* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download The Role of Attitude Accessibility in the Attitude-to
Group development wikipedia , lookup
Interpersonal attraction wikipedia , lookup
Group polarization wikipedia , lookup
Social tuning wikipedia , lookup
Carolyn Sherif wikipedia , lookup
Social perception wikipedia , lookup
Impression formation wikipedia , lookup
Introspection illusion wikipedia , lookup
Implicit attitude wikipedia , lookup
Elaboration likelihood model wikipedia , lookup
Vested interest (communication theory) wikipedia , lookup
Self-perception theory wikipedia , lookup
The Role of Attitude Accessibility in the Attitude-to-Behavior Process RUSSELL H. FAZIO MARTHA C. POWELL CAROL J.WILLIAMS* Attitudes toward a number of products and the accessibility of those attitudes as indicated by the latency of response to an attitudinal inquiry were assessed. Subjects with highly accessible attitudes toward a given product displayed greater attitude-behavior correspondence than did those with relatively less accessible attitudes. Furthermore, subjects with less accessible attitudes displayed more sensitivity to the salience afforded a product by its position In the front row, as opposed to the back row, than did subjects with more accessible attitudes. The implications of these data for a model of the process by which attitudes guide behavior are discussed. T he topic of attitudes clearly occupies a central position in research on consumer behavior (e.g., Engei and Blackwell 1982; Kassarjian and Kassarjian 1979). In part, the attention given to the attitude construct stems from the fact that much advertising can be described as social influence attempts aimed at creating positive attitudes toward the product. Indeed, considerable research has been concerned with understanding this persuasion process (e.g., Boyd, Ray, and Strong 1972; Lutz 1975; Petty, Cacioppo, and Schumann 1983). The assumption underlying such persuasive attempts is that the development of positive attitudes will produce a corresponding change in behavior. As a result, consumer research also has been concerned with understanding the relation between attitudes and subsequent behavior (e.g., Day and Deutscher 1982; Ryan and Bonfield 1975; Smith and Swinyard 1983). It is this attitude-behavior relation that constitutes the focus of the present article. Fazio and his colleagues proposed a model of the process by which attitudes guide behavior (Fazio 1986; Fazio, Powell, and Herr 1983). In brief, the model views behavior in any given situation as a func- tion of the individual's immediate perceptions of the attitude object in the context of the situation in which the object is encountered. "Perception" refers to the individual's current feelings about, or appraisal of, the object as experienced in the immediate situation. According to the model, attitudes guide such appraisals of the object, but only if they have been activated from memory upon observation of the object. Hence, the accessibility of the attitude from memory is postulated to act as a critical determinant of whether the attitude-to-behavior process is initiated. Various aspects of the model have received empirical support from both correlational and experimental investigations in such contexts as voting behavior (Fazio and Williams 1986), the evaluation of evidence related to a social policy issue (Houston and Fazio 1989), and "free-play" behavior with intellectual puzzles (Fazio et al. 1982; see Fazio 1989 for a review). The major goal of the present research was to assess the generality of the model to a consumer behavior domain. The research focuses upon the relation between attitudes toward specific products and product-selection behavior. The hypothesis here is that the status of individuals' attitudes along an attitude/non-attitude continuum moderates the attitude-behavior relation. Fazio and his colleagues (Fazio 1989; Fazio etai. 1986) have defined this continuum in terms of the strength of the association in memory between the object and the individual's evaluation of the object. At one end of the continuum is the non-attitude; no a priori evaluation of the object is available in memory. If asked to express an attitude, the individual needs to construct one on the spot. Such constructions presumably will involve some assessment of attributes of the object *Russell H. Fazio is Professor of Psychology, and Martha C. Powell and Carol J. Williams are Research Assistants, all in the Department of Psychology, Indiana University, Bloomington, IN 47405. This article was prepared while the first author was supported by the National Institute of Mental Health Research Scientist Development Award MH00452. The research was supported by a grant from the Ogilvy Center for Research and Development. The authors thank Clark Leavitt and Alexander Biel of the Ogilvy Center for their helpful consultation while the research was being designed and David Sanbonmatsu for his assistance in data collection. Portions of these data were presented at the 1986 Association for Consumer Research Annual Conference in Toronto. 280 ©JOURNAL OF CONSUMER RESEARCH •Vol. 16»December 1989 ATTITUDE ACCESSIBILITY that are recalled from memory and/or apparent in the external environment. However, at all other points along the continuum, an evaluation is available in memory. Moving along the continuum, the strength of the association between the evaluation and the object—and hence the accessibility of the attitude—increases. At the upper end of the continuum is a welllearned, strong association—sufficiently strong that the evaluation is capable of automatic activation from memory upon mere observation or mention of the object. Attitude-behavior consistency is expected to vary as a function of position along this attitude/ non-attitude continuum. Individuals who possess highly accessible attitudes toward a given product are expected to be more attitudinally consistent in their product-selection behavior than are individuals whose attitudes are relatively less accessible from memory. As mentioned earlier, evidence consistent with this hypothesized moderating role of attitude accessibility has been obtained in previous research. Most relevant is a correlational field study by Fazio and Williams (1986) in which it was found that the relation between attitudes toward Reagan and self-reported voting behavior in the 1984 presidential election varied as a function of the accessibility of the attitude. Attitude accessibility was measured via latency of response to the attitudinal inquiry. Individuals who were able to indicate their attitude relatively quickly (the high attitude accessibility group) displayed greater consistency between those attitudes and subsequent voting behavior than did individuals who responded relatively slowly (the low attitude accessibility group). In addition to examining the applicability of^ the model to product-selection behavior, the present research was intended to address an unavoidable shortcoming of the voting behavior study—the self-report nature of the behavior measure. As in any investigation of choices made within the confines of a voting booth, Fazio and Williams (1986) were forced to rely upon participants' reports of how they had voted. The respondents were telephoned within a day or two of the election and asked to reveal whether they had voted and, if so, for whom. Although we do not discern any plausible explanation for how such self-reports might have been biased by both the attitude expressed months earlier and the accessibility of the attitude, the shortcoming of this method of observing the attitude-to-behavior relation cannot be denied. The present investigation concerned actual overt behavior. Subjects selected items from a set of products arranged on a table. These selections, which subjects took home as "a token of our appreciation" for their having participated in the study, constituted the behavioral measure. The present investigation involves, as did the Fazio and Williams (1986) study, measurement of attitude accessibility via latency of response to an attitudinal 281 inquiry. Hence, it is important to review briefly what is known about the validity of this measure as an indication of the accessibility of attitudes and their position along the attitude/non-attitude continuum. First, the latency measure has been found to reflect what has been postulated to be the conceptual variable that determines the chronic accessibility of an attitude—namely, the strength of the association between the object and the evaluation. A number of experiments have manipulated the strength of this object-evaluation association by having subjects express their attitudes repeatedly. This research has indicated that such repeated expression enhances the speed with which individuals respond to later inquiries concerning their attitudes (Fazio et al. 1982; Powell and Fazio 1984). Second, and more important, the latency measure provides a good approximation of the likelihood that the attitude will be activated from memory automatically upon mere observation of the object. It has been demonstrated that attitude objects preselected on the basis of an individual's having responded quickly to an attitudinal inquiry are more likely to activate the attitude automatically upon subsequent presentation of the object than are attitudes characterized by relatively slow latencies of response to an attitudinal inquiry (Fazio et al. 1986). Sanbonmatsu and Fazio (1986) observed the same result in a study specifically concerned with the automatic activation of attitudes toward products. Thus, the latency with which one responds to an attitudinal inquiry is sensitive to the strength of the object-evaluation association and provides an indication of the likelihood that the attitude will be activated spontaneously upon one's encountering the object. Subjects in the present study responded to attitudinal inquiries concerning a large number of products. A subset of these products served as the target attitude objects and subsequently were made available as behavioral alternatives. According to the process model, these behavioral selections should be a function of the individual's perceptions of the object in the immediate situation. Is the individual's appraisal of the object at that particular moment favorable or unfavorable? An attitude that is highly accessible from memory and, hence, likely to be activated automatically upon the individual's observation of the object is apt to result in immediate perceptions that are congruent with the attitude. In contrast, when the attitude is not activated from memory, immediate perceptions are less likely to be influenced by a previously constructed evaluation that is available in memory. Instead, these immediate appraisals are likely to be based upon momentarily salient and potentially unrepresentative features (memory-based and/or stimulus-based) of the object that are not necessarily evaluated in a manner congruent with the attitude. As a result, greater attitude-behavior consis- 282 THE JOURNAL OF CONSUMER RESEARCH tency is expected when attitudes are highly accessible than when attitudes are relatively less accessible from memory. METHOD Subjects One hundred one individuals participated in the experiment. Sixty-one participated in partial fulfillment of an introductory psychology course requirement. The remaining subjects responded to an ad placed in the local newspaper and participated in return for monetary payment. Procedure Subjects were informed that the experiment concerned a new way of measuring attitudes via a microcomputer. Their initial task involved responding to each of 100 attitude objects, all of which were familiar and commonly available products. The subjects were instructed to press one oftwo keys, labeled "like" and "dislike," to indicate their feelings about each object. The subjects were told to "respond as quickly and as accurately as possible." The presentation was controlled by an Apple 11+ computer. The order in which the objects were presented was randomized for each subject. A three-second interval separated each trial. The subject's response was recorded, along with the latency of the response (from stimulus onset to response) to the nearest millisecond. To familiarize subjects with the procedure, subjects performed a series of practice trials involving different products than those used in the actual experimental list. Following the computerized task, subjects completed a number of questionnaires, only the first of which was relevant to the present purposes. The subjects provided a scalar attitudinal rating of each of the 100 objects. These ratings were made on a 1 (extremely bad) to 7 (extremely good) scale. Behavior Measure As their final task, subjects were given the opportunity to select five products from a set of 10 alternatives. The 10 products, which had been roughly equated for perceived value through informal pretesting, were a Snickers candy bar, a small bag of Fritos corn chips, two boxes of Sun-Maid raisins, a small can of Star-Kist tuna, a can of Dr. Pepper, a box of Cracker Jacks, a bag of Planters peanuts, two cans of V-8 juice, two 5-stick packs of Dentyne gum, and a Mounds candy bar. The products were arranged in two rows of five on a table and were covered by a tablecloth. At the appropriate time, the experimenter removed the tablecloth and told the subject to choose five of the items as a "token ofour appreciation" for having participated. The experimenter unobtrusively recorded the products selected and the order in which they were selected. The selected items were then placed in a paper bag for the subject to take home. RESULTS The mean latency of response across objects and subjects was 1.14 seconds. When a subject's dichotomous response during the computerized task was inconsistent with the subsequent scalar rating, it was considered an error. An error was defined as a subject's responding "like" during the computerized latency task but rating the product more negatively than the neutral point on the questionnaire scale, or responding "dislike" during the computer task but assigning a rating above the neutral point on the scale. Such errors were infrequent. On the average, subjects responded inconsistently on 7.3 percent of the 100 trials. The particular datum from any trial on which an error occurred was omitted from the analysis. Previous research on attitude accessibility has observed a relation between attitude accessibility, as measured by latency of response to an attitudinal inquiry, and attitude extremity, as indexed by deviation of a scalar value from the neutral point of the scale (Fazio and Williams 1986; Powell and Fazio 1984). The same relation was evident in the present data set. Faster response times were associated with more extreme ratings. The within-subject correlation between response latency and attitude extremity reached a level of significance for 46 percent of the subjects. The mean correlation across subjects was -0.18 (/(lOO) = 11.47, p< 0.001). As a result of this relation, it was necessary that all tests of the hypothesis regarding the moderating role of attitude accessibility be conducted in a manner that did not confound attitude accessibility and attitude extremity. Both within-subject and between-subject analyses were conducted. In both cases, the analysis that was performed examined the impact of accessibility as measured by response latency while considering equivalent attitude scores. Within-Subjects Analysis To overcome the naturally existing relation between attitude accessibility and extremity, the withinsubjects analysis focused upon any two or more target products for which a given subject had assigned the same attitude scalar rating. We restricted ourselves to these attitudinal "ties" to ensure that the role of attitude accessibility was examined in a manner that was independent of attitude scores. For each subject, the proportion of ties for which the subject's behavior concurred with a prediction based upon the subject's response latencies was determined. According to the 283 ATTITUDE ACCESSIBILITY hypothesis, subjects with highly accessible attitudes should behave more consistently than those with less accessible attitudes. Thus, for any ties at positions more positive than the neutral point, we predicted that the subject would either select the product associated with the faster response latency before selecting the other product or select the former product and not the latter. The prediction was reversed for any ties at attitudinal positions more negative than the neutral point. Any ties at the neutral point were not included in the analysis. Averaged across the 100 subjects with analyzable ties, the proportion of instances confirming the predictions was 0.59, significantly greater than the value of 0.50 expected by chance alone, (/(99) = 3.09, ;;< 0.0025). Between-Subjects Analyses The between-subjects analyses involved classifying subjects into groups of high, moderate, and low attitude accessibility for each product. As in the Fazio and Williams (1986) study, this group assignment was performed at each and every level of the attitude scale to avoid confounding attitude scores with attitude accessibility. For example, the subsample of subjects who had assigned a rating of 7 (extremely good) to a given target product were trichotomized on the basis of their accessibility scores into high, moderate, and low groups. The subsamples who had rated the product 6, 5, and so on were similarly trichotomized. This classification procedure was performed anew for each of the 10 target products. In this way, the attitude distributions for any given product were equivalent in the high, moderate, and low attitude accessibility groups. Unlike the Fazio and Williams (1986) investigation, however, attitude accessibility was not indexed simply by raw response latency. Instead, the accessibility measure employed for classification purposes was the z-score of a given subject's response latency for a given target product relative to that subject's mean and standard deviation of the latencies for the 90 filler products. This within-subject z-score serves to place a given subject's latency of response to a given object within his or her distribution of latencies for the filler products (see Fazio forthcoming for a discussion of indices of baseline speed of responding). This change was necessitated by the fact that interitem correlations among the 10 response latencies were substantial (average r = 0.42), whereas interitem correlations in the Fazio and Williams investigation were quite minimal (average r = 0.19; see Footnote 1 of Fazio and Williams 1986).' The use of raw 'Why the two studies differed in this regard is not clear. However, one difference in the attitude accessibility measurement procedures employed in the two studies is worth noting. The Fazio and Williams (1986) investigation involved subjects responding to an au- latency in the present case would have led to the consistent classification of individuals with tendencies to respond quickly to any inquiry as high accessibility subjects. In one of the analyses that was conducted, behavior was coded on a 0 to 5 scale, with products that were not selected by a given subject being assigned a score of 0 and the product that was selected first a score of 5. The remaining products were assigned intermediate scores indicative of the order of selection. For each product, the correlation between attitude and behavior was computed within the high, moderate, and low accessibility groups. These correlations were analyzed, following a Fisher's r-to-z transformation, via an a priori polynomial contrast examining whether they displayed a significant linear trend as a function of level of attitude accessibility. This prediction was confirmed (F( 1,9) = 8.01, p < 0.02).^ Averaged across the 10 products, the mean correlations (following a retransformation of the average zs back to rs) were 0.62, 0.54, and 0.50 for the high, moderate, and low attitude accessibility groups, respectively. The analysis involved the scoring of behavior as a function of order of selection and, thus, assumes that subjects chose first their most preferred alternative, then their second most preferred item, and so on. The data also were analyzed with behavior coded dichotomously as a function of whether the product was or was not selected. This analysis also confirmed the prediction. The mean correlations were 0.61, 0.59, and 0.51 in the high, moderate, and low attitude accessibility groups, respectively, and, as in the previous analysis, revealed a linear trend (F(l,9) = 5.09, p = 0.051). Thus, regardless of how the data are examined, the findings converge upon the notion that attitude accessibility moderated the attitude-behavior relation. The more accessible a subject's attitude was toward a given product, the more likely it was that product selection behavior was consistent with that attitude. Position EfFects As was mentioned earlier, the attitude-to-behavior process model views behavior as a function of the indiotaped statement, the end of which contained the electronic marker that initiated the timing. In contrast, the present study involved displaying the name of an object on a computer screen, the onset of which initiated the timing. The time necessary to read the visual display may have enhanced the amount of covariation observed. Reading time may be a fairly constant individual difference that formed a component of each response latency. ^Given the small number of units of analysis involved in this statistical test, i.e., the 10 products, it seemed desirable to also examine the data nonparametrically. The data were analyzed via the Page test for ordered alternatives (Page 1963; Seigel and Castellan 1988), which did reveal a significant linear trend. The same was true of all additional linear trends reported here. 284 dividual's perceptions of the object in the immediate situation. A highly accessible attitude is likely to be activated from memory automatically and is presumed to determine this immediate perception of the object. In contrast, an attitude that is relatively low in accessibility is less likely to be activated upon observation of the object. As a result of low accessibility, the immediate perception is likely to be influenced by momentarily salient features of the object. This hypothesized greater influence of a momentarily salient dimension for objects associated with a relatively inaccessible attitude than for objects associated with a highly accessible attitude was apparent in a rather surprising fashion in subjects' selection behavior. The 10 products were arranged in two rows of five for the subjects' viewing and selection.' On the assumption that the products positioned in the front row were more salient than those in the back row, we suggest that the selection of a given product will be more influenced by row status if the attitude is low in accessibility than if its accessibility is high. The data revealed precisely this pattern. Among the products in the front row (Snickers, Mounds, peanuts, Dentyne, and tuna), the lower the accessibility of their attitudes, the greater the likelihood that subjects selected the product. Indeed, across these products, the average proportion of subjects who selected the product was 0.55, 0.59, and 0.59 in the high, moderate, and low accessibility groups. For products in the back row, the reverse was true. The lower the accessibility of the attitude, the less likely subjects were to select the product (average proportions of 0.50, 0.46, and 0.40 for the high, moderate, and low groups, respectively). The row effects—the difference between the mean proportion of subjects who selected a product positioned in the front row and the mean proportion who selected an alternative that had been positioned in the back row—were 0.05, 0.13, and 0.19 for the high, moderate, and low attitude accessibility groups, respectively. Analysis of these data indicated that the row effects constituted a significant linear trend ) = 5.96,/? < 0.05)." ^Products were arranged randomly at the beginning of the experiment and then fixed across subjects. As a result, an overall preference for front-row products over back-row products (analogous to a main effect) may reflect either the influence of row status and/ or the products themselves. Obviously, no clear inferences can be drawn from data reflecting an overall preference for the products in one row over the products in the other row. However, this confounding does not affect our ability to examine the degree to which the exhibited preference for front-row products relates to the level of attitude accessibility (conceptually analogous to an interaction effect). •"The actual analysis was performed on scores for each product that represented the proportion of subjects who selected the product if it had been in the front row and the proportion who did not select it if the product had been in the back row. A greater influence of row status would be apparent if these scores increased linearly as THE JOURNAL OF CONSUMER RESEARCH Thus, the lower the attitude accessibility, the more selection behavior was influenced by the relative salience afforded a product by its positioning. In contrast, the linear trends presented earlier with respect to attitude-behavior correlations suggest that the greater the attitude accessibility, the more selection behavior was influenced by the attitude. DISCUSSION The Role of Attitude Accessibility The present findings are consistent with the hypothesis that attitude accessibility exerts a moderating role upon the attitude-behavior relation, just as is postulated by the process model. The more accessible the attitude, the more predictive it was of subsequent behavior. Furthermore, this evidence was obtained in a situation involving actual behavior as opposed to self-reports of behavior. Subjects chose, took home, and presumably consumed the products. Thus, the data are very consistent with the attitude-to-behavior process model. When interpreted within the context of the process model, the findings suggest that subjects scanned the set of available alternatives and arrived at immediate perceptions of each (or at least some subset) of the products. In the case of objects for which a strong evaluative associate existed in memory, this evaluation was activated automatically from memory upon the subject's observation of the object. That is, observation of the object automatically activated the attitude from memory if the chronic accessibility of the attitude, as estimated by a relatively fast latency of response to an earlier attitudinal inquiry, was high. Such activated attitudes strongly determined the subject's feelings toward the object in the immediate situation. If the object-evaluation association for a given product was weak and, hence, the chronic accessibility of the attitude was relatively low, then the immediate perception was likely to be influenced by momentarily salient thoughts or features of the object. The data suggest that the perception of such objects was affected by the greater salience afforded an object positioned in the front row as opposed to the back row. Other momentarily salient factors that might have operated include product attributes that were recalled accessibility decreased, which, as reported, they did (mean proportions of 0.52,0.56, and 0.60 for the high, moderate, and low accessibility groups). For ease of understanding, the data are presented in the text in terms of row effects, i.e., the difference in the proportion of subjects who selected front-row versus back-row products. Additional analyses on the frequency with which a given product was selected revealed that level of attitude accessibility was independent of the specific products that were positioned in each row. The data did not display any appreciable deviations from the overall linear trend that was apparent for the front-row (x^(8) = 4.15, p > 0.25) or back-row products (x^(8) = 1.34, p > 0.25). 285 ATTITUDE ACCESSIBILITY from memory, how thirsty or hungry the subject was, how recently the subject had eaten a particular food (e.g., a candy bar), what snack foods the subject recalled as having stored at home, and the like. In any case, such factors appear to exert a greater influence when the individual's attitude toward a given object is unlikely to have been activated automatically upon observation of the object than when the attitude is highly accessible from memory and capable of automatic activation. Product Positioning and the Consideration Set The differential influence of row status as a function of attitude accessibility is clearly evident from the data. The precise mechanisms through which this influence occurred is less clear. One possibility concerns the potential influence of salient positioning on the likelihood that a given alternative will receive much consideration in the selection process. Consumer behavior researchers have distinguished the "evoked" or "consideration" set—those alternatives that are actually considered for potential selection— from the pool of alternatives that are available (e.g.. Baker etai. 1986; Howard and Sheth 1969). Although the present investigation did not involve the collection of any data relevant to the question of which items constituted an individual's consideration set, the observed influence of both attitude accessibility and position may have operated at the level of inclusion or exclusion from the consideration set. For ease of discussion, imagine two individuals, one of whom (A) holds highly accessible attitudes toward some subset of the 10 alternatives and one of whom (B) does not have highly accessible attitudes toward any of the alternatives. When scanning the items, individual A may especially notice, and attend to, those products toward which s/he has a highly accessible attitude because the affective associate to the object will be activated automatically from memory upon observation of the object. Indeed, some very recent research indicates that objects toward which an individual holds highly accessible attitudes, be they positively or negatively valenced, are more likely to be noticed when presented in the visual field than are objects toward which the individual holds less accessible attitudes. Roskos-Ewoldsen and Fazio (1989) found support for this hypothesis in two investigations, one of which involved the measurement of attitude accessibility via response latency (as in the present study) and one of which involved the experimental manipulation of attitude accessibility. Thus, these data suggest that our hypothetical individual A will quickly and easily notice those products concerning which accessible attitudes are held even if those items are positioned in the relatively disadvantageous back row (see Alba and Hutchinson 1987 for a related discussion of the likelihood of product detection). Any products within this subset that are negatively valued will be quickly rejected from inclusion in the consideration set. In other words, the perception or appraisal of the product in the immediate situation, determined as it is by the negative attitude activated from memory, will lead to its rejection. In contrast, those products that are both positively valued and characterized by high attitude accessibility are likely to be included in individual A's consideration set. Construction of the consideration set may proceed quite differently for our hypothetical individual B. Lacking highly accessible attitudes, this individual's attention is unlikely to be attitudinally guided. Those products enjoying the relative salience afforded them by their position in the front row are more likely to be included in the consideration set than are those positioned in the back row. Thus, the probabilities that individual B will select a front-row product over a back-row product may be enhanced. Obviously, the role of attitude accessibility and product salience in defining the consideration set merits further investigation. The Attitude Accessibility Measure Our interpretation of the attitude-behavior consistency data rests on the validity of response latency to an attitudinal inquiry as a measure of the chronic accessibility of the attitude. As indicated earlier, evidence regarding such validity is provided by previous research (Fazio et al. 1986) demonstrating that this measure relates to the likelihood of automatic activation of the attitude upon exposure to the object. Those objects for which an individual could respond relatively quickly to a direct attitudinal query also were likely to activate the attitude from memory upon their mere presentation. In contrast, objects for which response latencies to an inquiry were slow showed little evidence of producing automatic attitudinal evaluation upon their presentation. Although we can be confident that the latency measure reflects the chronic accessibility of the attitude, it must be recognized that the present investigation is correlational in nature. Attitude accessibility was measured and not manipulated. Thus, whatever dimensions are naturally associated with attitude accessibility may have contributed to the differences that were observed with respect to attitude-behavior consistency. For example, familiarity may moderate the attitude-behavior relation. Because the present correlational investigation involved the measurement of attitude accessibility and because objects with which individuals are familiar are likely to be ones toward which they have accessible attitudes, the distinct moderating influences of attitude accessibility and familiarity cannot be ascertained from the present data. 286 However, two counterpoints should be noted. First, experimental work in which attitude accessibility was manipulated directly, by varying the number of times that subjects expressed their attitudes, has indicated a causal impact of attitude accessibility upon both attitude-behavior consistency (Fazio etai. 1982, Experiment 4) and attitude-judgment consistency (Houston and Fazio 1989). Although further experimental research is undoubtedly needed to isolate the causal influence of attitude accessibility, this finding does bolster our confidence that the present results are not due solely to the common dependence of attitude accessibility and attitude-behavior consistency on some third variable, such as familiarity. Second, we should note, as did Fazio and Williams (1986), what we view as the primary advantage of focusing upon attitude accessibility as opposed to some other strength-related dimension of attitude, such as confidence in one's attitude (e.g., Fazio and Zanna 1978), the manner of attitude formation (e.g., Fazio and Zanna 1981), affective-cognitive consistency (e.g., Norman 1975), or amount of information about the attitude object (e.g., Davidson et al. 1985). Unlike other indicants of attitude strength, the construct of attitude accessibility operates at an information-processing level of analysis. As a result, it is much more clearly relevant to the issue of the process by which attitudes guide behavior. Although other indicants of the strength of an attitude may be associated with attitude accessibility and, hence, with our classification of subjects in the present study, the construct of attitude accessibility has clear implications with respect to the basic mechanisms involved in the production of attitude-consistent behavior. Indeed, it has been postulated that various identified moderators of the attitude-behavior relation may exert their impact because they reflect the strength of the object-evaluation association and, hence, the accessibility of the attitude (Fazio 1986). This has been demonstrated with respect to one such moderator— the manner of attitude formation. Attitudes based upon direct behavioral experience with the object have been found to be both more predictive of later behavior (Fazio and Zanna 1981) and more accessible from memory (Fazio et al. 1982, 1983) than attitudes based upon indirect experience. Similar evidence has been obtained with respect to the moderating variables of personal importance of the attitude issue (Krosnick forthcoming) and of individual differences in self-monitoring tendencies (Kardes et al. 1986). The same may hold true for other variables that have been identified as moderators of the attitude-behavior relation. If so, the construct of attitude accessibility and the process model may provide a conceptual framework for integrating this set of moderators and for understanding how and why they influence attitude-behavior consistency. THE JOURNAL OF CONSUMER RESEARCH Affect Referral The attitude-behavior process model bears some similarity to Wright's (1975) theoretical statements regarding what he terms affect referral. Affect referral is a decision-making strategy in which the individual avoids reviewing any specific attribute information concerning the alternatives, but instead relies upon previously formed, global affective judgments of the alternatives. Although Wright did not detail any mechanisms or process by which such affect referral occurs, the affect referral strategy that he mentions is one that our model obviously endorses. Indeed, the model can be viewed as specifying the mechanism underlying, and the conditions necessary for, such affect referral. Position along the attitude/non-attitude continuum dictates the extent to which an affect referral strategy is possible. A strong object-evaluation association in memory and, hence, a highly accessible attitude is a prerequisite for such affect referral. In such a case, the attitude will be activated from memory and influence, if not completely determine, the individual's appraisal of the object in the immediate situation. The sparser the immediate environment is with respect to the availability of new information about the attitude object, the less critical the selective processing component ofour model becomes. In a situation in which no new information is presented (as in the present investigation), the selective processing amounts to an effortless application of the attitude stored in memory as one's appraisal of the object in the immediate situation—much as Wright appears to mean by his affect referral strategy. With a highly accessible attitude, the immediate appraisal is more likely to be congruent with the attitude stored in memory than it is when a relatively inaccessible attitude is involved. Active construction of an immediate appraisal is not necessary. In contrast, the selective processing component assumes more importance in an environment that is information rich. When the situation makes new information about the object available (as in the Fazio and Williams voting behavior study or in an ongoing social interaction), interpretations of this information will be colored by an activated attitude. As a consequence of what is potentially extensive selective processing in this case, immediate appraisals are more likely to be congruent with attitudes among individuals whose attitudes are highly accessible than among individuals whose attitudes are not. Thus, the extent of selective processing depends upon the degree to which the situation provides new information. In either case, the attitude-behavior process model points to the importance of correspondence between perceptions of the object in the immediate situation and attitudes toward the object. Such correspondence is more likely when the attitude involves a strong object-evaluation association and is capable of automatic activa- ATTITUDE ACCESSIBILITY tion from memory upon mere observation of the object. Final Implications The findings from the present investigation, as well as our discussion of them, attest to the relevance of the attitude-behavior process model to the consumer behavior domain. The accessibility of an attitude from memory does appear to moderate the extent to which that attitude guides product selection behavior. A major implication of this finding concerns the role of advertising as a social influence agent intended to promote the desired consumer behavior. Apparently, inducing a positive attitude toward the brand is not in and of itself sufficient to have much influence upon consumer behavior. The general suggestion offered by the model and the present findings is that the desired behavior is most likely to occur if the attitude is highly accessible from memory. Thus, if the goal of advertising is to influence purchase behavior, then one needs to be concerned not only with the valence of the resulting attitude but also with its accessibility from memory (see Herr and Fazio 1988 for a more general discussion of this issue). How social influence attempts can be designed to accomplish this end constitutes an important challenge for the future. [Received August 1988. Revised May 1989.] REFERENCES Alba, Joseph W. and J. Wesley Hutchinson (1987), "Dimensions of Consumer Expertise," Journal of Consumer Research, 13 (March), 411-454. Baker, William, J. Wesley Hutchinson, Danny Moore, and Prakash Nedungadi (1986), "Brand Familiarity and Advertising: Effects on the Evoked Set and Brand Preference," in Advances in Consumer Research, Vol. 13, ed. Richard J. Lutz, Provo, UT: Association for Consumer Research, 637-642. Boyd, Harper W., Michael L. Ray, and Edward C. Strong (1972), "An Attitudinal Framework for Advertising Strategy," Journal ofMarketing, 36 (April), 27-33. Davidson, Andrew R., Steven Yantis, Marel Norwood, and Daniel E. Montano (1985), "Amount of Information About the Attitude Object and Attitude-Behavior Consistency," Journal of Personality and Social Psychology, 49 (November), 1184-1198. Day, George S. and Terry Deutscher (1982), "Attitudinal Predictions of Choices of Major Appliance Brands," Journal ofMarketing Research, 19 (May), 192-198. Engel, James F. and Roger D. Blackwell (1982), Consumer Behavior, Hinsdale, IL: Dryden. Fazio, Russell H. (1986), "How Do Attitudes Guide Behavior?" in The Handbook of Motivation and Cognition: Foundations of Social Behavior, eds. Richard M. Sorrentino and E. Tory Higgins, New York: Guilford, 204-243. 287 (1989), "On the Power and Functionality of Attitudes: The Role of Attitude Accessibility," in Attitude Structure and Function, eds. Anthony R. Pratkanis et al., Hillsdale, NJ: Lawrence Eribaum Associates, 153179. (forthcoming), "A Practical Guide to the Use of Response Latency in Social Psychological Research," in Review of Personality and Social Psychology, Vol. 11, eds. Clyde Hendrick and Margaret S. Clark, Beverly Hills, CA: Sage. , Jeaw-Mei Chen, Elizabeth C. McDonel, and Steven J. Sherman (1982), "Attitude Accessibility, AttitudeBehavior Consistency, and the Strength of the ObjectEvaluation Association," Journal of Experimental Social Psychology, 18 (July), 339-357. , Martha C. Powell, and Paul M. Herr (1983), "Toward a Process Model of the Attitude-Behavior Relation: Accessing One's Attitude Upon Mere Observation of the Attitude Object," Journal ofPersonality and Social Psychology 44 (April), 723-735. , David M. Sanbonmatsu, Martha C. Powell, and Frank R. Kardes (1986), "On the Automatic Activation of Attitudes," Journal of Personality and Social Psychology, 50 (February), 229-238. and Carol J. Williams (1986), "Attitude Accessibility as a Moderator of the Attitude-Perception and Attitude-Behavior Relations: An Investigation of the 1984 Presidential Election," Journal of Personality and Social Psychology. 51 (September), 505-514. and Mark P. Zanna (1978), "Attitudinal Qualities Relating to the Strength of the Attitude-Behavior Relationship," Journal of Experimental Social Psychology, 14 (4), 398-408. and Mark P. Zanna (1981), "Direct Experience and Attitude-Behavior Consistency," in Advances in Experimental Social Psychology, Vol. 14, ed. Leonard Berkowitz, New York: Academic Press, 161-202. Herr, Paul M. and Russell H. Fazio (forthcoming), "The Attitude-to-Behavior Process: Implications for Consumer Behavior," in Ad Exposure, Memory, and Choice, ed. Andrew A. Mitchell, Hillsdale, NJ: Lawrence Eribaum Associates. Houston, David A. and Russell H. Fazio (1989), "Biased Processing as a Function of Attitude Accessibility: Making Objective Judgments Subjectively," Social Cognition, 7 (Spring), 51-66. Howard, John A. and Jagdish N. Sheth (1969), The Theory ofBuyer Behavior. New York: John Wiley. Kardes, Frank R., David M. Sanbonmatsu, Richard T. Voss, and Russell H. Fazio (1986), "Self-Monitoring and Attitude Accessibility," Personality and Social Psychology Bulletin, 12 (December), 468-474. Kassarjian, Harold H. and Waltraud M. Kassarjian (1979), "Attitudes Under Low Commitment Conditions," in Attitude Research Plays for High Stakes, eds. John C. Maloney and Bernard Silverman, Chicago: American Marketing Association, 3-15. Krosnick, Jon A. (forthcoming), "Attitude Importance and Attitude Accessibility," Personality and Social Psychology Bulletin. Lutz, Richard J. (1975), "Changing Brand Attitudes Through Modification of Cognitive Structure," Journal of Consumer Research, 1 (March), 49-59. 288 Norman, Ross (1975), "Affective-Cognitive Consistency, Attitudes, Conformity, and Behavior," Journal of Personality and Social Psychology, 32 (July), 83-91. Page, Ellis B. (1963), "Ordered Hypotheses for Multiple Treatments: A Significance Test for Linear Ranks," Journal of the American Statistical Association. 58 (March), 216-230. Petty, Richard E., John T. Cacioppo, and David Schumann (1983), "Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement," Journal of Consumer Research, 10 (September), 135146. Powell, Martha C. and Russell H. Fazio (1984), "Attitude Accessibility as a Function of Repeated Attitudinal Expression," Personality and Social Psychology Bulletin, 10 (March), 139-148. Roskos-Ewoldsen, David R. and Russell H. Fazio (1989), "Do Affectively-Laden Objects Attract Attention?" paper presented at the annual meeting of the Midwestern Psychological Association, Chicago. THE JOURNAL OF CONSUMER RESEARCH Ryan, Michael J. and E.H. Bonfield (1975), "The Fishbein Extended Model and Consumer Behavior," Journal of Consumer Research, 2 (September), 118-136. Sanbonmatsu, David M. and Russell H. Fazio (1986), "The Automatic Activation of Attitudes Toward Products," paper presented at the annual meeting of the Association for Consumer Research, Toronto, Ontario, Canada. Seigel, Sidney and N. John Castellan (1988), Nonparametric Statistics for the Behavioral Sciences, New York: McGraw-Hill. Smith, Robert E. and William R. Swinyard (1983), "Attitude-Behavior Consistency: The Impact of Product Trial Versus Advertising," Journal of Marketing Research, 20 (August), 257-267. Wright, Peter (1975), "Consumer Choice Strategies: Simplifying Vs. Optimizing," Journal of Marketing Research, 12 (February), 60-67.