Does retrospective thought measurement influence subsequent measures of cognitive structure in an advertising context?

Does Retrospective Thought Measurement Influence Subsequent Measures of Cognitive Structure in an Advertising Context?

This study examined the effects of taking thought measurements immediately after ad exposure on several aspects of post-exposure cognitive structure about the advertised brand. The results suggest that thought measurement influences subsequent measures of confidence in belief, attitude, and intention variables, and the strength of association between beliefs and attitude/intention measures--and particularly so for audiences high in message-response involvement. However, thought measurement produced no direct effects on traditional measures of ad effectiveness such as beliefs and attitudes about the advertised brand. Implications of these results for future advertising research utilizing the thought-measurement methodology are discussed. The cognitive response approach to persuasion (Cacioppo, Harkins, and Petty 1981; Greenwald 1968; Olson, Toy, and Dover 1982; Wright 1973, 1980) is currently enjoying considerable popularity among advertising researchers. This approach suggests that the spontaneous thoughts (i.e., cognitive responses) elicited by audience members as they view an advertisement are the primary mediators of ad effects on beliefs and attitudes about the advertised brand. Cognitive responses are usually measured by the method of retrospective thought listing--respondents are asked immediately after ad exposure to write down all the thoughts they had as they processed the ad. These thought-listing data have been used to test increasingly detailed theories and hypotheses about the cognitive mechanisms that mediate advertising effects. Recent applications of the cognitive-response approach include studies on time-compressed advertising (Vann, Rogers, and Penrod 1987), one-versus two-sided messages (Kamins and Assael 1987), ad repetition effects (Batra and Ray 1986; Belch 1982), and source credibility effects on attitude (Harmon and Coney 1982; Petty, Cacioppo, and Schumann 1983; Sternthal, Dholakia, and Leavitt 1978, etc.).

Given the widespread use of verbal reports as data in advertising research, it is important to establish that the procedures used to elicit these verbal reports are non-reactive. If the act of thought-listing immediately after ad exposure influences subsequent measures of brand-related cognitive structure (CS), then the validity of tests for experimental effects on these CS variables clearly has been compromised. The importance of this "reactivity" issue, and the threat it poses to the validity of cognitive-response research, is well recognized in the literature (Cacioppo, Harkins, and Petty 1981; Cacioppo and Petty 1981; Wright 1974). However, relatively few studies in the psychology literature (and none in advertising) have examined the effects of taking thought measurements on subsequent measures of cognitive structure. Consequently, little is currently understood about whether or not thought measurements produce reactive effects in advertising contexts, and if so, the extent of those effects.

In this paper, the effects of thought measurements on several post-exposure cognitive-structure variables that are of interest to advertising researchers are examined. The first section introduces a simple theoretical framework that (a) specifies the CS variables that are likely to be affected by thought measurement, (b) identifies the mechanisms or processes that mediate these effects, and (c) predicts the conditions under which these reactive effects are likely to be large. This framework is based on, and extends existing theoretical ideas and concepts in the literature on the thought-measurement methodology (Cacioppo and Petty 1981; Cacioppo, Harkins, and Petty 1981; Ericsson and Simon 1980; Wright 1980). Hypotheses generated from this framework are then tested using data from a laboratory experiment. The paper concludes with a discussion of the results and their implications for future advertising research using verbal reports as data.

Theoretical Framework

The thought-listing task is usually administered immediately after subjects have viewed the test advertisement. Subjects are instructed to list all thoughts, reactions, and feelings (i.e., cognitive responses) that they experienced as they processed the ad. Some of these cognitive responses are still available in short-term memory, and can be readily reported. However, a majority of the cognitive responses are unavailable in short-term memory and must be searched for and retrieved from long-term memory (Ericsson and Simon 1980; Wright 1980). Retrieval of information from long-term memory is best explained by the spreading activation theory of semantic processing (Collins and Loftus 1975). As subjects attempt to retrieve their cognitive responses, one or more cognitive-responses nodes in memory are initially activated. Activation then spreads to other neighboring concept nodes, i.e., to other cognitive responses as well as beliefs, attitudes/intentions, and ad claims that are stored in memory in close proximity to the cognitive responses. Note that this spread of activation is partly automatic and outside the conscious control of the subject (Collins and Loftus 1975). In other words, even though the subjects may be concentrating their attention on the task of retrieving their prior cognitive responses, some of the other brand-related concepts that are linked to those cognitive responses will also be activated. All concepts that are sufficiently activated enter short-term memory, and are available for reporting. Subjects must examine these activated concepts and identify those concepts that reflect their prior thoughts before producing a listing of their thoughts.

In sum, the retrospective thought-listing task can be conceptualized as a two-stage process: a search/retrieval stage and a reporting stage. Each of these processing stages can alter subsequent measures of brand-related cognitive structure. Here, the potential biasing effects at the reporting stage are considered first, since these have been clearly articulated in extant literature.

Additional Thought Processing During the Reporting Task

The thought-listing task is intended to produce a record of the spontaneous thoughts that subjects experienced earlier when they viewed the experimental ad. However, as noted by Wright (1980) and Cacioppo and Petty (1981), subjects may also engage in additional cognitive processing of the advertisement and its contents during this task. Such elaborative processing will result in new cognitive responses which, in turn, will induce …