A Preliminary study of double jeopardy in selected retailers.

Pleshko, Larry P. ; Souiden, Nizar

ABSTRACT

The authors investigate the "double jeopardy" (DJ) concept in the domain of retailing. The authors show that DJ is moderately evident within health clubs and medical clinics, but most likely does not exist within convenience store retailers. The authors attempt to test which of three alternative explanations for market share rankings is supported: (1) a familiarity effect, (2) an experience effect, or (3) a design effect. No evidence is found to support the design effect. However, either of the familiarity effect or the experience effect may be a viable explanation for market share (and DJ), depending on the type of retailer. For medical clinics, the familiarity effect may be the source, while for health clubs it may be the experience effect. The authors suggest two mitigating factors in the findings related to DJ across each of the retail types: (i) the strength of brand affiliation/names or (ii) the levels of inherent consumer involvement or effort.

INTRODUCTION

A company needs to focus at least a proportion of marketing efforts on the development, maintenance, or enhancement of customer loyalty (Dick & Basu 1994). This emphasis is important because a company with a large number of brand loyal buyers will be more secure in its markets and should have a higher market share than other firms without this vital customer asset (Raj, 1985; Robinson, 1979; Smith & Basu, 2002). Competitors are at a disadvantage when some firms have a larger number of brand loyal buyers than they have. The many advantages include: a greater response to advertising (Raj, 1982), larger purchase quantities per occasion (Tellis, 1988), and reduced marketing costs (Rosenberg & Czepial, 1983). The advantages garnered from loyalty are especially important since, as markets become more mature, increases in share become more expensive and improvements in the loyalty base might be a viable means of increasing and maintaining share (Gounaris & Stathakopoulos, 2004).

The fact that competitive markets oftentimes exhibit similar market structure characteristics (market share), which in turn was found to be correlated with the number of brand loyal buyers, was first noticed by McPhee (1963). This observance that brands with large market shares usually had the most brand loyal buyers (and vice versa) was termed "double jeopardy" (DJ) because it seemed unfair for smaller brands to suffer in both ways. Previous research related to DJ suggests its' applicability to a variety of consumer brands and setting. Additionally, some consumerspecific variables will exhibit a similar relationship with market share as consumer loyalty (Ehrenberg et al., 1990).

This study applies the aspects of DJ to the area of retailing, investigating convenience stores, health clubs, and medical clinics. The dynamic evolution of the 'quick-stop' shopping from small neighborhood grocery and service stations to today's multi-purpose convenience stores located in nearly every city block has transformed the way people all around the world shop for small repetitively purchased items. A continuing worldwide trend towards healthier lifestyles has spawned the boom in businesses providing gym and exercise services to maintain fitness, as well as to lose weight. Also, the healthcare industry has changed from an emphasis on family doctors to more versatile medical clinics over the past twenty years, altering the way most of us get medical treatment. Each type of retailer faces highly competitive environments in their markets and the establishment of a large and loyal customer base is vital for long-term survival. With estimates regarding the number of 'truly' brand loyal buyers for these and related consumer retailers hovering around 25% (Pleshko & Heiens, 1996, 1997), the presence or absence of the double jeopardy phenomenon in these retail segments might be critical to the retailers' decision making. It would be difficult for small-share firms to grow and show long term success with a strong DJ effect evident.

The authors first attempt to identify whether the DJ phenomenon is evident in each of the three retailer areas. This is tested by analyzing the relationship of loyalty to market share. The authors then attempt to test three alternative explanations for the DJ effect. The alternative theories are tested by investigating other factors which may be related to market share: product attributes and consumer choice sets. The article begins with a review of the concept of "double jeopardy" and the related constructs. A description of the data collection, measurement issues, analysis, and discussion follow and conclude the study.

THE DOUBLE JEOPARDY PHENOMENON

A firm's long-term success depends on both its ability to attract customers and its capability to retain these customers (McDowell & Dick, 2001). Jones (1990) points to this fact by stating that manufacturers should regard sales volume and market share as keys to the future, given that both involve sources of repeat business and scale economies. McDowell and Dick (2001) rightfully offer that a brand's market performance is driven by both the number of individuals buying a particular brand and the frequency of repeat purchases from these customers. The ability to manage these two factors determines the extent to which a firm maintains and sustains its customer base, as well as its market share. Indeed, Robinson (1979) and Raj (1985) state that the larger the number of loyal customers, the more secure will be the brand's market share. Therefore, as a priority, all companies must find ways to attract new customers to an existing user base and to retain these buyers over the long term. So, it must be that firms constantly battle with competitors to maintain or increase both the number of buyers and the loyalty of these customers. When a firm fails to hold a strong relative competitive position, it runs the risk of a widespread phenomenon called "double jeopardy" (DJ).

Double jeopardy is broadly characterized as a phenomenon whereby small-share brands attract somewhat fewer loyal consumers, who tend to buy the brand in smaller quantities, while larger-share brands are purchased more often by customers who exhibit more loyalty (Ehrenberg & Goodhardt, 2002; Badinger & Robinson, 1997; Donthu, 1994; Martin, 1973; Michael & Smith, 1999). Thus, less popular brands are punished twice for being small. That is, (i) they have fewer buyers who show (ii) less loyalty to the brands they buy. McPhee (1963) explained that DJ occurs when consumers select between two brands of equal merit, one having a larger market share and the other having a smaller market share. This does not signify a weak small brand or a strong large brand. Rather, it reveals that the smaller share brand is less popular than the larger share brand for some reason (Ehrenberg & Goodhardt, 2002; Ehrenberg, et al., 1990).

For any retail business, the firms must determine if DJ is an issue in their industry. Secondly, if DJ is evident, then small-share retailers must assess the causes of their brands' share woes and implement strategies to make their products better known and more often used. But managers must be warned that an increase in marketing efforts will not always have the preferred results. Marketing inputs do not greatly increase loyalty over an extended period unless the brand's penetration is significantly increased. Even then, the marketing mix factors rarely lead to important differences in brand loyalty (Ehrenberg & Goodhardt, 2002). However, marketing factors (such as product quality, attributes, or price) result in sales variations in the immediate term, which if maintained will eventually influence the market share (and the DJ) pattern (Ehrenberg & Goodhardt, 2002).

ALTERNATIVE EXPLANATIONS FOR DOUBLY JEOPARDY

Although long established, the DJ phenomenon has a variety of issues as yet unsettled (Ehrenberg & Goodhardt, 2002). For instance, though previous research has found an obvious relationship between brand share and loyalty, whether loyalty is a cause or a result of high share remains unclear (McDowell & Dick, 2001). Likewise, previous research has focused mainly on the issue of DJ for the product brand while the relevancy for the company brand, the retail brand, or the service brand is rarely discussed. Thus, the main issue--why two equally regarded brands or products differ in their relative shares of the market?--is still not truly defined in most settings. Three possibilities have been drawn from the literature as an explanation for double jeopardy: (i) a familiarity effect (c.f. Ehrenberg et al., 1990; McPhee, 1963), (ii) an experience effect (c.f. Tellis 1988; Raj, 1982; Brown & Wildt, 1992; Johnson & Lehman 1997; Narayana & Markin, 1975; Nedungadi, 1990; Pleshko et al., 1997), and (iii) a design effect (c.f. Ehrenberg et al., 1990; Keith, 1960). These three competing explanations are outlined below.

The experience effect suggests that buyers purchase those brands with which they have previous experiences. In other words, as buyers gain more experience with a brand, that brand is more likely to be purchased in the future. It is believed that longer-term or more permanent attitudes are probably not formed in most instances until after the buyer has used or tried the product (Vaughn, 1980). Also, retailers or brands can be grouped by a target market's past experience with the brand or, in other words, by the outcomes of a buyer's decision process as maintained in memory. So, as buyers gain more experience with a product-market, the category structures in memory become more detailed and developed leading to a larger impact of these remembered experiences on future consumer choices (Alba & Hutchinson, 1987). It might be said that many choices are determined, not by immediate stimuli and attributes, but rather by this previous knowledge and experience, stored as groupings of brands, called choice sets (Narayana & Markin, 1975; Spiggle & Sewall, 1987). Regarding experience, the relevant choice set is the reconsideration set: a grouping of brands/retailers that the buyer would consider in the future when making a purchase from a given product-market (Pleshko et al., 1997). Some might maintain that the evoked set is the appropriate indicator of past experience. However, the evoked set is comprised of brands that the buyer considers at a specific moment, and might not contain all the relevant products, due to situational influences. Therefore, if the 'experience effect' is viable, we would expect firms with larger market shares to also belong to more reconsideration sets. In other words, the customers are more likely to reconsider buying from those stores in the future which have the largest shares.

The familiarity effect suggests that one brand is more popular than another. Buyers will, for whatever reason select those brands with which they are familiar, regardless of whether they have knowledge or experience with the product or not. On any given day the number of imagebuilding or popularity-advertisements shown on television will be large, as the firms must believe consumers choose those brands that come to mind quickly. Repetitive advertising, with the purpose of quickly generating awareness and maybe image, is a useful tool to the firm in generating trial. However, the main result, once awareness is achieved, is to reinforce existing knowledge and beliefs of the consumer target towards the product (Ehrenberg, 1982). Thus, greater number of exposure to the brand should again result in a stronger, more detailed memory structure, with the result being the addition of the brand into a buyer's choice set structure, as outlined above. The learning through repeated exposure may occur via a number of ways: low-involvement learning (Krugman, 1965), exposure effects (Reibstein & Farris, 1995; Obermiller, 1985), or reasoning (Sheppard et al., 1988) to mention a few. The relevant choice set, in this instance, is the awareness-set. The awareness set includes all those brands for which the buyer is familiar or of which he is aware (Narayana & Markin, 1975). Therefore, if the 'familiarity effect' is viable, we would expect those firms with the larger market shares to also have higher levels of awareness, as indicated by inclusion in more buyer awareness sets. In other words, people will be most familiar with large share firms and they will most often select from these large share firms.

The design effect suggests that buyers prefer brands which have attributes that best match their wants because these attributes will best deliver the benefits for which the buyer is aiming. As the basic premise in marketing, the marketing concept, suggests: customers will most often select those firms' products whose design most closely matches their wants (Keith, 1960; Trout & Ries, 1972; Webster, 1993). Thus, through proper design and positioning, a firm should be able to develop over time a group of loyal buyers who prefer the firm's offering over those of another, less well designed competitor, on a variety of salient attributes. Brand attitudes, as often measured by attribute or characteristic evaluations, explain to a large extent the differences in success among brands (Badinger & Robinson, 1997; Farr & Hollis, 1997). Michael et al. (1999) report that the double jeopardy patterns for brand-name products emerge because larger, more-visible brands are closely associated with positive attributes by consumers. Following this line, Castleberry & Ehrenberg (1990) note that brand beliefs are positively correlated with usage or purchase frequency. They also find that consumers' opinions about a brand's given attributes (such as, "reasonably priced" or "high quality") tend to vary with the frequency with which they have bought the brand in the past. Also, users of a brand are more familiar with that brand and are thus more likely to have positive beliefs about its' attributes (Barnard & Ehrenberg, 1990). For this study it is not important whether the buyer has chosen the brand because of the specific attributes or whether those attributes have become more relevant with experience. Under the 'design effect' the buyer has made a selection based on the dominance of certain attributes of the chosen brand, and until another brand is more dominant (or variety-seeking is evoked), the buyer will continue to select the best-designed brand. Therefore, if the design effect is viable, we would expect firms with larger market shares to also have the higher rated attributes.

DATA COLLECTION

The data for the current study are gathered from a buyer group in a large university town in the southeastern USA. The sampling frame is comprised of undergraduate business students, a group of consumers who are frequent users of each of these types of retail businesses. Information is accepted only from consumers who buy from each type of retail outlets. Thus, individuals who do not use health clubs are excluded from the study of health clubs. The data are from self-administered questionnaires collected from three variations of the same research instrument for each type of retailer. The purpose of the instrument variations is to minimize any ordering effects in the collection of the data. Thus, there are nine versions of the questionnaire, three for each retailer type. Nine classes are selected for inclusion in the study from the offering at the university. Each class is assigned to one specific retailer type and the three different questionnaire versions are administered within each class. This process results in the following number of usable respondents: health clubs: 81, convenience stores: 90, and medical clinics: 71. As the non-users were screened during the sampling process and all those selected responded, few unusable surveys were found.

Many stores in each retailer category are included in the study. The retailers are identified by speaking with the buyers and looking through the yellow pages to locate outlets within an acceptable area of town. An 'others' category was included to catch those retailers not specifically listed on the questionnaire. The retailer types themselves are selected for two reasons. First, it was necessary to find a type of business that was used by the target group under study: students, who are oftentimes heavy users, especially in this university town, of the three types of retailers. Second, it was important to select different types of retailers in order to test the double jeopardy phenomenon across a variety of environments, as we would expect to find differences based on product-classes or retailer types (Chaudhuri & Holbrook, 2002). Thus, the Murphy and Enis (1986) taxonomy was used as a guide. They suggested classification of products based on risk perceptions/efforts undertaken and preferences of consumers. Our three retailer types vary across the risk/effortpreference details as follows: convenience stores--low risk/effort with few preferences (a convenience product), health clubs--high risk/effort with definite preferences (a specialty product), and medical clinics--high risk/effort with no true preferences initially (shopping product).

For health clubs there are sixteen original clubs included on the questionnaire. Eleven of these clubs are eliminated after data collection due to small numbers of users or small market shares. It seems that the student market is focused on only a few health clubs. This resulted in the five remaining health clubs for inclusion in the DJ analyses: club#6, club#9, club#3, club#10, and club#13, all with market shares above 5%.

For convenience stores there are twelve original outlets in the vicinity of the market that are included on the questionnaire. Only four of these stores are eliminated after data collection due to small numbers of users or small market shares. This resulted in the eight remaining convenience stores for inclusion in the DJ analyses: store#8, store#10, store#4, store#6, store#3, store#5, store#2, and store#9., all with market shares above 5%.

For medical clinics there are twelve clinics in the general area that are included on the questionnaire. Six of these clinics are eliminated after data collection due to either a small number of users or a small market share. This resulted in the six remaining medical clinics that are included in the final analysis. In order of market share from lowest to highest, the clinics are: clinic#1, clinic#4, clinic#6, clinic#8, clinic#2, and clinic#7. All six of the clinics have market shares above 3%.

MEASUREMENT

The study includes a variety of constructs for market share, loyalty and usage, choice sets, and salient store attributes. The overall indicators for the single market share indicator, the six loyalty and usage indicators, and the three choice sets indicators are derived by summing across multiple variable components for each indicator. These overall measures are all percentages and are consistent across the store types. The indicators for the salient store characteristics are specific to the store type, but have many commonalities across the types. The attributes were derived through focus group interviews with users of each store type. Most of the variables under study are percentages, except for the attributes which are mean ratings. Details of the indicators are shown in Table1, Table 2, and Table 3 4 for each of the relevant constructs. The variables re described below.

Market-Share is defined as the visits (uses) for store 'A' divided by the total visits (uses) for all stores. This is the primary variable in the study, as all other constructs are compared to market share. It is calculated for each store in each retail type. The respondents are asked how many 'times' they visit each retailer per month. These 'times' are summed for each store and overall within each store type. Thus, share A = (times for store A)/ (summation of times for stores A, B, C, N). The possible range of market share is from one to ninety-nine percent, but is not that high because many of the competitors all have significant shares.

Loyalty-%-of-use is defined as the % of total times (uses) the respondent uses each store if they are users of that store. This is one of three loyalty variables that are used to determine if DJ is evident. It is calculated for each respondent for each store used. Then an overall sample average percentage is calculated. Thus, for respondent X who uses stores A and B: L%A = times A/(timesA + timesB ... + timesN) and L%B = times B/(timesA + timesB ... + timesN). The possible range is from one to ninety-nine percent.

Most-used-% is defined as the percentage that each store is used as primary store in the category. This is the second of three loyalty variables that are used to determine if DJ is evident. Respondents indicate the number of 'times' they use each store. the store is the most used for a respondent when the largest number of times is indicated. Thus, for respondent X, if times A > timesB, timesC. ..., timesN, then storeA is assigned to respondent X as most used store. The indicator is a summation for each store of those respondents most using each store. The possible range is from one to ninety-nine percent.

Last-purchase-% is defined as the store that the respondent last purchased from in the category. This is one of three loyalty variables that are used to determine if DJ is evident. The indicator is a percentage calculated by summing, for each store, all the respondents who indicated they bought from that store last. Then this summation is divided by the total respondents to get the indicator. Thus, Last%A = (total purchased A last)/ (total respondents). The possible range is from one to ninety-nine percent.

Awareness-set is defined as the percentage of respondents who are aware of a store. This is included in order to test the familiarity effect. Respondents are asked to check boxes next to the stores of which they are aware. The indicator is a percentage calculated by summing, for each store, all the respondents who indicated they are aware of the store. Then this summation is divided by the total respondents to get the indicator. Thus, awarenumA = (total aware of A)/(total respondents). The possible range is from zero to one hundred percent.

Reconsider-set is defined as the percentage of respondents who would consider buying from a specific store in the future. This is included in order to test for the experience effect. Respondents are asked to check boxes next to the stores of which they would consider using again. The indicator is a percentage calculated by summing, for each store, all the respondents who indicated they would consider a store again. Then this summation is divided by the total respondents to get the indicator. Thus, reconsidernumA = (total reconsider A)/(total respondents). The possible range is from zero to one hundred percent.

Salient Attributes are ratings on single variables from one to seven anchored by 'not important at all' to 'very important' for each attribute. The attribute ratings are included to test for the design effect. For the health clubs, these include--good location, convenient hours, individual attention, friendly employees, cleanliness, good atmosphere, fair prices, variety of services, variety of equipment, clientele, no crowds, and employee quality. For medical clinics, these include--good location, fast service, convenient hours, quality of service, special services, doctor quality, friendly employees, cleanliness, atmosphere, friendly doctors, and fair prices. For convenience stores, these include--good location, fast service, convenient hours, easy access, friendly employees, cleanliness, atmosphere, fair prices.

ANALYSIS/RESULTS

The Spearman (1904) rank correlation coefficient is used to analyze the association between market share and the variables under investigation. Spearman's test statistic, rho or "r", is calculated with data taken from 'n' pairs (Xi,Yi) of observations from the respondents on the same objects, the retail brand outlets. In this study, market share (Xi) makes up one of the observational items in the pair, while the variable under study (Yi) makes up the other item. The observations within each pair of variables is then ordered from smallest to largest and assigned the respective ranks from one to n, where n is the number of retailers in the category. The construct values, rankings, test statistics, and 'p'-values are shown for each construct of interest and each store in Tables 1, 2, and 3. For example, in Table 1 for health clubs, the calculated market share values for outlet#10 is twelve percent and outlet#2 is six percent. Note also that the rankings for these share values respective to the other outlets are four (highest rank is five) for outlet#10 and one (lowest rank) for outlet#2. Following the same methodology leads to the rankings values for all the variables shown in the tables. Ties are assigned the average ranking value. These "rankings pairs" are then used to calculate the test statistic. See the example which follows for a detailed explanation of the rankings pairs and the test statistic. Due to the limited power of the test statistic, the following cutoff points are established for the 'p'-values: (i) strong relationship--p </=.05, (ii) moderate relationship--.10 >/= p >.05, and (iii) weak relationship - .15 >/= p > .10.

To calculate the "r", the "rankings pairs" are compared: this would include market share and the other variable under study. The test statistic, rho, is calculated as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

In the equation, 'n' equals the number of paired rankings and 'd' equals the absolute differences between the rankings for each outlet: (Xi-Yi). In this study, the number of paired rankings is equal to the number of retailers in each category: five for health clubs, six for medical clinics, and eight for convenience stores. Again, referring to Table 1, the pair of ranks for outlet #9 regarding market share with loyalty-%-of-use is (2,3). Thus, d for the pair equals 1. The d values are calculated for all the paired ranking for each variable combination for that loyalty variable. The calculation of the statistic of association, "r" follows for our example.

From Table 1, looking at the loyalty-%-of-use variable we see that r=-.300 indicating an insignificant relationship between the order of rankings of the two variables. This is calculated as follows:

r= 1- 6[[(1-5).sup.2] + [(2-3).sup.2] + [(3-1).sup.2] + [(4-1).sup.2] +[(5-4).sup.2]]/5(25-1) = 1-6(26)/5(24)=-.300

The test statistic ranges between +1 (perfect positive association) and -1 (perfect negative association). In this study, two-tailed tests are performed, giving the general hypotheses for the paired variables: Ho: independently ranked pairs or Ha: related ranked pairs. Remember, all the analyses use market share as the comparison variable.

Initially, we determine if double jeopardy is evident in each of the three retailer types by comparing market share rankings with the rankings of the three loyalty indicators. The reader is referred to the top of Tables 1, 2, and 3 for the analyses related to these variables: loyalty%-of-use, loyalty-%-self-report, most-used-%, and last-purchase-%. For health clubs, two out of three of the loyalty indicators exhibits a significant relationship. Thus, moderate evidence is provided that the double jeopardy effect does exist in health club retailers. For convenience stores, none of the loyalty indicators exhibit a significant relationship to share rankings. Thus, evidence is insufficient to support a DJ effect in convenience store retailers. For medical clinics, two out of three loyalty indicators exhibits a significant relationship to share rankings. Thus, moderate evidence is provided that the double jeopardy effect does exist in medical clinic retailers. To summarize, the evidence supports DJ in medical clinics and health clubs, but not in convenience store retailers.

Next, the authors move to the primary purpose of the study: to pinpoint the better explanation for the market share rankings and the associated double jeopardy effect. The rankings order for the two choice sets and also for all the attributes are tested versus the market share rankings order in each type of retailer. The reader is referred to the tables for the relevant analyses. Remember, a significant relationship with attribute variables would support the design effect. A significant relationship with the awareness set would support the familiarity effect. A significant relationship with the reconsider set would support the experience effect.

For health clubs, the reconsider set is significant, but not the awareness set or any of the attribute rankings. For convenience stores, which did not exhibit a DJ effect at all, the reconsider set is significant, while the awareness set and the attribute rankings are not related to market share. Finally, for the medical clinics, the awareness set is significant, while the reconsider set and all but one of the attributes are insignificant. The one attribute relationship, out of thirty-two, might be found simply by chance, so it will be ignored. Therefore, to summarize the relationships to market share: the design effect is not supported in this study, as only one out of many tests is found to be significant. Evidence is provided for the familiarity effect in medical clinics. Evidence is provided for an experience effect in both health clubs and convenience stores.

DISCUSSION

The findings of this study reveal that the concept of double jeopardy (DJ) does apply to Health Club retailers and Medical Clinics, but not to Convenience Store retailers for the given market segment. Other research has found DJ to be strongly evident in fast-food outlets, as well (Pleshko et al., 2006). Thus, it appears that double jeopardy is an important factor for retailing management to consider in strategic decision making. The current study uses estimates of behavioral data to investigate DJ, while previous studies show the DJ effect to be more predominant when using attitudinal measures (Bandyopadhyay & Gupta, 2004). Therefore, the findings of the current study may be more relevant, since behavioral data is possibly a better indicator than is attitudinal data. Additionally, the inclusion of three loyalty indicators is a definite improvement over previous efforts to study double jeopardy, again providing more confidence in the conclusions of the study, as related to DJ.

The absence of a DJ effect in Convenience Store retailers may be due to either of (i) the low involvement nature of the stores or (ii) the lack of strong brand affiliation with these types of businesses (c.f. Murphy & Enis, 1986). If we consider other research, Fast-Food outlets tend also to be low involvement, but with the presence of strong brand names or affiliations, which are mostly absent in Convenience Store retailing by definition. Thus, the presence of strong brands or affiliation with these brands might explain the finding of DJ in the fast-food product-market and the absence of it in the convenience businesses (c.f. Pleshko et al., 2006). This reasoning would explain the presence of DJ in Health Clubs as well, since these service providers also tend to exhibit strong brand names in a given market. But that does not explain the presence of double jeopardy in Medical Clinics, businesses generally not known for their strong brand names or affiliations. However, consumer decisions pertaining to Medical Clinics are often higher involvement, which might also be the case with Health Clubs. Maybe the Health Club involvement is more productclass involvement, associated with long-term interest in healthy and active lifestyles, rather than the brand-decision involvement with Medical Clinics: increased effort associated with an important choice (c.f. Celsi & Olson, 1988). Therefore, personal relevance or involvement might also explain the DJ effects. This would be contrary to the findings of Lin and Chang (2003), who suggest the DJ relationship to be stronger in low involvement products (such as fast-food). Maybe the previous studies only investigated low involvement products with inherently strong brand names or affiliation.

The primary focus of the study was to determine which theory correctly explains the market share variations in the retailers under study: (i) the design effect, (ii) the familiarity effect, or (iii) the experience effect. The absence of attribute relationships to market share in all three retailer types reveals the design effect to be questionable as an explanation for double jeopardy. The finding does fit with previous research that showed consumers' beliefs about various brand attributes are not always associated with market share (Castleberry & Ehrenberg, 1990). In general, one would not expect market share to vary with all the attributes of a retailer type. However, if the design effect was relevant, we should have found a few correlations with those attributes described as salient. This was not the evident. The finding that attributes or retailer design is irrelevant to share seems a bit naive. Maybe this null finding has something to do with the fact that all the respondents were users of these retailers, and not new to the market. It might be possible that over time, as buyers gain more knowledge of the brands, that design is secondary to other factors.

The familiarity effect is found to be a plausible explanation for share variations, but only in Medical Clinics. Other research has shown the familiarity effect to also be evident in Fast-Food Outlets (Pleshko et al., 2006). We would expect larger share companies to promote more and place more emphasis on advertising because they have more money to spend. Due to these efforts, buyers are reminded of larger share brands more often than smaller share brands. This has been the basic explanation of DJ up to this time (McPhee, 1963). But the question remains as to the conditions where the familiarity effect is important? We might conclude, from this and other studies, that the familiarity effect is most relevant for those products considered to be either shopping products or preference products (Murphy & Enis, 1986). Shopping products, such as Medical Clinics, require buyers to choose among many unfamiliar brands usually without full information. In those instances, the buyer rightfully selects a famous brand, assuming that it is popular because the brand is acceptable to many people. On the other hand, preference products, such as Fast-Food, have buyers who repetitively purchase from a few well-known brands in a given market. In this instance, the buyers have much information to make an unimportant decision because they are very familiar with each brand from intense advertising and promotion efforts of the businesses.

The experience effect is also found to be a plausible explanation for market share variation, both in Health Clubs and in Convenience Stores. Other research has found this to be true also in Fast-Food Outlets (Pleshko et al., 2006). We would expect buyers to purchase those brands with which they have more satisfying experiences than other brands. Again, the conditions under which the experience effect works are unknown. From this and other studies, we might expect the experience effect to be relevant to either shopping products, preference products, or specialty products. Shopping products, such as Medical Clinics, require intense decision making pertaining to important outcomes. Once buyers gain experience with specific retailers or brands in these conditions, they are likely to continue to purchase those products which meet the purchase goals, rather than to continually reevaluate at every occasion. Specialty products, such as Health Clubs, usually are characterized by buyers who have made an important decision and, having found a satisfying product, tend to stick with that favorite brand. The buyers of preference products, such as fast-food or soda, would also be known to simplify decisions from previous experiences, easily choosing from a select group of brands depending on the situation.

The readers must wonder if the current findings are indicative of general tendencies or simply a characteristic of this limited study. Larger studies with more respondents taken over time are probably needed to truly identify the scope of the double jeopardy phenomenon. Future research might include both different target respondents as well as different product-markets. Since involvement and/or brand preference was suggested to be a relevant factor for DJ, future studies might investigate this more thoroughly. Similarly, future studies might be designed to provide a better test of the three competing theories under a variety of different circumstances. Additionally, future efforts might address these issues in cross-national studies to determine where and under what conditions the double jeopardy construct applies globally. Finally, the question of life cycle issues might also be relevant for this area of study. What are the characteristics of double jeopardy, and is it evident, at various stages of the industry life cycle.

REFERENCES

Alba, Joseph W. & J. Wesley Hutchinson (1987). Dimensions of Consumer Expertise. Journal of Consumer Research, 13, 411-454.

Badinger, Allan L. & Joel Robinson (1997). The Jeopardy in Double Jeopardy. Journal of Advertising Research, 37 (6), 37-50.

Bandyopadhyay, Subir & Kunal Gupta (2004). Comparing Double Jeopardy Effects at the Behavioral and Attitudinal Levels. Journal of Product and Brand Management, 13 (3), 180-191.

Barnard, Neil R. & Andrew S.C. Ehrenberg (1990). Robust Measures of Consumer Brand Beliefs. Journal of Marketing Research, 27 (11), 477-484.

Brown Juanita J. & Wildt Albert R. (1992). Consideration Set Measurement. Journal of the Academy of Marketing Science, 20, 235-243.

Castleberry, Stephen B. & Andrew S.C. Ehrenberg (1990). Brand Usage: A Factor in Consumer Beliefs. Marketing Research, 2 (2), 14-19.

Celsi, Richard L. & Jerry C. Olson (1988). The Role of Involvement in Attention and Comprehension Processes. Journal of Consumer Research, 15, 210-224.

Chaudhuri, Arjun & Morris B. Holbrook (2002). Product-Class Effects on Brand Commitment and Brand Outcomes: The Role of Brand trust and Brand Affect. Journal of Brand Management, 10 (1), 33-59.

Dick, Alan & Kunal Basu (1994). Customer Loyalty: Toward an Integrated Framework. Journal of the Academy of Marketing Science, 22 (Spring), 99-113.

Donthu, Naveen, (1994). Double Jeopardy in Television Program Choice. Journal of the Academy of Marketing Science, 22 (2), 180-186.

Ehrenberg, Andrew S.C. & Gerald J. Goodhardt (2002). Double Jeopardy Revisited, Again", Marketing Research, 14 (1), 40-42.

Ehrenberg, Andrew S.C., Gerald J. Goodhardt, & T. Patrick Barwise (1990). Double Jeopardy Revisited. Journal of Marketing, 54 (3), 82-91.

Farr, Andy & Nigel Hollis (1997). What Do You Want Your Brand To Be When It Grows Up: Big And Strong?. Journal of Advertising research, 37 (6), 23-37.

Gounaris, Spiros & Vlasis Stathakopoulos (2004). Antecedents and Consequences of Brand Loyalty: An Empirical Study. Journal of Brand Management, 11 (4), 283-307.

Johnson, Michael D. & Donald R. Lehmann (1997). Consumer Experience and Consideration Sets for Brands and Product Categories. Advances in Consumer Research, 24, ed. Brucks & MacInnis, Provo, Utah, Association for Consumer Research, 295-300.

Jones, Jeffrey P (1990). The Double Jeopardy of Sales Promotions. Harvard Business Review, Sep/Oct, Vol.68 Issue 5, 145-152.

Keith, Robert J. (1960). The Marketing Revolution. Journal of Marketing, Jan., 37.

Krugman, Herbert (1965). The Impact of Television Advertising: Learning without Involvement. Public Opinion Quarterly, 29, 349-356.

Lin, M.Y. & L.H. Chang (2003). Determinants of Habitual Behavior for National and Leading Brands in China. Journal of Product and Brand Management, 12 (2/3), 94-108.

Martin Jr, Claude R., (1973). The Theory of Double Jeopardy. Journal of the Academy of Marketing Science, 1 (2), 148153.

McDowell, Walter S. & Steven J. Dick (2001). Using TV Daypart 'Double Jeopardy Effects' to Boost Advertising Efficiency. Journal of Advertising Research, 41 (6), 43-51.

McPhee, William N. (1963). Formal Theories of Mass Behavior, New York: The Free Press.

Michael, Judd H. & Paul M. Smith (1999). The Theory of Double Jeopardy: An Example from a Forest Products Industry. Forest Products Journal, 49 (3), 21-26.

Murphy, Patrick E. & Ben M. Enis (1986). Classifying Products Strategically. Journal of Marketing, 50, 24-42.

Narayana, Chem L. & Rom J. Markin (1975). Consumer Behavior and Product Performance: An Alternative Conceptualization. Journal of Marketing, 39 (October), 1-6.

Nedungadi, P. (1990). Recall and Consumer Consideration Sets: Influencing Choice without Altering Brand Evaluations. Journal of Consumer Research, 17, 263-276.

Obermiller, Carl (1985). Varieties of Mere exposure: The Effects of Processing Style and Repetition on Affective Response. Journal of Consumer Research, 12, 17-29.

Pleshko, Larry, Richard Brand, & Joseph Cronin (1997). Consumer-Specific Determinants of the Size of Retail Choice Sets: An Empirical Study of Four Service Types. Journal of Services Marketing, 11 (1), 19-38.

Pleshko, Larry P. & Richard A. Heiens (1997). An Empirical Examination of Patient Loyalty: An Application of the Customer Loyalty Classification Framework to the Healthcare Industry. Journal of Customer Service in Marketing and Management, 11 (2), 105-114

Pleshko, Larry P. & Richard A. Heiens (1996). Categories of Customer Loyalty: An Application of the Customer Loyalty Classification Framework in the Fast-Food Hamburger Market. Journal of Food Products Marketing, 3 (1), 1-12

Pleshko, Larry P., Nizar Souiden, Richard A. Heiens, & Michael J. Fekula (2006). The Operation of Double Jeopardy in the Fast-Food Services Market. Proceedings of the American Society of Business and Behavioral Science, 13(1), 1626-1637.

Raj, S.P. (1985). Striking a Balance between Brand Popularity and Brand Loyalty. Journal of Marketing, 49 (Winter), 53-59.

Raj, S.P. (1982). The Effects of Advertising on High and Low Loyalty Consumer Segments. Journal of Consumer Research, 9 (June), 77-89.

Reibstein, David & Paul Farris (1995). Market Share and Distribution: A Generalization, A Speculation and Some Implications. Marketing Science, 14 (3/2), G190-G202.

Rindfleisch, Aric & Jeffrey Inman (1998). Explaining the Familiarity-Liking Relationship: Mere Exposure, Information Availability, or Social Desirability? Marketing Letters, 9 (1), 1-15.

Robinson, Joel (1979). Brand Strength Means More than Market Share. Journal of Advertising Research, 19, 83-87.

Rosenberg, Larry & John Czepial (1983). A Marketing Approach to Customer Retention. Journal of Consumer Marketing, 2, 45-51.

Sheppard, Blair H., Jon Hartwick, & Paul R. Warshaw (1988). The Theory of Reasoned Action: A MetaAnalysis of Past research with recommendations for Modifications and Future research. Journal of Consumer Research, 15, 325-343.

Smith, Tasman & Kunal Basu (2002). A View from the Top: The Impact of Market Share Dominance on Competitive Position. Journal of Brand Management, 10 (1), 19-33.

Spearman, Charles E. (1904). The Proof and Measure of Association between Two Things. American Journal of Psychology, 15, 72-101.

Spiggle, Susan & Murphy A. Sewall (1987). A Choice Sets Model of Retail Selection. Journal of Marketing, 51 (April), 97-111.

Tellis, Gerard (1988). Advertising Exposure, Loyalty, and Brand Purchase: A Two-Stage Model of Choice. Journal of Marketing Research, 25 (May), 134-144.

Trout, Jack & Al Ries (1972). Positioning Cuts through Chaos in Marketplace. Advertising Age, May 1, 1972

Vaughn, Richard (1980). How Advertising Works: A Planning Model. Journal of Advertising Research, 20(5), 27-33.

Webster, Frederick E. Jr. (1993). Defining the New Marketing Concept. Marketing Management, 2(4), 2231.

Larry P. Pleshko, Kuwait University Nizar Souiden, Laval University

Table 1: Health Clubs

Items #2 #9 #3 #10 #13 r p

Market-share (%) 6 9 11 12 48
 Rank 1 2 3 4 5 n/a
Loyalty-%-of-use 100 86 70 75 89
 Rank 5 3 1 2 4 -.300 n.s.
Most-used-% 4 7 7 10 41
 Rank 1 2.5 2.5 4 5 .975 <.10
Last-Purchase-% 4 8 14 18 47
 Rank 1 2 3 4 5 1.00 .000
Awareness-set-% 68 52 95 79 80
 Rank 2 1 5 3 4 .600 n.s.
Reconsider-set-% 17 23 49 47 59
 Rank 1 2 4 3 5 .931 <.10
Location: mean 6.2 6.2 4.5 6.0 6.3
 Rank 3 4 1 2 5 .200 n.s.
Convenient hours: mean 6.2 5.5 4.2 5.2 6.3
 Rank 4 3 1 2 5 .100 n.s.
Individual attention: mean 2.2 3.1 3.5 2.4 3.7
 Rank 1 3 4 2 5 .700 n.s.
Friendly employees: mean 3.8 4.8 4.5 3.5 4.7
 Rank 2 5 3 1 4 .000 n.s.
Cleanliness: mean 4.4 5.9 5.5 4.8 7
 Rank 1 4 3 2 5 .600 n.s.
Atmosphere: mean 4.7 4.9 4.8 4.3 7
 Rank 2 4 3 1 5 .300 n.s.
Fair prices: mean 5.9 4.9 5.7 5.7 7
 Rank 4 1 2 3 5 .400 n.s.
Variety of Service: mean 5.8 5.2 4.7 4.8 6.3
 Rank 4 3 1 2 5 .100 n.s.
Variety of Equipment: mean 5.8 5.6 5.3 5.1 6.3
 Rank 4 3 2 1 5 .000 n.s.
Clientele: mean 4.8 4.1 3.0 3.5 5
 Rank 4 3 1 2 5 .100 n.s.
No Crowds: Mean 4.5 5.9 5 4.4 6
 Rank 2 4 3 1 5 .300 n.s.
Employee Quality: mean 3.6 3.9 3.7 2.9 6.7
 Rank 2 4 3 1 5 .300 n.s.
Modern Equipment: mean 3.6 4.9 4.3 4.2 6.3
 Rank 1 4 3 2 5 .600 n.s.

Table 2: Convenience Stores

Items #8 #10 #4 #6 #3 #5 #2 #9

Market-share (%) 6.6 7.2 7.6 8.8 9.1 9.5 12.1 19.1
 Rank 1 2 3 4 5 6 7 8
Loyalty-%-of-use 27 32 30 29 35 24 21 34
 Rank 3 6 5 4 8 2 1 7
Most-used-% 6 8 10 10 13 7 9 16
 Rank 1 3 5.5 5.5 7 2 4 8
Last-Purchase-% 11 13 4 8 9 9 9 18
 Rank 6 7 1 2 4 4 4 8
Awareness-set-% 81 83 92 94 71 82 86 97
 Rank 2 4 6 7 1 3 5 8
Reconsider set-% 54 59 63 71 58 62 71 72
 Rank 1 3 5 6.5 2 4 6.5 8
Location: mean 6 7 6.3 6.7 7 6.9 6.9 6.8
 Rank 1 7.5 2 3 7.5 5.5 5.5 4
Fast service: 5.5 5.6 5.5 5.7 6.1 4.8 6.3 5.1
 mean
 Rank 3.5 5 3.5 6 7 1 8 2
Conv. hours: mean 5.4 5.8 6 6 5.9 6.1 6 5.7
 Rank 1 3 6 6 4 8 6 2
Easy: mean 3.3 3.4 3.2 5 4.6 4.1 4.6 3.1
 Rank 3 4 2 8 6.5 5 6.5 1
Friendly empl: mn 4.3 3.3 4 3.9 4.6 2.9 3.9 3.5
 Rank 7 2 6 4 8 1 5 3
Cleanliness: mean 4.4 4.7 5 4.3 5.3 3.1 5.5 4.8
 Rank 3 4 6 2 7 1 8 5
Atmosphere: mean 4.4 4.8 4 3.4 4.9 3.3 4.3 3.6
 Rank 6 7 4 2 8 1 5 3
Fair prices: mean 3.2 4.1 7 3.9 5.4 4.3 5.1 4
 Rank 1 4 8 2 7 5 6 3

Items r p

Market-share (%)
 Rank n/a
Loyalty-%-of-use
 Rank -.024 n.s.
Most-used-%
 Rank .542 n.s.
Last-Purchase-%
 Rank .143 n.s.
Awareness-set-%
 Rank .381 n.s.
Reconsider set-%
 Rank .708 <.10
Location: mean
 Rank .321 n.s.
Fast service:
 mean
 Rank -.018 n.s.
Conv. hours: mean
 Rank .333 n.s
Easy: mean
 Rank .077 n.s.
Friendly empl: mn
 Rank -.286 n.s.
Cleanliness: mean
 Rank -.286 n.s.
Atmosphere: mean
 Rank -.405 n.s.
Fair prices: mean
 Rank .048 n.s.

Table 3: Medical Clinics

Items #1 #4 #6 #8 #2 #7 r p

Market-share (%) 3.7 3.7 4.5 16.7 17.5 42.3
 Rank 1.5 1.5 3 4 5 6 n/a n.s.
Loyalty-%-of-use 58 46 47 54 56 66
 Rank 5 1 2 3 4 6 .557 n.s.
Most-used-% 5 3 3 18 22 46
 Rank 3 1.5 1.5 4 5 6 .871 <.05
Last-Purchase-% 7 6 0 18 27 37
 Rank 3 2 1 4 5 6 .814 <.15
Awareness-set-% 30 48 53 42 70 76
 Rank 1 3 4 2 5 6 .786 <.15
Reconsider-set-% 18 44 23 31 73 70
 Rank 1 4 2 3 6 5 .700 n.s.
Location: mean 5 3.25 n/a 4.2 3.8 6.2
 Rank 4 1 3 2 5 .525 n.s.
Fast service: 5.6 2 n/a 5.1 3.9 4.8
 mean
 Rank 5 1 4 2 3 -.025 n.s.
Convenient hours: 5 3.75 n/a 4.6 4.4 5.2
 mean
 Rank 4 1 3 2 5 .475 n.s.
Service Quality: 6.8 5.5 n/a 6.4 6.5 5.3
 mean
 Rank 5 2 3 4 1 -.425 n.s.
Special service: 5.4 5.25 n/a 5 4.8 4
 mean
 Rank 5 4 3 2 1 -.925 <.10
Doctor quality: 6.4 6 n/a 6.3 6.5 4.9
 mean
 Rank 4 2 3 5 1 -.238 n.s.
Friendly 4 3.25 n/a 4.6 4.7 4.3
 employees: mean
 Rank 2 1 5 6 3 .350 n.s.
Cleanliness: mean 6.8 4 n/a 5.6 5.9 5.4
 Rank 5 1 3 4 2 -.075 n.s.
Atmosphere: mean 6 3.25 n/a 4.5 5 4.2
 Rank 5 1 3 4 2 -.075 n.s.
Friendly doctors: 5.8 3.75 n/a 5.5 5.2 4.4
 mean
 Rank 5 1 4 3 2 -.125 n.s.
Fair prices: mean 5.6 1.75 n/a 5.5 4.8 5.9
 Rank 4 1 3 2 5 .475 n.s.