A Nonreactive Measure of Inferential Beliefshomepages.uc.edu/~kardesfr/Kardes Articles/Kardes 1988...

ABSTRACTMethods for measuring inferential beliefs about products are reviewed and a newnonreactlve measurement procedure is proposed. This procedure is illustratedempirically in a study demonstrating that consumers deduce inferential beliefsfrom a set of subjectively related product claims even when there are minimalpressures to engage in inference formation. Implications for advertising researchan'd for public policy are discussed.

A Nonreactive Measure ofInferential Beliefs

Frank R. KardesMassachusetts Institute of Technology

The primary focus of this article is on the measurement of consumerinferences. An inference is defined as a conclusion that is derivedfrom a set of premises on the basis of a rule that associates thepremises to the conclusion in a subjectively logical fashion (Hastie,1983). Hence, inference formation involves going beyond directlyobservable events in accordance with the rules of induction (Hol-land, Holy oak, Nisbett, & Thagard 1986; Nisbett & Ross, 1980) ordeduction (Wyer & Hartwick, 1980).*

Although many consumer researchers agree that inferenceprocesses are important (e.g., Alba & Hutchinson, 1987; Bettman,1979; Huber & McCann, 1982; Johnson & Levin, 1985; Lynch &SruU, 1982; Meyer, 1981; Mizerski, Golden, & Keman, 1979; Olson,1978; Olson & Dover, 1978), few empirical investigations of con-sumer inference processes have been conducted. One reason for thisis that measurement difficulties have discouraged researchers frominvestigating this important topic. This is unfortunate because aclearer understanding of inference processes is needed now morethan ever. Consider the effects of current trends in advertising, for

Psychology & Marketing VoL 5(3): 273-286 (FaU 1988)© 1988 John Wiley & Sons, Inc. CCC 0742.6046/88/030273-14$04.00

A NONREACTIVE MEASURE OF INFERENTIAL BELIEFS

example. In recent years there has been a decrease in the use of con-crete, factual claims in advertising and an increase in the use ofabstract claims that are open to multiple interpretations (Shimp,1979; Shimp & Preston, 1981). As a consequence, the problem ofjdejitifying misleading advertising has shifted from determining if aparticular product claim is factually grounded to determining wheth-er or not a specific claim induces consumers to draw inferencesabout product performance that do not correspond to actual productperformance (Russo, Metcalf, & Stephens, 1981).

To illustrate the difficulties involved in measuring consumers'inferential beliefs, the most commonly employed procedures arebriefly reviewed and discussed. Preston (1967) and Preston & Schar-bach, (1971) were the first to develop a procedure for measuringconsumer inferences. This procedure involves exposing subjects toadvertising claims and providing them with a set of test items foreach claim. Two of the test items are accurate paraphrases (accordingto the experimenter's definition of accuracy), two are irrelevant, andthe critical test item is an "illogical" item. As an example of anillogical item, consider the Bell ad that stated "You'll get a royalwelcome when you call Long Distance" (if X then Y). One illogicalconversion of this claim is "If you don't call Long Distance youwon't get a royal welcome" (if X' then Y'). If subjects judge thisillogical item to be an accurate refiection of the stated or impliedcontent of the original claim, the claim is considered misleading.One difficulty with this procedure is that the rules of everyday logicmay not correspond closely to the rules of formal logic.

The "normative belief procedure (Dyer & Kuehl, 1978;Gardner, 1975) involves measuring consumers' normative beliefsabout the attributes associated with a particular product class. Thesemeasures are compared to measures of consumers' beliefs about theattributes associated with a specific brand, after exposure to an adpertaining to that brand. Discrepancies between normative productclass beliefs and specific brand beliefs suggest that the target ad ispotentially misleading. This procedure is appropriate only for prod-uct classes composed of relatively homogeneous brands.

The "salient belief procedure (Armstrong, Gurol, & Russ,1978) involves asking consumers to judge the validity of various testitems that are implied by claims presented in an ad. Each judgmentis weighted by an importance rating, where importance is opera-tionaiized in terms of the relevance of a belief for a purchase decision(Wilkie & Pessemier, 1973). If consumers' salient-belief scores on aset of false claims differ as a function of exposure or nonexposure toan ad, the ad is considered misleading.

Russo et al. (1981) developed a procedure that involves meas-

~274 ' FALL 1988


uring the beliefs of consumers who read an explicit claim, who didnot read an explicit claim, and who read a corrected (i.e., not misrleading) claim. If belief in an implied claim is greater in the explicit-claim group than in the no-claim and corrected-claim groups, thenthe explicit claim is judged to be misleading.

Schmittlein and Morrison (1983) developed a psychometricmodel that removes the effect of yea-saying or acquiescence bias inmeasuring comprehension. This procedure involves showing consum-ere explicit claims about a product. Immediately afterward, theyare asked to answer some true-false questions pertaining to theseclaims. The correct answer to some of these questions is true and thecorrect answer to the remaining questions is false. The average levelof true comprehension for a communication, the average level ofyea-saying for a communication, and the level of difficulty for eachquestion is estimated by equating the expected proportion of correctanswers for true and false questions with the observed proportion ofcorrect answers for true and for false questions, respectively.

The major shortcoming common to each of these procedures isthat they involve directly asking consumers about various inferencesthat they may have formed. In responding to this type of inquiry,consumers may either search memory for an answer or they mayformulate an answer on the spot. Thus, the measurement procedureitself may prompt inference formation, rather than assessing a pre-viously formed inference. For example, suppose that consumers wereexposed to an ad stating that Ocean Spray contains more foodenergy than orange or tomato juice. One false implication of thisclaim is that Ocean Spray is more nutritious than orange or tomatojuice (actually, it is higher in calories). If a researcher asks "Is OceanSpray more nutritious than orange and tomato juice?" and a con-sumer says it is, it is impossible to determine whether this belief wasformed before the consumer received the researcher's leading quesrtion, or if the question itself induced the consumer to form the falsebelief. In the latter case, the consumer would never have inferredthe false belief if he had not been exposed to the leading ques-tion.

One way to avoid this methodological difficulty is to use a non-reactive measurement procedure (Webb et al., 1966). In the presentstudy, a nonreactive measurement system was devised based on theability or inability of consumers to determine if a given piece ofinformation was explicitly stated in a product claim (explicit infor-mation) or if this information was merely implied by the claim (in-ferred information). To elaborate, any given piece of informationmust originate either from an external or from an internal source.External sources of information include sources such as advertise-

PSYCHOLOGY & MARKETING/VOL. 5, NO. 3 275


ments, package labels. Consumer Reports and other buying guides,salespeople, and other consumers. Self-generated inferences andelaborations, on the other hand, go beyond information that isacquired from external sources. Prior knowledge and experience isoften used to fill in missing details in externally provided informa-tion (Hastie, 1981; Taylor & Crocker, 1981) and explicit infonnationand inferred infonnation is often "unitized" into a single integratedcognitive representation (Bransford & Franks, 1971; Hayes-Roth,1977). This cognitive representation is later retrieved in an all-or-none fashion, and, as a consequence, individuals often experiencedifficulty in ascertaining if a given piece of infonnation originatedfrom an external or from an internal source. Henceforth, this type ofconfusion will be referred to as "origin" confusion.

Origin confusion can be useful in determining whether or not agiven iaferential belief was formed prior to measurement. If a con-sumer forms an inference while reading an ad but before exposure toa measurement instrument, he may subsequently be unable to deter-inine if this information was stated explicitly in the ad or if it wasinferred. If this belief was not inferred, however, origin confusion isunlikely.

It is important to note that origin confusion does not occurunder all circumstances in which inferences are formed. Sometimesindividuals can accurately discriminate between actually perceivedversus inferred information. One factor that may infiuence an indi-vidual's ability to determine whether or not a particular piece ofinformation was inferred is the manner in which that piece of infor-mation was encoded, or stored, in memory. According to Kahneman(1973), there is a limit on the amount of capacity available forperforming various cognitive activities. Moreover, some cognitiveoperations require a considerable amount of this limited capacity,whereas other operations require only a small amount. These cogni-tive operations are referred to as controlled versus automatic proc-esses, respectively. Automatic processes can operate in parallel withother activities and are not influenced by information load, whereascontrolled processes are strongly influenced by manipulations thatdivide attention (Bargh, 1984; Hasher & Zacks, 1979; Schneider &Shiffrin, 1977; Shiffrin & Schneider, 1977).

Consider the task of learning how to drive an automobile. Atfirst, this task is quite difficult and a great deal of effort is required.With more and more practice, however, driving becomes easier andeasier. This automatization process is useful because it enables theindividual to perform other behaviors (e.g., listening to the radio,engaging in a conversation) while driving. Performing several taskssimultaneously would not be possible if each of these tasks imposed

FALL 1988


a large demand on the limited capacity information processingsystem.

In an impressive series of studies, Johnson & Rahe (1981)demonstrated that accurate discrimination between perceived andinferred information is likely to occur when the processes mediatingthe derivation of inferred information from perceived informationare controlled and effortful. On the other hand, origin confusionis likely to occur when the processes mediating the derivation ofinferred information from perceived information are relativelyeffortless.

Thus, when a consumer exhibits confusion about the origin of apiece of information that was not stated expHcitly in an ad, a re-searcher may conclude that this infonnation was infened by theconsumer prior to measurement and that the cognitive mechanismsunderlying this inference process were relatively effortless.* How-ever, the absence of origin confusion is difficult to interpret. Originconfusion may fail to occur (1) because the inference was notformed prior to measurement, or (2) because the inference wasformed prior to measurement through a relatively effortful process.

THE PRESENT STUDY

In the present study, several sets of product claims were pre-sented to subjects. Each set consisted of two premises of the form"A implies B" and "B impUes C" (Loken, 1981; Loken & Wyer,1983). In addition, a conclusion of the form "A implies C" waspresented explicitly in half of the sets and omitted from the remain-ing sets.

To illustrate, subjects were asked to rate the validity of severalproduct claims of the form "Stresstabs contain B vitamins" (Aimplies B) and "B vitamins give the energy" (B implies C). On thebasis of these premises, subjects may infer the missing conclusionthat "Stresstabs give you energy" (A implies C).

A recognition/memory technique was employed to determinewhether or not subjects drew this inference. This technique involvedasking subjects to rate the validity of several sets of propositions,and subsequently asking them to complete a surprise recognitionquestionnaire. Some of the test conclusions in this questionnairewere identical to the conclusions presented earlier and some werefoils (i.e., conclusions that had not been presented earlier). Subjectswere asked to indicate whether or not they recognized each conclu-sion and they were also asked to judge how confident they were thateach conclusion either had or had not been presented earlier. Of

PSVCHOLOGY & MARKETING/VOL. 5, NO. 3 277


critical concern, then, was the recognition confidence associated withtest conclusions that had not been presented earlier.

Correct Recognition of Conclusions that Had Been Presented

It was reasoned that correct recognition of a conclusion thatactually had been presented earlier (a "hit") may refiect a straight-forward direct retrieval process. That is, an explicitly presentedconclusion may be encoded in a relatively unadulterated fashionwhen it is first considered. Later, during a recognition task, this in-formation may be retrieved directly.

One factor that may infiuence conect recognition is priorknowledge. Considerable evidence exists showing that prior knowl-edge facilitates the learning of new information and enhances mem-ory for this information (Bettman, 1979; Hastie, 1981; Johnson &Russo, 1984; Taylor & Crocker, 1981). Thus, recognition accuracymay be greater for conclusions pertaining to familiar than to un-familiar products.

False Recognition of Conclusions that Had Not Been Presented

False recognition of a conclusion that had not been presentedearlier (a "false alarm"), on the other hand, implies that this conclu-sion was inferred from previously considered premises, and that theinferred and presented propositions were integrated into a single,coherent cognitive unit. Later, during a recognition task, the entirecogmtive unit is retrieved from memory and the subject experiencesdifficulty in distinguishing between explicit and inferred informa-tion. This type of confusion indicates that the inference was formedprior to measurement and that automatic cognitive operations medi-ated the formation of this inference.

One factor that may infiuence false recognition is the related-ness of the propositions that imply the missing conclusion. Deduc-tive prmciples may gmde inference processes only when the premisesare logically related. To test this prediction, control conditions wereconstructed for the purpose of comparing false recognition responsesto a conclusion when logically related premises were presented (e.g.,"A implies B" and "B implies C") to false recognition responses to aconclusion when logicaDy unrelated premises were presented (e.g.,"A implies D" and "E implies C"). These premises were presented

278 FALL 1988


in a random order, with the constraint that no two premises fromthe same set could occur consecutively. Hence, each premise wassunounded by premises from other sets and conclusions sometimesoccuned prior to their relevant premises. This procedure was em-ployed to minimize pressures on the subject to engage in deductivereasoning.

To summarize, a 2 (presented or omitted conclusion) X 2(familiar or unfamiliar product) X 2 (logically related or unrelatedpremises) factorial design was employed. Recognition confidenceserved as the dependent measure. It was predicted (1) that recogni-tion confidence for targets (i.e., presented conclusions) may begreater for conclusions pertaining to familiar products than forconclusions pertaining to imfamihar products, and (2) that recogni-tion confidence for foils (i.e., omitted conclusions) may be greaterfor conclusions that are supported by logically related premises thanfor conclusions that are associated with logically unrelated prem-ises.

METHOD

Overview

Respondents were 36 undergraduates attending a large midwest-em university who volunteered to participate in a study about"beliefs and opinions." Respondents first judged the subjectiveprobabilities of 44 product claims. Next they received a set of mathproblems and were instructed to solve them as quickly as possible.This task was employed to prevent rehearsal and to eliminate short-term memory effects on the final task, a surprise recognition con-fidence task.

The Belief Questionnaire

Examples of the product claims contained in the belief ques-tionnaire are presented in Table 1. The belief questionnaire consistedof 16 two-statement sets of logically related and unrelated premises.Half of these sets contained a conclusion and half did not. Further,half of the sets pertained to familiar products (e.g., Nytol, Excedrin,Geritol, Rolaids, Stresstabs, No Doz, Alka Seltzer, and Tylenol) andhalf pertained to unfamiliar products (e.g., Lemphorin, Allogen, Adri-nyl, Seratol, Myacin, Cedatdl, Tolarin, and Gelutal). Postexperimental

PSYCHOLOGY & MARKETING/VOL. 5, NO. 3 279

A NONREACTIVE MEASURE OF INFERENTIAL BEUEFS

TABLE 1Examples of tiie Bdief Statement Sets

Logically Related Propositions

A implies B: Taking Excedrin will relieve tension headaches.

B implies C: Medication that relieves tension headaches may causestomach upset.

A implies C: Taking Excedrin may cause stomach upset.

Logically Unrelated Propositions

A implies D: Seratol helps relieve insomnia.

E implies C: Intestinal disorders often cause a loss of appetite.

A implies C: Seratol often causes a loss of appetite.

interviews revealed that subjects did not suspect that these unfamil-iar products were fictional.

Two sets of syllogisms were presented in each cell of the 2 X2 X 2 experimental design. In addition, four conclusions with nopremises were included as filler product claims. These claims per-tained to familiar and unfamiliar products not mentioned in theother sets. Subjects were asked to judge the validity of each claimon a scale ranging from - 4 (extremely unlikely to be true) to +4(extremely likely to be true).

It should be noted that subjects were asked to judge the prod-uct claims without being explicitly asked to leam them or thinkabout them in relation to one another. Moreover, the claims werepresented in a random order, and no two claims belonging to thesame set occurred together. This procedure was used to mimimizepressures to engage in deductive reasoning.

The Recognition Confidence Task

Immediately after completing the belief questionnaire, subjectssolved math problems for 10 minutes. This was done to eliminateshort-term memory effects. Next they received a surprise recogni-tion task in which they were asked to rate the confidence withwhich they believed they either had or had not seen each of 16test-conclusions. Half of these statements had been presented earlierand half had not. Subjects were asked to indicate how confidentthey were that they either had or had not seen each statementearlier in the session on a scale ranging from - 4 ("I am extremelycertain I have seen the statement") to +4 ("I am extremely certain Ihave seen the statement).

280 FALL 1988


RESULTS

A 2 (presented or omitted conclusion) X 2 (familiar or un-familiar product) X 2 (logically related or unrelated premises)repeated measures analysis of variance was performed on recogni-tion confidence ratings of the conclusion statements. This analysisindicated that recopiition confidence was ^ a t e r when the conclu-sion had been presented than when it had not been presented F(l,35) = 196.17, p < 0.001, when the product was familiar than whenit was not familiar, F( l , 35), = 6.68, p < 0.02, and when the prem-ises were logically related than when they were not related, F(l , 35)= 6.28, p < 0.02. Moreover, significant presence X familiarity,F( l , 35) = 4.65, p < 0.04, and presence X relatedness, Fil, 35) =7.15, p < 0.02, interactions were found. The familiarity X related-ness interaction (F = 1.95) and the three-way interaction (F < 1)were not significant. To ir.terpret these interactions, separate analy-ses were performed on reccjnition confidence ratings of conclusionsthat were explicitly presented (correct recognition) and of conclu-sions that were omitted (false recognition).

Correct Recognition

Correct recognition confidence ratings as a function of producttype and premise type are presented in the first and second columnsof Table 2. A 2 (familiarity) X 2 (relatedness) repeated measuresanalysis of variance performed on these ratings yielded a familiarityeffect, F(l , 35) = 8.54, p < 0.01, and no other main effects or inter-actions. As predicted, recognition confidence was greater for conclu-sions pertaining to familiar products (M = 1.83) than for conclusionspertaining to unfamiliar products (M = 0.86).

False Recognition

False recognition confidence ratings as a function of producttype and premise type are presented in the third and fourth columnsof Table 2. A 2 (familiarity) X 2 (relatedness) repeated measuresanalysis of variance performed on these ratings revealed a relatednesseffect, F( l , 35) = 16.43, p < 0.001, and no other main effects orinteractions. As hypothesized, recognition confidence was greaterfor conclusions that were implied by logically related premises(M = — 1.95) than for conclusions that were associated with logicallyunrelated premises (M = —2.92). This finding suggests that when

PSYCHOLOGY & MARKETING/VOL. 5. NO. 3 281


TABLE 2Recognition Confidence for Conclusions as a Function

of Product Familiarity and Type of Premises

Types ofpremises

RelatedUnrelated

Presented Conclusions

Familiarproducts

1.6781.99

Unfamiliarproducts

1.060.65

Omitted

FamiliarProducts

-1.93-2.74

Conclusions

Unfamiliarproducts

-1.96-3.10

^Higher numbers indicate greater recognition confidence.

logically related product claims are presented, consumers can con-strue the implications of these claims with relatively little effort.

DISCUSSION

The findings indicate that recognition confidence ratings canbe useful for extrapolating consumers' inferential beliefs withoutdirectly measuring or influencing these beliefs. The present nonre-active measurement procedure enables researchers to measure con-sumers' inferential beliefs without prompting consumers to engagein deductive reasoning. Instead, inferential beliefs that were formedspontaneously prior to measurement are assessed.

The results are also consistent with the "reality monitoring"model of memory (Johnson & Rahe, 1981), which focuses on theprocesses by which people determine whether a given piece of in-formation retrieved from memory originated from an external or aninternal source. Consumers experience little difficulty in determin-ing the origin of a piece of information when it was inferred throughan effortful process. When an inference process is not effortful, onthe other hand, consumers cannot make the external/internal sourcediscrimination with a high degree of confidence.

Although subjects were never explicitly asked to form infer-ences, and although inference formation was assessed by employingan unexpected recognition test, subjects formed inferential beliefsbefore they were exposed to the recognition test. Moreover, productclaims were presented in a random order and no two claims aboutthe same product were presented contiguously. Nevertheless, subjectsspontaneously deduced the inferential implications of logically rela-ted product claims.

Before discussing the implications of the present research, thedistinction between spontaneous and automatic processes should be

282 FALL 1988


clarified. A spontaneous process requires little cognitive effort,whereas an automatic process requires no effort. Moreover, anautomatic process is activated "whenever a given set of external in-itiating stimuli are presented, regardless of a subject's attempt toignore or bypass the distraction" (Shiffrin & Dumais, 1981, p. 117).An automatic process does not require conscious awareness or inten-tion and it cannot be prevented or suppressed. Thus, the key featureof an automatic process is its inescapability. As the effort requiredto perform a given cognitive operation decreases, the ability ofconsumers to avoid performing this operation also decreases.

Many advertisers create ads that do not require consumei toengage in effortful inference processes (Preston, 1977; Wilkie,McNeill, & Mazis, 1984). Instead, many advertising communicationsare designed to prompt consumers to draw inferences spontaneously.Misleading ads that lead consumers to draw invalid inferences spon-taneously and inescapably, then, are particularly insidious, becauseconsumers are defenseless against such advertising practices. Thepresent nonreactive measurement procedure offers a means foridentifying advertising claims that prompt consumers to form in-valid inferences spontaneously. This procedure possesses the advan-tage of enabling the researcher to determine if a particular inferencehad been formed before the consumer is exposed to any measures orqueries. Further, the present procedure may prove to be a usefultechnique for addressing empirically difficult but important issuespertaining to how consumers generate inferences about products.

FOOTNOTES

* Several inferential rules have been identified, such as covariation, augmenta-tion, and discounting rules for causal inference (Kelley, 1973); averaging(Anderson, 1981), adding (Fishbein & Ajzen, 1975), and Bayesian (Fischhoff& Beyth-Marom, 1983) rules for integrating information; and various rulesbased upon schema theory (Abelson, 1981; Hastie, 1981; Taylor & Crocker,1981). Moreover, Wyer and Hartwick (1980) have described deductionquantitatively:

where Pg refers to the subjective probability with which conclusion B is true;PA and i*^ • refer to the subjective probabilities with which premis/1 is trueand not true, respectively; and P^IA and PB/A' refer to the conditionalsubjective probabilities with which B is true if 4 is true and is not true,respectively.The inference processes elicited by most paper-and-pencil measurementinstruments tend to be relatively controlled and effortful (Webb et al., 1966).Moreover, the term "relatively" is used because many everyday inference

PSYCHOLOGY & MARKETING/VOL 5. NO. 3 ^"


processes are not represented by the extreme ends of the automatic-con-trolled continuum (Kahneman & Treisman, 1984; Shiffrin, in press).

REFERENCES

Abelson, R. P. (1981). Psychologic., atus of the script concept. AmericanPsychologist, 36, 715-729.

Alba, J. W., & Hutchinson, J. W. (1987). Dimensions of consumer expertise.Journal of Consumer Research, 13, 411-454.

Anderson, N. H. (1981). Foundations of information integration theory. NewYork: Academic Press.

Armstrong, G. M., Gurol, M. N., & Russ, F. A. (1978). Detecting and correctingdeceptive advertising./ouma/o/Co«JMmer.RcjeflrcA, 6, 237-246.

Bargh, J. A. (1984). Automatic and conscious processing of social information.In R. S. Wyer & T. K. Srull (Eds.), Handbook of social cognition, VoL 3,(pp. 1-43). Hillsdale, NJ: Lawrence Erlbaum.

Bettman, J. R. (1979). An information processing theory of consumer choice.Reading, MA: Addison-Wesley.

Bransford, J. D., & Franks, J. J. (1971). The abstraction of linguistic ideas.Cognitive Psychology, 2, 331-350.

Dyer, R. F., & Keuhl, P. G. (1978). A longitudinal study of corrective advertis-ing./owrna/o/Mflrfcer/n^/?ciejrcA, 15, 39-48.

Fischhoff, B., & Beyth-Marom, R. (1983). Hypothesis evaluation from a Bay-esian perspective. Psj c/ioZogjco//Review, 90. 239-260.

Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: Anintroduction to theory and research. Reading, MA: Addison-Wesley.

Gardner, D. M. (1975). Deception in advertising: A conceptual approach./our-nal of Marketing, 39, 40-46.

Hasher, L., & Zacks, R. T. (1979). Automatic and effortful processes in mem-ory. Journal of Experimental Psychology: General, 108, 356-388.

Hastie, R. (1981). Schematic principles in human memory. In E. T. Higgins,C. P. Herman, & M. P. Zanna (Eds.), Social cognition: The Ontario sympo-sium. Vol. 1 (pp. 39-88). Hillsdale, NJ: Lawrence Erlbaum.

Hastie, R. (1983). Social inference. In M. R. Rosenzweig & L. W. Porter (Eds.),Annual review of psychology. Vol. 34, (pp. 511-542). Palo Alto, CA:Annual Reviews. -

Hayes-Roth, B. (1977). Evolution of cognitive structures and processes. Psycho-logical Review, 84, 260-278.

HoUand, J. H., Holyoak, K. J., Nisbett, R. E., & Thagard, P. R. (1986). Induc-tion: Processes of inference, learning, and discovery. Cambridge, MA: MITPress.

Huber, J., & McCann, J. (1982). The impact of inferential beliefs on productevaluations. Journal of Marketing Research, 19, 324-333.

Johnson, E. J., & Russo, J. E. (1984). Product familiarity and learning new ixi-toTmation. Journal of Consumer Research, 11, 542-550.

Johnson, M. K., & Rahe, C. L. (1981). Reality monitoiing. Psychological Re-view, 88, 67-85.

Johnson, R. D., & Levin, I. P. (1985). More than meets the eye: The effect ofmissing information on purchase evaluations. Journal of Consumer Re-search, 12, 169-177.

Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice-HaU.

284 FALL 1988


Kahneman, D., & Triesman, A. (1984). Changing views of attention and auto-maticity. In R. Parasuraman & R. Davies (Eds.). Varieties of attention(pp. 29-61). New York: Academic Press.

Kelley, H. H. (1973). The processes of causal attribution. American Psycholo-gist, 28, 107-128.

Loken, B. (1981). Memory for logically-deduced conclusions. In A. A. Mitchell(Ed.), Advances in consumer research. Vol. 9, (pp. 348-353). Ann Arbor,MI: Association for Consumer Research.

Loken, B., & Wyer, R. S. (1983). Effects of reporting beUefs in syllogisticaUyrelated propositions on the recognition of unmentioned propositions.Journal of Personality and Social Psychology, 45, 306-322.

Lynch, J. G., & Srull, T. K. (1982). Memory and attentional factors in consumerchoice: Concepts and research methods. Journal of Consumer Research,9, 18-36.

Meyer, R. J. (1981). A model of multiattribute judgements under uncertaintyand informational constraints. Journal of Marketing Research, 18, 428-441.

Mizerski, R. W., Golden, L. L., & Keman, J. B. (1979). The attribution processin consiuner decision making. Journal of Consumer Research, 6, 123-140.

Nisbett, R., & Ross, L. (1980). Human inference: Strategies and shortcomings ofsocial judgment. Englewood Cliffs, NJ: Prentice-HalL

Olson, J. C. (1978). Inferential belief formation in the cue utilization process.In H. K. Hund (Ed.), Advances in consumer research. Vol. 5, (pp. 706-713). Ann Arbor, MI: Association for Consumer Research.

Olson, J. C, & Dover, P. A. (1978). Cognitive effects of deceptive advertising.Journal of Marketing Research, 15, 29-38.

Preston, I. L. (1967). Logic and illogic in the advertising process. JournalismQuarterly, 44, 231-239.

Preston, I. L. (1977). The FTC's handling of puffery and other selling claimsmaie'hy iia^^cation.' Journal of Business Research, 5, 155-181.

Preston, I. L., & Scharbach, S. E. (1971). Advertising: More than meets the eye?Journal of Advertising Research, 11, 19-24.

Russo, J. E., Metcalf, B. L., & Stephens, D. (1981). Identifying misleading ad-vertising./ourna/o/Consumer/JejearcA, 8, 119-131.

Schmittlein, D. C, & Morrison, D. G. (1983). Measuring miscomprehension fortelevised communications using true-faise questions. Journal of ConsumerResearch, 10, 147-156.

Schneider, W., & Shiffrin, R. M. (1977). ControUed and automatic human in-formation processing: I. Detection, search, and attention. PsychologicalReview, 84, 1-66.

Shiffrin, R. M. (in press). Attention. In R. C. Atkinson, R. J. Hermstein, G.Lindzey, & R. D. Luce (Eds.), Stevens' handbook of experimental psy-chology (2nd ed.). New York: John Wiley & Sons.

Shiffrin, R. M., & Dumais, S. T. (1981). The development of automatism. InJ. R. Anderson (Ed.), Cognitive skills and their acquUition (pp. 111-140).Hillsdale, NJ: Lawrence Erlbaum.

Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human in-formation processing: II. Perceptual learning, automatic attending, and ageneral theory. Psychological Review, 84, 127-190.

Shimp, T. A. (1979). Social-psychological (mis)representations in televisionadverting. Jounrnl of Consumer Affairs, 13, 28-40.

Shimp, T. A., & Preston, I. L. (1981). Deceptive and nondeceptive consequencesof evaluative advertising./ottma/o/^arfc«j>l^, 45, 22-32.

PSYCHOLOGY & MARKETINGTVOL. 5. NO73 285~


Taylor, S. E., & Crocker, J. (1981). Schematic bases of social information proc-essing. In E. T. Higgins, C. P. Herman, & M. P. Zanna (Eds.). Social cogni-tion: The Ontario symposium. Vol. 1 (pp. 89-134). Hillsdale, NJ: Law-rence Erlbaum.

Webb, E. J., Campbell, D. T., Schwartz, R. D., &Sechrest, L. (1966). Unobtru-sive measures: Nonreactive research in the social sciences. Chicago: RandMcNally.

Wilkie, W. L., McNeill, D. L., & Mazis, M. B. (1984). Marketing's "scarlet let-ter": The theory and practice of corrective advertising. Journal of Mar-keting, 48, 11-31.

Wilkie, W. L., & Pessemier, E. A. (1973). Issues in marketing's use of multi-attribute attitude models./owrwa/o/it/arfccftn^/?ejearc/i, 10, 428-441.

Wyer, R. S., & Hartwick, J. (1980). The role of information retrieval and condi-tional inference processes in belief formation and change. In L. Berkowitz(Ed.), Advances in experimental social psychology. Vol. 13 (pp. 241-284). New York: Academic Press.

The author wishes to thank Russell Fazio, Perilou Goddard, David Mick, RichardOlshavsky, Steven J. Sherman, Robert Smith and four anonymous reviewers fortheir helpful comments on an earlier version of this manuscript.

Requests for reprints should be addressed to Frank R. Kardes, Sloan School ofManagement, Massachusetts Institute of Technology, Cambridge, Massachusetts02139.

286 FALL 1988

A Nonreactive Measure of Inferential Beliefshomepages.uc.edu/~kardesfr/Kardes Articles/Kardes 1988...

Documents

Transcript of A Nonreactive Measure of Inferential Beliefshomepages.uc.edu/~kardesfr/Kardes Articles/Kardes 1988...