Introductory marketing textbooks emphasize the need for firms to segment their potential customers in an effort to develop a ma

Extremely Frequent Behavior in Consumer Research:
Theory and Empirical Evidence for Chronic Casino Gambling
Ralph Perfetto, University of Rhode Island Send correspondence to Ralph Perfetto, University of Rhode Island, Ballentine Hall, Suite 216, Kingston, RI, 02881 (Email: ). Arch G. Woodside, Boston College, Carroll School of Management, Department of Marketing, 140 Commonwealth Avenue, ABSTRACT
The present study informs understanding of customer segmentation strategies by extending Twedt’s heavy-half propositions to include a segment of users that represent less than two percent of all households – consumers demonstrating extremely frequent behavior (EFB). Extremely frequent behavior (EFB) theory provides testable propositions relating to the observation that few (2%) consumers in many product and service categories constitute more than 25 percent of the frequency of product or service use. Using casino gambling as an example for testing EFB theory, an analysis of national survey data shows that extremely frequent casino gamblers do exist and that less than 2 percent of all casino gamblers are responsible for nearly 25 percent of all casino gambling usage. Approximately 14 percent of extremely frequent casino users have very low-household income, suggesting somewhat paradoxical consumption patterns (where do very low-income users find the money to gamble so frequently?). Understanding the differences light, heavy, and extreme users and non-users can help marketers and policymakers identify and exploit “blue ocean” opportunities (Kim and Mauborgne, 2005), for example, creating effective strategies to convert extreme users into nonusers or nonusers into new users. Keywords: segmentation, consumer, retailing, consumption Extremely Frequent Behavior in Consumer Research:
Theory and Empirical Evidence for Chronic Casino Gambling
INTRODUCTION
Lotte, South Korean’s biggest department-store chain, opened a huge department store in the Wangfujing shopping district in Beijing in July 2008. The wealthiest customers will be granted special parking spots and will be guided around the store by personal attendants. (Appealing to the very rich works well for Lotte at home: its richest 1% of customers accounted for 17% of its $5.8 billion in sales Marketing executives often use consumer demographics, psychographics, and behavioral characteristics for crafting market segmentation strategies. Common demographic segmentation categories include age, income, ethnicity, family life cycle, and gender. Income is particularly attractive as a segmentation basis because income helps define consumers buying power. As such, extremely frequent users are segmentable further into those with low, moderate, and high- income levels. When a product category’s users include both high-income and low-income users, their demographic profiles are unique, and their consumption profiles across many product With copious amounts of information now available through internet purchasing and customer loyalty programs, volume segmentation has become an increasingly useful segmentation strategy. Volume segmentation assigns people into groups via usage categories— non-users, light users, and heavy users (Twedt, 1964). According to Twedt, the heavy users represent the top 50 percent of the users in a product category after ordering users by their amount of product consumption. Thus, if 40 percent of the adult U.S. population consumes beer and 60 percent do not consume beer, a median split by beer consumption among the 40 percent results in 50 of the users being identified as the light-half users and 50 percent of the heavy-half users. The heavy-half users often account for nearly 90 percent of the total consumption for many fast moving consumer goods (FMCGs) (Twedt, 1964). Twedt argues that firms should consider these consumers to be most important (Twedt, 1964). Many segmentation studies support Twedt’s heavy-half propositions, although the number of actual user categories varies a bit (Goldsmith and Litvin, 1999; Spotts and Mahoney, 1991). In addition to demographic and volume differences among customers, individuals often reveal insights into their personalities, social status, self-identity, and profession through the products they consume (Belk, 1988; Belk, Mayer, and Bahn, 1982; Holman, 1980, 1981; Levy, 1964; Lowery, Englis, Shavitt, & Solomon, 2001; Shavitt & Nelson, 2000). Solomon (1988) argues that most psychological treatments of product symbolism focus on the individual product category level and that “social behavior is often accompanied by the joint consumption of many disparate products and services that, when taken together, appear to define a social role. Products are not consumed in a vacuum, but instead often play an integral part in consumers’ social lives” (Solomon, 1988, p. 244). Clusters of disparate products and services make up what Solomon and Assael (1987) refer to as “consumption constellations.” The present study informs understanding of customer segmentation strategies by extending Twedt’s heavy-half propositions to include a segment of product or service users that represents 1 to 2 percent of all potential customers – the small number of customers demonstrating extremely frequent behavior (EFB). Extremely frequent behavior (EFB) theory in consumer research builds and extends on prior work (Cook and Mindak, 1984; Twedt 1964) to provide testable propositions relating to the observation that very few (2%) customers in most product and service categories constitute 20+ percent of the usage frequency of the product or service. More generally, the present report contributes to customer portfolio analysis literature (Woodside and Soni, 1991; Woodside and Trappey, 1996) in identifying and examining the antecedents of customers segmented by usage levels and categories of products consumed. This article uses casino gambling behavior in an application of the extremely frequent behavior theory. Most chronic casino gambling (defined here as gambling most-to-all weeks annually) likely represents one example of compulsive buying. Compulsions are "repetitive and seemingly purposeful behaviors that are performed according to certain rules or in a stereotyped fashion" (American Psychiatric Association 1985, p. 234). “Results indicate people who buy compulsively are more likely to demonstrate compulsivity [n = 380 from a mail survey of compulsive buyers living in California] as a personality trait, have lower self-esteem, and are more prone to fantasy than more normal consumers [n = 250 from a mail survey of adults living in 3 Illinois cities]. Their [compulsive buyers’] primary motivation appears to be the psychological benefits derived from the buying process itself rather than from the possession of purchase objects. The consequences of compulsive buying include extreme levels of debt, anxiety and frustration, the subjective sense of loss of control, and domestic dissension” “Pathological gambling research, and chronic casino gambling research in particular, is still in very early stages of development….Pathological gambling (PG) is a major psychiatric disorder and public health issue that is just beginning to receive public and scientific attention. For many individuals who gamble, the activity is an occasional form of entertainment that does not negatively influence their lives, but PG is different. The American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders, third edition (DSM-111) first recognized PG as a psychiatric disorder in 1980. PG has been categorized as an impulse-control disorder not elsewhere classified since that time (Hollander, Buchalter, and DeCaria 2000, p. 629). The general population’s estimated rate of gambling [all types] is more than 80%. Epidemiologic surveys distinguish between PG and problem gambling. Problem gambling refers to all individuals with gambling-related problems, including mildly affected individuals who do not meet clinical diagnostic criteria for PG as defined by the DSM-IV. The prevalence of PG is estimated to be 1% to 3% of the population, and the male-to-female ratio traditionally has been reported as 2: 1. In adolescent populations, the prevalence of PG has been reported to be as high as Three points worth emphasizing include the following propositions. First, not all extremely frequent casino gamblers are necessarily PG individuals even though they are classifiable as chronic (persistent, inveterate, unremitting) casino gamblers. Yet, chronic casino gambling versus casual-to-no casino gambling is likely to associate positively to PG diagnosis. Second, the available research on the psychology of gambling and profiles of casino gambling and gamblers offer unclear and shallow profiles of such behavior and individuals (Hope and Havir 2002; Kusyszyn, 1984). Third, representative national samples of households may be useful in particular for identifying individuals with compulsive behaviors such as chronic casino gamblers versus occasional and non casino gamblers. The present report builds from these three points to provide in-depth profiles of three distinct types of chronic casino gamblers and compares these types to each other and with occasional and non casino gamblers. Rather than using a variable-oriented approach, the approach presented in this paper is based on property-space methods (Barton 1955, Lasarsfeld 1965) and explanatory typological literature (George and Bennet 2005, Elman 2005). Property- spaces and explanatory typologies allows for a case-based analysis of survey data while capturing interaction effects in an simplier manner than large-n regression techniques (Bennett and Elman 2006). Furthermore, conjuntural conditions, whereby the same outcome may exist for differing typologies, may be identified. Extending the Heavy Half Proposition – Extremely Frequent Behavior (EFB)
Dik Twedt (1964) encourages the process of volume segmentation; Loudon and Della Bitta Volume segmentation attempts to identify frequent users of a product category or brand. Marketers often refer to the “20-80” thesis. That is, that 20 percent of the market accounts for 80 percent of sales of their product. Although the exact proportion may vary and the rule may not universally apply, it does indicate the importance of a firm’s product or service. Marketing researchers historically identify consumers dichotomously as either users or non-users (Twedt 1964). Tests of marketing manipulations seldom focus on nonusers and such tests often assume all users are equally important (Twedt, 1964). Some marketers weight light users the same in importance as high-volume users. Twedt gives the following example to emphasize the point of considering alternative customer importance weighting methods: A household that consumes two six-packs of soft drinks in a given month would be weighted twice as heavily as a household that consumers only one six-pack. And if another household were to consume 30 six-packs during the same time period, then to the manufacturer the opinions of that household should be 30 times as important as those represented by the household that buys only one package. (Twedt, 1964, p. 71) According to Twedt, high-volume consumers are more valuable than low-volume consumers. In one experiment, Twedt studies a sample of Chicago households and finds that 42 percent did not purchase lemon-lime beverages (0% of the volume purchased). The remaining 58 percent of the sample (those households that did purchase lemon-lime beverages) were ordered into two sub-groups, based on a median split of purchasing volume. Twedt calls these the “light using-half” and the “heavy-using half” and notes that “one heavy-half household is equal in purchase volume to nine households in the light-half” (Twedt, 1964). Extending this analysis to other products, Twedt informs that in all product categories (except toilet paper), the top 50 percent of the consumers (users of a product category) account for most of the consumption. Twedt concludes his findings by suggesting that “what can be said is that the heavy-using household buys more, buys more often, and buys more different brands” (Twedt, 1964, p. 71 ). Heavy users are an attractive segment for many firms, since they are a relatively small group that accounts for a very large volume (Goldman & Litvin, 1999; Goldsmith, 2000; Goldsmith & d’Hauteville, 1998; Woodside, Cook, & Mindak, 1987). Some researchers characterize the heavy user in product categories such as fashionable clothing (Goldsmith, 2000), wine consumption (Goldsmith & d’Hauteville, 1998), casino gambling (Moufakkir, Singh, Moufakkir-van der Woud, & Holecek, 2004) and travel (Spotts & Mahoney, 1991; Woodside et al., 1987). Twedt (1964) separates users into three usage categories – non-users, light-users, and heavy-users. Other researchers have arbitrarily chosen to use a different number of categories. Goldsmith and Litvin (1999) segment users into only two segments; light users and heavy users. Spotts and Mahoney (1991) report that most users fall in the middle of the frequency distribution of the usage variable and, as such, chose to segment users into three user groups; light users, medium users, and heavy users. Spotts and Mahoney (1991) argue that the researcher must select user categories in a manner that best allows the characteristics of the market segments to be uniquely distinguishable from one another. Extremely Frequent Behavior Theory
Extremely frequent behavior (EFB) theory in consumer research extends from typological theory and Twedt’s ‘heavy-half’ proposition (Cook & Mindak 1984; Twedt 1964) to provide testable propositions relating to the observation that very few (2%) customers for many product and service categories constitute more than 25 percent of the frequency of product or service use. Typological theory provides propositions on possible groups related to what Lazarsfeld (1965) refers to as “property-space” (see also Barton 1955). This approach “allows researchers to translate a multi-dimensional attribute space into a handful of types” (Ragin 2000, pp. 66) and helps illuminate ways in which the same outcome can arise from different conjunctive paths. Furthermore, property-space analysis can help identify the existence of extreme cases as well as paradoxical relationships that might exist. Property-space analysis starts with identification of conjunctive variables that are likely related to the frequency of a given behavior. In the case of extreme casino gambling, income is particularly attractive variable because it helps define consumers buying power. Figure 1 shows the property-space contingency table and identifies seven unique types of households by Extremely frequent behavior (EFB) theory includes the proposition that (P1) for FMCGs and comparable services a very small share of potential customers constitute more than one-fifth of the total frequency of use. In casino gambling, for example, EFB users may gamble in a casino at least once or more per week. The wealthiest (or high income) EFB users are often sought after by casino owners, since their consumption habits associate with extraordinarily high spending on gaming activities (Moufakkir et al., 2004). This segment of casino gamblers is so important to revenues that casino owners fittingly refer to these consumers “whales” since their wealth affords them an appetite for consumption that is so much larger than any other consumer As Figure 1 demonstrates, not all EFB casino visitors have high income. Rather, extremely frequent consumers vary greatly by demographic make-up and include not only high- income users but also very low-income users as well. Much like the whales, these low-income EFB casino users gamble just as often, but with much smaller pocketbooks. If the high-income extremely frequent casino gamblers are considered whales, then perhaps very low-income casino gamblers are more like “jumbo shrimp” since they represent high volume users within the group of very small income levels. Those extremely frequent gamblers with low to moderate-income fall some where in between – “big fish” that have more buying power than the jumbo shrimp, Whales likely can afford to gamble in casinos, but how do the jumbo shrimp afford to gamble at the same frequency? When both low-income and high-income EFB users exist in the same product category, paradoxical consumption patterns for one or more EFB user segments may exist. More formally, P2: EFB users include a seemingly paradoxical sub-segment of extremely low-income consumers, for example, customers who visit gambling casinos weekly with extremely low incomes (jumbo shrimp). Non-users are an important segment of consumers and are often overlooked by marketers. Typically, marketers test advertising copy with users, not the non-users. As Twedt Ordinarily, marketing researchers screen respondents for product usage before proceeding with questions about attitudes, opinions, or behaviors. Thus, a non- smoker would rarely be included in a test of cigarette advertising copy, nor would a woman who never bakes cakes be asked to state her flavor preference for a cake Given the potential importance of nonusers in crafting what Kim and Mauborgne (2005) refer to as, “blue ocean strategies”, the total number of households represents the base in EFB theory - not the total number of users. The aim of a blue ocean strategy is not to outperform the competition by constantly fighting bloody battles over current product and brand user bases (i.e., “red ocean strategies”). Rather, a blue ocean strategy makes a firm’s competitors irrelevant by going after the non-users and converting them with special offers and “lighter versions” of product offerings (i.e., product designs offering a few exceptional benefits usually at a very low price point). In essence, blue ocean strategy creates a whole new market space (Kim and Mauborgne, 2005). Understanding the differences between users and non-users can help firms identify and exploit blue ocean opportunities. P3: Non-users with moderate income represent the single largest segment in numbers of customers (or households)—this segment provides a nonequivalent comparison group (see Cook and Campbell, 1979) for examining the unique profiles of EFG groups. With such different demographic profiles, high-income and low-income extremely frequent customers are likely to see the world very differently not only from each other, but also from high-income product non-users, low-income product non-users, as well as everyone else. P4: EFB users sub-segmented by income (extremely low versus low-to-moderate versus high) differ from (a) each other, (b) as well as from other potential customer segments in their demographic profiles (i.e., unique conjunctive streams made up of gender, race, household size, Consumption Constellation of EFB Users
Prior studies show that demographic variables alone are, in general, poor predictors of consumer behavior and are less than optimal for segmentation strategies (see Halley, 1968; Frank, 1967; Frank, Massy, & Harper, 1967). In an attempt to better understand consumer behavior and decision making, market researchers often include behavioral and psychometric variables that help to group consumers based on values, personalities, AIOs, and consumption Levy (1959) informs that products are consumed not only for functional reasons but also for their symbolic meanings. “A consumer’s personality can be seen as the peculiar total of the products he consumes” (Levy, 1964, p. 149). Consumers often divulge their individual personalities, social status, gender, self-identity and profession through the products they consume (Belk, 1988; Belk, Mayer & Bahn, 1982; Levy, 1964; Lowery et al., 2001; Holman, 1980, 1981; Shavitt & Nelson, 2000). Heavy versus light use of complementary products and services are captured in what Solomon and Assael (1987) refer to consumption constellations - “clusters of complementary products, specific brands, and/or consumption activities” (p.235) - to define and communicate social roles. For example, the consumption constellation of the social group called “Yuppies” may include products such as a Rolex watch, Gucci loafers, and a Burberry trenchcoat (Solomon, 1988). Essentually, a consumption constellation constitutes a set of products and services which seem to “go together” with a particular stereotypical lifestyle category (Englis and Solomon, 1995; Lowery et al., 2001;). While past research on consumption constellations informs our understanding of lifestyle categories, reference groups and the way in which consumers cognitively categorize objects (Englis and Solomon, 1995), marketers often focus attention on product comparisons within a product category (Sujan and Bettman 1989; Ward and Loken 1986), such as comparing three brands of cars or three brands of cameras. Lowery et al. (2001) notes that academic researchers mostly ignore joint product interdependencies, instead choosing to focus on single-brand choice models. Advertisers and psychographic researchers, on the other hand, have been fully aware of cross-category relationships and often report product interaction effects in their studies (Lowery et al., 2001). Consumer “behavior is often accompanied by the joint consumption of many disparate products and services that, when taken together; appear to define a social role. Products are not consumed in a vacuum, but instead often play an integral part in consumers’ social lives.” (Solomon, 1988, p. 244). For example, one study reports that heavy consumers of Kentucky Fried Chicken also consume large amounts of other products such as eye makeup, nail polish, soft drinks, gum and TV dinners (Tigert, Lathrope, and Bleeg, 1971). Consumption constellations help marketers to position their products into desirable lifestyle categories - “constructed on the basis of the purchase behaviors and, to a lesser extent, on psychographic information such as opinions, attitides, and personality” (Englis et al., 1995, p. 15). It follows that EFB users differ not only in demographic makeup, but also in their More formally, consider the following propositions. P5: EFB users sub-segmented by income differ substantially from (a) each other, (b) as well as other potential customer segments in their consumption constellation (Solomon, 2006) profiles—each segment EFB sub-segment has unique aspects of nonuse, use, and extreme use of other products and services. P6: EFB users sub-segmented by income differ substantially from (a) each other, (b) as well as other potential customer segments in their psychographic profiles (e.g., they differ dramatically views toward government, children, religion, gun use, abortions, and/or other attitudes and beliefs. Mass media, in the form of newspapers, magazines, television and the Internet, plays an important and influential role in the socialization of consumers by positioning products as symbols of cultural lifestyles and connecting these symbols with reference groups (Bearden and Etzel 1982; Cocanougher and Bruce 1971; Englis and Solomon, 1995; Stafford 1966). Differing frequencies of media consumption by EFB users likely lead to differing consumption patterns. P7: EFB users sub-segmented by income differ substantially from (a) each other, (b) as well as other potential customer segments in their media use behavior; such as the frequency of reading newspapers and magazines, watching daytime and evening television, and listening to country, rock, classical, or other radio stations. The data analyzed in this study came from the annual DDB Needham Life Style Survey. Similar versions of the questionnaire were administered annually during 1975 to 1999, consisting of 300- 400 questions that address respondents’ demographic profile, perceived personality traits, shopping habits, political beliefs, media habits (e.g., newspapers, TV, radio), religious beliefs, international affairs and overall satisfaction with life. The data set contains approximately 3000 respondents per year and over 87,000 respondents in total. Data for years more recent than 1999 has not been released publicly by DDB Needham. Although the DDB Needham lifestyle survey data are rich, some survey questions changed over the 25-year period of administration. The present study analyzes a subset of the survey data for years 1993 to 1998. These years are chosen because survey responses were available for demographic variables such as income, race, education level, household size, gender, and casino gambling during these years. The age variable is segmented into five groups consisting of: less than 35, 35 to 44, 44-54, 55-64, and those 65 and older. Table 1 shows additional demographic variables recoded into meaningful groups. DDB survey data includes a question on the frequency of gambling in a casino over the past 12 months. Respondents were asked to indicate the number of times they gambled in a casino during the past year (none, 1-4 times, 5-8 times, 9-11 times, 12-24 times, 25-51 times, 52+ times). The seven response levels were transformed into values at the midpoints with the value of 60 used for the highest level. The values used include 0, 2.5, 6.5, 10, 18, 38, and 60. The use of alternative values of the highest level (52 and 56) does not substantively change the findings. Attitudes, interests, and opinions were measured on a 6-point scale (definitely disagree, generally disagree, moderately disagree, moderately agree, generally agree, and definitely agree). This study employs a quasi-experimental design using a non-equivalent group design. In the nonequivalent group design (NEGD) (Cook and Campbell, 1979; Reichardt, 1979), persons or intact groups (e.g., non-gamblers, very-low income gamblers,) are arbitrarily assigned to either the program or comparison condition. Non-gamblers with low to moderately high income level represents the largest group (n = 9,225). This group serves as the basis for the comparison group. Comparing the mean of the EFB users to low to moderate-income non-users (the largest consumer segment) helps effectively discriminate between patterns of group behavior. Following the suggestions of Armstrong and Andress, 1970) and Bass, Tigert, and Lonsdale (1968), the present study also uses tree analysis and cross-classification analysis with variable stacking. Armstrong & Andress (1970, p. 491) suggest that tree-analysis is superior to regression analysis “in situations where there are large sample sizes and are subject to interaction, non-linearties or causal priorities”. Bass, Tigert, and Lonsdale (1968) show that a cross-classification analysis is more useful than regression analysis when user segmentation variables include socioeconomic characteristics. FINDINGS
A total of 20,658 survey responders were analyzed for the period during 1993 to 1998. Table 2 shows a frequency distribution of casino gambling behaviors indicated by responders. The largest share of households is the non-gambler group, representing 66 percent (13,756) of all respondents. When income level is factored in, the largest user segment is the low to moderate- income non-gambler group (n = 9,225), representing 67 percent of all non-gamblers and nearly 45 percent of all respondents (see Table 3). For all income levels, moderate gamblers account for 32.2 percent (6,658) of all respondents. The study includes a total of 244 extremely frequent casino gamblers — 35 (14%) have very low income, 183 (75%) have low to moderate income, and 26 (11%) indicate high-income levels. Approximately 33% of DDB households gambled in a casino at least one time during the prior 12 months and took an average 6.2 trips per year. These findings are consistent with AGA findings that suggest 26% of U.S. households gambled in a casino and made 161 million trips to casinos in 1999; an average of 5.6 trips annually (AGA 1999). Although actual household spending data are not included in the DDB survey, the AGA report estimates that U.S. households spent approximately $20 billion at commercial casinos in 1998; slightly less than spending for basic cable ($23B) and slightly more than spending for coffee ($18B) and home video rentals and sales ($15B) (AGA 1999). This implies that U.S. households spent approximately $124 per visit ($20B divided by 161 million visits), or nearly $700 to $750 per The following limitation is worth considering when examining the findings. The results of this study may be limited by the lack of diversity in the survey data. For the years 1993 to 1998, nearly 82 percent of the households responding to the survey are white/Caucasian. This fact may influence the findings with respect to the demographic profile of extremely frequent users. Future studies should attempt to replicate these findings with a more diverse sample of Support for P1: Extreme Frequent Gamblers Exist
Consistent with proposition 1, extremely frequent gamblers (those that gamble in casinos 25 or more times annually) do exist and represent less than 2 percent (244) of all casino gamblers. Yet, their casino use accounts for 27 percent of all casino gambling visits (see Table 4). Table 3 shows the percent share of casino gambling relating with each gambling group. High-income extremely frequent casino gamblers account for the highest ratio of share- of-gambling to share–of-people. High-income extremely frequent gamblers account for the highest casino usage of all casino user groups. In general, the ratio of the share of extremely frequent casino gamblers (all income levels) to the share of people is 10 times greater than moderate gamblers. Although extremely frequent gamblers (all income levels) account for less than 2 percent of the overall households, they visit casinos approximately 10 times more often Support for P2 and P3
A cross tabulation of casino gambling frequency and income level reveals that extremely frequent casino gamblers includes very low, low to moderate, and high income households (see Table 4). Consistent with proposition 2, some very low-income households demonstrate extremely frequent behavior. Of the 244 extremely frequent casino gamblers, 35 (14%) are from very low-income households. Non-gamblers with low to moderate-income levels represent the largest segment of households. Consistent with proposition 3, 9,225 (67%) households with low- to-moderately high income indicate that they do not gamble in casinos at all. This group serves here as the nonequivalent comparison group (see Cook and Campbell, 1979; Kusyszn, 1984; Demographic Tree Analysis: Support for P4
To further explore the demographic differences among very low, low-to-moderate, and very high income extremely frequent gamblers, the study includes a tree analysis using five demographic variables including gender, race, household size, education level, and age. Figures 2 and 3 show demographic profiles of two extremely frequent casino gamblers groups. In support of proposition 4, extremely frequent casino gamblers differ significantly in their demographic profiles based on gender, age, race, education, household size and age. Extremely frequent casino gamblers with very low-income (Figure 2) are predominantly white, females, with a high school education or less. One third of all low-income extremely frequent gamblers are over the age of 65, while nearly another third (28%) are less than 35 years old. Overall, household size for very low income and extremely frequently casino visiting females is small, typically less than two people. However, non-white females tend to have larger households (3+) than white females. Non-white females account for 23 percent of all females reporting. Although the largest female age group is 65 and older, it represents only 33 percent of the total female group (extremely low income casino gamblers). Men are also predominantly white, with a much smaller percentage of non-whites (11%) than the female group. Unlike the female group, white males tend to have more education, with nearly 63 percent reporting more Extremely frequent casino gamblers with extremely high-income (Figure 3) are predominantly white, male, with college experience. Unlike the very low-income group, only 11 percent of the high-income extremely frequent gamblers are over the age of 65 and only 13 percent are under the age of 35. Most of the high-income extremely frequent gamblers are between 35 and 55 years old (60%). High-income extremely frequent casino visiting white males also differ from their very low-income counterparts in household size. Females represent a small percentage of high-income extremely frequent casino gamblers (27%). Most report ages between 35 and 64 years of age (75%). The number of high-income extremely frequent casino gamblers over the age of 64 is very small (less than 12%). Large households (3+) represent a higher percentage of the high-income group (57%) than their low-income counterparts (35%). Finally, no non-white females are found in the group of high-income extremely frequent casino Those individuals reporting low to moderate-income levels represent the largest group of extremely frequent casino gamblers (see Table 3). Females represent only slightly more than males (53% females, 47% males). More than 60 percent report having more than a high school education and more than half are less than 45 years of age. Non-whites represent about 17% of this group, which is slightly more than the high-income group, but less than the low-income group. More than half of this group have households larger than 3. Consumption Constellations: Support for P5 and P6
Figures 4 and 5 shows consumption constellations for the seven groups. Consistent with propositions 5, extremely frequent casino gamblers (EFCGs) differ in their consumption activities, interests, and opinions not only from each other but also from the non-gamblers (see Figure 4). Note that very low-income non-gamblers attend church nearly twice as often as low income EFCGs (extremely frequent gamblers). Similarly, high income EFCGs cook outdoors and go out for breakfast nearly two to four times more often than non-gamblers. Figure 5 show attitudes, opinions and interests of EFCGs and other groups. In support of proposition 6, EFCGs agree more favorably with the attitude that a drink or two at the end of the day is a great way to relax, as compared with other non-EFG groups. Unlike the other groups, low income non-gamblers are less likely to favor legalized abortion. Low income EFCGs enjoy parties, games, shows and anything for fun more than all other groups. This group also feels more strongly that guns should be in every home, as compared with all other groups. Media Usage: Support for P7
In support of proposition 7, EFCGs also differ in their media usage (Figure 6). Seventy three percent of high income EFCGs report reading most or all of the business section of the newspaper; an amount ten percent greater than high income non-gamblers (62.4%). Fifty three percent of high income EFCGs report that television is their primary source for entertainment, as compared to only thirty nine percent of the high income non-gambling group. For the low- income EFCGs, sixty percent report that TV is their primary source of entertainment (the highest percentage of all groups) and twenty three percent report using the newspaper as their main source of daily news (the lowest of all groups). These finding suggest that groups differ in their media usage and would likely respond more favorably to targeted rather than mass media CONCLUSIONS AND MARKETING STRATEGY IMPLICATIONS
Volume segmentation can be an effective tool to help firms focus their marketing efforts toward customers that use their products or service the most. The present study falls in the category of a rather delayed response to Twedt’s prediction that, “It seems most likely that volume of product usage will eventually replace standardized demographic breaks in marketing survey research.” (Twedt, 1964). While Twedt’s (1964) heavy-half proposition informs us that 50 percent of the customers are responsible for most product usage, the theory of extremely frequent behavior extends these findings to the heaviest users – the top 2 percent of the customers that are often responsible for more than one-fifth of all product usage. The present study uses survey data and property-space theory to explore the propositions that extremely frequent behavior exists and that EFB users differ in their demographic makeup, attitudes, interests and consumption constellations. Using casino gambling as an example, we show that extremely frequent casino gamblers do exist and that less than 2 percent of all casino gamblers are responsible for nearly 25 percent of all casino gambling usage. Tree analysis and cross classification analysis (Armstrong & Andress, 1970; Bass, Tigert & Lonsdale, 1968) is useful for showing that the EFB users differ significantly in their demographic configurations. The segment of EFB users consist of very low, low-to-moderate, and high income users. These demographic differences lead to paradoxical consumption patterns. For example, low-income extremely frequent casino gamblers are not supposed to have money to afford this type of behavior. Yet, very-low income extremely frequent casino gamblers not only visit casinos more often, but also eat out in restaurants and take airplane trips for personal reasons more often than other low-income non-gamblers. The following question left for future research: where do they find the resources for this type of consumption? Extremely frequent behavior theory also proposes that EFB users differ in their attitudes, interests, and consumption constellations. Using a nonequivalent group design (see Cook and Campbell, 1979; Kusyszn, 1984; Reichardt, 1979), the present study compares the means of the EFB users to the low-to-moderately high income non-users (the largest segment of households), which helps to effectively discriminate between patterns of alternative group behaviors. The findings suggest that EFB users indeed differ from non-users in their consumption activities, interests and opinions. Identifying such differences increases understanding of EFB users and also the non-users of their products or services. Understanding the differences between users and non-users can help firms identify and exploit blue ocean opportunities (see Kim and Mauborgne, 2005) and inform industry, government, and near government organizations’ efforts in designing effective social reform programs that attempt to control anti-social and self-abusive behavior (Campbell 1969; Palmgreen, Lorch, Donohew, Harrington, Dsilva, & Helm 1995) Chronic casino gamblers vary in demographics, consumption constellations, and AIOs. However, even though no one profile fits all chronic casino gamblers, a limited number of profiles are identifiable and most chronic casino gamblers are classifiable meaningfully into one of these profiles. Jumbo shrimp, big fish, and whales are category names that include different chronic casino gamblers with unique demographic, AIO, and consumption constellation/media usage patterns. Members of these three unique chronic casino categories will most likely differ in their responses to specific pathological-gambling-control intervention programs. The effectiveness (i.e., rate of success and specific impact metrics) of alternative treatment programs for pathological casino gambling addictions will likely vary substantially across the substantially different sub-segments of chronic casino gamblers. The necessity of targeting treatment programs—government and near government organizations’ “demarketing” programs aiming to enable extremely frequent users to reduce or eliminate their consumption dependencies—to reach extreme users does receive a modest amount of attention in the behavioral science literature (e.g., Palmgreen et al. 1995; Wolfgang 1988). For example, Wolfgang considers undergraduate college students responses to a sensation seeking scale by gender and the students’ expectations of future participation in four leisure activities that usually involve betting money. However, the present article is unique in developing and applying an extremely frequent consumption theory; applying the theory empirically using national survey data enriches understanding of subcategories of chronically Such knowledge confirms the proposition that considering sub-segments of extremely frequent casino gamblers is likely necessary—one demographic, lifestyle, and media-use profile does not fit all extremely-frequent casino gamblers. Chronic casino gamblers are segmentable meaningfully into a few unique sub-segments and thick demographic and lifestyle descriptions of each segment are possible. Designing a few, unique, treatment programs to control/eliminate chronic dependencies—with each program separately targeting one sub-segment of extremely frequent casino gamblers—is likely to be a more effective strategy than designing one program that targets all chronic casino gamblers. Creating and testing such multiple-treatment program designs reflects Campbell’s (1969) wisdom and proposals for viewing reforms as experiments. References
American Psychiatric Association (1985), Diagnostic and Statistical Manual of Mental Disorders, Washington, D.C.: American Psychiatric Association. Armstrong, J. Scott and James G. Andress (1970), “Exploratory analysis of marketing data: Tree vs. Regression,” Journal of Marketing Research, 7(4): 487-492. Barton, AH. 1955. “The concept of property-space in social research.” In Language of Social Research, ed. PF Lazarsfeld, M. Rosenberg, pp. 40-53. Glencoe: Free Bass, Frank, Douglas Tigert, and Ronald Lonsdale (1968), “Market Segmentation: Group versus Individual Behavior,” Journal of Marketing Research, 5(3): 264-270. Bearden, William O. and Michael J. Etzel (1982), "Reference Group Influence on Product and Brand Purchase Decision," Journal of Consumer Research, 9 (March), 183-194. Belk, Russell, Robert Mayer, and Kenneth Bahn (1981), "The Eye of the Beholder: Individual Differences in Perceptions of Consumption Symbolism," in Advances in Consumer Research, 9, ed. Andrew Mitchell, Ann Arbor: Association for Consumer Research, 523- Belk, Russell W. (1988). "Possessions and the Extended Self," Journal of Consumer Research, Campbell, Donald T. (1969), “Reforms as Experiments,” American Psychologist, 24, 409-429. Clancy, Kevin and Robert Shulman (1994), Marketing Myths that Are Killing Business, Cocanougher, A. Benton and Grady D. Bruce (1971), "Socially Distant Reference Groups and Consumer Aspirations," Journal of Marketing Research, 8 (August), 79-81. Cook, Thomas and Donald Campbell (1979), Quasi-experimentation: Design & analysis issues for field settings, Dallas, TX: Houghton Mifflin Company. Cook, Victor. Jr. and William Mindak (1984), "A search for constants: the 'heavy user' revisited!" Journal of Consumer Marketing, 1, (Spring): 79-81. Economist (2008), “Retailing: Lotte ambition,” 387 (8586), June 28 – July 4, 72. Elman, C. (2005), “Explanatory typologies in qualitative studies of international politics,” Int. Englis, Basil and Michael Solomon (1995), “To Be and Not to Be: Lifestyle Imagery, Reference Groups, and The Clustering of America,” Journal of Advertising, 24 (Spring), 13-28. Frank, Ronald, William Massy and Harper Boyd (1967), “Correlates of Grocery Product Consumption Rates,” Journal of Marketing Research, 4 (2), 184-190. Frank, Ronald, (1967), “Correlates of Buying Behavior for Grocery Products,” Journal of Marketing, Vol. 31 (October), pp 48-53. George, AL, Bennett A. (2005), “Case Studies and Theory Development in the Social Sciences,” George, AL, Bennett A. (2006), “Qualitative Research: Recent Developments in Case Study Methods,” Annual Rev. Polit. Sci.,9, 455-476 Goldsmith, Ronald, (2000), “Characteristics of the heavy user of fashionable clothing,” Journal of Marketing Theory and Practice, 8(4), 21. Goldsmith, Ronald and Francois d’Hauteville (1998), “Heavy Wine Consumption: Empirical and Theoretical Perspectives,” British Food Journal, 100 (4), 184-190. Goldsmith, Ronald, Leisa Flynn, and Mark Bonn (1994), “An Empirical Study of Heavy Users of Travel Agencies,” Journal of Travel Research, 33 (1), 38-43. Goldsmith, Ronald and Stephen Litvin (1999), “Heavy Users of travel agents: A Segmentation Analysis of Vacation Travelers,” Journal of Travel Research, 38(2), 127-133. Goudriaan, Annm Jaap Oosterlaan, Edwin de Beurs, and Wim van den Brink (2005), “Decision Making in Pathological Gambling: A Comparison between Pathological Gamblers, Alcohol Dependents, Persons with Tourette Syndrome, and Normal Controls,” Cognitive Hartley, R., (1976), “Use of Customer Analysis for Better Market Penetration,” Industrial Marketing Management, 3 (February): 57-62. Hollander, Eric, Alison Buchalter and Concetta DeCaria (2000), “Pathological Gambling,” The Psychiatric Clinics of North America, 23 (3), 629-642. Holman, Rebecca M. (1980), "Clothing as Communications: An Empirical Investigation," Advances in Consumer Research, 7, ed. Olson, Ann Arbor: Association for Consumer Holman, Rebecca M., (1981), "Product Use as Communication: An Appraisal of a Venerable Topic," in Review of Marketing, eds. Enis and Roering, Chicago, IL: American Hope, Janet and Linda Havir (2002), “You Bet They're Having Fun! Older Americans and Casino Gambling,” Journal of Aging Studies, 16 (2), 177-197. Kim, W. Chan and Renee Mauborgne (2005), Blue Ocean Strategy, Boston: Harvard Business Kusyszyn, Igor. (1984), “The Psychology of Gambling,” Annals of the American Academy of Political and Social Science, 474 (July), 133-145. Lazarsfeld, Paul (1965), “Qualitative Measurement in Social Science: Classification, Typologies, and Indices,” in Daniel Lerner and Harold D. Lasswell, eds., The Policy Sciences, Standford: Standford University Press, 155-192 Levy, Sidney, (1964), “Symbolism and Life Style,” in Toward Scientific Marketing, Stephen A. Greyser, ed., Chicago: American Marketing Association, 140-150. Loudon, David and Albert Della Bitta (1993), Consumer Behavior: Concepts and Applications, Lowrey, Tina, Basil Englis, Sharon Shavitt, and Michael Solomon (2001), “Response Latency Verification of Consumption Constellations: Implications for Advertising,” Journal of Moufakkir, Omar, A.J. Singh, Afke Moufakkir-van der Woud, and Donal Holecek (2004), “Impact of Light, Medium and Heavy Spenders on Casino destinations: Segmenting gaming Visitors Based on Amount of Non-gaming Expenditures,” UNLV Gaming Research & Review Journal, 8(1): 59-71. O’Guinn, Thomas and Ronald Faber (1989), “Compulsive Buying: A Phenomenological Exploration,” Journal of Consumer Research, 16 (2), 147-157. Palmgreen, Philip, Elizabeth Lorch, Lewis Donohew, Nancy Harrington, Margaret D’silva, and David Helm (1995). Reaching At-Risk Populations in a Mass Media Drug Abuse Prevention Campaign: Sensation Seeking as a Targeting Variable. Drugs & Society, Ragin, Charles C. (2000), Fuzzy-Set Social Science, Chicago: The University of Chicago Press. Reichardt, Christain, (1979), The Design and Analysis of the Non-Equivalent Group Quasi- Experiment, Unpublished doctoral dissertation, Northwestern University, Chicago. Sevin, Charles, (1965), Marketing Profitability Analysis, McGraw-Hill, New York. Shavitt, Sharon and Michelle Nelson (2000), “The Social Identity Function in Person Perceptions: Communicated Meanings of Product Preferences,” in Why We Evaluate: Function of Attitudes, G. R. Maio and Jerry M. Olson, eds. Mahwaj, NJ: Lawrence Solomon, Michael, (1988), “Mapping Product Constellations: A Categorization Approach to Symbolic Consumption,” Psychology & Marketing, 5 (3), 233-258. Solomon, Michael and Henry Assael (1987), “The Forest or the Trees? A Gestalt Approach to Symbolic Consumptions,” in Marketing and Semiotics: New Directions in the Study of Sings for Sale, Jean Umiker-Sebeok, ed., Berlin: Mouton de Gruyter, 189-218. Spotts, Daniel, Edward Mahoney (1991), “Segmenting visitors to a destination region based on volume of their expenditures,” Journal of Travel Research, 29(4), 24-31. Stafford, James E. (1966),"Effects of Group Influences on Consumer Brand Preferences," Journal of Marketing Research, 3 (1), 68-75. Sujan, Miat and James Bettman (1989),”The Effects of Brand Positioning Strategies on Consumers' Brand and Category Perceptions: Some Insights from Schema Research,” Journal of Marketing Research 26(4): 454-467. Tigert, Douglas, Richard Lathrope and Michael Bleeg (1971), “The Fast Food Franchises; Psychographic and Demographic Segmentation Analysis,” Journal of Retailing, 47 Twedt, Dik W., (1964), “How Important is the ‘Heavy-User’?” Journal of Marketing, 28 (1), 71- Ward, James, and Barbara Loken (1986), “The Quintessential Snack Food: Measurement of Product Prototypes, in Advances in Consumer Research, 13, Richard J. Lutz, ed., Provo, UT: Association for Consumer Research, 126-131. Wells, William, (1968), Backward Segmentation, in Insights into Consumer Behavior, Johan Wolfgang, Ann (1988), Gambling as a Function of Gender and Sensation Seeking. Journal of Woodside, Arch, Victor Cook and Williams Mindak (1987), “Profiling the Heavy Travel Segment,” Journal of Travel Research, 25 (4), 9-14. Woodside, Arch and Praveen Soni (1991), “Customer Portfolio Analysis for Strategic Development in Direct Marketing,” Journal of Direct Marketing, 5 (2), 6-19. Woodside, Arch and Randolph Trappey (1996), “Customer Portfolio Analysis among Competing Retail Stores,” Journal of Business Research, 35: 189-200. Table 1. Demographic Variables
Demographic
Category 1
Category 2
Category 3
Household Size
Education
Table 2. Casino Gambling – Annual Frequency
Gambling
Cumulative
Frequency Count
Table 3. Share of People and Share of People and Share of Casino Gambling
Note. Confirming the core extremely frequent behavior proposition, 1.2% of the users represent 27.4% of all casino gambling visits. A total of 70.9% of the population report never gambling is a casino each year. Table 4. Cross Tabulation of Casino Gambling Groups by Income Level
Income Groups*
Casino Gambling Groups
* Low Income <$20,000/yr, Moderate Income $20,000 to $79,999/yr, High Income >= $80,000/yr Property Space Configuration for “Extremely Frequent Casino Visitors” (Percent distribution of U.S Households based on 1993-1998 BBD survey data, n=20658) Demographic Tree Analysis for Low Income Extremely Frequent Gamblers (Jumbo Shrimp)
Very Low Income Extremely FrequentGamblers Note. Key: H = high school or less education; C = some college to college graduate; P = postgraduate courses to postgraduate degree; <35, 35-44, 55-64, and 65+ = age categories. Numbers 1, 2, and 3+ represent household size. Total numbers of low income extremely frequent gamblers appear in boxes and represent 94% of all low income extremely frequent casino gamblers. Demographic Tree Analysis for High Income Extremely Frequent Gamblers (Whales)
Note. Key: H = high school or less education; C = some college to college graduate; P = postgraduate courses to postgraduate degree; <35, 35-44, 55-64, and 65+ = age categories. Numbers 1, 2, and 3+ represent household size. Total numbers of low income extremely frequent gamblers appear in boxes and represent 94% of all low income extremely frequent casino gamblers. Activities and Consumption Constellations for EFCGs and Other Groups (Average frequency across groups statistically significant at p<.001.) Average
Frequency
Behavior
No Casino Visits
Extreme Casino Visits
Attitudes, Opinions and Interests for EFCGs and Other Groups Average agreement across groups statistically significant by analysis of variance Agreement
No Casino Visits
Extreme Casino Visits
Average agreement across groups statistically significant by analysis of variance TV is main source ofdaily news (phi= .122) Agreeing
Read most or all of thebusiness section of the TV is primary source ofentertainment (phi= .162) No Casino Visits
Extreme Casino Visits

Source: http://trfdc.asia.edu.tw/stratagem_data/Dr.Woodside%E5%88%86%E4%BA%AB%E8%B3%87%E6%96%99%E4%BA%94.pdf