOVA and meta-regression analyses were conducted to reveal the variability in alpha coefficients. According to the findings, scale types significantly affect the average reliability value(p<.01). In addition, according to the meta-regression findings, the reliability coefficient of the statistical anxiety scales increases as the scale mean score and standard deviation value increases. As a result, it can be concluded that when statistical anxiety scales are used in different cultures and samples, reliable measurements can be made and the source of differences can be revealed.