A Reliability Generalization Meta-Analysis of the Padua Inventory of Obsessions and Compulsions

The Padua Inventory (PI) of obsessions and compulsions is one of the most usually applied tests to assess obsessive-compulsive symptomatology in research contexts as well as for clinical and screening purposes. A reliability generalization meta-analysis was accomplished to estimate the average reliability of the PI scores and to search for characteristics of the samples and studies that can explain the variability among reliability estimates. An exhaustive literature search enabled us to select 39 studies (53 independent samples) that reported alpha and/or test-retest coefficients with the data at hand for the PI total score and subscales. An excellent average coefficient alpha was found for the PI total score (M = .935; 95%CI = .922–.949) and for Impaired Mental Control subscale (M = .911; 95%CI = .897–.924), being good for Contamination (M = .861; 95%CI = .841–.882) and Checking (M = .880; 95%CI = .856–.903), and fair for Urges and Worries (M = .783; 95%CI = .745–.822). The average test-retest reliability for PI total score was also satisfactory (M = .835; 95%CI = .782–.877). Moderator analyses showed larger coefficients alpha for larger standard deviation of the PI total scores (p = .0005; R 2 = .46), for adapted versions of the test (p = .002; R 2 = .32), and for samples composed of clinical participants (p = .066; R 2 = .10). The practical implications of these results are discussed as well as the need for researchers to report reliability estimates with the data at hand.