| Abstract: | Correction Notice: An erratum for this article was reported in Vol 137(6) of Psychological Bulletin (see record 2011-24066-002). In the Samples section of Meta-Analysis 1: Reliability, third paragraph, the number of studies reporting data on clinical samples is incorrect. The sentence “Four studies included clinical samples, and eight studies included nonclinical samples” should read “Twenty-four studies included clinical samples, and eight studies included nonclinical samples.”] In 2 meta-analyses involving 58 studies and 59,575 participants, we quantitatively summarized the relative reliability and validity of continuous (i.e., dimensional) and discrete (i.e., categorical) measures of psychopathology. Overall, results suggest an expected 15% increase in reliability and 37% increase in validity through adoption of a continuous over discrete measure of psychopathology alone. This increase occurs across all types of samples and forms of psychopathology, with little evidence for exceptions. For typical observed effect sizes, the increase in validity is sufficient to almost halve sample sizes necessary to achieve standard power levels. With important caveats, the current results, considered with previous research, provide sufficient empirical and theoretical basis to assume a priori that continuous measurement of psychopathology is more reliable and valid. Use of continuous measures in psychopathology assessment has widespread theoretical and practical benefits in research and clinical settings. (PsycINFO Database Record (c) 2011 APA, all rights reserved) |
