Advances in Consumer Research Volume 1, 1974 Pages 29-37
CONSUMER PERCEPTIONS: OVERCHOICE IN THE MARKET PLACE
Robert B. Settle, California State University, San Diego
Linda L. Golden, University of Florida
In a free enterprise economy the allocation of resources and the distribution of goods are "regulated" by consumer choices in a free market. In order for the market to operate effectively, the consumer must have an adequate number of choice alternatives. From the passage of the Sherman Anti-Trust Act to the present day, Congress has passed numerous acts designed to foster competition and ultimately to ensure adequate consumer choice. The basic philosophy underlying this legislation and the regulatory bodies created to enforce them has been that the greater number of choice alternatives the more effectively the system will operate.
In more recent years, however, there has been some recognition that the individual is limited by time and ability in the number of alternatives from which he can choose effectively. For example, the Truth in Packaging legislation enacted in 1966 seeks to limit the number of choices of package sizes for a wide variety of goods. This objective implies both a recognition of the costs incurred by the consumer (and society) in seeking and analyzing information during the purchase decision process, as well as the costs associated with the dissatisfaction created by the elimination of some alternatives. The legislation and the mechanisms that it initiated were designed to create some balance between the costs of too few and too many choices of package sizes.
In Future Shock, Alvin Toffler devotes a chapter to "The Origins of Overchoice"(l). He contends that most social critics and observers fear under choice and the homogenization of the consumer because they base their predictions on the trend toward standardization during the transition from a craft economy to industrialization. His basic thesis is that this is an illegitimate extrapolation into the future, that technological changes result in discontinuities, and that the super-industrial revolution will result in extreme overchoice. Toffler suggests that, "Ironically, the people of the future may suffer not from an absence of choice, but from a paralyzing surfeit of it" (p. 264). Citing the trend toward the introduction of more and more varieties of a given product by one company, he suggests, "Two factors encourage this trend: first, consumers have more money to lavish on their specialized wants; second, and even more important, as the technology becomes more sophisticated, the cost of introducing variations declines" (p. 266). Rather than imposing standardization, Toffler feels that automation ". . . frees the patch to endless mind-numbing diversity." (p. 266)
This study addresses itself to the question of whether or not overchoice does, in fact, exist today. More specifically, two types of overchoice are examined: perceived overchoice--the degree to which the estimated number of choice alternatives exceeds the ideal number, and actual overchoice--the degree to which the actual number of choice alternatives available in a modern supermarket exceeds the ideal number. If perceived and actual overchoice can be shown to exist in today's supermarket, in terms of current consumer's preferences and estimates, this would constitute support for the overchoice hypothesis of Toffler.
THE FIELD SURVEY
To determine if overchoice currently exists, three types of information are needed concerning a sample of typical supermarket items: the ideal number of choice alternatives, the estimated number now available at the respondent's favorite supermarket and the actual number available. Twenty items, shown in Table I, were selected for the study. Meats, produce, frozen foods, canned goods, and non-grocery items were included in the list to provide a wide range of products. Some, such as breakfast cereal, provided a very large number of choice alternatives, while others, such as eating apples, provided only a few. All items were thought to be very likely to appear on the typical family shopping list, so that respondents would be more likely to have had some experience in buying every product, and this proved to be the case.
The data were obtained from a field survey with personal interviews of housewives in their homes. A questionnaire was constructed which listed the twenty items. After determining the respondent's favorite store, she was first asked, "Ideally how many different brands or types of -------would you like to have available at your supermarket, from which you could choose?" After obtaining the ideal number for the first item, sliced bacon, which was merely a filler item to avoid bias on the initial item, the respondent was cautioned, "Remember now, we are only interested in the number of brands or types, but not the number of different sizes that might be available." She was then questioned in turn about each item. When this list was completed, the respondent was asked, 'Now, how many different brands or types of ----do you think are now available at your favorite store?" Again, the respondent went through the list. After completing it, data were obtained on the occupation of the head of the household, the last year of school the housewife completed, the year she was born, and the approximate annual family income to the nearest thousands of dollars. The respondent's name and address were recorded to permit check-back on the field interviewers.
Four supermarkets in upper middle class neighborhoods of San Diego were chosen for the study. Eighteen student interviewers were instructed, trained and rehearsed in conducting personal interviews, and were assigned to one store area. A "convenience" sample of respondents was adequate for the study, so interviewers began on a residential street near the store, and called on each single family dwelling consecutively on the block until a quote of 10 interviews was obtained. A total of 174 usable interviews was obtained.
The actual number of choice alternatives for each store was obtained when three different enumerators independently counted the different brands and types of each product at each store. Wide discrepancies occurred on two items at one store, and these were reconciled when the definitions were classified and the items were recounted by other enumerators. In general, few definitional problems were encountered in either the actual counting or the interviewing of consumers.
THE SURVEY RESULTS
The mean estimated and ideal number of choice alternatives and the percent of perceived overchoice is shown in Table I. The estimated number exceeded the ideal number in all but four cases, indicating perceived overchoice was prevalent. The significance of the differences in the two mean vectors, for estimated and ideal numbers of alternatives, was tested with the Mahalanobis D2, and the probability of such difference occurring by chance proved to be less than .05. This statistic can be considered the multivariate generalization of the simple t-test. It comprehends intercorrelations of items and permits heterogeneity of variance in the items.
MEAN ESTIMATED AND IDEAL NUMBER OF CHOICE ALTERNATIVES AND PERCENT OF PERCEIVED OVERCHOICE
The percentage of perceived overchoice was calculated by subtracting the ideal number from the estimated number and then dividing this difference by the ideal number, for each individual datum and taking the mean for these percentages for the 174 respondents to yield the column shown in Table I. Thus, the percentage overchoice does not necessarily agree with the percent obtained by dividing the mean ideal by the mean difference.
While the Mahalanobis statistic indicated the two vectors of means were significantly different, the statistic says nothing about the individual items which may or may not be significantly different. To examine the items individually, each was submitted to a simple t-test. In the calculations of the t-values, the covariance of the items was used to adjust for intercorrelations. Tn addition, the alpha level received a Bernoulli adjustment to compensate for the fact that 20 individual tests were being made, rather than a single test. Table I indicates that 13 of the 20 items were significantly different, and none of the four items showing under choice proved to have significant differences. The conclusion appears to be that the consumer does perceive overchoice to exist in the market.
The average actual and ideal numbers of choice alternatives and the percentages of actual overchoice are shown in Table 2. All differences were positive, indicating overchoice in the case of every item. In general, differences were much larger than those between ideal and estimated numbers. The Mahalanobis D2 was much larger, and the associated F-statistic highly significant. Individual t-values were significant in every case but one, for frozen strawberries. The percentage of actual overchoice ranged from a low of 35% for that item to a high of 940% for cake mix. This could be interpreted to mean that approximately nine times as many choice alternatives actually existed for cake mix as the consumer considered ideal.
Stepwise multiple regression was used to measure the relationship between perceived overchoice and the four demographic characteristics: age, education, occupation, and income. A summary of the results is depicted in Table 3. Age is expressed in years, and education according to the last year of school completed. Occupation of the head of the household was divided into four categories: professional, administrative and clerical, skilled & semi-skilled labor, and unskilled or common labor. Family income was Riven to the nearest thousands of dollars.
The amount of perceived overchoice explained only 5% of the variance in age, with two items positively related and two negatively. Only four items were entered in the equation because no others provided an F-ratio to enter greater than one (the cut-off value), and none of the four items proved to have a significant F-ratio to enter. The F to enter can be regarded as a measure of the significance of the increase in the explained variance due to the addition of that particular variable. In general, the results indicate that perceived overchoice is not systematically related to ate.
The amount of overchoice explained about 25% of the variance in education, and three items, margarine, eating apples and aluminum foil had a significant F-value to enter. Overchoice in margarine had a positive relationship to education, while the other two were negatively related, indicating no generalizations concerning overchoice and education are permissible.
MEAN ACTUAL AND IDEAL NUMBER OF CHOICE ALTERNATIVES AND PERCENT OF ACTUAL OVERCHOICE
Margarine overchoice provided a significant negative relationship with occupation, while laundry detergent and cake mix also had a significant F-value to enter. In total, the regression of perceived overchoice on the demographics does not yield a clear picture of a systematic relationship. The regression of actual overchoice on demographics is summarized in Table 4. In general, a much higher percent of variance in the four demographics is explained by actual overchoice than was explained by perceived overchoice. The parity of the relationships is again mixed, however examination of individual items appears to correspond to expectations in some cases. For example, the degree of overchoice found in facial tissue is higher for Older people, and lower for the better educated, higher status occupations, and upper income respondents. Frozen strawberries has a similar pattern for age, occupation and income. Analysis of these patterns provides only very limited and inconclusive support for the hypothesis that overchoice is a systematic function of the demographic variables cited. Given the results for both perceived and actual overchoice, it seems that the relationship with demographic characteristics is predominately item specific. Probably differential preference patterns, rates of purchase and use, and brand loyalty importantly affect the relationship.
SUMMARY OF STEPWISE MULTIPLE REGRESSION OF PERCENT OF PERCEIVED OVERCHOICE ON DEMOGRAPHIC CHARACTERISTICS
SUMMARY OF STEPWISE MULTIPLE REGRESSION OF PERCENT OF ACTUAL OVERCHOICE ON DEMOGRAPHIC CHARACTERISTICS
The distributions of response on the demographic characteristics revealed a relatively homogeneous upper and middle class respondent group. Replication of the study over a wider range of demographic characteristics might be expected to yield more significant and pervasive differences in overchoice.
Summary of Results
The results of this study may be summarized as follows:
1. Both perceived and actual overchoice exist in the marketplace now.
2. Actual overchoice far exceeds perceived overchoice.
3. Overchoice appears to exist for a wide variety of supermarket items.
4. The higher the number of actual choice alternatives, the greater the percentage of perceived overchoice and actual overchoice.
5. Generalizing over all items studies, overchoice is not systematically related to age, education, occupation, or income of the respondent.
6. For specific items, overchoice is systematically related to the demographic characteristics.
IMPLICATIONS OF THE STUDY
The study measured the degree of under choice or overchoice, but it did not comprehend the amount of negative effects associated with under choice or overchoice. That is, overchoice conditions may cause confusion in the consumer's mind and inefficiency in the buying process. To the degree that this negative effect is present, the free market mechanism will operate less efficiently. It seems likely that there would be more negative effects associated with under choice than there would be with overchoice at any given percentage level. Despite this asymmetry, overchoice conditions should provoke concern, because they prove to be more prevalent.
With the trend toward the diversification and proliferation of products and services, and given the overchoice conditions revealed by this study, it appears that the multiplicity of choice alternatives may become an important problem in the near future. There are several negative consequences associated with overchoice for both the consumer and the marketer. From the marketer's point of view, the necessity for stocking and handling "superfluous" brands and types of each product will increase costs. Inventory levels would increase directly with the increased number of types or brands of products. In addition, the likelihood of stock-outs for any particular brand or type would be increased, and the physical and clerical handling costs would be expected to increase when additional brands and types are offered for sale.
The existence of overchoice in the marketplace implies that the marketer, and especially the retailer, might avoid these costs by measuring the optimum number of choice alternatives for the consumer in various product categories. When positioning a new product, the manufacturer might benefit from seeking areas where under choice or low overchoice conditions exist, rather than positioning a product such that it would increase the alternatives in an already crowded area. By the same token, a manufacturer whose products are already on the market might be able to "defend" his market chair by increasing the number of brands or types of products that he offers.
From the consumer's point of view, the cost of overchoice can be defined in terms of increased effort and confusion in the buying decision process. The time and effort required to obtain the necessary information about each brand or type of product would obviously increase with the number of choice alternatives. It may be that under extreme overchoice conditions consumers may abandon the decision process in favor of a relatively simple decision rule such as to buy the least expensive or remain completely loyal to one particular brand. In the extreme for low priced items with high product homogeneity, the consumer may revert to making a selection strictly on a stochastic basis.
The confusion resulting from overchoice may, in fact, change the prescribed role of some retail outlets by the incorporation of an additional characteristic. Consumers, if distressed and confused by the large number of choice alternatives, may depend on the retail outlet to act as their agent in the selection of products. In an unconscious effort to limit their alternatives, some consumers may restrict their buying behavior to stores offering a more limited number of alternatives. For example, the supermarket may offer fifty brands or types of crackers while a convenience store or a "Ma and Pa" store may carry only five. Those who do tend to prefer a limited number of choice alternatives may frequent such stores for a substantial part of their total purchase mix.
Toffler, A. Future Shock. New York: Random House, 1970, 263-283.