Methods to Research Shoppers' Knowledge of Supermarket Prices

Citation:

Peter R. Dickson and Alan G. Sawyer (1986) ,"Methods to Research Shoppers' Knowledge of Supermarket Prices", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 584-588.

Advances in Consumer Research Volume 13, 1986      Pages 584-588

METHODS TO RESEARCH SHOPPERS' KNOWLEDGE OF SUPERMARKET PRICES

Peter R. Dickson, The Ohio State University

Alan G. Sawyer, University of Florida

Much marketing research studies the effect of relative price or price reductions or promotions in the grocery store. These studies implicitly presume that consumers pay careful attention to price in order for it to affect their purchase behavior; Psychological studies of price perception explicity posit the existence of referent prices in consumers' minds. These referent or usual prices are posited to form the basis for perceptions of new prices as low or high. Knowledge of referent prices also helps the consumer to recognize both the presence and extent of a price discount. Of course, any attempts to help educate consumers to be more conscious of costs and values would benefit from knowledge of the accuracy of shoppers' knowledge of prices of purchased items.

The purpose of this paper is to review past research about consumers' knowledge and use of price information. After a review and a critique of past methods, a new procedure is described. This method has been designed to correct the problems of past research and preliminary results indicate the method is a useful improvement.

PAST RESEARCH

Gabor and Granger

In their frequently quoted study, Gabor and Granger (1961) examined consumer ln-home price awareness. In November, 1958 they studied a random sample of housewives living in the City of Nottingham, in England. In 82% of the 5,276 purchases across a sample of 428 households, the housewife named a price. This price "consciousness" was higher than the researchers expected and was reported to be inversely correlated with a social status measure. The accuracy of the price estimate was checked for seven of the 15 commodities that had been purchased in the last week. Overall, 57% of the estimates were correct, 25.4: were wrong and 17.6% gave no answer. The exact price recall accuracy was highest for tea (79.3%), coffee (68.1%) and sugar (67.0%) and lowest for margarine (46.1%), flour (35.6%) and breakfast cereal (34.8%). If latitudes of | 10% around the actual price were allowed, the overall percentage of correct price estimates would have risen to at least 60: and 70%, respectively.

At the risk of being accused of taking some cheap shots at a truly classic study undertaken over two decades ago, it is important to discuss some of its weaknesses, precisely because of the impact it has had on later researchers' and theorists' thinking about consumer price awareness. Gabor and Granger's measure of willingness to simply volunteer a response to a price estimate question was a suspect operationalization of "price awareness." The measure clearly depended very much on the consumer's desire to cooperate as well as her price knowledge and would have included many "best guesses." The extent of this "desire to cooperate" bias might have been inferred by observing the nonresponse variability across the 44 student interviewers some of whom no doubt elicited more "cooperation" than other interviewers.

Regardless, the observed relationship between reporting a price and social status was, in fact, not statistically significant. Gabor and Granger did not report any statistical tests but a chi-square test and correlational test that can be computed from the percentages given in Table 1 (Gabor and Granger 1961, p. 177) and both fail to reject the null hypothesis despite more than adequate statistical power.

Measurement and analysis criticisms aside, the fact remains that over half of the subjects accurately recalled the price. This finding, however, must also be qualified. On the one hand, no allowance was made for special offer prices, deviations from the manufacturers' recommended price and the fact that shop assistants may have quoted and charged an incorrect price. Consequently the consumer's price estimate may have been correct and the "actual" checked price incorrect. But on the other hand, eight of the 15 commodities were not used in the price accuracy study because they were sold at "such a variety of prices." This made it impractical to check the answers. It also introduced a serious commodity selection bias that led to an overstatement of price awareness. Consumers will be more accurate in the recall of commodity prices that are stable across time and place. It should also be noted that this recall accuracy was for items purchased in the last seven days. The researchers acknowledge that they were really interested in studying attention paid to the price at the time of purchase and that even someone who paid great attention to price would likely forget it after more than seven days. In other words, their findings speak more to attention paid to price in the past and recall of that price rather than to attention paid to price at the point of purchase or whether price last paid will be accurately recalled at the time of the next purchase.

A further concern is the uniqueness of some of the shopping circumstances and consequent behavior. An unidentified percentage of the shopping was undertaken at stores where the commodities were not directly accessible to the buyer and were not priced on the package. The buyer had to ask for the particular commodities often by brand but they were in the habit of asking for the product by price in some instances. For example, Gabor and Granger note that housewives frequently described and ordered their favorite tea by the price and not by a brand name. It is therefore not surprising that tea had the highest recall accuracy (79.3%), some 10% higher than the next commodity.

A budget conscious buyer has a greater need to know the price of a commodity outside the store in circumstances where such knowledge cannot be readily gathered inside the store. Alternatively the consumer could set a limit on the price she is willing to pay for an item and ask the shop assistant to provide an item at, or below, that price. Either way the unique shopping situation studied by Gabor and Granger encouraged the learning of the prices of frequently purchased goods. The low inflation rate and stable pricing climate that existed at the time maintained the relevance of these learned prices and ensured frequent rehearsal of the unchanging prices in memory. All in all, the relevance of the findings of the study to supermarket shopping in the United States in the 1980's is very suspect.

The Food Marketing Institute

In a study undertaken in the mid 1970s, interviewers intercepted 1,500 shoppers at the checkout (Allen, Harrell and Hutt 1976). They randomly selected three items from the shopping cart and asked the shopper to name the price. The percentage of exactly correct response for the three items was 55.8% and the average price estimates were off by 8.0%. The same shoppers were also asked to select and price three items that they had purchased on previous shopping trips (but not the day of the interview) from a display of 40 commonly purchased products. Only 16.8% of the shoppers were able to correctly name the price even with the selection bias that inflated the perception accuracy. These same researchers interviewed another sample of shoppers at the point of purchase but the price accuracy estimates of this group were grossly inflated (to the point that they were meaningless) because the interviewers allowed the shopper to check the price on the item or on the shelf label before answering.

A third phase of this study involved noting products at the time of the checkout interview. Some two weeks later subjects were asked to recall the price of these items. The shopper was not aware items had been noted so item specific sensitization was avoided. However, the interviews and some expectation of future questioning may have increased general sensitivity to item prices in the home. The in-home price awareness was 31.4% and the price estimates were off by an average of 14.5%. The results of this study are particularly interesting for they suggest that price awareness is higher the closer you get to the time and place of the past purchase.

Progressive Grocer

A variation of the Gabor and Granger price awareness study was undertaken in the United States in 1963. Colonial Stores and Progressive Grocer (1964) selected 59 frequently advertised and very price competitive items. Tables were set up in the stores and several thousand customers played a version of "The Price is Right" TV game. They had to estimate the price of a subset of six of the 59 items. The shoppers were reported to have been cooperative and interested. They were not very accurate at estimating price. For all but one item, fewer than 40% of the shoppers could name the exact price. Less than 20% could name the exact price for 41 of the items. The extraordinary exception was that 86% of the customers correctly named the price of a six pack of Coca-Cola.

When some slight error in the estimate was allowed, the percentage of correct answers increased but even then for only 7 of the 59 items could a majority of the shoppers give a price estimate within + 52 of the actual price. The variation in their price accuracy measure across the item was considerable (x = 31.4%, s.d. - 14.9%). The research reported that item users (and presumably therefore item buyers) were, overall, twice as knowledgeable about the correct price as non-users. While the evidence for this assertion and even the criteria upon which it was based was not explained, it does make the findings more consistent with the British study. Sex age and household income of the shopper was not related to the accuracy of the estimates.

The relationship between exact price accuracy and approximate price accuracy was not as high as might be expected. Excluding the Coca-Cola result (because of the ceiling effect) the correlation, across the 59 items, between the percentage of shoppers who could exactly name the price of the item and the percentage of shoppers who could name the approximate (within + 5%) price of the item was only 0.58. The following examples explain why the correlation was low. Twenty-three percent of the shoppers could exactly state the price of Pillsbury Pancake Mix. When the accuracy was relaxed, 25% of the shoppers could estimate the price within 5%. By contrast 202 of the shoppers could exactly price 5 lbs. of Domino Sugar, but a much higher 67: could estimate the price within 5%. In short, shoppers' exact and approximate price accuracy are not necessarily related. This could be due to the high likelihood that frequent purchase of the product group will result in a high approximate accuracy and frequent purchase of a brand in a product group will result in high exact accuracy. But these two frequencies are not necessarily related, in part because brand loyalty and brand share within product groups varies.

The 1964 study was replicated in 1974 using subsets of 11 product brands from a total of 44 items and 560 shoppers (Heller 1974). The consumers who said they bought the item were on average more likely to give the correct price (within 5%). However, for 7 of the 44 items the percentage of the non-buyers who gave a correct price estimate was surprisingly higher for non-buyers than buyers. As might be expected, the more price conscious buyers of private labels were more aware of prices than buyers of national brands.

The variation for item buyers in accuracy of price estimates (within 5% of the actual price) across the products was again extraordinary, ranging from highs of 71% for Marlboro cigarettes, 54% for Land-O-Lakes butter and 52% for Scott paper towels down to 10% for Crest toothpaste, 9% for Maxwell House Instant Coffee and 8% for Saran Wrap. The prices of cornflakes, frozen orange juice, canned corn and fruit cocktail were heavily overestimated. The prices of instant coffee, shortening, sugar, salad dressing, and facial tissue were heavily underestimated. Although it might have been expected that price accuracy would be greater for higher priced grocery items, this was not found; the 10 private label items enjoyed higher price recognition than the 34 nationally branded products. These private labels may have been more notable in consumers' memories because of their low price.

The accuracy of price perceptions was lower in 1974 than in 1963. Only 8% of the 1974 shoppers estimated the exact price compared to 20% exact in 1963. Across the matched pairs of 35 common items there was a significant (p < 0.05) 6% average decline in price estimation accuracy, estimated within 5%. This effect was product specific and may have reflected changes in price associated with the recent lifting of price controls.

Two further price awareness studies have been undertaken by Progressive Grocer. A study undertaken in late 1976 (Dietrich 1977) found that only 30% of the time could buyers of 12 popular items name the price within 5%. For five of the items a significantly greater number of buyers overestimated the price but for another five items a significantly greater number underestimated the price. In the most recent study (Zbytniewski 1980), 27% of the buyers were able to estimate the price of 16 items, within 5%. The most accurately recalled price was for Perrier Bottled Mineral Water (46% amongst buyers compared to 14% amongst non-buyers). The least accurately recalled price was for Maxwell House Instant Coffee (11% amongst buyers and 12% amongst non-buyers).

Only 8% of the shoppers in the 1977 and 1980 studies knew the exact price. Age and shopping tine did not influence awareness but self-labelled price aware shoppers were a little more accurate (24: to 20.5%, within + 5% of actual price). A comparison of price awareness changes for particular products between 1980 and 1977 revealed that Scott paper towels dropped from 38% to 29% (it was 52% in 1974), Coca-Cola went from 39% to 25% and Tide XK detergent from 34% to 25%. Price accuracy did not increase for any of the eight comparable items.

Over recent years Progressive Grocer has provided a good deal of evidence that price has become a much more important consideration in supermarket shopping. The percentage of households using particular items has changed as household have switched from expensive product groups to cheaper substitutes. The fact that the incidence of one-brand users has dropped over the years suggests less brand loyalty and greater price sensitivity. Low prices was ranked the fifth most important characteristic in choosing a supermarket in 1973, rose to the third most important in 1975, became the second most important in 1979 and was number one in 1981 (Progressive Grocer 1975, 1981; Zybtniewski 1979). Shoppers have also greatly increased their reported use of shopping lists, ad reading and coupon clipping.

Despite all of this evidence of increasing focus on the price of products, buyers have not become more aware of prices. There are several possible explanations for this apparent anomaly. It can be argued that inflation and the increase in price specials have wade the task of price estimation much more difficult. It is true that price awareness dropped significantly between the 1963 and 1974 study and 1976 was a year when, according to Progressive Grocer, even the store personnel had trouble remembering the latest operative shelf price. However, the effect of inflation should have shown itself in a general tendency for buyers to underestimate the price of items. Across the 59 items in the 1974 study, 38% of the buyers underestimated the price but 33% overestimated the price. This is not a substantial difference, and the higher percentage who overestimated cannot account for the overall drop in accuracy between 1963 and 1974. It should also be noted that estimation was more likely to be under than over the correct price for 26 of the 44 items, a result only slightly higher than chance. The 1977 and 1981 studies both reported tendencies for buyers to overestimate rather than underestimate the price. One possible reason for this could be a tendency for shoppers to intentionally guess high as a way to include any recent inflation. The increase in couponing and offering of price specials would also be expected to produce a bias toward understating rather than overstating price. Consequently, the results do not obviously support increased price promotions as an explanation for the lack of improvement in price awareness over a period when the salience of price has increased.

The compelling conclusion is that many buyers do not learn the current exact price of the many hundreds of items they purchase. They probably do not attempt to learn and retain this information because they do not need to learn and retain this information even if they are very price conscious. The point of purchase shopping setting provides all of the price information they need for making an item or brand choice within a product group. This reduces the need to bring any information about exact prices paid in the past to the point of purchase. The increased volatility of prices resulting from inflation and the increased use of price promotions has encouraged shoppers to rely on price information at the point of purchase. Such a theory assumes that evaluations of an item's price are made relative to the immediate set of alternatives. Buyers use an external rather than internal reference frame to evaluate the price of an item within a product group. Buyers do use an internal reference frame to evaluate the general price level of a product group.

In-store Price Checking Behavior

Consumers have a number of ways of becoming aware of the price of items. Advertising often presents information about the price of an item. But a reduced special price is frequently featured rather than the standard price. Such advertising may actually lower consumers' price expectations for a product group or brand to a level below the normal price. Of course, if most purchases are made at a special price, then that price rather than the manufacturer's list price should be rightly considered the "normal" price. It is then the manufacturer and not the consumer who has a price expectation problem. This situation probably exists today for many products, ranging from home appliances to soft drinks, where a significant proportion of retail sales are made at a Sale or Special price.

Prices may also be noted in the home when actually using an item. If price awareness were solely based on such in-home price checking, then consumer's price awareness might tend to lag behind actual prices of products be cause the stock item was purchased some time ago and prices have risen over the inventory holding period.

It is most logical to assume, however, that price information is obtained at the point of purchase. It is current, unbiased and relevant. This information may be so useful that it dramatically reduces the need to remember specific prices and for consumers to possess fine-tuned internal price standards. The question of immediate interest is what patterns of search behavior do occur at the point of purchase.

Wells and Lo Sciuto (1966) undertook one of the first, if not the first, observational study of supermarket shopping behavior. In the research tradition of Barker (1965), 1,500 behavioral episodes were recorded. An episode began when the shopper entered an aisle and ended when the shopper left the aisle. The advantages of an observational study over a self-report study are obvious. It does not depend on a subject's ability to recall a rather mundane individual purchase from among many. Nor does it depend on the subject's understanding of the meaning of a question. It is also not biased by the subjects' desire to cast themselves in a favorable light and delude both themselves and the researcher.

The observers attempted to assess how much attention shoppers paid to prices. It was admitted that it was hard to tell whether the shoppers were looking at price when examining a package, the implication being that the results perhaps understated price checking behavior. The authors also imply that prices were not provided on the shelf facings of the supermarkets. This is an important issue in evaluating the relevance of their findings to today's supermarket shopping environment. Three product purchases (cereal, candy and detergent), two types of stores (urban, suburban) and four shoppers characteristics (adult male, adult female, adult couple with or without children, children alone) were studied. The highest observed price inspection activity was among adult females shopping for detergent in an urban store; one third were observed checking the price. The lowest reported price checking activity was observed for couples shopping for cereal in a suburban store (8%). The store and shopper differences were less than the product differences. Adult females did the most checking (in 17% of the events averaged across the products and stores) and children the least (10%). Surprisingly, a brand loyalty explanation for the lower price search activity for the cereal purchase was not supported by other evidence.

Allen, Harrell and Hutt (1976) reported a number of measures of price search behavior obtained from subjects at the point of purchase. In contrast to the Wells and Lo Sciuto study, their statistics may overstate the price search activity because they are self-report and subject to a possible self-presentation bias. Seventy-one percent of the shoppers in a conventional supermarket reported checking the price of the item before placing it in their shopping basket. However, only 31.5% reported checking the prices of other items and 12.9% reported using unit price information. The shoppers, interviewed at the checkout, were asked whether they had referred to an item already in their cart when considering another later purchase. Thirty-one percent of shoppers indicated that they had made such between product comparisons and, in 81% of such cases, a price comparison was made. Another question revealed that 86% of the shoppers had checked the prices of substitutes within a product category (i.e., canned, fresh or frozen peas).

In summary, the above two studies present conflicting findings on point of purchase price checking activity. The study undertaken in the mid-sixties reports a low incidence of price checking compared to the study undertaken a decade later. The difference may be partly the result of the research approach but it also could be a consequence of changes in the shopping environment. By 1975, supermarkets were presenting the price of items on the shelf facing along with, for some stores, the unit price of items. Where previously an item often had to be handled to establish its price, now the price of all of the brands and sizes could be easily assessed in a few seconds. Reducing the cost of seeking price information (in terms of effort) seems to have encouraged greater use of point of purchase price information. As already mentioned, the spurt of inflation that occurred in 1974-75 and a higher incidence of price specials may also have made the shoppers in the later study more conscious of price and yet less certain about the price of individual items, leading to more point of purchase price checking. To summarize, the increased value and reduced cost (in effort) have very probably over time contributed to more point of purchase price.

PROPOSED METHOD

A survey procedure was designed to provide the following advantages:

1. Ask shoppers about their knowledge of and use of price information in the stove at the point of-purchase immediately after they had selected the item in question. This would alleviate most of the problems due to decay in memory of delayed questions. It is very reasonable to expect rapid decay immediately after and also minutes after choice since most consumers will be focusing their attention on their next choice and have little need and to retain incentive the price information.

2. Combine observation of shopper's behavior at the point of purchase with verbal reports.

Observers were stationed at the point of purchase with a clipboard and were instructed to give the appearance of undertaking some stocktaking activity, a not unusual sight in a supermarket. On the next tenth minute of the hour the observer recorded the behavior of the first shopper to select a brand and size of the target product. This involved estimating the time spent at the point of purchase (the interval between when the shopper turned her attention to the display and when the chosen item was placed in the shopping cart or basket), noting the number of different brands physically touched and inspected (including the chosen brand), and observing the number of different sizes physically touched and inspected (again including the chosen size). Any other unusual behavior was noted.

The shopper was intercepted immediately after she had placed the item in her cart. The interviewer pointed to an Ohio State lapel label and said "Excuse me, I am from Ohio State University. May I ask you a few questions about the item you just chose. In return as a token of our appreciation I would like to give you this one dollar bill." The refusal rate was less than 1% and was primarily because the shopper was in a hurry. The interviewer then asked: "Off the top of your head, without checking, can you tell me what the price is of the (coffee/margarine/cereal/toothpaste) that you just chose?" This meant that the shopper was asked to recall the price of the item just selected within about 15 seconds of having placed the item in her shopping basket.

Any answer was coded as to its immediacy and was followed by an open ended question: "How did you know that price?" If no answer to the latter was offered, respondents were prompted with a list of possibilities. Respondents were then asked if the price of the just selected item was lower than the usual price. A positive answer to that question was followed by asking how much the price was below the regular price. Subjects were then asked several questions about their recognition and usage of shelf price information including the meaning of the shelf tags which were color-coded by the store chain to indicate whether the price was a regular price, an everyday low price, or an even further discounted special price. Respondents were then asked about their shopping patterns, usual use of price information and why they did or did not check the price of the item in question. The full interview took some 4-5 minutes. On completion the interviewer recorded the actual price of the item, whether it was on special and the amount of any price reduction.

Four products were chosen after discussion with the supermarket chains's regional marketing and advertising managers. They were meant to be representative of relatively low (coffee and toothpaste) and high (margarine and cold cereal) turnover products and infrequently (cold cereal and toothpaste) and frequently (margarine and coffee) promoted product groups. Very low turnover items (such as shoe polish) were not chosen because we expected that Price recall accuracy for such items would be very poor and a very long time would be required to gather a sample size of 50 observations per product per store. An attempt was made to balance the time of observation according to peak shopping times so that the sample reflected the same proportions of shoppers by time period as can be observed shopping over the week in the stores. Effectively this meant that very few observations were made through Monday-Wednesday and most observations were made in the afternoon and early evening on Thursday and Friday and morning and afternoon on Saturday and Sunday. Store were chosen to represent different extremes of demographic composition of customers and store display of price information.

Except for measures of purchase frequency and use of coupons, we did not ask for or record any demographic information. In hindsight this was perhaps a mistake but we faced store imposed concerns over the inconvenience to the shopper of undertaking an extended personal interview at the point of purchase. In the future we would recommend giving the subject a questionnaire to complete at a later date and drop mail to receive a further gift. Apart from some fairly predictable effects of age and extremes of income. previous studies have found demographics not to be very powerful determinants of instore, supermarket shopping (e.g., Zeithaml and Fuerst 1983; Zeithaml 1984) and since our intent was to focus on product effects this also discouraged us from asking a battery of demographic and psychographic questions.

DISCUSSION

We have implemented and analyzed the results of a study using the procedure described above. The method appears to work quite well. Subjects seem to be relatively free from attempts to present themselves as more rational than actual. The few seconds interviewing between when the respondent placed the item in the shopping cart and when the interviewer asked the first question about the price of the chosen item all but eliminated memory loss since the purchase as a problem. Analysis of answers to the several questions in addition to the ones about knowledge of the price paid enabled several insights about the antecedents of price knowledge and how it relates to other aspects of information processing (Dickson and Sawyer 1985).

This method could be improved. One aspect that went unmeasured in this study was the possession of coupons for chosen items and the intention to use them. Demographics and other unmeasured characteristics of the shoppers might be best assessed in a followup mail questionnaire. Finally, this measurement procedure might be combined with manipulations of such marketing variables as coupons, featured item advertising, product displays, shelf items and flags, and price specials.

REFERENCES

Allen, J. W., G. D. Harrell and M. D. Hutt (1976), Price Awareness Study, Washington D. C.: The Food Marketing Institute.

Barker, Roger (1965), "Explorations in Ecological Psychology," American Psychologist, 20, 1-14.

Dickson, Peter R. and Alan G. Sawyer (1985), "Point of Purchase Behavior and Price Perceptions of Supermarket Shoppers," Working Paper, Marketing Science Institute.

Dietrich, Robert (1977), "Poor Price-Quiz Scores Give Shoppers no Cause for Price," Progressive Grocer, January. 33

Gabor, Andre and C. W. J. Granger (1961), "On the Price Consciousness of Consumers," Applied Statistics, 10, November, 170-188.

Progressive Grocer (1964), "How Much do Customers Know About Retail Prices?" Progressive Grocer, February, 104-106.

Progressive Grocer (1975a), "Then and Now: Shopping Behavior 10 Years Apart,"Progressive Grocer, October, 37-41.

Progressive Grocer (1975b), "What Customers Say About Scanning Without Prices,"Progressive Grocer, December, 58.

Progressive Grocer (1981), "What Shoppers Want Now," Progressive Grocer, September, 136-137.

Progressive Grocer (1982), "Shopping Habits: Prices Aren't, Afterall Everything," Progressive Grocer, April, 161-166

Wells, William D. and Leonard A. Lo Sciuto (1966) "Direct Observation of Purchasing Behavior," Journal of Marketing Research, 3, August, 227-233.

Zbytniewski, Jo-Ann (1979), "How Do Shoppers Choose a Supermarket,"Progressive Grocer, August, 105-207.

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

Zeithaml, Valarie A. (1984), "The New Demographics and Market Fragmentation," Journal of Marketing, 49.3, Summer, 64-75.

Zeithaml, Valarie A. and Fuest, William L. (1983), "Age Differences in Response to Grocery Store Price Information," Journal of Consumer Affairs, Winter.

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