Positivity and Negativity Effects in Inferences About Products

ABSTRACT - A pilot experiment was conducted to examine the effects of initial judgments and perceived technicality on inference processes. Respondents received technical information about a product. After reading this information, they answered leading questions designed to induce either a positive or a negative initial judgment of the product. The results suggest that initial judgments affect inference processes in conditions of moderate perceived technicality, but not under conditions of either low or high perceived technicality. Moreover, both positivity ant negativity effects were observed. The conditions under which positivity effects are likely and the conditions under which negativity effects are likely are discussed.

Citation:

Frank R. Kardes and William M. Strahle (1986) ,"Positivity and Negativity Effects in Inferences About Products", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 23-26.

Advances in Consumer Research Volume 13, 1986      Pages 23-26

POSITIVITY AND NEGATIVITY EFFECTS IN INFERENCES ABOUT PRODUCTS

Frank R. Kardes, Indiana University

William M. Strahle, Indiana University

ABSTRACT -

A pilot experiment was conducted to examine the effects of initial judgments and perceived technicality on inference processes. Respondents received technical information about a product. After reading this information, they answered leading questions designed to induce either a positive or a negative initial judgment of the product. The results suggest that initial judgments affect inference processes in conditions of moderate perceived technicality, but not under conditions of either low or high perceived technicality. Moreover, both positivity ant negativity effects were observed. The conditions under which positivity effects are likely and the conditions under which negativity effects are likely are discussed.

INTRODUCTION

Recently a number of prominent consumer researchers have called for more research in the area of inference processes (Bettman 1979, Huber and McCann 1982, Loken 1981, Lynch and Srull 1982, Meyer 1981, Olson 1978). Unfortunately, it is not clear how one should study inference processes and as a consequence, little research has been conducted on this important topic. One notable exception, however, is a study conducted by Huber and McCann (1982). The present experiment was designed to extend Huber and McCann's (1982) findings.

The Negativity Effect

Huber and McCann (1982) conducted an experiment in which product price ($1.50, 1.75, 2.00, 2.25, or 2.50) and product taste (better than 10, 30, 50, 70, or 90% of the competitors' products) information was varied. Either price, taste, or price and taste information was omitted systematically and respondents were either prompted or not prompted to draw inferences about the omitted information. The results revealed that both prompted and unprompted inferences influenced respondents' judgments of the product. Moreover, respondents did not substitute missing information about an attribute with the mean value of that attribute. Rather, respondents assumed that missing information about a particular attribute meant that this attribute had a below-neutral value. This was called the negativity effect.

Similar results were found in a study conducted by Meyer (1981). In this study, respondents received information pertaining to three attributes of hypothetical pizza restaurants: pizza quality, decor ant music, and price. Attribute information was omitted systematicaLly. The results indicate that a below-neutral value vas assigned to an attribute for which respondents had 30 information. This value was integrated with the values of attributes for which information was provided to form an overall evaluation.

Under what conditions are positive inferential beliefs formed and under what conditions are negative inferential beliefs formed? To answer this question, one must Consider the factors that may influence the inference process. The present experiment examines two such factors, initial judgments and perceived technicality.

The Role of Initial Judgments in the Inference Process

One factor which may influence the inference process is aa individual's initial judgment or "first impression" of a product. Early judgments of a stimulus influence subsequent judgments of the stimulus and early judgments may even affect judgments of attributes for which no information was made available (Carlston 1980, Lingle ant Ostrom 1981). To test this notion experimentally, one could manipulate respondents' initial judgments of a target product by using a methodology that was developed by Carlston (1980). Carlston reasoned that when an individual is asked to judge a stimulus, the individual does not search memory for all of the information he or she has stored about the stimulus. Rather, only a small subset of one's stored knowledge is retrieved. The particular subset that is selected often depends upon the wording of the question one is attempting to answer. If the question pertains to a positive characteristic of the stimulus, the individual may form a positive initial judgment. If the question pertains to a negative characteristic, the individual may form a negative initial judgment.

If an initial judgment is processed deeply (see Craik and Lockhart 1972, ant Greenwald ant Leavitt 1984), the initial judgment mag drive subsequent judgments ant inferences. That is, deeply processed positive initial judgments may lead to positive inferences, whereas deeply processed negative initial judgments may lead to negative inferences. This has been called the initial judgment effect (Wyer, Srull, and Gordon 1984).

The present experiment was designed to determine the conditions under which the initial judgment effect is likely to occur spontaneously. Previous research has shown that the initial judgment effect occurs when respondents are asked to form an impression of a stimulus (Carlston 1980, Wyer et al. 1984). However, whether or not respondents will form an impression ant exhibit the initial judgment effect in the absence of instructions to form an impression of a stimulus is unclear. We reasoned that spontaneous inference processes may depend upon levels of perceived technicality.

The Role of Perceived Technicality in the Inference Process

The characteristics of consumers may affect the depth of processing of new information about a product. For example, recent research has shown that consumers who are low in prior knowledge and experience, with respect to a particular product domain, are unlikely to process new information about that product class very deeply (Bettman and Park 1980, Johnson ant Russo 1984). This is because they lack the prior knowledge ant experience that is necessary for processing new information about the product class. As a consequence, they are forced to rely on what little information they have available in memory when asked to judge an instance of the product class.

Consumers who are high in prior knowledge and experience, on the other hand, possess the prior knowledge ant experience that is necessary for processing new information. However, they may be unmotivated to process much new information because they way feel that they do not need to (Bettman and Park 1980, Johnson and Russo 1984). Instead they may simply rely on the large store of information they have available in memory when asked to judge a product.

Consumers who possess a moderate amount of prior knowledge and experience have the ability to process new information and they may be motivated to process new information deeply. This is because they have enough prior knowledge and experience to process new information effectively, but they to not have so much prior knowledge and experience that they feel that they can afford to rely solely on information available in memory when asked to judge a product (Bettman and Park 1980, Johnson and Russo 1984).

To elaborate, if a product description is perceived as high in technicality, an individual may consider the information provided in the description as confusing and, perhaps, overwhelming. As a result, the individual may be unwilling and/or unable to spend the time and energy required to make sense of this information. Consequently, the individual may fail to process this information deeply.

If a product description is judged to be not very technical, an individual may be very knowledgeable and as a consequence, he may be unmotivated to exert a great deal of energy in order to understand the description. Searching memory for information is faster and easier than deeply processing new information provided externaLly.

If a product description is perceived as moderately technical, however, an individual may consider the information provided in the description to be interesting ant comprehensible. In this case the individual may be willing and able to process new information deeply. Thus, information may be processed more deeply in conditions of moderate perceived technicality than in conditions of either low or high perceived technicality.

The Present Study

In the present study, all respondents were exposed to identical information about a product. This information consisted of favorable information about one attribute and unfavorable information about another. After reading this information, respondents answered questions about either the positive attribute or the negative attribute. These questions were used to manipulate respondents' initial judgments. One week later, respondents rated the attributes which were mentioned in the original product description and they also judged an attribute which was never mentioned in this description. It was predicted that respondents' inferences about this unmentioned attribute would be congruent with their initial judgments only when the product information was processed deeply (i.e., in moderately technical conditions).

METHOD

Overview

Forty-four male and female undergraduate students were randomly assigned to one of two initial judgment conditions (positive or negative). Hence, there were 22 respondents per condition. Further, respondents were divided into one of three perceived technicality conditions (low, moderate, high) on the basis of their ratings of the technicality of the product description. The primary dependent measure was respondents' inferences about an attribute which was not mentioned in the product description.

The Product Description

All respondents read the same information about an ostensibly new stereo system which we referred to as the T85 stereo system. Two attributes of the product were described, the turntable ant the loudspeakers. The information pertaining to the turntable had positive implications ant the information pertaining to the loudspeakers had negative impLications. Respondents were told, simply, to read the product description. They were told that they would be asked to rate the description for clarity so that it could be administered to another group of respondents. Respondents did not expect to judge the target product, nor did they expect to rate the product's attributes in a second session. After they read the product description, the description was collected and a questionnaire was administered.

The Initial Judgment Manipulation

Respondents in the positive initial judgment condition rated the favorability of the performance of the turntable and one component of the turntable, the tonearm.

Respondents in the negative initial judgment condition rated the performance of the loudspeakers and one component of the loudspeakers, the tweeter. All items were rated on seven-point semantic differential scales.

The Perceived Technicality Measure

Respondents were divided into three groups (low, moderate, and high perceived technicality) on the basis of their ratings of the technicality of the product description. A seven-point semantic differential scale with endpoints labelled "Not very technical" and "Very technical" was used.

Session Two

One week later, a second set of questionnaires were administered. Respondents' global judgments (e.g., s = ed ratings of respondents' overall impression of and liking for the target product), inferences (e.g., ratings of the favorability of the receiver's performance), and purchase intentions were measured on seven-point semantic differential scales.

RESULTS

Manipulation Check

A 2 (Positive or negative initial judgment) x 3 (Low, moderate, or high perceived technicality) fixed effect between-groups analysis of variance was performed on respondents' initial judgment ratings to determine whether or not the leading questions were effective in manipulating respondents' initial judgments. Equal weights were used to handle unequal cell sizes (Kenny 1985). The analysis yielded a significant main effect for initial judgment condition, F(1, 38) = 36.75, p < .001, and no other effects (all F's < 1). As Table 1 indicates, initial judgments were more positive in positive initial judgment conditions than in negative initial judgment conditions. Thus, the initial judgment manipulation was clearly effective.

TABLE 1

INITIAL JUDGMENTS AS A FUNCTION OF THE JUDGMENT CONDITION AND PERCEIVED TECHNICALITY

Inferences About the Performance of an Unmentioned Attribute

A 2 x 3 analysis of variance was performed on respondents' judgments of the performance of the stereo system's receiver. Postexperimental interviews revealed that respondents failed to reaLize that information about the receiver was not provided in the product description. Nevertheless, respondents' judgments of the receiver varied as a function of initial judgment ant perceived technicality. A significant main effect for perceived technicality was found, F(2, 38) = 5.96, p < .01. Respondents who perceived the product as moderately technicaL or highly technical formed more positive inferences than respondents who perceived the product as not technical. The main effect for initial judgment was not significant (F - 1.30), but the initial judgment x perceived technicality interaction was marginally significant, F(2, 38) = 3.11, p < .06.

As predicted, initial judgments tit not influence respondents' inferences in low (F = 2.07, ns) or in high (F < 1) perceived technicality conditions. In moderate perceived technicality conditions, however, initial judgments exerted a powerful influence on respondents' inferences, F(1, 21) 8.42, p < .01.

As Table 2 indicates, initial judgments had no effect on inferences in low or high perceived technicality conditions. In moderate perceived technicality conditions (see the middle column in Table 2), inferences were more favorable in positive than in negative initial judgment conditions. Hence, when initial judgments are processed deeply, subsequent inferences are formed which are evaluatively congruent with the initial judgments.

TABLE 2

INFERENCES ABOUT TEE PERFORMANCE OF AN UNMENTIONED ATTRIBUTE AS A FUNCTION OF INITIAL JUDGMENT AND PERCEIVED TECHNICALITY

Ancillary Analyses

A 2 x 3 analysis of variance was performed on respondents' global judgments of the target product. This analysis yielded a significant initial judgment x perceived technicality interaction, F(2, 38) = 5.76, p < .01. The initial judgment effect tended to occur only in conditions of moderate perceived technicality.

Respondents' inferences were correlated with their initial judgments, r = .27, p < .04. Hence, initial judgments can be used to predict inferences. Moreover, respondents' inferences about one previously unmentioned attribute of the target product tended to correlate with purchase intentions, r = .24, p < .06. This correlation is surprisingly high, given that this is a correlation involving only one inference and respondents may have formed many inferences. However, global judgments were the single best predictor of purchase intentions, r = .44, p < .01.

DISCUSSION

The results of the present experiment indicate that inference processes are not characterized solely by negativity effects. When initial judgments are processed deeply, positive initial judgments lead to positive inferences and negative initial judgments lead to negative inferences.

It is important to note that respondents were not told to form an impression of the target product while reading the product description. Further, respondents did not expect to judge the product or its attributes. Moreover, respondents did not expect to participate in a second session conducted one week after the first session. Nevertheless, respondents were able to express initial judgments and in some cases, these initial judgments affected respondents inferences one week later. These inferences pertained to an attribute that was not mentioned in the product description. Therefore, it is clear that the product description could not have directly influenced respondents' judgments of this attribute. Any effect of the product description on these judgments must have been mediated, at least partially, by initial judgments.

The present study raises some interesting issues for future research. Technicality was measured in the present study. Future work in which technicality is manipulated is needed. Moreover, a study in which instructions are manipulated (e.g., impression versus recall set) and in which appropriate control conditions are included could be useful in delimiting the initial judgment phenomenon.

Just as it is the case that seemingly trivial factors (e.g., subtle changes in a decision situation) can have a large impact on judgment and choice (see Payne 1982), so too it is the case that seemingly trivial factors (e.g., subtle changes in the wording of a question) can affect inference processes. In our view, similar processes may be involved in inference, judgment, and choice. The field of consumer behavior could profit greatly from an integration of these issues and future research should be directed toward such an integration.

REFERENCES

Bettman, J. R. (1979), An Information Processing Theory of Consumer Choice. Reading, MA: Addison-Wesley.

Bettman, J. R. and C. W. Park (1980), "Effects of Prior Knowledge and Experience and Phase of the Choice Process on Consumer Decision Processes: A Protocol Analysis," Journal of Consumer Research, 7 (December), 234-248.

Carlston, D. E. (1980), "The Recall and Use of Traits and Events in Social Inference Processes," Journal of Experimental Social Psychology, 16 (July), 303-328.

Craik, F. I. M. and R. S. Lockhart (1972), "Levels of Processing: A Framework for Memory Research," Journal of Verbal Learning and Verbal Behavior, 11 (December),

Greenwald, A. G. and C. Leavitt (1984), "Audience Involvement in Advertising: Four Levels," Journal of Consumer Research, 11 (June), 581-592.

Huber, J. and J. McCann (1982), "The Impact of Inferential Beliefs on Product Evaluations," Journal of Marketing Research, 19 (August), 324-333.

Johnson, E. J. and J. E. Russo (1984), "Product Familiarity ant Learning New Information," Journal of Consumer Research, 11 (June), 542-550.

Kenny, D. A. (1985), "Quantitative Methods for Social Psychology," in Handbook of Social Psychology, Vol. 1, eds. G. Lindzey and E. Aronson, New York. Random House, 487-508.

Lingle, J. H. ant T. M. Ostrom (1981), "Principles of Memory and Cognition in Attitude Formation," in Cognitive Responses in Persuasion, eds. R. E. Petty, T. M. Ostrom, and T. C. Brock, Hillsdale, NJ: Erlbaum, 399-420.

Loken, B. (1981), "Memory for Logically-Deduced Conclusions," in Advances in Consumer Research, Vol. 9, ed. A. A. Mitchell, Ann Arbor: Association for Consumer Research, 348-353.

Lynch, J. G. and T. K. Srull (1982), "Memory ant Attentional Factors in Consumer Choice: Concepts ant Research Methods," Journal of Consumer Research, 9 (June), 18-36.

Meyer, R. J. (1981), "A Motel of Multiattribute Judgments Under Attribute Uncertainty and Informational Constraint," Journal of Marketing Research, 18 (November), 428-441.

Olson, J. C. (1978), "Inferential Belief Formation in the Cue Utilization Process," in Advances in Consumer Research, Vol. 5, ed. H. K. Hunt, Ann Arbor: Association for Consumer Research, 706-713.

Payne, J. W. (1982), "Contingent Decision Behavior," Psychological Bulletin, 92 (September), 382-402.

Wyer, R. S., T. K. Srull, and S. Gordon (1982), "The Effects of Predicting a Person's Behavior on Subsequent Trait Judgments," Journal of Experimental Social Psychology, 20 (January), 29-46.

----------------------------------------