This experiment tested the results of a meta-analysis of message sidedness (Jackson & Allen, 1987). The results are consistent with the prior meta-analysis. Two-sided refutational messages were more persuasive than one-sided messages, which were in turn more persuasive than two-sided nonrefutational messages. Testing the generalizability of this finding using 17 messages found consistent effects across the messages. F OR more than 40 years, research has investigated the impact of message features in persuasion. In the last 10 years that research has changed radically. The methodology of meta-analysis, i.e., accumulating results across studies, has been improved and is widely accepted in the social sciences. The results of meta-analyses on persuasive message features demonstrate consistencies in bodies of literature previously thought to be inconsistent. One recent example serves as the focal point for this research. A recent meta-analysis of message sidedness research in persuasion discounted popular explanations for sidedness effects and advanced a previously unconsidered explanation (Jackson & Allen, 1987).The basic issue in the message sidedness literature is whether it is more persuasive to present only arguments favoring the position advocated by the source (a one-sided message) or both opposing and supporting arguments (a two-sided message). Past explanations for sidedness effects asserted that the effects were moderated by the message recipient's initial attitude toward the source's position, topic familiarity, and intelligence (Hovland, Lumsdaine, & Sheffield, 1949). Specifically, one-sided messages were posited to be more effective than two-sided messages when the recipient initially agreed with the source's position, was unfamiliar with the topic, or unintelligent. When the recipient disagreed with the source, was familiar with the topic, or intelligent, a two-sided message was posited to be more effective. While these popular explanations were frequently advanced, the considerable experimental data were inconsistent with them.In an effort to discern consistencies in the literature, Jackson and Allen (1987) meta-analyzed 38 message sidedness studies. From the meta-analytic data they reached two important conclusions. First, the data were at odds with previously hypothesized effects for one's initial attitude, familiarity with the topic, and intelligence. Second, they found sidedness effects were moderated by the type of two-sided message utilized in any given study.In scrutinizing the sidedness literature, Jackson and Allen (presence of three types of messages in which the source advocates his or her position on some issue: (a) a refutational two-sided message in which the possibility of another position on the issue is raised and refuted, (b) a nonrefutational two-sided message where the existence of another position on the issue is acknowledged and noted to be undesirable, but where no attempt is made to demonstrate why it is undesirable, and (c) one-sided messages where only the source's position on ...