]CHIal -[uty‘s \’-1(1/-k.r-<el-16.Jdcker, j,ner:\-01{(l,in, {//t?l-hi-‘.’4on?1.ia -:il-,.[.4 noru-(I.n.!en-2.)r1-1.“’r.-‘t?).7,“.~.Message-Evoked Thoughts: PersuasionResearch Using Thought VerbalizationsPETER WRIGHT*Verbalizations of the thoughts evoked by an advocacy message are used asdata in many persuasion experiments. In this article, the theory and meas-urement principles underlying such research are examined, and a review ofempirical evidence on the relationship between experimental treatments, ver-balized thoughts, and ensuing topical judgments is presented. The conclu-sion is that verbalized thoughts may be valid indicants of some audienceprocessing activities.Persuasion theories typically postulate cognitiveactivities as mediators between message trans-mission and message-caused changes in behavior. Inthis paper, research on message-evoked thinking asrevealed in a receiver’s thought verbalizations is re-viewed. In the last decade, several dozen experimentsusing thought verbalizations as data have been re-ported, many by researchers concerned with consumerreactions to advertising. However, these have ap-peared in diverse places, and have not shared a unifiedconceptual or methodological basis. Moreover, thevalidity of any self-report on mental activity is con-troversial (Ericcson and Simon 1978; Nisbett and Wil-son 1977; Smith and Miller 1978; Weitz & Wright 1979;Wright & Rip 1980b). The intent here is to examinethe particular uses of such data in persuasion researchdnd to take stock of what has been learned about suchmeasures per se and, through such measures, aboutpersuasion processes,The way in which thought samples have been usedin persuasion experiments is reviewed and thought-sampling procedures are evaluated. Then studies bear-ing on the validity of such data are examined; thesedeal with the relationship of the reported thoughts toantecedent variables, like the transmission medium,or to other measures of mental behavior, such as re-ported topical attitudes.* Peter Wright is Associate Professor of Marketing, GraduateSchool of Business, Stanford University, Stanford, CA 94305. Thisresearch was supported in part by a grant from Proctor and Gamble.The author thanks Bobby Calder, Joel Cohen. Mike Ray, and PeterIUp for helpfui comments on an earlier draft of this paper.USES OF THOUGHT VERBALIZATIONSIN PERSUASION RESEARCHVerbal reports on mental events have always beena central measure in persuasion research, especiallyscale responses to belief or attitude questions or re-ported recall of message content. Free reporting ofmessage-evoked thoughts has been used only since thelate 1960s. Its use was stimulated in part by interestin cognitive counterarguing as a mediator of distractioneffects on message impact (Festinger and Maccoby1964; C)sterhouse and Brock 1970), by Greenwald’s(1968) argument that people store in memory personalreactions to a message, not the text per se, and byMcGuire’s (I %9a,b) argument that richer messageprocessing models were needed to order the disparatefindings typical in persuasion studies. More generally,thought sampling was an out-growth of the detailedtheorizing about cognitive events apparent in consist-ency theories (Kelman and Baron 1%8), attitude for-mation theories (Fishbein and Ajzen 1972), and prob-lem-solving theories (Newell and Simon 1972). Interestin exposing natural mental events was high, and verbalreports represented one possibly valid source of data.The use of verbalization data in persuasion researchhas differed somewhat from its use in problem-solvingresearch. The Iatter—heaviy dominated by the Newelland Simon (1972) influence-involved an exploratoryphase in which verbalizations were searched for somegeneral constructs or operations to serve as corner-stones in a general theory of human problem solving.Once articulated, this theory’s stmcture strongly in-fluenced most later analyses of verbalizations. The in-itial goal in persuasion research was more modest—totest theorizing about limited types of mental activitiesthought to mediate effects of particular variables (e.g.,1510 JOURNAL OF CONSUMER RESEARCEI . Vol. 7. Septem&r 19S0ii557152distraction) on short-term attitude change. There wasreally no strictly exploratory phase, little belief thatthe data exposed enough to guide a general model ofpersuasion processes, and no monolithic theoreticalfilter through which the data have been passed. Per-suasion researchers have tested whether the data be-haved as predicted according to propositions about theantecedents and consequences of the events the re-ports supposedly reveal. typically in a small-scale ex-periment, or have used them to inspire new hypothesesfor test ing on data other than t bought verbalizations.To put prior applications of thought-sampling in per-spective, it is first useful to distinguish four episodesof cognitive activity. 11. Message response episode—an episode in whichmessage assertions are processed, rehearsed, storedor forgotten, and in which immediate message-evokedthoughts are generated.2. Attitude determination episode—an episode in whicha global evaluation of the advocated option isproduced.3. Preference determination episode—an episode inwhich a preference judgment between the advocatedoption and other options is produced.4. Action planning episode-an episode in which a se-quence of personal physical actions involving theadvocated option is planned.These may occur as four distinct episodes of cog-nitive activity, separated in time by unrelated activi-ties. For example, some message response activitiesmay occur, then a receiver turns attention to an un-related task without initiating immediate attitude orpreference determination activities or action planningactivities. Or message response activity may feed intoan immediate attitude or preference judgment episodeor into mental action planning, in a two-step sequenceor a more jumbled sequence (e.g., a little message re-sponse. then some preference judgment activity, thenmore message response, etc. ). The conditions that de-termine these scenarios are not yet well understood.And. of course, sometime after message processing,receivers presumably engage in thinking episodes dur-ing which they try to produce an attitude toward theadvocated option, to judge preference, (o plan actionsequences, or during which several of these activitiesoccur.Subjects in persuasion studies have been asked toreport the thoughts that occur while