Meta-Analytic Reliability Generalization Study of Perceived Stress Scale in Türkiye Sample

The aim of the study is to examine the meta-analytic reliability generalization of the 14, 10, and 8-item forms of the Perceived Stress Scale, which was developed by Cohen, Kamarck and Mermelstein in 1983 and translated into Turkish by different researchers (Erci, 2006; Yerlikaya & & Idot;nan & ccedil;, 2007; Eskin et al. 2013, etc.) between 2006 and 2013, for the theses produced in T & uuml;rkiye. For this purpose, how different moderator variables affect the reliability coefficients and publication bias were also examined. A total of 81 Cronbach Alpha coefficients from 78 studies, selected in accordance with the established criteria, were included in the meta-analysis. Reliability generalization was applied in the data analysis. The analyses were conducted using the random effects model with coefficient values converted through the Bonett method. In the study, the effect size value was found to be .82 (95% CI: .80, .83), and it was concluded that the sample type (alpha= .81) and the study area (alpha= .81) moderator variables had a statistically significant effect on the reliability estimation in terms of heterogeneity of effect sizes. This revealed that these two moderator variables affected the estimations of the reliability coefficients. In addition, it has been observed that other moderator variables such as age, gender, and the number of items in the scale are not sources of heterogeneity and have no effect on reliability estimation. From this, it was deduced that the scale works well enough to generalize to different contexts with different item numbers. Finally, according to the results of the analysis for the determination of publication bias, it was seen that there was no publication bias in the study.