A meta-analytic reliability generalization of the Physical Self-Description Questionnaire (PSDQ)

ObjectivesThis study aims at providing meta-analytically derived information on the reliability of test scores obtained with the Physical Self-Description Questionnaire (PSDQ). The following research questions are addressed: (1) What is the mean internal consistency reliability (coefficient Alpha) of PSDQ scores across studies to date? (2) Which factors influence the reliability estimates?DesignWe conduct an extensive literature search for studies using the PSDQ in relevant databases. Overall, 49 studies using either the whole PSDQ or a selection of PSDQ subscales are analyzed.MethodsWe use the technique of reliability generalization (RG) to pool internal consistency estimates from extant applications of the PSDQ. Moderator analyses are carried out to identify the impact of selected study variables on the reliability estimates. We use a total of six moderator categories (original vs. translated version, complete test vs. selected subscale(s), gender composition of the sample, age group, sample size, and year of publication).Results54.4% of the studies analyzed report reliability data for the sample at hand. Mean reliability estimates within all subscales are above α = .8, with the subscale Health presenting the lowest mean reliability estimate at α = .823 and the subscale Body fat the highest at α = .943. The moderator analyses reveal influences of moderators on distinct PSDQ scales.ConclusionOverall, the PSDQ accomplishes good internal consistency based on the extant empirical evidence. Most importantly, however, the moderator analyses show clear evidence of a reduction in the reliability coefficients due to using a translated questionnaire version.

► We conduct a reliability generalization study of the Physical Self-Description Questionnaire (PSDQ). ► Overall, all PSDQ subscales reveal good internal consistency (Cronbach's α). ► Moderator analyses show that the reliability of certain subscales is affected by a number of factors.