Daniel J. O’Keefe
Department of Communication Studies
Northwestern University


            One of the defining problems in communication research is understanding the role of persuasive messages in human decision-making. It was one of the first major foci for social-scientific research on human communication and for over fifty years has been a significant and central area of inquiry. Given the large number of persuasion studies generated, it is striking that so little attention has been devoted to research synthesis in this area. My work has taken up the project of organizing and synthesizing this substantial body of work on the effects of messages on persuasion and hence addresses the distinctive problems associated with the development of dependable generalizations about persuasive message effects. This involves both methodological and substantive work, the latter especially, though not exclusively, in the form of meta-analytic research.
            In general my work on persuasion research synthesis might be seen as representing a “bottom-up” approach, in the sense that it seeks to derive and integrate findings from the large number of extant persuasion effects studies, in contrast to an approach that seeks synthesis through the application of a single general theoretical framework to various diverse research phenomena. These two enterprises, of course, are naturally interdependent; the identification of dependable generalizations both feeds and is fed by general theorizing. But (to invoke Isaiah Berlin’s distinction) I am by inclination a fox rather than a hedgehog.
            One requirement for sound generalizations about persuasive message effects is evidence derived from multiple messages (message replications). In primary research on persuasion, the widespread use of single-message research designs—perhaps encouraged by an underelaborated conception of message structure and features—has created barriers to generalization (e.g., because such designs are insensitive to the possibility of between-message variation in effect). Given commonly observed message-to-message variability in effects, only replications (whether between or within studies) provide evidentiary security for generalizations about messages.
            Against this backdrop it may be easy to appreciate the special attractiveness that meta-analytic methods have for research synthesis in persuasion effects, as these methods offer not only systematic means of providing quantitative summaries of a body of research but also specifically a means of addressing weaknesses attendant to single-message designs. Thus although I have pursued the substantive task of persuasion-effects research synthesis through several avenues, including a book that provides a general review of the persuasion literature (O’Keefe, 2016, Persuasion) and broad summary pieces (O’Keefe, 2001, in Encyclopedia of rhetoric; O’Keefe, 2004, in Perspectives on persuasion, social influence, and compliance gaining; O’Keefe, 2006, in Handbook of communication skills; O’Keefe, 2008, in International encyclopedia of communication; O’Keefe, 2013, in Communication Yearbook 36; O’Keefe, in press, in Concise encyclopedia of communication), one important focus of my work has been the application of meta-analytic methods to the persuasion effects literature. To date this meta-analytic work has addressed a variety of specific phenomena, including the effects of variations in gain-loss message framing (O’Keefe & Jensen, 2006, in Communication Yearbook; O’Keefe & Jensen, 2007, in Journal of Health Communication; O’Keefe & Jensen, 2008, in Communication Studies; O’Keefe & Jensen, 2009, in Journal of Communication, O’Keefe & Nan, 2012, in Health Communication; O’Keefe & Wu, 2012, in International Journal of Environmental Research and Public Health), message sidedness (O’Keefe, 1999, in Communication Yearbook), conclusion omission and conclusion specificity (O’Keefe, 1997, in Argumentation and Advocacy; O’Keefe, 2002, in Advances in pragma-dialectics), justification explicitness (O’Keefe, 1998, in Argumentation and Advocacy), guilt appeals (O’Keefe, 2000, in Communication Yearbook; O’Keefe, 2002, in The persuasion handbook), cultural adaptation of advertising appeals (Hornikx & O’Keefe, 2009, in Communication Yearbook), and the door-in-the-face strategy (O’Keefe and Hale, 1998, in Communication Yearbook; O’Keefe and Hale, 2001, in Communication Research Reports). A number of additional meta-analytic reviews of persuasion phenomena are underway.
            The long-term objective of this research is a deeper understanding of persuasion processes and effects. Meta-analytic research is of course synthetic work, aiming at integrating primary research findings in a given line of inquiry. But my synthetic interest also reflects a broader motivation not only to summarize this or that individual body of studies but also to connect otherwise-separated lines of work; for example, I have sought links between research on the door-in-the-face influence strategy and research on guilt arousal as an influence mechanism (e.g., O’Keefe and Figge, 1997, in Human Communication Research; O’Keefe and Figge, 1999, in Communication Monographs; O’Keefe, 2000, in Communication Yearbook; O’Keefe, 2002, in The persuasion handbook; see also O’Keefe, 1999, in Communication Studies). The more general hope is that stitching together the fabric of research findings across diverse phenomena will help unravel the puzzles of persuasion.
            This focal interest in the synthesis of persuasion effects research is naturally aligned with several other related lines of work. One explores the implications of persuasion research for other domains (e.g., O’Keefe and Medway, 1997, in Journal of School Psychology; Kreuter et al., 2007, in Annals of Behavioral Medicine). Of special interest here is the relationship of persuasion research to argumentation studies, in which I have long-standing interests (e.g., O’Keefe, 1977, in Journal of the American Forensic Association; O’Keefe, 1982, in Advances in argumentation theory and research). My work in this area has especially emphasized the interplay of descriptive and normative considerations in persuasion and argumentation (e.g., O’Keefe, 2003, in Proceedings of the fifth conference of the International Society for the Study of Argumentation; O’Keefe, 2006, in Considering pragma-dialectics; O’Keefe, 2007, in Argumentation; O’Keefe, 2012, in Argumentation).
            A second area of related research is conceptual work focused on problems of data analysis and research design, especially as these arise in the study of persuasion. This work addresses issues such as the assessment of persuasive effects (O’Keefe, 1993, in Communication Studies; O’Keefe, 2013, in Communication Yearbook 37), the identification of design features that permit dependable generalization (Jackson, O’Keefe, and Brashers, 1994, in Journalism Quarterly; O’Keefe, 1999, in Document Design), the appropriate analysis of data in common persuasion-effects research designs (O’Keefe, 2003, in Communication Theory; O’Keefe, 2003, in Human Communication Research; O’Keefe, 2007, in Communication Methods and Measures), and the conduct of meta-analyses in this domain (O’Keefe, 1991, in Communication Monographs).
            Two threads run through much of this conceptual work. One is the importance of the close analysis of message properties (e.g., O’Keefe, 1994, in Communication Theory; O’Keefe and Jackson, 1995, in Argumentation and values; O’Keefe, 2003, in Communication Theory; O’Keefe, 2013, in Communication Yearbook 36). The other is the questioning of some familiar and long-standing—but, upon inspection, dubious—research practices. It’s common to think that one should always have message manipulation checks—but one shouldn’t (O’Keefe, 2003, in Communication Theory). It’s common to think that one should use Bonferroni-like alpha-adjustment procedures when multiple significance tests are conducted—but one shouldn’t (O’Keefe, 2003, in Human Communication Research). It’s common to think that when one’s research question concerns the relative persuasiveness of two messages, one should be careful to distinguish attitudinal, intention, and behavioral assessments—but one shouldn’t (O’Keefe, 2013, in Communication Yearbook 37).

Daniel J. O’Keefe home page
Department of Communication Studies
School of Communication
Northwestern University