The Accuracy of Price Knowledge: Issues in Research Methodology

Jerry N. Conover, Northern Arizona University
ABSTRACT - Two studies exploring shoppers' awareness of prices of products they had selected found that price awareness was better than indicated by previous research. Methodological considerations in such research are addressed, and preliminary findings concerning price awareness are reported.
[ to cite ]:
Jerry N. Conover (1986) ,"The Accuracy of Price Knowledge: Issues in Research Methodology", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 589-593.

Advances in Consumer Research Volume 13, 1986      Pages 589-593

THE ACCURACY OF PRICE KNOWLEDGE: ISSUES IN RESEARCH METHODOLOGY

Jerry N. Conover, Northern Arizona University

ABSTRACT -

Two studies exploring shoppers' awareness of prices of products they had selected found that price awareness was better than indicated by previous research. Methodological considerations in such research are addressed, and preliminary findings concerning price awareness are reported.

INTRODUCTION

The idea that consumers know the prices of products on the market is a key assumption of economic theory. The idea is also implicit in marketers' manipulation of price as a key strategic tool. m is assumption has been challenged, however, by data from consumer surveys (Harrell, Butt and Allen 1976; Heller 1974; Progressive Grocer 1964, 1982; Zbytniewski 1980). These studies reported a surprisingly large number of shoppers who could not very accurately report the prices of common grocery products, even when asked about products they use regularly (typical figures indicated about two thirds or more of shoppers were off by more than 5% in their price judgments).

Given the importance of price as a major marketing variable, it would be very useful to gain a clearer understanding of what consumers know about prices, and how they use that information in their purchase decisions. Unfortunately, the literature concerning price knowledge is quite sparse (see Dickson and Sawyer [1986] for a review), and several scholars have called for research to clarify this area (Conover 1984; Monroe 1973, 1976; Olson 1980; Zeithaml 1984). The two studies presented below were conducted to address these questions.

Both studies were designed to assess the accuracy of consumers' price knowledge and to determine whether such accuracy is related to shopper characteristics or practices, marketing practices, and product variables. me present paper reviews the procedures employed in each study, and discusses methodological issues in this type of research. Additionally, some key findings concerning price knowledge accuracy are discussed. However, a full elaboration of the complex relationships between price knowledge and the numerous other factors investigated is beyond the scope of this paper, and will appear in a later report.

STUDY 1

The first of these investigations was designed to assess the accuracy of supermarket shoppers' price knowledge for a variety of products. In addition, the study explored several aspects of participants' perceptions, preferences motivations, demographic status, and general shopping behavior, as possible factors influencing their price knowledge

Method

Subjects. Participants were recruited from adult shoppers in a large supermarket that is part of a major national grocery chain. This particular store was chosen for the representativeness of its clientele, upon the advice of the store's regional manager. The 168 shoppers whose data are reported below were systematically sampled to provide a balanced cross section of the local market population in terms of age (median = 32 years), education (median = "some college"), household size (median = 2.2) and income (median = $18,354). Females accounted for 80.8% of the total, a figure approximating that of the store's overall clientele Sixty-six percent of respondents worked full or part-time, and 58.6% were married. Distributions of each of these classification variables (except sex) were quite similar to those in the metropolitan area in which the store is located.

Procedure. Data from shoppers were obtained through individual personal interviews in the store. Shoppers were eligible to participate if they had in their cart at least one product from any of nine categories (white milk, white bread, mayonnaise, cola, stick margarine, ground coffee, powdered laundry detergent, frozen orange juice concentrate, or paper towels). Efficiency in data collection was enhanced by the selection of these high volume categories (as indicated by store management) as the target products. Only a small proportion (less than 10%) of shoppers who were asked to participate declined, usually due to time constraints.

Shoppers were recruited while within the aisles of the store. The time elapsed from selection of a target product until they were intercepted for the interview was not explicitly monitored, but probably ranged from several seconds (when product selection was directly observed) to perhaps 20-30 minutes.

Upon consenting to participate in the survey, the shopper was asked a variety of questions concerning grocery shopping habits (store patronage; shopping frequency; use of newspaper ads and coupons; general preferences for national, private, and generic brands; etc.). The shopper was then asked several questions about the target product (from the categories above) she/he had already selected. These questions began with the shopper recalling, without looking at the item, the price marked on that product. Next, the shopper named all the brands he/she could think of in that product category, and specified the factors he/she considered important in choosing such a product.

The shopper was then asked to rank all the brands carried by the store in that category from low to high in price. This task was accomplished by sorting cards, each bearing a different brand name, on a board that had a long line labeled, at opposite ends, "Most Expensive Brands" and "Least Expensive Brands". Shoppers were permitted to omit brands about whose price they had no idea, and to place multiple brands at the same position on the ranking line if those brands were thought to be identically Priced.

Following the ranking task, the shopper answered several questions about his/her perceptions of the product, factors influencing this particular choice, and usage patterns for the product. Finally, the shopper completed a brief demographic questionnaire. Upon completion of the interview, the shopper was thanked and received a coupon redeemable in the store for a free dozen eggs. Each interview took from seven to ten minutes to complete.

RESULTS

Exact Price Recall. The absolute value of the difference between the subject's estimate of the product price ("exact price recall") and the actual marked price was divided by the actual price, to provide a measure of the proportion of error in price recall. Each price judgement was also classified as an overestimate, an underestimate, or a correct recall. These results are shown in Table 1.

The mean error in price judgments, pooled across all products, was 6.1% of the actual price. Mean exact price recall was mast accurate for margarine and milk, and least accurate for colas. However, pooling the over- and under-estimates into a single mean error score obscures some interesting variations across products in the nature of errors. For example, the percentage of shoppers who correctly recalled the exact prices of milk and margarine were among the lowest for any product. Therefore, Table 1 also shows the mean size of over- and underestimates for each product, as well as the percentage of shoppers who have such erroneous responses.

TABLE 1

EXACT PRICE RECALL FOR NINE GROCERY PRODUCTS (STUDY 1)

Of all the products, shoppers were most likely to correctly recall the price of brand (65% correct). The price of the product with the highest percentage of errors, milk, was correctly recalled by 36.8% of shoppers. And correct prices were recalled by at least half the shoppers for six out of the nine products. Thus, the exact price recall by shoppers of these products was substantially better than that of shoppers in earlier studies cited above. Erroneous price recalls were fairly evenly divided across products between over- and underestimates, Although no one underestimated the price of bread.

Relative Price Recall. The price rank order of brands as sorted by the shopper was assessed for accuracy against actual price rankings by computing a Spearman rank-order correlation coefficient for each shopper. In these computations, any brands omitted by the shopper were also omitted in determining the actual rankings. Thus, each shopper's relative price recall accuracy was indexed by this correlation coefficient. Overall, as shown in Table 2, shoppers' relative price recall was rather accurate; the mean correlation, with all products pooled, was O.579, p<.0001. There was some variability among products in relative price accuracy, but all correlation coefficients were significant at p<.0001. Moreover, the variability in relative price recall does not seem to be an artifact of the number of brands available in a product category to be ranked (see Table 2?.

Relative and Exact Price Recall Compared. Shoppers' accuracy in relative price recall was only slightly related to their accuracy in exact price recall (r=0.121 across products, p<.06). No substantial product differences were found in the correlation between exact and relative price recall. Pooled across products, shoppers who correctly recalled the exact price had a mean relative price recall coefficient of 0.550; those who overestimated exact price had a mean relative price recall score of 0.597; those who underestimated exact price had a relative price recall mean of 0.626. These values are not significantly different f.WE each other (P[2,165]) = 1.30, p>.30). Thus, exact price recall was not a reliable predictor of relative price recall.

Discussion

Implications of the above results are addressed following discussion of the second study, below. The present discussion focuses on issues of measurement and analysis of price recall accuracy.

Assessing Exact Price Recall. The proportional deviation of price recall from the actual price, expressed as an absolute value, has been used by Harrell et al. (1976) and Zeithaml (1982) to index the accuracy of exact price knowledge. This measure has the advantage that it is easily interpreted as the relative deviation of price recall from the correct price. However, several potentia] limitations should also be considered:

1. The absolute value obscures the directionality of any errors.

2. The proportional nature of the measure makes its magnitude dependent on the base (correct) price, whereas it might be informative to explore the actual deviation in cents of the recalled price from the correct price.

3. The number of shoppers who correctly recall the exact price is unavailable from such a measure. Thus, it obscures qualitative differences that may exist between those who err and those who correctly recall prices.

These concerns should not eliminate the exact price recall measure from use. Rather, it should be supplemented by measures that do reveal something about the nature of recall errors, such as percent of shoppers over- and underestimating the price.

TABLE 2

CORRELATIONS BETWEEN JUDGED AND ACTUAL PRICE RANKINGS (STUDY 1)

Assessing Relative Price Recall. The brand sorting task for indicating the shopper's relative price judgments has both advantages and disadvantages. It is easy for the shopper to understand, and it can reveal the degree of differentiation of price levels within a product category perceived by the shopper. The only previous measure of price recall error employed a conceptually similar task, but computed an index from the sum of the differences between each brand's correct and recalled ranks (Zeithaml 1982). The present correlation coefficient measure has an advantage over Zeithaml's "price comparison error" index in that it can be easily compared across categories with different numbers of brands. Drawbacks of this method of assessing relative price recall include:

1. The procedure is cumbersome to administer. The board must be large enough to hold many brand cards, and shuffling the cards between interviews can be awkward. Also, considerable time is spent in recording the rank ordering of the brands, especially for large categories.

2. Establishing the "correct" ranks of the various brands is confounded by the fact that many brands come in multiple sizes and prices. The correct rank of a brand thus depends on its size. One could compare it only against other brands of the same size, but not every category has standardized package sizes. me present study determined correct ranks on the basis of unit prices, averaging across sizes for each brand. (It should be noted that this problem is not limited to this one measure; it also applies to other similar approaches to assessing relative price recall when multiple package sizes exist.)

In conclusion, Study 1 produced substantial data showing considerable ability of shoppers to recall both exact and relative prices. Study 2, reported below, extended the investigation of consumer price knowledge in several ways.

STUDY 2

Like Study 1, the second study was designed to investigate the influence of numerous factors on the recall and use of price information. Again, the complex pattern of interrelationships among these factors will not be addressed in the present paper. Instead, this report will focus on the general nature of price recall findings, and also will addresses methodological considerations in such research.

Given the unknown effect of time delay between initial processing of price information and subsequent questioning in Study ], the second study was designed to control that effect by closely monitoring consumer choice processes; price recall questioning began immediately after product choice. Additionally, this study sought to determine any possible change in price recall over time by including both immediate and delayed recall tests. Another extension beyond Study 1 was provided by obtaining price processing and recall measures for several products from each shopper, thus allowing investigation of the generalizability of price awareness across products. Finally, Study 2 explored the nature of processing of price information both at the time of choice and at recall.

Method

Grocery shoppers were recruited for Study 2 in the same supermarket described in Study 1. They participated in a simulated shopping task that allowed good control aver the choice situation and close monitoring of responses.

Subjects. Sixty-six adult women participated in this study. Their demographic profile closely matched that of the market area as a whole: median age = 36.5; median educational bracket = "some college"; median income = $18,438: median household size = 2.5. full- or part-time workers comprised 68.2% of the sample, and 59.1% were married.

Procedure. Shoppers entering the store were randomly selected and asked to participate in a study of "grocery shopping habits". Those who consented (the large majority of those approached) were led to an isolated room in which was located a section of store shelves stocked with several common products. Products were taken from the actual store shelves, and current prices were marked on both shelves and individual packages. Each product category included national brands, private label and generic products, and at least one item in each category was identified by a shelf talker as being on a temporary price deal.

Facing the shelves, the shopper was instructed to choose one brand from a particular product category (mayonnaise), basing her decisions on the same considerations she would normally apply in such a choice. She was instructed to "think out loud" while she made her selection, mentioning anything she thought about with respect to the decision. Her words were recorded on tape for later analysis.

This first brand choice was a practice trial, to confirm comprehension of the instructions and get the shopper used to the procedures. Once the procedures were established, the shopper was asked to make similar choices from four other product categories (colas, flour, toothpaste, and peanut butter), one at a time upon instruction from the experimenter. As each brand was selected, the shopper placed the item in a shopping cart and proceeded to the next choice. Order of item selection was systematically varied across subjects so that each product category was chosen first, second, third, and fourth approximately the same number of times.

Immediately following the last item choice, with the tape recorder still running, the shopper was asked to state, without looking at either the cart or the shelves, the price of the item she had just placed in her cart. She then recalled the prices of any other brands in that category that she could remember, and indicated whether she in fact had noticed any prices at all. She was also asked whether she had noticed whether any brands were offered at a special price. Next, she identified (from a list of brands) all the brands that she had actually considered in making her choice.

The shopper indicated her relative price recall by a procedure somewhat different than in Study 1. A similar board labeled "most" and "least expensive brands" was used, but no cards representing the brands were sorted. Instead, the line from "least" to 'most expensive" was evenly graduated with a series of numbers from 1 to 9. The shopper was asked to point to the place on that line representing where the price of her chosen brand stood relative to all the other brands in the category. The nearest number on the line was recorded as an index of the brand's perceived relative price.

This same procedure was replicated for each of the other three product categories, in reverse chronological order. Verbal protocols were recorded throughout.

Following the exact and relative price recall tasks, the shopper was asked a series of questions about the brands and categories used in the tasks. For each category, she indicated brand repeat purchase behavior and preference, category purchase frequency, perceived similarity among brands, lowest-price buying tendency, and perceived price volatility over time. The questioning then proceeded to general shopping behaviors and tendencies. The shopper rated her store loyalty; percentage of the household's grocery shopping that she does; general national/store/ generic brand preferences; and tendencies to notice grocery prices in general, to read grocery ads, to use coupons, to notice grocery price changes, and to remember prices after buying groceries. Finally, she completed a standard demographic questionnaire. The entire procedure took approximately 20 minutes. Upon completion, participants were debriefed, thanked, and given a coupon for a free dozen eggs. (Shoppers did not keep the items they had selected.)

TABLE 3

IMMEDIATE AND DELAYED EXACT PRICE RECALL FOR FOUR GROCERY PRODUCTS (STUDY 2)

Two days after completing the above tasks, each shopper was telephoned, without advance warning, and was asked to recall the exact prices of each of the four target products she had selected. The response latency in seconds was also recorded at that time for each recalled price.

Results

Exact Price Recall. Shoppers' exact price recall was analyzed as in Study 1. As shown in Table 3, errors in this task were somewhat larger than in Study 1, averaging 16.5% of the actual prices across products during immediate recall, and 13.4% during delayed recall. Correct recalls were provided by no more than 44.4% of the shoppers for any product, and more than 90% of the shoppers erred in immediate recall of the price of peanut butter.

The magnitudes of errors in 2-day delayed recall were generally similar to those in immediate recall, except in the case of flour, for which delayed recalls improved considerably. In no case is there evidence for a substantial worsening of price recall after a 2-day delay. The correlations between immediate and delayed recall were: cola = .?74; flour = .607; toothpaste = .518; and peanut butter = .604; all of which are highly significant. Thus, these shoppers who were more accurate in immediate exact price recall tended to be more accurate in delayed recall as well.

The question of whether individuals vary in a general tendency to recall prices accurately was addressed by comparing each shopper's immediate exact price recall for each of the four products. Table 4 reveals the correlations between these recall figures, and shows that there was very little consistency in recall accuracy across products. None of the correlations was significant at =.05.

Relative Price Recall. The accuracy of shoppers' recall of relative price of each brand they chose was computed from the correlation between their judged position of the brand on the numbered scale from "lowest" to "highest" price and its actual position. The actual relative position for each brand was converted to a 9-point scale by the formula:

EQUATION

where Pb = the actual price of the brand, Ph = the highest price in its category, and P1 = the lowest price in its category.

The resulting correlation coefficients for relative price judgments were: cola = .687; flour = .843; toothpaste = .691; and peanut butter = .881. All are significant at =.00001, indicating very good relative price recall by these shoppers.

Discussion

Many measures collected in Study 2 are not reported above, as they are beyond the scope of this report. The discussions that follow again focus and issues of measurement and analysis, as well as presenting same preliminary interpretation of the findings.

Relative Price Index, Study 2. The revised approach to assessing the shopper's relative price knowledge used in this study is much easier and quicker to administer than the procedure used in Study 1. Consequently, it probably is more enjoyable for the consumer, as well. Given the similar magnitudes of relative price recall correlations in the two studies (the one common product, cola, produced very similar results), the advantages of the latter procedure should be strongly considered for future such research.

The design of Study 2 also eliminated another problem found in assessing relative price recall in Study 1. By including in the simulated choice situation only products of the same size within each category, no ambiguity as to correct relative prices occurred.

GENERAL DISCUSSION

Jacoby and Olson (1977) pointed cut that price information may be retained in memory in various possible ways, including exact representation of the actual price and also a representation of the product's price relative to other brands. The present research confirms that consumers can and do maintain price knowledge in both of these fashions. However, as those authors noted, price information as tested by recAll might be directly retrieved from a stored representation, or it might instead be reconstructed on the basis of other product knowledge at the time of testing. Data from the present study are currentlY being analYzed to shed some light on this issue.

Whatever the nature of production of responses to price recall questions, the measures employed in this study seem to offer reasonably valid, manageable procedures for assessing consumers' price knowledge. Certainly, as Monroe et al. (1986) argue, there may be cases in which recall measures will not reveal memory traces of prices that consumers could possibly maintain, and the additional use of more sensitive measures such as recognition tests is encouraged. However, recognition tests cannot provide, as can recorded protocols from recall tests, much clue to the nature of price information encoding and retrieval.

Finally, some comments should be made about the price recall results of the present studies. There data suggest that shoppers are more knowledgeable about prices of products they have recently selected than earlier studies indicate. The poorer price recall revealed in Study 2 may reflect the artificiality of the situation (the shoppers were not given any forewarning that price was a focus of the study, and they were clearly being closely observed, which may have altered their normal decision-making style somewhat). Or it may result from sampling of different product categories than in Study 1. The failure to find a drop in price recall over two days in Study 2 may reflect heightened attention to those prices once price questions began, and rehearsal of them afterward.

The impact of methodological variations on price recall measures is a topic that should clearly be addressed in future research. a e more obtrusive the data collecting procedure, especially if it affects the choice process itself, the more likely price recall will be influenced by factors unrelated to the development and use of price knowledge in real shopping situations. Extending the present studies with such investigations, and exploring the impact that stored price knowledge has on consumer behavior, are worthy objectives for future research.

REFERENCES

Conover, Jerry N. (1984), "Price Effects on Consumer Behavior: A Status Report," Advances in Consumer Research, Volume 11, T. C. Kinnear (Ed.), 633-635.

Dickson, P. R. and Sawyer, A. G. (1986), "Price Perceptions of Supermarket Shoppers at the Point of Purchase," Advances in Consumer Research, Volume 13, R. Lutz (Ed.).

Harrell, G. D., M. D. Hutt, and J. W. Allen (1976), Universal Product Code East Lansing MI: MSU Business Studies.

Heller, Walter H. (1974), "What Shoppers Know - And Don't Know - About Prices," Progressive Grocer, (Nov.), 39-41.

Jacoby, J. and J. C. Olson (19 n), "Consumer Response to Price: An Attitudinal, Information-Processing Perspective," in Y. Wind & M. Greenberg (Eds.), Moving Ahead with Attitude Research, Chicago: American Marketing Association.

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

Monroe, Kent B. (1976), "The Influence of Price Differences and Brand Familiarity on Brand Preferences' Journal of Consumer Research, 3, 42-49.

Monroe, K. B., C. P. Powell, and P. K. Choudhury (1986), "The Influence of Price Awareness on Price Perceptions," Advances in Consumer Research, Volume 13, R. Lutz (Ed.).

Olson, Jerry C. (1980), "Implications of an Information Processing Approach to Pricing Research," in C. W. Lamb & P. Sunne (Eds.), Theoretical Developments in Marketing, Chicago: American Marketing Association, 13-16.

Progressive Grocer (1964), "How Much do Shoppers Know about Retail Prices?", February, C104-C106.

Progressive Grocer (1982), "Shopping Habits: Prices Aren't, After All, Everything," April, 141-144.

Zbytniewski, Jo-Ann (1980), "Shoppers Cry 'Remember the Price' -- But Do They Practice what they Screech?", November. 119-122.

Zeithaml, Valarie A. (1982), "Consumer Responses to In-Store Price Information Environments," Journal of Consumer Research, 8, 357-369.

Zeithaml, Valarie A. (1984), "Issues in Conceptualizing and Measuring Consumer Response to Price," Advances in Consumer Research, Volume 11, T. C. Kinnear (Ed.), 628-633.

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