Consumer Use of Information and Confidence in Making Judgements About a New Food Store: an Attribution Theory Analysis

ABSTRACT - Kelly's ANOVA model is offered as an explanation for consumer uncertainty in making new product judgments and its relevance is demonstrated with a scenario typical of many new product evaluations. Next, an experiment testing the model's predictions is discussed. The results, directions for future research, and implications for marketing are provided.

Citation:

Richard F. Yalch and Suzanne Yoshida (1983) ,"Consumer Use of Information and Confidence in Making Judgements About a New Food Store: an Attribution Theory Analysis", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 40-44.

Advances in Consumer Research Volume 10, 1983      Pages 40-44

CONSUMER USE OF INFORMATION AND CONFIDENCE IN MAKING JUDGEMENTS ABOUT A NEW FOOD STORE: AN ATTRIBUTION THEORY ANALYSIS

Richard F. Yalch, University of Washington

Suzanne Yoshida, University of Washington

ABSTRACT -

Kelly's ANOVA model is offered as an explanation for consumer uncertainty in making new product judgments and its relevance is demonstrated with a scenario typical of many new product evaluations. Next, an experiment testing the model's predictions is discussed. The results, directions for future research, and implications for marketing are provided.

INTRODUCTION

With few exceptions (e.g., Hansen and Stonner 1978, Mizerski 1978), most consumer research on attribution theory has been in an advertising context and has focused on source credibility (Settle and Golden 1974, Smith and Hunt 1978), or some other aspect related to source perceptions (Robertson and Rossiter 1974, Sparkman and Locander 1980, Swinyard 1981). [Other research has examined consumer self-perceptions allowing responses to requests (e.g., Scott 1977, Tybout 1979), but this has typically been designed to test self-perception theory rather than the four principles of attribution theory developed by Kelley (1967).] It is our feeling that this narrow perspective on attributions about the advertiser has ignored a useful application: consumers' analysis of their own knowledge about a new product or service.

This paper will first review four principles of attribution theory in the context of a consumer's judgment of a product after an initial exposure to it, illustrate how these principles offer an excellent description of the sources of uncertainty in the early stages of product evaluations, and suggest how they may also guide search behavior. Next, it will be argued that consumers may evaluate their information about a product consistent with attribution theory principles and the result of this evaluation should affect their judgments about the product and their confidence in these judgments. The results of an experiment testing these predictions will be reported and discussed.

AN ILLUSTRATION OF KELLEY'S ANOVA MODEL

The aspect of attribution theory of current relevance is Kelley's ANOVA model, which describes how individuals process information to understand cause-and-effect relationships. Space does not permit a full discussion of these principles and interested readers are directed to papers by Mizerski, Golden, and Kernan (1980) and Sparkman and Locander (1980). However, there appears Lo be sufficient confusion about the operationalizations of these principles to warrant a discussion with consumer examples.

Kelley (1967) proposed that individuals act as naive scientists when making judgments about cause-and-effect relationships (answering the question, "Why did this occur?" with an analysis of the available relevant data). Four tests or questions may be posed to judge an hypothesis regarding how much an entity caused an individual's behavior These are:

1. Distinctiveness -- To what extent is the effect present only when the entity is present and absent when other entities are present?

2. Consistency over time -- To what extent is the individual's reaction to the entity the same each time the entity is present?

3. Consensus -- To what extent do other individuals react to the entity in a similar way?

4. Consistency over modality -- To what extent do other methods of interacting with the entity result in similar reactions?

In addition, Kelley (1967) indicates that different patterns of behavior or information will result in different attributions regarding the causal importance of the three possible causes of the observed response. For example, when distinctiveness, consistency over time, consistency over modality, and consensus are all high, the entity should be judged to be the major cause of the response. On the other hand, when three of the analyses are high but consensus is low, the attribution should be to the individual. And, when all are low, the attribution should be to the circumstance (i.e., conclude that the entity and the individual had little causal influence).

One way to understand these abstract principles is to consider a hypothetical purchase situation. Assume that you have Just accepted a position in a different city and must travel to the city to purchase a house. You select a real estate agent and begin your search. The agent shows you one house and asks whether you would like to make an offer on it. Few consumers would feel comfortable in this situation and most would decline to make the offer.

Why is the information provided by the inspection not considered adequate? The uncertainty can be explained in terms of a series or questions likely to be present explicitly or implicitly. They are organized for purposes of illustration in terms of four attribution principles.

1. Distinctiveness -- Is the price/quality relationship good for this house? That is, what are comparable homes selling for or what else could I buy for this amount of money?

2. Consistency over time -- Will the house look as nice tomorrow as it does now? How will it seem at night? Will it be comfortable in the winter and summer?

3. Consensus -- Will other persons react as favorably toward the house as I have? Will my family like it? My friends? Why is the owner selling the house?

4. Consistency over modality -- The house looks nice, but what are the heating and maintenance expenses? What are the taxes? What is the neighborhood like? Are there any strange smells or sounds?

These questions should direct the search and evaluation of additional information about the house. [The link between attributions and search is complex. Interested readers are directed to two articles for studies examining this relationship, Hansen (1980) and Major (1980).] For example, consistency over time is easily tested by viewing this house on another day or time of day. Consensus can be tested by asking other real estate agents for their opinion, determining why the owner is selling, and having a professional appraisal. Consistency over modality can be tested by asking to see the heating bills, taxes, and maintenance costs, and by looking in the attic and at the foundation to examine the physical structure, walking over the floors to listen for squeaks, and smelling for odors in the basement and bathrooms. Finally, distinctiveness is tested by examining other homes to determine whether a better one is available for the same amount of money and whether similar homes are less expensive.

The analysis of this information will yield different conclusions depending on the pattern of the answers. Thus, if after looking at other houses the first one still stimulates the most positive reaction (high distinctiveness), if after several visits evaluations are still positive (consistency over time), if after looking at the heating, maintenance and tax expenses and inspecting the physical structure (high consistency over modality), and if all other individuals react positively (high consensus), the consumer should conclude that the house is the true cause of the positive reaction and be prepared to make an offer with confidence. On the other hand, if all the questions yield low consistency and distinctiveness, the individual should conclude that the initial positive reaction was a quirk and decline to make the offer with confidence. However, if there is high distinctiveness and consistency over time and modality, but low consensus, the individual should conclude that he has a true positive reaction to the house but for idiosyncratic reasons. In this case, the judgment may remain positive but probably with less confidence than in the situation where there is high consensus.

Sparkman and Locander (1980) conducted the most complete test of all four principles in a consumer context. In their study, after being exposed to one paragraph providing information about an advertising spokesperson's decision to appear in a campaign, subjects were asked to make an attribution about the spokesperson's decision, judging how much this was caused by his liking of the advertised product. Different paragraphs were developed to vary the amount of consistency and distinctiveness. The results were that consensus information was the only type to significantly affect these attributions and that the modality effect was in the opposite direction from that predicted. None of the possible interactions was reported.

The operationalizations and dependent measures may have affected these results. For example, consistency over modality resulted in an effect opposite to that predicted. High consistency over modality was represented by the statement that "he will appear in magazine ads, TV commercials, and also make personal appearances in dealer showrooms." Low consistency over modality was not described in the article but probably was communicated by stating that Ralph would only appear in television advertising for the product. The problem with this operationalization is that one might infer that Ralph is making more money by agreeing to appear in the various aspects of the promotional campaign (and thus has a substantially greater self-interest served by agreeing to endorse the product). This confound might have been alleviated by indicating in the message that Ralph was making the same amount of money in both campaigns. Another problem is that the different ways of promoting the product do not represent different ways of interacting with the product, but merely represent repeated public pronouncements that the product is good. A better method would be to focus on the different aspects of the product. For example, in the high consistency over modality condition, Ralph could have stated that he thought that the ride was smooth, the car had a beautiful appearance, and that the sound of the doors closing indicated that it was solidly built. In the low consistency, Ralph would have stated approval for only one of the dimensions of the automobile.

Sparkman and Locander's focus on attributions about the spokesperson, although consistent with attribution research, does not consider the effects of the attribution manipulations on the confidence in the attributions made by the audience. This is surprising because Kelley (1967) initially conceived the model as an explanation for the amount of confidence that an individual might have in the attributions as well as to change the attributions. That is, it is possible that one does not attribute a greater cause-and-effect relationship to the entity's impact on the individual but rather merely becomes more confident that the hypothesized cause-and-effect relationship is correct. Hansen (1980) presents the results of five experiments showing the relationship between confidence and the use of information presented in an attribution framework.

In consideration of the limitations of the Sparkman and Locander study and the fact most other attribution studies were concerned with effects on advertiser credibility rather than the consumer's use of information about a product and confidence in judgments made on the basis of this information, an experiment was conducted to test the following two research questions:

1. Do consumers utilize information when making product judgments consistent with attribution theory's description of how information is evaluated?

2. Do consumers report greater confidence in their judgments when their information is consistent with attribution theory's principles for evaluating information than when it is not?

These two questions were tested in a laboratory study of consumer judgments about a new food co-op store. Individuals were presented with favorable and unfavorable information about the store and this information was systematically varied in terms of the attribution principles of consistency over time, and modality, consensus, and distinctiveness. Then, they were asked to evaluate the new store and to indicate how confident they were in their judgments.

It was believed that favorable store information attributable to the store (i.e., high in consistency over time and modality, consensus, and distinctiveness) would result in the most favorable evaluations. Further, it was thought that this information would result in the consumers being most confident in their judgements relative to the other information presentations. These predictions are based on Kelley's ANOVA model, which specifies that a stimulus attribution (the store's features are the cause of the positive information) is most likely when the observations or information about it are high on all four dimensions. On the other hand, when there is high consistency over time and modality, high distinctiveness, but low consensus, the attribution should be to the person. Further, when consistency over time and modality, distinctiveness, and consensus are all Low, the attribution should be to a unique circumstance. In these two cases, the evaluation or the store should be less favorable. However, confidence might be high because interpretation of the information is unambiguous.

METHOD

Questionnaire

A three-page questionnaire was distributed in classrooms and dormitories to undergraduate students at a large urban university. Usable surveys were completed by 90 persons. The first page explained the study as a market research study about the feasibility of opening a food cooperative that would be student owned and operated. The food co-op was used because it was actually being considered and because it was a retail operation that did not currently exist in the campus area. This would ensure that the respondents would base their judgments on the information presented to them, though clearly some subjects would have beliefs from past experiences with food stores and co-ops.

The second part of the first page was an evaluation of the current food outlet patronized by the subject. This was included for possible use as a covariate to adjust for differences in the evaluations of food stores and to determine whether variations from the initial evaluation would be a better measure of the judgment variation attributable to the information presented about the co-op. Subsequent analysis revealed that although there was a significant relationship between these evaluations and the evaluation of the co-op, the effects of the attribution message were unchanged. Therefore, it was decided to report the actual judgments to simplify interpretation of the cell means.

The second page of the questionnaire was a two-paragraph description of other consumers' experience with a food co-op. The first paragraph was included in all the questionnaires and provided a general description of food cooperatives. For example, it discussed how the members volunteered their time to work in the store. The second paragraph included the four attribution manipulations. randomly sorted among the distributed questionnaires.

The third page of the questionnaire asked subjects to indicate their evaluation of the proposed co-op on the same dimensions as their current store (fresh produce, fresh meat, service, wide selection, low prices, convenience of location, brand name goods, and short checkout lines). Next, they were asked to indicate how confident they were in making these evaluations (seven-point scale with labels like "very unsure' to "very sure"). Then, they were asked if they had been involved in a rood co-op, whether they would participate, and what did or did not interest them about the co-op. Finally, they provided their year in school and sex.

Independent Factors

The independent factors in this experiment were the four attribution considerations: consistency over time, consistency over modality, consensus, and distinctiveness. Each was presented in a high and low level as follows, with underlined sections of the high version replaced by the text in parenthesis for the low version.

1. Consistency over time -- For those persons who have been members for five or more years, most reported that they were always (usually but not always) satisfied.

2. Consistency over modality -- Asked about the characteristics or a rood store, co-op members reported a liking or the selection. adequacy or parking, cleanliness, uncrowded and friendly atmosphere (a liking of the selection, quietness, and friendly atmosphere but expressed some dislike of the parking, cleanliness, and crowdedness).

3. Consensus -- About 70% (30%) of the individuals who reported shopping at a food co-op considered them to be desirable places to shop.

4. Distinctiveness -- Compared to other food stores in the area, the sample thought that the co-op was unique (about the same as the others) in terms of its selection, prices, convenience and friendliness.

It should be noted that the operationalizations in this study are different from many others. Some attribution studies vary the amount of information on a dimension rather than the amount of consistency (e.g., Major 1980, Sparkman and Locander 1980). For example, consistency over time might vary from seeing one occurrence of the effect (low consistency) to seeing many occurrences (high consistency). However, this tends to confound consistency with the amount of observations. It would seem more true to the formulation of attribution theory if the amount of consistency was varied and the number of observations held constant. For example, one might observe the person and the entity together on five occasions and note that on four of the five occasions (high consistency), the person's reaction was the same, or that the same reaction was observed on only two of the five occasions (low consistency).

RESULTS

An analysis of the evaluations of the subject's usual food store revealed no differences across the four treatments (p > .1), and therefore it was assumed that the random assignment procedure was successful.

The responses to the eight food store evaluations were factor analyzed. A three-factor solution was selected (eigenvalues greater than 1.0) and subject to a varimax rotation. The pattern of the factor loadings suggested that there were three dimensions to the judgments. One was selection, low prices, convenience, and brand names; the second was fresh meat and fresh produce; and the third was service and checkout times. Summed raw scores for these three measures and confidence in the judgments were tested using ANOVA. The results and cell means are reported in Tables 1 and 2.

The first prediction stated that the product evaluations would be higher when the information was consistently favorable across time and modality, reported by a consensus, and distinctive to the store than when this was not true. This was tested in several ways. First, the main effects of consistency over time and modality, consensus, and distinctiveness were determined. The only significant main effect was for consistency over modality for the judgment of the service and checkout times (F[1,74] = 10.7, p < .05). Second, the cell for which the information was at the high level for all four dimensions (Group 16) was compared to the other groups. Their judgments were more favorable only for the selection/convenience/brand names/low price measure (mean = 15.2 versus 12.5 for the other 15 cells, F[1,74] = 4.80, D < .05)

The second prediction stated that consistently favorable information would result in higher confidence compared to the inconsistent information and that the highest confidence would occur when all four attribution dimensions were at the high level. As Table 1 shows, the groups exposed to the information that was consistent over time reported significantly greater confidence in their judgments than those exposed to the low consistency over time messages (F[ 1,74] = 4.8, p < .05). The greatest confidence was reported by the group exposed to the information which was at a low level for all four dimensions (Group 1), though this group's confidence was not significantly greater than the all high level group's (Group 16). Unexpectedly, the group exposed to the information representing high consistency over time and modality, high distinctiveness. but low consensus (i.e., the group expected to make a person attribution), reported one or the lower levels of confidence. Intentions to shop at the co-op were also analyzed using ANOVA, and the only significant effect was for distinctiveness.

TABLE 1

ANALYSIS OF VARIANCE OF JUDGMENTS AND CONFIDENCE BY CONSISTENCY OVER TIME and MODALITY, CONSENSUS, AND DISTINCTIVENESS

TABLE 2

CELL SIZES AND MEANS FOR JUDGMENTS AND CONFIDENCE BY CONSISTENCY OVER TIME AND MODALITY, CONSENSUS, AND DISTINCTIVENESS

A sign test was used to test the direction of the effects of information presented in the high consistent and distinctive manner relative to the low consistent and distinctive manner, using the cell means for the 12 main effects (three judgment measures and four indpendent factors). fine or the 12 means were higher for the high level, which is marginally statistically significant (p < .07).

DISCUSSION

The results of the experiment provided only minimal support that consumers do evaluate their information about a new product in a manner consistent with attribution theory principles of consistency over time and modality, consensus. and distinctiveness. The information had a small effect on the consumers' judgments of the new product and their confidence in their ratings. However, most of the effects were in the predicted direction.

If one compares the three judgment measures with the messages, and particularly the distinctiveness and consistency over modality variations, some interesting patterns can be observed. First, the distinctiveness messages differed claiming uniqueness for the selection, prices, convenience, and friendliness of the co-op. Selection, convenience, and low prices were the attributes included in the first measure and the difference in judgments for those exposed to the low and high levels were in the expected direction, though not statistically significant. There were no differences for the other two measures, but subjects exposed to the high distinctiveness version expressed a significantly greater intention to shop at the co-op and a nearly significantly difference for their confidence in their judgments (p < .1). Thus, it seems that for those attributes featured in the distinctiveness messages, consumer use of the information was enhanced when the attributes seemed uniquely descriptive of the new product.

The information presented to vary consistency over modality also seemed to be effective when examined on an attribute-by-attribute basis. The high consistency version claimed that the surveyed members were satisfied with the selection, parking, cleanliness, quietness, and friendliness, but dissatisfied with parking, cleanliness, and crowdedness. The judgment measures (service and checkout times) that evaluate these factors were significantly lower for those exposed to the low consistency version, whereas there were no differences for the other measures, including selection. This seems to indicate that the lack of consistency over modality reflected by the negative evaluations on certain dimensions is accepted more readily by individuals than the positive information. This finding is consistent with other research showing the great impact of negative product information. See Weinberger, Allen and Dillon (1980) for a review of this research.

This experiment represents only a preliminary investigation of the question as to how consumers evaluate their knowledge about a new product and how this evaluation may affect their judgments and confidence in making judgments about the product. Certain aspects of the design have limited the power or the test. For example, 14 percent or the subject population reported prior experience with a rood co-op, and that might have stabilized their Judgments making it difficult to demonstrate changes because of the attribution manipulations. Also, the operationalizations may not have provided sufficient differences between the high and low versions to demonstrate the importance of each dimension. The information was presented in a brier, condensed version and represented the experiences of other patrons or the co-op. A personal experience format, in which the subjects would actually interact with the new product in a manner controlled to ensure high or low consistency over time and modality, high and low consensus, and high and low distinctiveness would provide more vivid information and should enhance the differences.

On a positive note, most of the differences were in the expected direction, and as predicted, the effects seemed greater for confidence than for the judgments themselves. This finding suggests that information presented in accord with these attributions principles might be particularly useful as a means to stabilize judgments once they have been formed. Other presentation schemes might be used to establish first impressions and attribution-styled messages would then be used-to develop loyalty on the part of triers of the new product. Another interesting area for future research would be to test the evaluation of unfavorable information about a product in terms of the attribution principles.

REFERENCES

Hansen, Ranald D. (1980), "Commonsense Attribution," Journal of Personality and Social Psychology, 39, 996-1009.

Hansen, Ranald D. and Stonner, David M. (1978), "Attributes and Attributions: Inferring Stimulus Properties, Actors' Dispositions, and Causes," Journal of Personality and Social Psychology, 6, 657-667.

Kelley, Harold H. (1967), "Attribution Theory in Social Psychology,' in D. Levine ( d . ), Nebraska Symposium on Motivation, Lincoln: University of Nebraska Press.

Major, Brenda (1980), "information Acquisition and Attribution Processes," Journal of Personality and Social Psychology, 39, 1010-1023.

Mizerski, Richard W. (1978), "Causal Complexity: A Measure of Consumer Causal Attribution," Journal of Marketing Research, 15 (May), 220-228.

Mizerski, Richard W., Golden, Linda L., and Kernan, Jerome (1979), "The Attribution Process in Consumer Decision Making," Journal of Consumer Research, 6, 123-140.

Robertson, Thomas S., and Rossiter, John R. (1974), 'Children and Commercial Persuasion: An Attribution Theory Analysis," Journal of Consumer Research, 1, 13-20.

Scott, Carol A. (1977), "Modifying Socially-Conscious Behavior: The Foot-in-the-Door Technique," Journal of Consumer Research, 1, 20-30.

Settle, Robert B., and Golden, Linda L. (1974), "Attribution Theory and Advertiser Credibility," Journal of Marketing Research, 11, 181-185.

Smith, Robert E., and Hunt, Shelby D. (1978), "Attributional Processes and Effects in Promotional Situations," Journal or Consumer Research, 5, 149-158.

Sparkman, Richard M., and Locander, William B. (1980), Attribution Theory and Advertising Effectiveness," Journal of Consumer Research, 7 219-924.

Swinyard, William R. (1981), "The Interaction between Comparative Advertising and Copy Claim Variation," Journal of Marketing Research, 18, 175-186.

Tybout, Alice M. (1978), "Relative Effectiveness or Three Behavioral Influence Strategies as Supplements to Persuasion in a Marketing Context," Journal of Marketing Research, 15, 229-242.

Weinberger, Marc, Allen, Chris T., and Dillon, William R. (1981), "Negative Information: Perspectives and Research Directions," in Rent Monroe (ed.), Advances in Consumer Research, Vol. 8, Ann Arbor: Association for Consumer Research.

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