Advertised Comparative Price Effects on Buyer Perceptions and Behavior: a Model and Empirical Test

Joel E. Urbany, University of South Carolina
William O. Bearden, University of South Carolina
Dan C. Weilbaker, Bowling Green State University
ABSTRACT - The use of comparative reference prices to enhance consumer perceptions of advertised price reductions is a frequent practice among retailers. In the present study, the effects of reference prices on perceived offer value, benefits of search, and search behavior are investigated. The results of a structural equation analysis of data obtained from a 1 X 4 between subjects experiment, in which the presence of a reference price at three levels is compared with a control (no reference price) condition, are presented. Tests of an hypothesized model indicate that reference prices can impact perceived offer value and the benefits of search and that offer value and search benefit are also affected indirectly through the effects of advertised discounts (from previously charged prices) on estimates of average market prices and normal retailer prices.
[ to cite ]:
Joel E. Urbany, William O. Bearden, and Dan C. Weilbaker (1988) ,"Advertised Comparative Price Effects on Buyer Perceptions and Behavior: a Model and Empirical Test", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 332-340.

Advances in Consumer Research Volume 15, 1988      Pages 332-340

ADVERTISED COMPARATIVE PRICE EFFECTS ON BUYER PERCEPTIONS AND BEHAVIOR: A MODEL AND EMPIRICAL TEST

Joel E. Urbany, University of South Carolina

William O. Bearden, University of South Carolina

Dan C. Weilbaker, Bowling Green State University

ABSTRACT -

The use of comparative reference prices to enhance consumer perceptions of advertised price reductions is a frequent practice among retailers. In the present study, the effects of reference prices on perceived offer value, benefits of search, and search behavior are investigated. The results of a structural equation analysis of data obtained from a 1 X 4 between subjects experiment, in which the presence of a reference price at three levels is compared with a control (no reference price) condition, are presented. Tests of an hypothesized model indicate that reference prices can impact perceived offer value and the benefits of search and that offer value and search benefit are also affected indirectly through the effects of advertised discounts (from previously charged prices) on estimates of average market prices and normal retailer prices.

INTRODUCTION

A number of empirical studies have examined from various perspectives the managerial and public policy issues involved with the use of advertised reference prices and their effects on a variety of consumer responses (e.g., Ahmed and Gulas 1982; Barnes 1975; Berkowitz and Walton 1980; Blair and Landon 1981; Della Bitta, Monroe, and McGinnis 1981; Fry and McDougall 1974; Keiser and Krum 1976; Liefeld and Heslop 1985; Sewall and Goldstein 1979; Winer 1986). The objective of the present effort is to extend this research through the development and test of a model which considers the influence of advertised reference prices on estimates of average market prices and the advertiser's regular prices, the perceived benefits of search, and a measure of search behavior in addition to the typically studied perceptions of perceived offer value. In the study, the hypothesized relationships are examined for data obtained from a controlled laboratory experiment in which advertised reference prices were manipulated at three levels: (1) an average or expected reference price; (2) an above average reference price; and (3) an extreme reference price well above expected market prices.

THE ROLE OF ADVERTISEr REFERENCE: PRICES AND PRIOR RESEARCH

Role of Advertised Reference Prices

Advertised reference or comparative prices represent important contextual cues that affect consumer processing of price stimuli. The influence of these comparison cues can be explained to a large extent by adaptation-level theory and Thaler's (1985) transaction theory. Regarding the former, individuals are assumed to judge a given stimulus (price) relative to an adaptation level (or internal standard) which the individual expects or has become accustomed to (i.e., has "adapted to") (Monroe and Petroshius 1981, p.49). Adaptation-level theory suggests that price perceptions depend on the actual price and the individual's internal reference price or adaptation-level (Petroshius and Monroe 1987, p. 511). Provision of a reference price, such as a previously charged price, a manufacturer suggested price, or a competitor's price, represents a means of establishing or influencing the consumer's basis for evaluating an -advertised offer.

Thaler's model (1985) proposes that advertised reference prices serve to increase the perceived value of a purchase through their effects on transaction utility. Transaction utility depends upon the merits of the deal and is positive if the actual price is less than the consumer's internal reference price. Consistent with Thaler's model, advertising the usual or regular price along with a lower asking price is an attempt to provide buyers with a price frame of reference, in the spirit of transaction utility, to augment buyers' perceptions of value (Monroe and Chapman 1986). As such, an advertised reference price may make the lower advertised sale price appear more attractive and, thus, increase transaction utility (Thaler 1985, p. 212).

Prior Research

The effects of reference prices alone and in conjunction with other contextual cues have been investigated in a number of empirical studies. Much of this research has been reviewed by Monroe and Petroshius (1981) and Della Bitta et al. (1981). Liefeld and Heslop (1985, p. 868) provide the following summary:

The independent variables employed in these studies include a variety of actual prices (e.g., sale price, offer price, regular price, manufacturer suggested list price), a variety of semantic cues (e.g., percent off, dollar off, compare at), and variations in price levels, the size of implied discounts, the medium in which the advertisement was presented, and the identification of the advertiser.

The dependent variables have been of two general types: (1) measures of consumer reactions to price statements such as believability or credibility, and (2) measures of the effects of reference price statements on perceptions of implied savings or value for the money, perceived price reduction, or motivation to buy.

The findings from this research reveal that reference prices, in spite of some discounting by consumers of actual savings, can indeed raise consumer perceptions of perceived value.

HYPOTHESIZED MODEL OF REFERENCE PRICE EFFECTS

The model of reference price effects investigated here is presented in Figure 1A. In the model, the presence of a reference price is hypothesized to positively influence consumer estimates of the advertiser's normal price, expected average market price, and perceived offer value. Indirect effects of reference prices on perceived offer value are also posited through the effects of reference prices on estimated average regular prices and expected average market prices. Increased perceived offer value is hypothesized to lower the benefits of search which, in turn, are predicted to be inversely related to direct patronage (i.e., no shopping of competing stores) of the advertiser.

Prior research supports the direct and indirect effects of reference prices on perceived offer value. For example, Blair and Landon (1981) found that an advertised reference price had the effect of raising consumer estimates of the advertiser's normal selling price over an ad which did not contain a reference price. The research of Thaler (1985) and Monroe and Chapman (1986) suggests that a consumer's internal reference price (reflected by expected average market price) can be influenced by an advertised reference price. The direct effects of reference prices on perceived offer value are also consistent with Thaler's (1985) premise that an advertised reference price can make a lower advertised sale price appear more attractive and, hence, increase transaction utility. Higher internal reference prices (i.e., expected average market prices) and higher expected retailer prices caused by the presence of a reference price in conjunction with a sale price should also increase perceived offer value. The first of these two predictions is consistent with Monroe's (1975; 1977) interpretation of Helson's (1964) adaptation-level theory. The second prediction (i.e., that higher estimates of the advertiser's normal price relative to a sale price can increase perceived offer value) is based upon earlier empirical research regarding the effects of larger discount levels (e.g., Della Bitta et al. 1981).

Although the effect of reference pricing practices on consumer search behavior has been identified as a critical research issue (Della Bitta et al. 1981), no studies exist which provide direct evidence regarding that issue. To the extent that an advertised reference price increases consumers' internal reference prices and the perceived value of the advertiser's offer, the perceived benefits of search will be reduced. As a result, there should be a greater likelihood that the consumer will purchase the product at the advertiser's store.

Extreme Reference Prices

The justification for the model shown in Figure 1A is based upon theory and research in which plausible reference prices (and discount levels) have been considered. In the ensuing experiment, three reference price conditions are paired with a constant sale price "only" condition. These reference price conditions are: (1) a plausible average reference; (2) a plausible above average reference price; and (3) a high reference price well above expected market prices. This latter condition represents a novel treatment and the corresponding predicted effects warrant some brief explanation.

Assimilation-contrast theory predicts that a reference price judged to be implausible will be rejected (Monroe and Petroshius 1981, p.50). This effect may result in the reference price being ignored and/or consumers lowering perceptions of the offer value and the retailer's credibility (cf., Barnes 1975; Fry and McDougall 1974). An alternative prediction is that, while consumers may "discount" the advertised price reduction, the reference price may still positively affect perceptions (and subsequent behavior) compared to a condition in which only a sale price is presented.

METHOD

One hundred and fifteen undergraduate business students, in groups of 8 to 16, participated in a lab experiment designed to test the hypothesized model of reference price effects. The study involved the purchase of a particular television brand (19" RCA Colortrak) from a market consisting of nine retail stores. Subjects within each session were randomly assigned to one of four treatment conditions: (1) sale price ($319) only (i.e., a no reference price control condition); (2) sale price ($319) plus a $359 reference price; (3) sale price ($319) plus a $419 reference price; and (4) sale price (5319) plus a $799 reference price. These prices were selected based upon pretests to represent: (1) a sale price that would be considered reasonably low but not so low as to discourage search; (2) plausible average ($359) and above average ($419) reference prices; and (3) a high reference price ($799) well above expected market prices. The model was examined for three "samples:" (1) the no reference price control group combined with the $359 reference price group (n=57); (2) the control group combined with the $419 reference price group (n=57); and (3) the control-group combined with the $799 reference price group (n=58).

Procedure

Data were collected via subject interaction with personal computers on which they were provided with instructions and performed a practice task prior to the television shopping and purchase task. Subjects were told that prices used in the study were taken from a real (but disguised) market. Subjects were told that the local Better Business Bureau (BBB) published an annual retail survey for assisting consumers in shopping and that the research in which they were participating was designed evaluate the effectiveness of the retail price survey. A practice task, prior to the television purchase, involved shopping and purchasing a vacuum cleaner. After initial interaction with the computer, in which the purpose of the study was explained, subjects were exposed to the electronic edition of the annual price survey for home entertainment products. These procedures enabled the establishment of marketplace retailers and provided information regarding local price structure and the cost search. General information regarding several products (but not TV"s) described fairly large price differences within the market. After exposure to the retail price survey, subjects responded to several Likert items on the computer regarding their perceptions of the price survey and their understanding of the shopping system.

Next, the subjects were exposed to the "one" advertisement during the previous week for the needed 19-inch television. Subjects were told that only one store (of the nine comprising the relevant market) had advertised the particular TV during the previous weekend and that the ad was provided as a matter of realism (i.e many consumers check retail newspaper ads prior to shopping). Upon exposure to the ad, perceptions of the offer and beliefs regarding market TV prices were assessed. Upon completion of these items, subjects shopped the retailers as needed until a purchase was made, responded to a final post-task series of question< which evaluated their perceptions of the marketplace at the experimental setting, and were then dismissed and debriefed in a later session.

The Shopping System. Participants were allowed to telephone for information or travel directly to a store without shopping to make their purchase. Five of the nine stores in the computerized marketplace (including the advertising store) carried the desired brand. Two stores had prices that were lower ($299 and $279) than the advertised $319 sale price while two stores had prices that were higher ($349 and $449). Subjects were given a beginning bank balance of $500, which was used to provide motivation for subject performance. The purchase task was timed by the computer and $3.00 was deducted for each minute spent shopping. In addition, $10.00 was charged for traveling to a store. The shopping system included time delays that simulated waiting for information and reaching the correct store department. Once the subjects finished a telephone call and obtained (or failed to obtain) the price information, they were given the option of calling another store or traveling to a store to make a purchase. The interactive computer system calculated an ending balance for each subject which included costs incurred for time and travel (i.e., search costs) and for buying the television set. Economic incentives were included in the study to encourage high ending balances (and hence to incorporate the costs and benefits of search).

Operational Measures

The exogenous variable (i.e., the presence of a reference price) was operationalized as a zero-one dummy variable (cf., Bagozzi 1977) in which the control condition is paired separately with each of the three reference price treatment conditions. Consequently, the subsequent tests of the hypothesized causal structure involved three nonindependent (i.e., the control group is included in each analysis) sets of analyses. Estimates of expected average market price and the advertiser's normal price (both obtained after the retail ad was reviewed) were included as endogenous variables. Perceived offer value was operationalized as the sum of the responses to three disagree-agree statements (cf., Berkowitz and Walton 1980; Della Bitta et al. 1981). The coefficient alpha estimate of internal consistency reliability for this three item measure was .79. Perceived benefit from search was defined as the difference between the sale price ($319) and each subject's expected lowest price (prior to the shopping task). The last endogenous variable, purchase without search, was operationalized as a zero-one categorical measure where one represented those individuals purchasing at the advertised store without shopping.

TABLE 1

OVERALL MODEL FIT STATISTICS

RESULTS

Data Checks

Forty-four percent of the study participants reported prior shopping experience for a television. This experience was not related to the treatment conditions. On average, more than four stores were shopped, a number which is higher than store search estimates obtained in field studies. Responses to the post experimental task measures indicated that the subjects were motivated in the task and felt it was realistic. For example, 85 percent of the subjects felt that the price they paid was the lowest available and virtually 100 percent of the subjects agreed that they had tried their best in the payoff task (mean = 6.75 on a 7 point scale). Further, debriefing of subjects and post-experimental inquiry revealed that demand artifacts due to large price differences or the experimental methods were not problematic.

Structural Equation Analysis

LISREL (Joreskog and Sorbom 1984) was used to analyze the data and the standardized parameter estimates and overall fit statistics are based upon analysis of the correlation matrices. Since all constructs were measured with a single indicator, our interests focused upon the structural equation relationships.

Original Model. The results of analyzing the original hypothesized model are summarized in Tables 1 and 2. (Ignore for the moment the estimates associated with the revised model.) The overall fit statistic (Table 1) and the estimated path coefficients (Table 2) depict mixed support for the hypothesized model. That is, the overall estimates suggest that the model does not fit the data adequately while the direction and significance of the path coefficients are largely consistent with the predicted relationships. (Efforts to improve model fit are described in our subsequent discussion of the revised model.) Regarding the overall estimates of model fit, the rather large significant chi-square values and low adjusted goodness-of-fit statistics suggest that the model is an inadequate representation of the data. The latter estimates were .62, .57, and .68 for the $159, $419, and $799 analyses, respectively. The corresponding total coefficient of determination estimates were .17, .21, and .46. These increasing values are undoubtedly due to the differential impact of the varied reference price treatment levels.

The standardized path coefficients for the original model across the three separate analyses are presented in Table 2. With only several exceptions, the estimated relationships were as predicted [These exceptions include the insignificant paths between the reference price manipulation and perceived average market prices (gamma 1,1) and between average market prices and perceived offer value (beta 3,1) for the average reference price analysis. Also, in the same $359 condition, the path between the reference price manipulation and the estimate of the advertiser's normal price was unexpectedly negative. This was due to the fact that, prior to seeing the ad, subjects generally expected that the average market price would be above $359. As such, the presence of the 5359 reference price actually lowered subjects' estimate of the advertiser's regular price.]. The paths in the two larger discount levels ($419 and $799) were largely as anticipated; however, the negative relationship found between the measure of search benefit and the (0,1) measure of search was weaker than anticipated.

TABLE 2

PATH COEFFICIENTS FOR ORIGINAL AND REVISED MODEL

The major findings can be summarized as follows. First, the presence of a reference price can increase estimates of average market price and the advertiser's normal price in addition to increasing the more frequently studied offer value perceptions. Second, the strength of these effects (as evidenced by the size of the path coefficients) increase with the size of the discount. Third, these effects hold even when the advertised discount is well-above normal expectations.

As predicted, a consistent negative path was found between perceived offer value and the perceived benefits of search. This supports earlier findings that increasing offer value through the use of advertised discounts from higher previous prices can lower the benefits to search. Also, a negative but weak relationship was found between the benefits of search and search behavior. This latter finding, while restricted by the dichotomous coding scheme, supports the premise that decreased benefits from search increase the likelihood of direct patronage of the advertised retailer.

Revised Model. Analysis of omitted paths within the original model through tests of a series of nested models revealed that the model could be improved by the addition of several structural paths (cf., Ryan 1982; MacKenzie, Lutz, and Belch 19, 6). Further, these paths provide additional insight regarding the effects of reference prices and the mediating roles of expected advertiser normal price and market price perceptions. The overall fit statistics and the standardized path coefficients for a revised model are shown in Tables 1 and 2 along with the original model results.

The first comparison involved adding individually five paths: (1) from expected average market price to search benefit; (2) from expected normal price to search benefit; (3) from expected normal price to expected average market price; (4) from expected average market price to purchase behavior; and (5) from expected normal price to purchase behavior. Comparisons of these less restricted models nested within the original model enabled chi-square difference statistics with one degree of freedom. Some comment is warranted regarding the direction of the path between expected normal price for the retailer and expected average market price. The test of the path from expected normal price to expected average market price (beta 1,2) in lieu of the reciprocal path (beta 2,1) was based upon an intuitive argument. That is, if subject estimates of the average market price are raised by the presence of the retailer's regular price in an ad, that effect should occur in part because the advertised reference price raises subject estimates of the regular price (than if no reference price were provided). In other words, information about a retailer's normal selling price should be used by subjects as one piece of evidence in judging market prices.

FIGURE 1

HYPOTHESIZED ORIGINAL AND REVISED MODEL OF REFERENCE PRICE EFFECTS

Two of the additional paths in these less restricted models were consistently significant. These results suggest the addition of direct effects of, first, expected normal price on expected average market price and, second, expected average market price on search benefits. The former path was particularly robust across all three treatment conditions. The additions of these paths added substantially to model fit. Models in which multiple paths were allowed were also examined. The even less restricted models involved inclusion of the two significant paths identified in the previous nested model results plus the path between expected normal price and the benefits of search. Given the strong relationship between expected normal price and expected average market price identified in the individual path tests described above, other paths involving expected normal price perceptions were deemed appropriate. The chi-square significance tests involving the addition of these multiple paths were statistically significant and suggested then the revised model (with now five degrees of freedom) shown in Figure 1B.

As shown in Table 1, the overall chi-square goodness-of-fit statistics are now not significant. The adjusted goodness-of-fit statistics have improved from .62, .57, and .68 to .85, .91, and .91 for the $359, $419, and $799 samples, respectively. A number of effects were observed consistently across the three reference price conditions. First, consistent with earlier research regarding the effects of reference prices, the presence of a reference price directly impacts perceived offer value. And, for reference prices promoting sizable discounts, the presence of a reference price significantly increases the perceived normal price of the advertiser. Further, increased perceived offer value is inversely related to search benefit. The effects of the reference price manipulations studied here were found to operate significantly through changes in perceived regular prices of the advertiser. That is, perceptions of the retailer's normal price were found to be directly related to search benefits and perceptions of average market prices which, in turn, were also found to be related to the benefits of search.

Differences Across Reference Price Levels. Examination of the estimated relationships across the treatment levels revealed a number of differences. For example, the effects of the reference price manipulation on average market price perceptions for the above average and extreme reference prices was found to operate largely through its influence on perceptions of the advertiser's normal price. In support of this finding, the path between average market price estimates and perceived offer value was found significant only for the two large advertised discounts. In contrast, the effect of the expected advertiser normal price on perceived offer value was not significant in the extreme condition. This latter finding may suggest some degree of discounting of the extreme reference price claim.

DISCUSSION

Multiple tests of the model presented here, in which a sale price only condition was paired with three reference price treatments (cf., Bagozzi 1977) largely supported the predicted relationships. Further, a series of nested model analyses suggested that reference prices also impact the perceived benefits of search through their effects on estimated average market prices and normal retailer prices.

A number of caveats are in order and, certainly, the generalizability of our findings should not be overstated. The study was conducted in a laboratory setting in which student subjects were required to shop for a particular brand of a single product. Additional study in more realistic research environments employing other subject groups and products is warranted. However, in spite of these limitations, a number of interesting findings regarding the general effects of reference prices and, specifically, the effects of an extreme or exaggerated price reduction were found. To summarize these results:

1. The presence of a reference price enhances perceived offer value and the size of this effect seems to increase as the advertised discount increases (even to an exaggerated level). The fact that the $799 reference price claim was perceived to be exaggerated was supported by declining scores on measures of offer "believability" which were collected with the measures analyzed here.

2. The perceived benefits of search are enhanced, for a given sale price, by increased estimates of normal retailer prices. However, the benefits of search are reduced when perceived offer value is higher and when expected average market price is lower

These findings suggest that even exaggerated reference prices can influence consumer beliefs about advertised products and market prices. This conclusion is further supported by additional analyses which indicate that the extreme reference price was apparently discounted rather than totally dismissed. The weak predictive ability of the hypothesized antecedents for search behavior, however, was disappointing. In spite of consistent negative estimates as anticipated, the estimated relationships (in both the original and revised model tests) were small and not significant. This weak predictive finding is undoubtedly due, in part, to the dichotomous nature of the variable and the uneven distribution of subjects across the direct purchase/no direct purchase categories. Future research regarding the determinants of search behavior is needed and should be of interest to researchers with public policy and/or managerial interests.

REFERENCES

Ahmed, Sadrudin A. and Gary M. Gulas (1982), "Consumers' Perception of Manufacturers' Suggested List Price," Psychological Reports, 50 (2), 507-518.

Bagozzi, Richard P. (1977), "Structural Equation Models in Experimental Research," Journal of Marketing Research, 14 (May), 209-226.

Barnes, James G. (1975), "Factors Influencing Consumer Reaction to Retail Newspaper 'Sale' Advertising," Proceedings of the American Marketing Association, 471-477.

Berkowitz, Eric N. and John R. Walton (1980), "Contextual Influences on Consumer Price Responses: An Experimental Analysis," Journal of Marketing Research, 17 (August), 349-358.

Blair, Edward A. and E. Laird Landon, Jr. (1981), "The Effects of Reference Prices in Retail Advertisements," Journal of Marketing, 45 (Spring), 61-69.

Della Bitta, Albert J., Kent B. Monroe, and John M. McGinnis (198-1), "Consumer Perceptions of Comparative Price Advertisements," Journal of Marketing Research, 18 (November), 416-427.

Fry, Joseph N. and Gordon H. McDougall (1974), "Consumer Appraisal of Retail Price Advertisements," Journal of Marketing, 38 (July), 64-74.

Helson, Harry (1964), Adaptation Level Theory, New York, NY: Harper and Row.

Joreskog, Karl G. and Dag Sorbom (1984), LISREL VI: Analysis of Linear Structural Relationships by the Method of Maximum Likelihood, Mooresville, Indiana: Scientific Software, Inc.

Keiser, Stephen E. and James R. Krum (1976), "Consumer Perceptions of Retail Advertising with Overstated Savings," Journal of Retailing, 52 (Fall), 27-37.

Liefeld, John and Louise Heslop (1985), "Reference Prices and Deception in Newspaper Advertising," Journal of Consumer Research, 11 (March), 868-876.

MacKenzie, Scott 8., Richard J. Lutz, and George E. Belch (1986), "The Role of Attitude Toward the Ad as a Moderator of Advertising Effectiveness: A Test of Competing Explanations," Journal of Marketing Research, 3 (May), 130-143.

Monroe, Kent B. (1973), "Buyers' Subjective Perceptions of Price," Journal of Marketing Research, 10 (February), 70-80.

Monroe, Kent B. (1977), "Objective and Subjective Contextual Influences on Price Perception," in Consumer and Industrial Buyer Behavior, A.G. Woodside, J.N. Sheth, P.D. Bennett, eds., New York: North Holland, 287-296.

Monroe, Kent B. and Joseph D. Chapman (1987), "Framing Effects on Buyer's Subjective Product Evaluations," in Advances in Consumer Research, Vol. 14, Association for Consumer Research.

Monroe, Kent B. and Susan M. Petroshius (1981), "Buyers' Perceptions of Price: An Update of the Evidence," in Perspectives in Consumer Behavior, 3rd Edition, Harold H. Kassarjian and Thomas S. Robertson, Eds., Glenview, Illinois: Scott, Foresman, and Company, 43-55.

Petroshius, Susan M. and Kent B. Monroe (1987), "Effect of Product-Line Pricing Characteristics on Product Evaluations," Journal of Consumer Research, 13 (March), 511-519.

Ryan, Michael J. (1982), "Behavioral Intention Formation: The Interdependency of Attitudinal and Social Influence Variables," Journal of Consumer Research, 9 (December), 263-278.

Sewall, Murphy A. and Michael H. Goldstein (1979), "The Comparative Price Advertising Controversy: Consumer Perceptions of Catalog Showroom Reference Prices," Journal of Marketing, 43 (Summer), 82-92.

Thaler, Richard (1985), "Mental Accounting and Consumer Choice," Marketing Science, 4 (Summer), 199-214.

Winer, Russel S. (1986), "A Reference Price Model of Brand Choice for Frequently Purchased Products," Journal of Consumer Research, 13 (September), 250-256.

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