[ Article ]
International Journal of Tourism and Hospitality Research - Vol. 31, No. 10, pp.5-17
ISSN: 1738-3005 (Print)
Print publication date 31 Oct 2017
Received 08 Feb 2017 Revised 04 Mar 2017 Accepted 05 Mar 2017

# Are heavy players really loyal customers in the Korean casino industry?

Namhyun Kim* ; Choong-Ki Lee
*Assistant professor, Department of Hotel and Tourism Management, Dongguk University-Gyeongju, Gyeongju, 38066 nkim@dongguk.ac.kr

Correspondence to: Professor, Department of Tourism, Kyung Hee University, e-mail: cklee@khu.ac.kr

## Abstract

This study explores whether heavy spenders in casinos are really loyal customers. It applies a volume segmentation technique in order to identify segments of the Korean casino market and then examines differences among emergent segments in terms of service quality, satisfaction, and loyalty. An onsite survey was conducted for casino gamblers at K casino, which allowed the research team to interview gamblers in a live gambling situation. A total of 470 usable questionnaires were collected. A two-step cluster analysis identified three emergent segments: recreational players, high spending players, and high frequency players. The results show that high frequency players were less satisfied with all four dimensions of casino service quality than recreational players. Ratings for satisfaction, loyalty, recommendation, and word-of-mouth were significantly lower for high frequency players and high spending players than for recreational players. These findings imply that, interestingly, traditional assumptions regarding customer loyalty do not seem to apply to the casino market. Casino marketers should focus on identifying truly loyal customers rather than merely heavy players to ensure their sustainable operation. This study addressed that volume segmentation appeared to be important in the casino industry and revealed unique characteristics of the Korean casino market.

## Keywords:

Casino gamblers, Volume segmentation, Service quality, Satisfaction, Loyalty

## Ⅰ. Introduction

Casino gambling has been popular as a leisure activity for a long time; however, there are still ongoing debates regarding its social and economic consequences (Lee, Lee, Bernhard, & Yoon, 2006). Building casino facilities has been one of the national strategies to improve economic conditions in countries with suffering economies. South Korea is no exception. Although gambling has historically been prohibited in Korea, the legalization of gambling in casinos has generated tourism receipt as well as provided nightlife for foreign tourists. Further, Korea has recently approved foreigner-only casinos as integrated resort (IR) to compete with Macau and Singapore. While some people have welcomed the development of IR in Korea to attract more tourists, many people have expressed concerns about the negative impacts of gambling on society. In the casino industry, successful operation often depends on a casino’s ability to attract heavy gamblers, such as high rollers or whales, who wager large amounts of money and frequently visit casinos. However, frequent visits to casinos may foster gambling addiction; thus, heavy gamblers may become problem or pathological gamblers. This is the greatest concern of the gambling industry, as an increased level of problem or pathological gambling would become a substantial social problem with increasing social costs. Thus, problem or pathological gambling is a concern for not only society but also casino operators. A difficult and critical question for casino operators is whether heavy players are really loyal customers who contribute to a casino’s revenue and reputation. To answer this question, this study employs a volume segmentation technique and a service quality-satisfaction- loyalty framework.

Recent studies have proposed market segmentation as a useful tool to help casino operators identify the target spectrum of customers (e.g., Lee et al., 2006). As one of its main contributions, the market segmentation literature has explored how to identify a certain group and which criteria should be used for market segmentation. Kotler and Armstrong (1996) suggested twenty-five segmentation criteria for consumer markets. The most common segmentation approaches divide customers into heterogeneous groups based on demographics, psychological characteristics, socio-economic characteristics, or behavioral characteristics (Park et al., 2002). Recently, researchers have revisited volume segmentation, as suggested by Twedt (1964), which categorizes a market based on the volume of customer usage or expenditures. Twedt (1964) advocated for the importance of volume segmentation because heavy users or spenders contribute to a large percent of a company’s revenue. Empirical research in the tourism literature, for example, have supported this notion by showing that the expenditures of heavy spenders (users) account for from 78% (e.g., Spotts & Mahoney, 1991) to 90.5% (e.g., Moufakkir, Singh, Moufakir-van der Woud, & Holecek, 2004) of the total expenditures of a sample of spenders (users).

Sustainable operation in any business depends on satisfying customers and further developing loyal customers. Loyal customers spend more money and more frequently purchase certain goods or services because they are satisfied with products and, to some extent, are addicted to them. Hence, marketers’ primary task is to identify heavy users or spenders as loyal customers. In the context of casinos, many casino users repeatedly visit casinos, and repeat visitors may spend more money than casual visitors. The question then arises, are repeat visitors loyal customers? or are they just addicted to gambling? Further, should marketers treat all heavy players as loyal customers in their operations? Despite the importance of targeting heavy spenders, research on whether heavy spenders are really loyal customers remains limited. The basic idea of volume segmentation is to identify loyal customers who substantially contribute to a company’s revenue and then to develop marketing strategies targeting this particular group. From a service quality-satisfaction perspective, it is well established that customers who perceive a high level of service quality are more likely to be satisfied with the service provided and that this high level of satisfaction then fosters repurchase intention and positive word of mouth (Cronin & Taylor, 1992; Lee, Graefe & Burns, 2004). Recent studies in the casino industry (e.g., Prentice, 2013; Shi, Prentice, & He, 2014) also suggest that the perception of service quality differ according to different market segments, showing that loyal customers perceive a high level of service quality. Accordingly, volume segmentation research should be able to explain why loyal customers are different from their counterparts in terms of their perceived quality of, satisfaction with, and loyalty to company, products, or services. However, most volume segmentation studies have focused on exploring differences among segments with regard to socio-demographic characteristics, tripographics, or behavioral characteristics (e.g., Lima, Eusébio & Kastenholz, 2012; Pizam & Reichel, 1979; Spencer, 2010; Spotts & Mahoney, 1991). Although the number of tourism studies focused on the segmentation of tourists to identify heavy spenders (users) has been increasing, none of these studies, to the best of our knowledge, has accounted for service quality or customer loyalty.

In addition to the lack of studies concerning service quality and loyalty, previous studies in the tourism field have suffered from several limitations. First, most volume segmentation studies have used one variable, either expenditures or usage, as a criterion, and have used tertiles or quartiles of the frequency distribution of expenditures to divide the segments. Thus, these studies usually focus on three groups: heavy, medium, and light spenders (e.g., Díaz-Pérez, Bethencourt-Cejas, & Álvarez-González, 2005; Laesser & Crouch, 2006; Legohérela & Wong, 2006). Such classifications are simple to use but do not necessarily characterize a market. Second, volume segmentation studies have focused on general holiday tourists (e.g., Goldsmith & Litvin, 1999; Spotts & Mohaney, 1991), golf travelers (e.g., Shani et al., 2010), mountain tourists (e.g., Lima et al., 2012), or rail-trail users (Spencer, 2010). Little research has examined casino customers by applying this technique (e.g., Moufakkir et al., 2004). Further, most segmentation research on casino customers has focused on motivation or involvement as a criterion for market segmentation (e.g., Park et al., 2002; Lee et al., 2006, 2009).

Thus, it is necessary to further investigate the application of volume segmentation to gambling markets. Identifying real loyal customers in the casino context may be more demanding because these customers may or may not be problem gamblers who tend to spend more time and bet more money than casual visitors. This study specifically aims to (1) identify segments of the Korean casino gambling market by applying a volume segmentation technique and (2) determine whether heavy players (spenders) are really loyal customers of casinos by examining differences among emergent segments in terms of service quality, satisfaction, and loyalty. This study contributes to the volume segmentation literature by applying this technique to the casino industry, and it expands the market segmentation literature by integrating a service quality-satisfaction framework into the volume segmentation approach. This study also contributes to marketing practice by providing implications regarding whether heavy users are really loyal customers and whether they differ from their counterparts.

## II. Literature review

Marketers in many industries have long been aware that customers are heterogeneous and that the customers within a certain group share similarities. Thus, market segmentation has been a useful tool for understanding markets and guiding marketing strategies in order to target certain groups of customers (Dolnicar, 2004). Segmentation facilitates business managers’ ability to develop effective marketing strategy by identifying similarities among customers and the characteristics of certain customer segments. By targeting a specific group of customers, market segmentation can reduce the costs of marketing and increase the efficiency of marketing activities.

Many researchers have employed various segment variables to classify markets. Variables that are frequently used include demographics, socio-economic characteristics, psychographics, and behavioral characteristics. Kotler and Armstrong (1996) proposed twenty-five variables for market segmentation, including usage rate. According to Twedt’s theory, volume segmentation is based on the idea that heavy users or heavy spenders contribute more to companies’ sales and revenue as loyal customers than other customers (Twedt, 1964). Research on tourism marketing recently revisited the usefulness and viability of volume of tourists’ expenditures or usage as a segmentation variable (e.g., Goldsmith & Litvin, 1999; Koc & Altinay, 2007; Lima et al., 2012; Mok & Iverson, 2000). For example, Lima et al. (2012) reviewed 43 market segmentation studies in the field of tourism and suggested that the volume of tourists’ expenditures is an important segmentation criterion in a Portuguese mountain tourist market. However, few studies have assessed volume segmentation based on visitors’ expenditures or usage (Lima et al., 2012).

There are three issues in the literature on tourism volume segmentation that should be raised. The first issue concerns the selection of a variable for dividing markets. It is imperative to choose an adequate variable that can better distinguish one group from others. General market segmentation using demographics or psychological variables generally tends to be a probabilistic approach, which does not elucidate how many groups would be obtained and what characteristics would be represented for the emerging groups. In contrast to general market segmentation, volume segmentation using expenditures or usage is a goal-oriented approach, as the purpose of volume segmentation is to identify loyal customers who contribute to a company’s revenue. Customer loyalty has been defined as repeat purchase/visit behavior or as consistent purchase/visit behavior resulting from a psychological decision-making process (Jacoby & Kyner, 1973). By definition, loyal customers should be identified based on both their volume of expenditures and their frequency of purchases (usage). Most studies on volume segmentation, however, have used only expenditures as a segmentation criterion. However, as notable exceptions, Goldsmith and Litvin (1999) and Goldsmith, Flynn, and Bonn (1994) used travel agency usage to partition the Singaporean vacation market. Hence, additional research using both expenditures and usage for volume segmentation to identify loyal customers is needed.

The second issue concerning volume segmentation is related to the adopted methodology. Most studies have been based on simple arithmetic division for segmentation (see Table 1). With such an approach, samples are divided into three groups by partitioning the frequency distribution of expenditures at tertiles, yielding groups of heavy, medium, and light spenders. Most of these studies have adopted Spotts and Mahoney’s method (1991), with some exceptions (e.g., Díaz-Pérez, Bethencourt-Cejas, & Álvarez-González, 2005; Laesser & Crouch, 2006; Legohérela & Wong, 2006; Lima et al., 2012). In addition, the samples of volume segmentation studies have comprised general holiday visitors (e.g., Goldsmith & Litvin, 1999; Spotts & Mohaney, 1991), golf travelers (e.g., Shani et al., 2010), mountain tourists (e.g., Lima et al., 2012), or rail-trail users (Spencer, 2010). Little research has aimed to segment casino customers by applying a volume segmentation technique (e.g., Moufakkir et al., 2004). Spotts and Mahoney (1991), for example, addressed the viability of expenditure-based segmentation in regional destination marketing. They classified a sample of summer travel parties in a three-county region in Michigan into three groups (heavy, medium, and light spenders) based on the amount of their per-trip expenditures in the study area. The results showed that the three groups differed in terms of their tripographics: recreational activities, trip purposes, and travel party characteristics.

Volume segmentation studies in tourism

Mok and Iverson (2000) identified three segments of the Taiwanese market in Guam based on total expenditures and compared demographics and tripographics among the segments. The results showed that compared to their counterparts, heavy spenders stayed longer in Guam and were more likely to be under 50 years of age. Jang, Ismail, and Ham (2001) defined three groups of Japanese outbound travelers based on their per-trip expenditures and compared these groups on socio-demographic variables and trip-related variables. The results showed that compared to their counterparts, heavy spenders were more likely to be older and married; to travel with a spouse, a boy/girlfriend, or their parents; and to travel to the US, Canada, Europe, or Oceania. Spencer (2010) adopted volume segmentation to classify rail-trail users in the Black Hills of South Dakota, USA, into heavy, medium, and light spenders based on expenditures. He focused on comparing socioeconomic characteristics and tripographics and found that heavy spenders had higher income, stayed longer, and were more likely to have been mountain biking aficionados, compared to other two segments. Lima et al. (2012) also studied the mountain destination tourist market in Portugal and identified four distinct segments: light spender, medium spender, lodging & activities oriented, and food & shopping oriented visitors. They performed a cluster analysis by using daily expenditures per capita by product as a criterion and found differences between the emergent segments in terms of socio-demographics and travel behavior characteristics.

The third issue concerns whether the level of customers’ expenditures or usage explains their level of satisfaction and loyalty. Volume segmentation has commonly been used to identify target segments because heavy spenders are assumed to substantially contribute to companies’ sales and to become loyal customers of tourism destinations. High-spending customers are thus targeted to gain economic benefits, and targeting these customers is successful in the way to satisfy them during their stay at a destination. Nevertheless, as shown above, none of the previous studies in the area has attempted to explain whether high spending or high usage customers are actually satisfied with the services they receive. Rather, the main focus of previous studies has been differences between heavy spenders and other segments regarding socioeconomic and/or demographic characteristics and tripographics (e.g., Mok & Iverson, 2000; Moufakkir et al., 2004; Spencer, 2010; Spotts & Mahoney, 1991). In addition, these expenditure-based segmentation studies have focused on the proportion of heavy spenders among the studied sample to determine the importance of heavy spenders among all visitors. For example, empirical research in the tourism literature has shown that the expenditures of heavy spenders (users) account for from 78% (e.g., Spotts & Mahoney, 1991) to 90.5% (e.g., Moufakkir et al., 2004) of the total expenditures of a sample of spenders (users).

In summary, to address the above issues, this study used casino visitors as a sample to expand the casino segmentation literature. Previous studies on casino segmentation exist, but most segmentation research has focused on motivation or involvement as a criterion for market segmentation (e.g., Park et al., 2002; Lee et al., 2006, 2009). Regarding methodology, simple arithmetic techniques are easy to use and are often a logical method for dividing samples, but when more than one variable is used and when more reliable methods are required, it is more appealing to adopt a statistical method, such as cluster analysis, which is a common method in general segmentation studies (Lima et al., 2012). This study also adopts a service quality-satisfaction framework in order to extend the volume segmentation literature by taking into account customer loyalty, which is explained by the level of service quality and satisfaction.

## III. Methodology

### 1. Measurement

First, respondents were asked to report their expenditures during their visit to the casino when the survey was conducted (e.g., “how much money did you spend for this visit?”) and their number of visits to the casino during the previous year. Second, the respondents were asked to respond to items concerning casino service quality, satisfaction, and loyalty. The final section of the questionnaire recorded the respondents’ demographic characteristics, including gender, age, education, and income. A casino service quality (CSQ) scale was adapted from Wong and Fong (2010) after a comprehensive literature review of the casino gaming literature (Johnson et al., 2004; Lucas, 2003; Mayer et al., 1998; McCain et al., 2005; Wong & Fong, 2010; Yi & Busser, 2008). This scale was refined through pilot test to determine whether these items were relevant to the casino industry in Korea. The pilot tests were conducted with a group of casino experts consisting of academics, industry managers, and government officers in Korea. The research team also asked these experts to provide additional service quality items. The final list of CSQ items consisted of thirteen items. All of these items were measured with 5-point Likert-type scales (1= extremely unsatisfied and 5=extremely satisfied). The customer satisfaction measure (e.g., “overall satisfaction with the casino”) was adapted from Wong and Fong (2010). Customer loyalty reflects committed behavior and was measured by using 3 items regarding behavioral intention on 5-point Likert-type scales (1=strongly disagree and 5=strongly agree). The loyalty variables included intention to revisit, recommendation, and word of mouth (Baker & Crompton, 2000; Wong & Fong, 2010).

### 2. Data collection

Data were collected by administering the on-site survey at K casino, which allowed the research team to interview casino gamblers in a live gambling situation. The survey was conducted on both weekdays and weekends in September 2010, and a temporary booth was placed on the first floor of K casino. When casino gamblers voluntarily approached the booth, six trained field researchers explained the purpose of the research project and then invited them to participate in the survey. A self-administered questionnaire was distributed to each respondent. To increase the response rate and to assure the quality of responses, a small gift was given to respondents who completed the survey questionnaire. A total of 524 questionnaires were collected, representing the response rate of 93.5%. After eliminating 54 questionnaires because of missing and/or patterned data, 470 questionnaires were coded for analysis.

### 3. Data analyses

All collected data were analyzed by using SPSS 18 and AMOS 18. First, descriptive analysis including frequencies, means, and standard deviations was performed for all survey questions. Second, to determine the dimensionality of the CSQ scale exploratory factor analysis was firstly conducted. After the validity of CSQ scale was confirmed, confirmatory factor analysis (CFA) was conducted on the 13 items of the scale. The validity and reliability of the CSQ dimensions were tested and confirmed by CFA. Third, to identify heterogeneous groups of casino players based on their volume of expenditures and frequency of visits, a two-step cluster analysis was employed to automatically reveal natural clusters (or groupings) within a data set. Similarity between clusters was determined by using a log-likelihood method, and the “best” number of clusters was chosen with the Schwarz's Bayesian Information Criterion (BIC) as the clustering criterion.

Two-step cluster analysis is a relatively new technique for clustering samples (Hsu, Kang, & Lam, 2006). It includes two steps. In the first step, a cluster features (CF) tree is constructed to group all the cases into several nodes. The construction of this tree begins by placing the cases into several nodes and then adding each successive case to an existing node or forming a new node based on its similarity to existing nodes and using the distance measure as the similarity criterion. In the second step, subclusters (nodes) resulting from the first step are grouped into a desirable number of clusters by using an agglomerative clustering algorithm based on the BIC. To ensure the external validity of the clusters, a discriminant analysis was performed.

Fourth, a series of multivariate analysis of variance (MANOVA) procedures were performed to test for any significant differences in the CSQ dimensions among the clusters obtained by the two-step cluster analysis. In the MANOVA, the factors from the CFA were used as the dependent variables, and the emergent groups were used as the independent variables. The Tukey HSD post-hoc test was used to detect differences among the segments. Finally, a MANOVA with the Games-Howell post-hoc test was used to test for significant differences among the segments on a series of customer loyalty variables (and behavioral variables). The Games-Howell test was used instead of another common post-hoc test (e.g., Tukey) because the homogeneity of variance assumption was violated in the MANOVA (i.e., the Box test of the equality of covariance matrices was significant at the .05 level in the analysis). Research suggests that the Games-Howell test is appropriate when (1) the sample sizes within cells are unequal and/or (2) the homogeneity of variance assumption is violated (Games & Howell, 1976; Sullivan, Riccio, & Reynolds, 2008).

## Ⅳ. Results

### 1. Volume segmentation of casino players

The auto-clustering algorithm of the two-step cluster analysis indicated that a three-cluster solution was the best model because it minimized the BIC value (265.859) with a relatively large ratio of BIC change (.307) and a relatively large distance measure (2.554) between the adjacent numbers of clusters. Clusters I, II, and III consisted of 365, 32, and 73 cases, which corresponded to 77.7%, 6.8%, and 15.5% of the cases, respectively.

The results of the MANOVA showed that two dependent variables (expenditures and number of visits) contributed to differentiating the three volume segments, at p<.001 (see Table 2). Before the MANOVA, a correlation analysis between dependent variables was conducted, indicating moderate correlation with each other (.464). In addition, Games-Howell post-hoc tests were employed to examine any differences between the three segments with respect to the two variables as segmentation criteria. The results of the Games-Howell tests revealed statistically significant differences between the emergent clusters, indicating that the categorization of subjects was appropriate.

Summary statistics of the cluster analysis of volume segmentation

As shown in Table 2, cluster I includes customers with the lowest average spending in the casino for their current visit when the survey was conducted (US$2,402) and the lowest average frequency of visits during the last year (15.47 times). The first cluster group was labeled “Recreational players (RPs)” since the expenditures and frequency of visits were lowest for this group among the three groups. The average spending of the second cluster was US$48,381, which is the highest amount of expenditures among the three groups. Hence, this cluster was labeled “High spending players (HSPs)”. The third cluster group visited the casino most frequently during the last year, with an average number of visits of more than 110 times. Thus, this cluster was named “High frequency players (HFPs)”.

Results of the discriminant analysis of casino volume segments

To ensure the external validity of the above clusters, a discriminant analysis was performed. Discriminant analysis is usually used to assess the adequacy of classification by calculating classification matrices to explain the degree to which subjects are correctly grouped in their clusters (Lee et al., 2006). As shown in Table 3, Wilks’ Lambda indicated the significance of discriminant functions (p<.001). The classification results (Table 3) revealed that 96.6% of the respondents were classified correctly into the three groups. These results further confirm the validity of the clusters.

### 2. Results of the factor analysis of casino service quality (CSQ)

Studies on service quality and satisfaction have stressed that a high level of service quality fosters customer satisfaction, which leads to customer loyalty with respect to the products or services provided. However, casino products/services may have different characteristics in this framework because the most frequent visitors may be problem gamblers. Before the test of differences in CSQ, a CFA was conducted to identify CSQ dimensions. Wong and Fong (2010) proposed three dimensions of casino service quality consisting of 9 items: game service, service environment, and service delivery. In this study, one more construct as “service facilities” dimension was added because many casinos try to provide facilities for entertainment and accommodation in addition to food services and because the results of the pilot tests with casino experts indicated that other service facilities are also important. The result of the CFA indicated that a four-factor measurement model had adequate validity and reliability, with average variance extracted (AVE) greater than or equal to the .5 threshold, Cronbach’s α greater than or equal to .7, and composite reliability greater than or equal to .7 (see Table 4). As one exception, the service environment dimension was revealed to have an AVE of .47; however, we did not drop this dimension because other indicators and the overall model-fit statistics all met the recommended values (comparative fit index (CFI)=.968, goodness of fit (GFI)=.957, root mean square error of approximation (RMSEA)=.056, root mean square residual (RMR)=.045, and standardized root mean square residual (SRMR)=.037). In addition, this inadequate AVE result is not critical in this study because the purpose of this study is not to develop a CSQ scale but to compare the level of CSQ among emergent segments.

CSQ scale validity and reliability

### 3. Difference in casino service quality (CSQ) by segments (MANOVA)

As a means of checking multicollinearity, correlation between the dependent variables was tested. The results of correlation analysis showed moderate correlation with each other ranging from .45 to .677. To test for differences between the emergent segments on CSQ, a MANOVA was performed. As shown in Table 5, the results showed that the three segments differed on all four dimensions of CSQ (Wilks’ lambda=.063, F=188.313, p<.001). The overall ratings on all CSQ dimensions were relatively lower than 3 on 5-point Likert scales, indicating that service quality of K casino was considered to be below average.

Casino service quality by segment

The highest score was 2.97 for service delivery, as rated by RPs, and the lowest score was 2.24 for game service, as rated by HFPs. RPs gave higher ratings than the other groups on all four dimensions, ranging from 2.64 to 2.97, while HFPs were dissatisfied with all four dimensions of CSQ, reporting the lowest mean value for each dimension among the groups. Because the assumption of homogeneity of variance-covariance was met Tukey HSD post hoc procedure was conducted. The differences in CSQ between RPs and HFPs were statistically significant at p<.05. That is, RPs were more satisfied on CSQ dimensions except service facility than HFPs. A post-hoc test did not reveal significant differences in CSQ either between RPs and HSPs or between HSPs and HFPs.

### 4. Difference in loyalty by segments (MANOVA)

Based on the conventional definition of customer loyalty, HSPs and HFPs should be considered really loyal customers who substantially contribute to casino revenue. Further, it is assumed that these two groups should have higher scores for overall satisfaction, recommendation, and word of mouth than RPs. To test for significant differences in the customer loyalty variables by segment, a MANOVA was conducted with the Games-Howell post-hoc test. Before the MANOVA, correlation between the dependent variables was tested. The results of correlation analysis showed low and moderate correlation with each other ranging from .278 to .685. The results of MANOVA showed that the three segments differed on three of the four variables (Wilks’ Lambda=.063, F=186.366, p<.001). As shown in Table 6, the results showed that RPs had the highest scores among the three segments for all of the customer loyalty variables except the intention to revisit variable. Further, RPs reported a higher level of overall satisfaction (2.61) than HFPs (1.88).

Customer loyalty by segment

High spending customers and high frequency customers are generally considered loyal customers in most types of businesses because high expenditures and frequent revisits directly contribute to business revenue. However, interestingly, this traditional assumption of the marketing literature regarding customer loyalty does not seem to apply to casinos. Although the respondents in the HSP and HFP groups spent more money in the casino and visited the casino more frequently, respectively, than the respondents in the RP group, their level of loyalty was significantly lower than that of RPs, who spent less and less frequently visited the casino. HFPs had the lowest mean values for recommendation and word of mouth (1.52 and 1.51, respectively), and the mean differences between RPs and HSPs and between RPs and HFPs on these variables were statistically significant. The overall values for intention to revisit (ranging from 3.29 for HSPs to 3.47 for HFPs) were the highest among the customer loyalty variables, and HFPs had the mean value for this variable among the groups. However, no significant differences in behavioral intention to revisit were found among the three segments. Overall, the results showed that respondents in all three segments are not satisfied with the casino services at K casino but that they are likely to visit the casino again. This finding indicates that they would like to visit the casino regardless of their satisfaction, which is a unique characteristic of casinos likely arising from gambling addiction. In addition, ratings for recommendation and word of mouth are below average, implying that casinos are not perceived to be socially desirable in Korea.

## V. Conclusion and implications

This study aimed to identify segments of the Korean casino market by applying volume segmentation in order to understand any differences among the segments in terms of service quality, satisfaction, and loyalty. Specifically, the study examined whether high spenders or frequent visitors are loyal customers.

The findings from the volume segmentation appeared to be important in the casino industry and revealed the unique characteristics of the Korean casino market. Ratings for overall satisfaction, CSQ, and loyalty were relatively low for all three groups, while one of loyalty variables, intention to revisit, was among the highest rated variables. Recreational players reported relatively higher levels of satisfaction, CSQ, and loyalty than high spending players and high frequency players, who are traditionally considered loyal customers in other business contexts. This finding contradicts the traditional view in the marketing literature regarding customer loyalty and recent results of casino research (e.g., Prentice, 2013; Shi, Prentice, & He, 2014). Thus, conventional marketing theory on customer loyalty might not be applicable to the casino market, at least the Korean casino market. The results also showed the uniqueness of the Korean casino market. Korean people tend to have negative perceptions of casino gambling because of Confucianism. They perceive casino gambling to be socially undesirable. These negative perceptions were apparent in this study, as ratings for overall satisfaction, recommendation, and word of mouth were below average while ratings for revisit intention were above average. Another unique finding of this study is that the demographic characteristics were not different among the three segments identified by volume segmentation (see Appendix A). This finding is inconsistent with previous research on volume segmentation in other tourist markets (e.g., Goldsmith & Litvin, 1999; Koc & Altinay, 2007; Lima et al., 2012; Mok & Iverson, 2000). Thus, it would be necessary to investigate this issue in future research.

Composition of demographic profiles and the differences by segments

In summary, managers should be aware that repeat visitors or high spenders are not true loyal customers in the casino industry. Although casino companies may prefer high frequency players and high spending players in terms of revenue, they are not really loyal customers in terms of satisfaction with CSQ. The results of this study showed that high frequency players and high spending players were dissatisfied with CSQ and that their loyalty was low. These customers might repeatedly visit the casino because they might be addicted. Hence, handling these groups of players should be a priority for casino management. In Korea, problem gamblers are the greatest concern in the casino industry, and they may harm the casino’s reputation. If the casino transforms their operations toward an IR and targets more recreational players, they might improve their reputation. As the findings showed, recreational players are more satisfied than other players with regard to service facilities such as resorts, shopping centers, and performance centers. If the casino wants to have more satisfied customers and to improve overall satisfaction among its customers, casino managers should pursue diversification among their services and products. Such diversification could help the casino improve its reputation as well.

This study has some theoretical implications. The results show that volume segmentation can provide casino managers with a new perspective in terms of customer loyalty. Specifically, casino managers can use volume segmentation to identify whether frequent visitors or high spending visitors are really loyal customers. Most volume segmentation studies have focused on exploring differences in socio-demographic characteristics among emergent segments. This study extends the segmentation literature by integrating the service quality-satisfaction-loyalty framework into volume segmentation research. This framework is particularly important in the casino context because repeat visitors or heavy spenders in casinos may be misunderstood as loyal customers. The portion of repeat visitors might be higher in the casino market than in general product markets merely because of problem gamblers. As the results indicated, not all frequent visitors are loyal customers and they tend to visit casinos repeatedly regardless of their satisfaction with the casino services or loyalty to the casino. Hence, further research is recommended to determine whether the applicability of the traditional view regarding customer loyalty depends on the context of the market or the characteristics of the product.

A second implication of this study concerns the methodology of volume segmentation. We addressed two issues related to volume segmentation in terms of methodology. This study aimed to improve the reliability of the volume segmentation technique by applying two variables, namely, expenditures and usage, and by using cluster analysis in order to classify the Korean casino market by a statistical method rather than merely a simple arithmetic method. The results revealed reasonable customer segments; thus, this study contributes to the identification of loyal customers and their characteristics.

Finally, the findings of this study may not be generalized since only one casino was conduct for study. Therefore, future research is recommended to conduct similar research in other jurisdictions. The study focused on volume segmentation in the Korean casino industry. Although the results show that volume segmentation is an applicable tool for the casino industry, the result of segmentation showed no differences in demographics among segments, and this is inconsistent with previous studies in volume segmentation. Therefore, further research is required to identify whether this is unique characteristics of casino customers or gamblers.

## References

• Baker, D. A., & Crompton, J. L., (2000), Quality, satisfaction and behavioral intentions, Annals of Tourism Research, 27(3), p785-804. [https://doi.org/10.1016/s0160-7383(99)00108-5]
• Craggs, R., & Scholfield, P., (2009), Expenditure-based segmentation and visitor profiling at The Quays in Salford, UK, Tourism Economics, 15(1), p243-260. [https://doi.org/10.5367/000000009787536753]
• Cronin, J. J. Jr., & Taylor, S. A., (1992), Measuring service quality: A reexamination and extension, Journal of Marketing, 56(3), p55-68. [https://doi.org/10.2307/1252296]
• Díaz-Pérez, F. M., Bethencourt-Cejas, M., & Álvarez-González, J. A., (2005), The segmentation of Canary Island tourism markets by expenditure: Implications for tourism policy, Tourism Management, 26, p961-964. [https://doi.org/10.1016/j.tourman.2004.06.009]
• Dolnicar, S., (2004), Beyond “Commonsense segmentation”: A systematics of segmentation approaches in tourism, Journal of Travel Research, 42(3), p244-250.
• Games, P. A., & Howell, J. F., (1976), Pairwise multiple comparison procedures with unequal N's and/or variances: A monte carlo study, Journal of Educational Statistics, 1(2), p113-125. [https://doi.org/10.3102/10769986001002113]
• Goldsmith, R. E., Flynn, L. R., & Bonn, M., (1994), An empirical study of heavy users of travel agencies, Journal of Travel Research, 33(2), p38-43. [https://doi.org/10.1177/004728759403300105]
• Goldsmith, R. E., & Litvin, S. W., (1999), Heavy users of travel agents: A segmentation analysis of vacation travelers, Journal of Travel Research, 38(2), p127-133. [https://doi.org/10.1177/004728759903800205]
• Hsu, C. H. C., Kang, S. K., & Lam, T., (2006), Reference group influences among Chinese travelers, Journal of Travel Research, 44(4), p474-484. [https://doi.org/10.1177/0047287505282951]
• Jacoby, J., & Kyner, D. B., (1973), Brand loyalty. vs. repeat purchasing behavior, Journal of Marketing Research, 10(1), p1-9. [https://doi.org/10.2307/3149402]
• Jang, S., Ismail, J.A., & Ham, S., (2001), Heavy spenders, medium spenders, and light spenders of Japanese outbound pleasure travelers, Journal of Hospitality & Leisure Marketing, 9(3-4), p83-106. [https://doi.org/10.1300/j150v09n03_07]
• Johnson, L., Mayer, K. J., & Champaner, E., (2004), Casino atmospherics from a customer’s perspective: A re-examination, UNLV Gaming Research & Review Journal, 8(2), p1-10.
• Koc, E., & Altinay, G., (2007), An analysis of seasonality in monthly per person tourist spending-market segmentation perspective, Tourism Management, 28(1), p227-237.
• Kotler, P., & Armstrong, G. M., (1996), Principles of marketing, Prentice Hall PTR.
• Laesser, C., & Crouch, G. I., (2006), Segmenting markets by travel expenditure patterns: The case of international visitors to Australia, Journal of Travel Research, 44(4), p397-406. [https://doi.org/10.1177/0047287505282952]
• Lee, C. -K., Lee, Y. -K., Bernhard, B. J., & Yoon, Y. -S., (2006), Segmenting casino gamblers by motivation: A cluster analysis of Korean gamblers, Tourism Management, 27(5), p856-866. [https://doi.org/10.1016/j.tourman.2005.05.009]
• Lee, C. -K., Lee, B., Bernhard, B. J., & Lee, T. K., (2009), A comparative study of involvement and motivation among casino gamblers, Psychiatry Investigation, 6(3), p141-149. [https://doi.org/10.4306/pi.2009.6.3.141]
• Lee, J., Graefe, A., & Burns, R., (2004), Service quality, satisfaction, and behavioral intention among forest visitors, Journal of Travel & Tourism Marketing, 17(1), p73-82. [https://doi.org/10.1300/j073v17n01_05]
• Legohérel, P., & Wong, K. K. F., (2006), Market segmentation in the tourism industry and consumers’ spending: What about direct expenditures?, Journal of Travel and Tourism Marketing, 20(2), p15-30.
• Lima, J., Eusébio, C., & Kastenholz, E., (2012), Expenditure-based segmentation of a mountain destination tourist market, Journal of Travel & Tourism Marketing, 29(7), p695-713. [https://doi.org/10.1080/10548408.2012.720155]
• Lucas, A. F., (2003), The determinants and effects of slot servicescape satisfaction in a Las Vegas hotel casino, UNLV Gaming Research & Review Journal, 7(1), p1-19.
• Mayer, K. J., Johnson, L., Hu, C., & Chen, S., (1998), Gaming customer satisfaction: An exploratory study, Journal of Travel Research, 37(2), p178-183. [https://doi.org/10.1177/004728759803700210]
• Mok, C., & Iverson, T. J., (2000), Expenditure-based segmentation: Taiwanese tourists to Guam, Tourism Management, 21(3), p299-305. [https://doi.org/10.1016/s0261-5177(99)00060-6]
• Moufakkir, O., Singh, A. J., Moufakkir-van der Woud, A., & Holecek, D. F., (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), p59-71.
• McCain, S. L., Jang, S. C. S., & Hu, C., (2005), Service quality gap analysis toward customer loyalty: Practical guidelines for casino hotels, International Journal of Hospitality Management, 24(3), p465-472.
• Park, M., Yang, X., Lee, B., Jang, H. C., & Stokowski, P. A., (2002), Segmenting casino gamblers by involvement profiles: A Colorado example, Tourism Management, 23(1), p55-65. [https://doi.org/10.1016/s0261-5177(01)00063-2]
• Pizam, A., & Reichel, A., (1979), Big spenders and little spenders in U.S. tourism, Journal of Travel Research, 18(1), p42-43. [https://doi.org/10.1177/004728757901800108]
• Prentice, C., (2013), Service quality perceptions and customer loyalty in casinos, International Journal of Contemporary Hospitality Management, 25(1), p49-64. [https://doi.org/10.1108/09596111311290219]
• Shani, A., Wang, Y., Hutchinson, J., & Lai, F., (2010), Applying expenditure-based segmentation on special-interest tourists: The case of golf travelers, Journal of Travel Research, 49(3), p337-350. [https://doi.org/10.1177/0047287509346852]
• Shi, Y., Prentice, C., & He, W., (2014), Linking service quality, customer satisfaction and loyalty in casinos, does membership matter?, International Journal of Hospitality Management, 40(July), p81-91. [https://doi.org/10.1016/j.ijhm.2014.03.013]
• Spencer, D., (2010), Segmenting special interest visitors to a destination region based on the volume of their expenditures: An application to rail-trail users, Journal of Vacation Marketing, 16(2), p83-95. [https://doi.org/10.1177/1356766709357486]
• Spotts, D. M., & Mahoney, E. M., (1991), Segmenting visitors to a destination region based on the volume of their expenditures, Journal of Travel Research, 29(4), p24-31. [https://doi.org/10.1177/004728759102900405]
• Sullivan, J. R., Riccio, C. A., & Reynolds, C. R., (2008), Variations in students’ school- and teacher-related attitudes across gender, ethnicity, and age, Journal of Instructional Psychology, 35(3), p296-305.
• Twedt, D. W., (1964), How important to marketing strategy is the ‘heavy user’?, Journal of Marketing, 28(1), p71-72. [https://doi.org/10.2307/1249232]
• Wong, I. A., & Fong, V. H. I., (2010), Examining casino service quality in the Asian Las Vegas: An alternative approach, Journal of Hospitality Marketing & Management, 19(8), p842-865. [https://doi.org/10.1080/19368623.2010.514553]
• Yi, S., & Busser, J. A., (2008), The influential attributes that affect resident slot gamblers’ repatronage intentions and willingness to recommend casinos to others, Journal of Hospitality Marketing & Management, 16(4), p343-367. [https://doi.org/10.1080/10507050801949712]

### Table 1.

Volume segmentation studies in tourism

Study Year Market Type of trip Criterion adopted Classification method Emergent segments
Pizam &
Reichel
1979 US travelers (Domestic tourism) Holiday
visitors
Expenditures Arithmetic division by quartiles of frequency distribution Big and little spenders
Spotts &
Mahoney
1991 Visitors to Michigan’s Upper Peninsula (Michigan domestic tourism) Holiday
visitors
Expenditures Arithmetic division by tertiles of frequency distribution Light, medium, and heavy spenders
Goldsmith
et al.
1994 Users of travel agents in the US (Domestic tourism) Holiday
visitors
Usage Survey responses to a travel agency usage question Heavy and light users
Goldsmith
& Litvin
1999 Users of travel agents in Singapore (Domestic tourism) Holiday
visitors
Usage Survey responses to a travel agency usage question Heavy and light users
Mok &
Iverson
2000 Taiwanese visitors to Guam (Guam inbound tourism) Holiday
visitors
Expenditures Arithmetic division by tertiles of frequency distribution Light, medium, and heavy spenders
Jang et al. 2001 Japanese travelers (Japan outbound tourism) Holiday
visitors
Expenditures Arithmetic division by tertiles of frequency distribution Heavy, medium, and light spenders
Moufakkir
et al.
2004 Casino visitors (Domestic tourism) Casino
visitors
Expenditures Arithmetic division by tertiles of frequency distribution Light, medium, and heavy spenders
Díaz-Pérez
et al.
2005 The Canary Islands tourism market (Canary Islands inbound tourism) Holiday
visitors
Expenditures CHAID technique High and low spenders
Laesser &
Crouch
2006 International visitors to Australia (Australia inbound tourism) Holiday
visitors
Expenditures Hedonic (log-linear) regression N/A
Legohérela
& Wong
2006 International visitors to Hong Kong (Hong Kong inbound tourism) Holiday
visitors
Expenditures CHAID technique N/A
Koc &
Altinay
2007 International visitors to Turkey (Turkey inbound tourism) Holiday
visitors
Expenditures N/A N/A
Craggs &
Schofield
2009 Visitors to the Quays in the UK Holiday
visitors
Expenditures Arithmetic division by tertiles of frequency distribution Light, medium, heavy, & no expenditure segments
Shani et al. 2010 Golf travelers in the US (US domestic tourism) Golf
travelers
Expenditures Arithmetic division by tertiles of frequency distribution Light, medium, heavy spenders
Spencer 2010 Rail-trail users in Black Hills (US domestic tourism) Rail-trail
users
Expenditures Arithmetic division by tertiles of frequency distribution Light, medium, heavy spenders
Lima et al. 2012 Portuguese Mountain tourist market (Portugal domestic tourism) Mountain
tourists
Expenditures Hierarchical cluster analysis Light, medium, lodging & activities oriented, and food & shopping oriented

### Table 2.

Summary statistics of the cluster analysis of volume segmentation

Cluster I
(n=365)
Cluster II
(n=32)
Cluster III
(n=73)
F-valueb Games-Howell post-hoc test
I-II I-III II-III
Notes: ***p<.001, a.1USD=1,100KRW, b.MANOVA was conducted by using two dependent variables with logarithmic transformation to ensure their normality
Expendituresa 2,402 48,381 5,609 12,666.38*** *** *** ***
Number of visits in the past 12 months 15.47 47.72 115.96 913.40*** *** *** ***
Cluster name Recreational players High spending players High frequency players Pillai's trace = 1.394***

### Table 3.

Results of the discriminant analysis of casino volume segments

Discriminant function Eigen value Canonical correlation Wilks' Lambda χ2
Notes: ***p<.001, In all, 96.6% of the originally grouped cases were correctly classified; 96.6% of the cross-validated grouped cases were correctly classified
Volume 1 .735 .651 .443 380.269***
Segmentation 2 .302 .482 .768 123.245***
Standardized canonical discriminant function coefficients Function 1 Function 2
Log (Expenditures) .107 1.083
Log (Number of visit) .953 -.526

### Table 4.

CSQ scale validity and reliability

CSQ Number of items Mean SD Cronbach's α Composite reliability AVE
Notes: AVE: average variance extracted, Fit statistics: chi-square(df=59)=148.768 (p=.000), CFI=.968, GFI=.957, RMR=.045, RMSEA=.056, and SRMR=.037
Game service 3 2.56 .86 .80 .80 .57
Service environment 3 2.72 .90 .76 .72 .47
Service delivery 4 2.86 .91 .83 .80 .50
Service facilities 3 2.73 .93 .79 .75 .50

### Table 5.

Casino service quality by segment

CSQ Mean (SD) Tukey
HSD
F-value R2
Recreational
players (n=365)
High spending
players (n=32)
High frequency
players (n=73)
Notes: ***p<.001, For ANOVA, df=3, Wilks' Lambda=.063, F=188.313, p<.001, CSQ Scale: 1=extremely unsatisfied and 5=extremely satisfied, a, b. Means in the same row are significantly different at p<.05 in the Tukey test
Game service 2.64a
(.87)
2.48
(.78)
2.24b
(.80)
a>b 1415.81*** .90
Service environment 2.81a
(.90)
2.47
(.90)
2.35b
(.80)
a>b 1491.00*** .91
Service delivery 2.97a
(.89)
2.67
(.89)
2.38b
(.84)
a>b 1646.86*** .91
Service facilities 2.79
(.91)
2.69
(.96)
2.43
(.94)
- 1380.36*** .90

### Table 6.

Customer loyalty by segment

Customer loyalty Mean (SD) Games-Howell
test
F-value R2
Recreational players
(n=361)
High spending
players (n=31)
High frequency
players (n=73)
Notes: ***p<.001, For ANOVA df=3, Wilks' Lambda=.063, F=186.366, p<.001, Customer loyalty scale: 1=strongly disagree and 5=strongly agree, a, b, c.Means in the same row with different superscript are significantly different at p<.05 in the Games-Howell test
Overall satisfaction 2.61a
(1.12)
2.26
(1.21)
1.88c
(.94)
a>c 794.43*** .84
Intention to revisit 3.39
( .98)
3.29
(1.13)
3.47
(.99)
- 1826.55*** .92
Recommendation 2.16a
(1.15)
1.55b
( .93)
1.52c
(.84)
a>b, a>c 538.85*** .78
Word of mouth 2.27a
(1.14)
1.68b
( .98)
1.51c
(.88)
a>b, a>c 92.70*** .79

### Appendix A.

Composition of demographic profiles and the differences by segments

Characteristics N(%) RPs(n=365) HSPs(n=32) HFPs(n=73) F-value /X2-value p value
Notes: The numbers indicate percentages that vertically sum to 100%, a1USD=1150KRW
Gender 469 365 32 72 2.75 .253
Male 60.3% 58.4% 68.8% 66.7%
Female 39.7% 41.6% 31.3% 33.3%
Age 470 365 32 73 18.028 .054
20-29 3.6% 4.1% 3.1% 1.4%
30-39 22.1% 24.9% 12.5% 12.3%
40-49 28.9% 29.6% 28.1% 26%
50-59 34.5% 32.1% 46.9% 41.1%
60-69 10.2% 8.5% 9.4% 19.2%
over 70 .6% .8% - -
Education 470 365 32 73 14.419 .071
under elementary 1.7% 1.1% 3.1% 4.1%
Middle and high school 32.1% 29.9% 53.1% 34.2%
College 21.1% 20.5% 18.8% 24.7%
University 38.3% 41.4% 21.9% 30.1%
Graduate school 6.8% 7.1% 3.1% 6.8%
Income 468 363 32 73 14.183 .077
less than US$870 8.8% 8.8% 9.4% 8.2% US$870-1,730 16% 14.3% 31.3% 17.8%
US$1,739-3,470 35.7% 34.7% 21.9% 46.6% US$3,478-5,209 24.8% 26.4% 18.8% 19.2%
over US\$5,217 14.7% 15.7% 18.8% 8.2%