Testing three models for the sleeper effect

For forty years scholars have disagreed about the long term effect a low credible source has on the attitude change of message recipients (the “sleeper” effect). Some have suggested that in the long term a low credible source may actually attain greater attitude change in message recipients than a high credible source (Watts and Holt, 1979; Watts and McGuire, 1975). This review uses meta‐analysis to test three models that have been suggested by researchers. Our analysis concludes that the sleeper effect does exist. The review offers suggestions for practitioners and theorists.