Beliefs in Quality Differences and Brand Choice

Carl Obermiller, University of Washington
John J. Wheatley, University of Washington
ABSTRACT - In a laboratory setting and in a field study subjects who believed or were let to believe that there were large quality differences among brands in several product categories tended to use price as a produce quality cue. Conversely, chose who did not believe or were led not to believe in quality differences tended not to use price information in the same manner.
[ to cite ]:
Carl Obermiller and John J. Wheatley (1985) ,"Beliefs in Quality Differences and Brand Choice", in NA - Advances in Consumer Research Volume 12, eds. Elizabeth C. Hirschman and Moris B. Holbrook, Provo, UT : Association for Consumer Research, Pages: 75-78.

Advances in Consumer Research Volume 12, 1985      Pages 75-78


Carl Obermiller, University of Washington

John J. Wheatley, University of Washington


In a laboratory setting and in a field study subjects who believed or were let to believe that there were large quality differences among brands in several product categories tended to use price as a produce quality cue. Conversely, chose who did not believe or were led not to believe in quality differences tended not to use price information in the same manner.


Evidence of the existence of a link between beliefs about quality variation among brands in a product class and preference for the higher priced alternative has been suggested by several researchers (Leavitt 1954, Tull, Boring and Gonsior 1964, Lambert 1972). While suggestive, these earlier studies were not quite convincing because no attempt was mate to measure preference when "real" money, or something approximating it was involved.

In a recent study, Obermiller and Wheatley (1983) attempted to answer the question, "Why do consumers buy the higher prices brand in produce choices among brands that may be similar?" They proposed that evidence of similarity, either supplied information or use experience, should increase preference for the lower priced of two similar (identical) brands, but the effects of such information would be mediated by the strength of prior beliefs in brand quality differences. They conducted an experiment in which subjects received evidence of the similarity of two brands from two sources--a Consumer Reports mock-up and taste experience. Prior beliefs were manipulated by replicating across two products, cola represented a product about which consumers had strong prior beliefs in brand quality difference and popcorn, weak prior beliefs. The results provided support for their hypotheses: cola showed greater initial preference for the higher priced brand and greater resistance to evidence of similarity of two brands.

The two studies reported here extend the Obermiller and Wheatley investigation. Two questions are raised: (1) While the Obermiller and Wheatley study did attempt to introduce external validity with respect to the issue of a monetary transaction being involved it was flawed by a possible confounding of prior belief strength and product x subject characteristics. To eliminate this problem and isolate strength of prior belief in brand quality difference as the independent variable, a lab study (study 1) was conducted. (2) Subjects in the original Obermiller, Wheatley study were undergraduate students. Study 2 determined whether the effects would generalize to other samples, patrons of a supermarket, who were expected to be less educated, less perspicacious in the ways of experimental research. The second study also replicated across a new product category.


In the previous study Obermiller and Wheatley reported greater initial preference for the higher priced brand and more resistance to information of similarity of brand qualities for cola than for popcorn. They then argued that the effects were the result of stronger prior beliefs in brand quality differences for cola than for popcorn. In fact, subjects were pre-screened to insure differences in prior belief strengths. There are two problems with this procedure. The first is a problem of construct validity that is unavoidable whenever a specific object, task, or instance is chosen to represent a position on a theoretical dimension. Cola and popcorn likely do represent products characterized by high and low strength of consumer beliefs in brand quality differences; but they may also represent different positions on other dimensions that could provide alternative explanations. One possibility is that cola is characterized by high social risk, since it is frequently consumed with the brand visible and is heavily promoted in terms of life style, whereas popcorn is neither. The second problem is selection bias. By screening subjects to insure strong prior beliefs for cola and weak prior beliefs for popcorn, they may have confounded the treatment with some other individual difference characteristics of the subjects. An expression of strong prior beliefs in brand quality differences may have been less associated with the two produces as with such characteristics as dogmatism or cognitive complexity. Dogmatists may insist on their ability to detect differences and be less willing to accept contrary information; cognitive complexity might reduce willingness to express any strong beliefs and result in openness to new information.

In order to avoid both confounds of stimulus characteristics and selection biases, an experiment was designed in which subjects were randomly assigned to belief strength conditions. Belief strength was manipulated by giving subjects information about brands in a given category that suggested either that brands were similar or dissimilar in quality. Subsequently, subjects chose between two new brands at different prices. The research hypothesis was that subjects who were given data consistent with a belief in large brand quality differences would make a price-quality inference and prefer the higher priced alternative and that subjects who were given data consistent with a belief in low brand quality would not infer quality from price and would, therefore. prefer the lower priced alternative.


Fifty-five undergraduate business students acted as subjects in a classroom situation. Subjects were given booklets with decision problems, each followed by a choice situation and a rating scale. The booklet was titled Management Decision Analysis and described as an exercise in management decision-making with incomplete information. Directions indicated that students should base their decisions only on the information given, however limited, not on their personal experiences. Each problem described someone faced with a purchase decision and presented some information about "other" brands; the decision was a choice between two new brands. No information was given about the two new brands except that one was priced 20% (15% in one case) higher than the other.

The information provided about other brands was in ended to manipulate prior belief in brand quality differences. Each of the four problems provided a different basis for the formation of these beliefs. In the first problem, the produce was an industrial precision cutting tool, and the information was 3 years of data on the % successful cuts and life expectancies of three "other" brands. For the low difference belief condition, the data were similar (mean success = 94.3%. a success = 1.8: mean life = 210,000 units, G life = 4200). For the high difference belief condition, the data were dissimilar (mean success = 95%, : success = 8.7; mean life = 209,000 units, : life = 19,500). For the second problem, the product was institutional bedding supplies and the information was an expert opinion, a memo from a former purchasing agent that indicated that brands available in the past had either no difference or substantial differences in quality. The third problem was a choice between electric blow dryers, and the information described the brands as either differing in familiarity due to remembered advertising or uniformly unfamiliar. The final problem was a choice between boat trailers. The information was from an expert source a quote from Consumer Reports that stated there is either no difference or a great deal of difference in quality from one brand of boat trailer to another.




The dependent measures included a manipulation check and a choice. The results of both are displayed in Table 1. The manipulation check was a scale rating the difference among brands, with higher numbers indicating greater difference. The mean responses to the quality difference questions indicate a successful manipulation overall (2.0 for low difference vs 3.1 for high) and stronger effects for three of the four conditions. The use of different recall of advertising in the blow dryer problem did not result in different beliefs in brand quality differences. The failure of this manipulation suggests dropping these responses from the analysis. T-tests showed the remaining three groups to be statistically significantly different in the quality difference score (laser cutting tool, t53 = 5.15; institutional bedding, t53 = 5.72; boat trailer, t53 = 5.75; for all, p < .001).

Overall preference reflected the manipulation of prior bases for quality difference beliefs. Overall, when prior information suggested little difference among brands, subjects preferred the higher priced alternative only 10 times vs 97 times for the lower priced brand. In contrast, when prior information suggested larger brand quality differences, subjects preferred the higher to the lower priced brand by 60 to 46. (If one ignores the blow dryer data, these differences are even more pronounced, 8 vs 73 for low difference and 52 vs 25 for high difference.) These results were analyzed with a repeated measures Anova. Each subject responded to four preference questions, two high difference and two low difference situations. The main effect of prior bases for quality difference beliefs was statistically significant (F54,1 = 53.5; P < .001).


Beliefs in brand quality differences may be formed in many ways. In this experiment, manipulations of performance data, expert opinion, and the conclusion of an expert source were successful in creating different strengths of prior beliefs in brand quality difference. The manipulation of extent of recalled advertising was unsuccessful. Level of advertising may, nonetheless be an important basis for quality difference beliefs; it is likely that a statement that brands were remembered as differentially advertised or not is a poor substitute for actual differences in advertising exposures. A stronger and more correspondent manipulation of advertising level differences would be required to verify the relationship.

Another explanation for the failure of the advertising manipulation rests upon subtler response to ads. While it is possible that large advertising expenditures are a way of signaling consumers that the advertised product is of high quality (Nelson 1974), such efforts may not lead to meaningful changes in consumers' beliefs. In the case of low cost consumer products in particular, it is quite likely that many consumers process ads peripherally (Petty, Cacioppo, and Schuman 1983). Thus, attitudes and preferences may be influenced by heavy advertising, but the lack of involvement with the ads diminishes the probability the beliefs about the product category will be affected.

Preference between brands of different prices was shown to be largely determined by prior belief in brand quality differences. When people were led to believe that product alternatives were similar, they exhibited a strong preference for the lower priced brand, and when they were led to believe that product alternatives were dissimilar, they exhibited a strong preference for the higher priced brand. These results suggest that inferences from one attribute to another do not occur automatically as the result of a generalized heuristic. Price-quality inferences occur as a function of product category beliefs. People will only infer one from the other when they believe the two to be correlated. If quality does not vary, and price does, the two cannot be correlated. If both quality and price vary, the two may or may not be correlated, but when choices are made under uncertainty, variation in both attributes may be sufficient evidence for one to infer correlation and rely on a generalized price-quality heuristic.


The strength of study 1 is the isolation of the construct of interest. Its weakness is also the isolation of that construct; we are not much surprised that with minimal variation in any mundane factors, changes in the bases for quality variation beliefs about a product category are reflected in preferences between two alternatives. A stronger demonstration of the generalizability of this effect would be differences in preferences for real products in real choice situations that could be explained by measures of prior belief in brand quality differences for that product. Study 2 was designed as such a demonstration.


Eighty four customers were intercepted in two locations of a large grocery chain in Seattle. After being identified as a user of the product in question (Margarine), each subject was given a response booklet that included measures of belief in differences among brands, brand usually purchased, and usage frequency. After completing the preliminary measures, subjects were shown two samples of the product, labeled Xb and Wt. Brand Xb was given a price that reflected the high end of the price range for that product in the store; brand Wt had a price from the low end. Subjects were asked to rate the two brands on quality and select between prizes of one pound of brand Xb or one pound of brand Wt plus the difference between the Xb price and the Wt price in cash. These measures were taken before any similarity evidence was presented.

Informational similarity was presented on the next page of the booklet for every other subject. This information stated that scientific experiments conducted by the National Food Survey indicated that it was impossible for consumers to distinguish among brands of margarine and that Consumer Reports found all margarines to be very similar. The remaining half of the subjects also received information, but it described margarine nutrition, and packaging and was irrelevant to brand differences. Thus, one group received informational evidence of brand similarity; the other did not.

At this point all subjects were given an opportunity to taste the two brands, which were, in fact, identical. Thus, all subjects were given experience evidence of similarity. Following the taste, each subject was given a second choice between Xb and Wt plus the cash. Then subjects filled out measures of their abilities to detect differences, self-rating of expertise, influence of price in their choice, comprehension checks for the information they had been given, agreement with that information, beliefs in price as an indicator of quality, and demographics.


The results shown in Table 2 reveal that 59% of those subjects who felt that there were only small differences among brands of margarine preferred the less expensive brand while only 33% of the subjects who said that there were large differences among the various brands selected the cheaper brand and the 30c in cash which represented the difference in price between the two. A x2 test of these data yields a value of 3.19 p = .08.

Those consumers who said that they felt that there were large differences among brands of margarine also said that brand Xb was of higher quality than brand Wt. This evaluation was based on a visual inspection of identical products in identical dishes. The difference in means, Xb = 2.85 vs Wt = 3.72 (lower score means higher quality here); was statistically significant, p < .05. As might be expected, the subjects who said that they felt that the difference among brands of margarine was small rated the two brands almost identically, Xb = 3.00 vs Wt= 3.05.



On the other hand, when these subjects were subsequently questioned directly about whether they agreed with the statement that they "personally believed that the retail price of margarine was a good indicator of its quality" or, more generally, whether "prices are a good indicator of the quality of most products," there was no statistically significant difference between the two groups. Once again, it is apparent that consumer preference or choice is not the result of a simple rule-of-thumb; it is beliefs about differences among brands in a product class that seem to result in inferences about quality based on price information.

Table 3 shows the effects of sampling the product and reading a message that said most consumers could not tell the difference between brands of margarine because most of them were very similar in terms of sensory cues. Only 44% of the 43 subjects who were not exposed to the message and not given an opportunity to caste the two samples preferred the lower priced brand. However, 66% of the 41 subjects who had the opportunity to sample the two products and who were exposed to the message picked the cheaper alternative. This difference is statistically significant, (2 = 3.15 p < .02.




The results of study 2 and those of study I provide evidence of convergent validity. Prior beliefs about quality differences among brands apparently influence product choice. When presented with price information, a sample of in-store customers, all of whom were margarine users, tended to pick the lower priced alternative when they believed that there were small differences in the quality of margarine brands. Conversely, when consumers felt that there were large differences among brands they were more inclined to choose the higher priced brand. When given an opportunity to sample the product and information suggesting the similarity of the two brands another sample of consumers responded by showing a strong inclination to choose the lower priced alternative. From a social standpoint, it is, unfortunately, a very expensive proposition for a seller to furnish such a demonstration in the marketplace.


While avoiding the problem of confounding prior belief strength and product-subject characteristics, all or the subjects in both studies were still confronted with a forced choice situation and no information, other than price, was presented to them. Buyers are, of course, likely to use other informational cues when making quality judgments or purchase.

The effect of price as a quality cue may have been diminished, however, because of the type of subjects used in both studies. Shapiro (1975) has reported that well educated persons are less prone to rely on price information as a product quality cue than chose with lower educational attainments. Both the college student s in study 1 and the adult subjects in study 9 were significantly better educated than the general population.


Although it seems unlikely to hold true in every case, the hypothesis that prior beliefs about large quality differences among brands in a product class lead a number of consumers to select the higher priced of two alternatives seems tenable. It is also clear that many consumers, even those with strong prior beliefs, can be persuaded to choose the lower priced alternative when they are allowed to sample the product and are provided with information about it.

Perhaps the most intriguing finding in these two studies is the sizable proportion of subjects who still pick the higher priced alternative after strong encouragement to do the opposite. Approximately one third of the supermarket consumers selected the higher priced margarine despite sampling the identical products,. reading reports from two credible sources pointing out that the products were indistinguishable, a 304 incentive, and, one would suppose, a suspicion that they were being urged to choose the lower priced alternative. Similar results were reported by Obermiller and Wheatley (1983) and Wheatley, Yalch and Chiu (1980). Distortion of quality perceptions is one explanation. The use of price-quality heuristic under uncertainty is another

The two studies provide evidence for the importance of beliefs in quality differences as a product specific individual difference variable that mediates inferences from price to quality. A continued investigation of the characteristics of consumers and products that influence such inference processes is necessary if we are to make sense of the confusion of findings in the price-quality literature and if we are to view obstinate preference for the higher-priced brand as psycho-rational behavior.

Finally, the studies provide weak evidence that an inference from one attribute to another does not require specific evidence of a correlation between them. Given a generalized heuristic (such as from price to quality), subjects in study 1 made the inference when evidence suggested variation in quality, without any specific evidence of a price-quality correlation. Thus, the requirements for an influence in a specific situation may be relatively weak; evidence need not be sufficient for an inference of a correlation between two attributes, provided it is necessary.


Lambert, Z. V. (1972), "Price and Choice Behavior," Journal of Marketing Research, IX, 35-40.

Leavitt, H. J. (1954), "A Note on Some Experimental Findings About the Meaning of Price," Journal of Business, 27, 201-10.

Nelson, P. (1974), "Advertising as Information," Journal of Political Economy, 78 (July-August), 729-54.

Obermiller, C. and J. J. Wheatley (1983), "Price Effects on Choice and Perceptions Under Varying Conditions of Experience, Information and Beliefs in Quality Differences," Advances in Consumer Research, XI, 453-58.

Petty, R. E., J. T. Cacioppo, and D. Schumann (1983), "Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement," Journal of Consumer Research X, 135-46.

Tull, D. S., R. A. Boring, and M. H. Gonsior (1964), "A Note on the Relationship of Price and Imputed Quality," Journal of Business, 37, 186-91.

Wheatley, J. J., R. F. Yalch, and J. S. Chiu (1980), "In Search of the Economists Consumer: The Effects of Product Information, Money and Prices on Choice Behavior," Advances in Consumer Research, VIII, 583-87.