ARTICLE IN PRESS

Tourism Management 27 (2006) 466480 www.elsevier.com/locate/tourman

Examining structural relationships among perceived impact, benet, and support for casino development based on 4 year longitudinal data
Choong-Ki Leea,, Ki-Joon Backb
b

College of Hotel & Tourism, Kyunghee University, #1, Hoiki-dong, Dongdaemun-Ku, Seoul 130-701, South Korea Department of Hotel, Restaurant, Institution Management and Dietetics, Kansas State University, 148C Justin Hall, Manhattan, KS 66506-1404, USA Received 4 May 2004; accepted 17 November 2004

Abstract The objectives of this study were two-fold. First, this study examined the underlying relationship among impact factors, benets, and support level in a longitudinal pattern based on the social exchange theory. Second, this longitudinal study sought to explore the changing attitudes of residents toward casino development over a 4-year period, using a structural equation model. The results of the study show that residents had signicantly different perceptions about the casino before and after its development. Although the residents changed their perceptions before and after the casino opened, they tended to maintain these perceptions for 2 years postopening. The casinos positive economic impact was consistently cited as the most signicant factor in predicting perceived benet. Also, the benet factor was found to be the most important factor in affecting residents support level, which was consistent with the social exchange theory. r 2005 Elsevier Ltd. All rights reserved.
Keywords: Structural relationship; Casino; Resident perception; Longitudinal study; Social exchange theory

1. Introduction Several recent studies on the impacts of gaming legalization have been reported in the literature (Hsu, 2000; Long, 1996; Perdue, Long, & Kang, 1995; Roehl, 1999). Hsu (2000) stated that most of these gaming impact studies are based on the tourism impact studies of the 1970s. Their main interest was residents perceptions of the social, economic, and environmental impacts of gaming on their community (Lee & Back, 2003). For example, researchers have considered both positive and negative impacts of gaming in creating employment opportunities, encouraging inbound tourism, demoralizing family values, and motivating gambling.

Corresponding author. Tel.: +82 2 961 9430; fax: +82 2 964 2537.

E-mail addresses: cklee@khu.ac.kr (C.-K. Lee), back@humec.ksu.edu (K.-J. Back). 0261-5177/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.tourman.2004.11.009

Specically, Carmichael, Peppard, and Boudreau (1996) stressed the importance of residents positive and negative attitudes toward casino development. Longs (1996) study revealed that the majority of respondents did not recommend legalized gambling to other communities because of the severe negative impacts of casinos. On the other hand, Giacopassi, Nichols, and Stitt (1999) studied seven casino communities and found that the majority of respondents favored the casino in the community and believed that it enhanced quality of life by providing positive impacts on the economy. Thus, the dual faces of casino development depend on the interests of the different parties. Despite the published research, gaming impact studies in South Korea are rare due to the short history of the casino industry there. Although thirteen casinos have been in operation since the late 1960s and early 1970s, there was no casino for domestic players until 2000. The rst casino, the Kangwon Land, was made legally

ARTICLE IN PRESS
C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480 467

available to domestic customers in the Kangwon province in October 2000. Similar to those in Colorado, the geographic location of this casino is a rundown mining area that suffered a devastating economic situation after coal was replaced by oil and gas. The government rst legalized a casino in Chongsun in 1995, after residents repeatedly urged ofcials to legalize gaming for domestic customers in order to revitalize this dilapidated mining town. The unique aspect of this approval was that only one domestic use casino was allowed in Korea. The small casino was opened in October 2000, and included 30 table games and 480 slot games and was attached to a deluxe hotel that had 199 guestrooms. This small casino was planned for a capacity of 700 people, but recorded 3200 visitors per day, almost ve times its capacity (Kangwon Land Casino, 2001). Prior to opening the casino, the government expected to experience numerous positive economic impacts, including an increased employment rate, disposable income, and sales revenue in local businesses. Local businesses, lodging facilities, restaurants, taxies, and gas stations increased their sales volumes by 50200 percent (Kangwon Province, 2000). In March 2003, the main casino was opened with 100 table games and 960 slot machines. The main casino was also attached to a hotel with 477 guestrooms and a theme park. In 2003, more than 1.5 million people visited the casino and casino revenues amounted to 564 million dollars (Kangwon Land Casino, 2004). Also, the casino hired 37 percent of its employees from the community. Despite the many positive impacts of casino development, a considerable number of residents have expressed their concerns about its negative impacts, and most specically about social and environmental problems. Like casinos in the U.S. market or in other nations, the residents of the casino community are experiencing problems with gambling addiction, crimes, prostitution, and trafc congestion. The number of pawnshops rapidly increased just after the casinos opening and stories about nancial crises and problem gamblers were frequently reported by the mass media. However, much of the qualitative data, including residents attitudes toward the casino and social perspectives, have not been well reported. Mason and Cheyne (2000) stress that There are few studies on the perceived impacts of tourism either prior to any development or when it is not seen to be a signicant economic area of activity for a region. The majority of research since the mid-70 s, however, has been in the form of snapshots taken at a particular time, in a particular location, with most of the studies taking place where it was already economically important (p. 392). Perdue, Long, and Kang (1999) noted that developing a longitudinal research design is a major challenge in examining changes in residents perceptions

in boomtowns since most research is conducted after the beginning of the boom. This is why developing a base line for investigating changes is extremely difcult. Other authors also emphasized the necessity of a longitudinal study in order to investigate changes in residents attitudes and perceptions over the various development phases of a tourist site (Ap, 1990; Getz, 1994; Carmichael et al., 1996). Furthermore, little research has been conducted on the structural relationship among impact, benet and support by residents based on longitudinal data in the eld of gaming. In this respect, a semi- or quasi-longitudinal study was developed to investigate the impacts of casino development 6 months before and after casino was opened (Lee & Back, 2003). The major limitation of the previous study was that respondents might have experienced history bias due to the short period between pre- and post-data. Thus, a further longitudinal study was strongly recommended to investigate residents perceptions of casino development over time. This study was conducted to reconrm the underlying structures of residents attitudes toward casino development in terms of social, economic, and environmental impacts and to examine causal relationships between the impact variables and benets as well as support over 4 years by using a structural equation model.

2. Theoretical background A review of related literature indicates that two major theories were appropriate for explaining residents attitudes toward the impacts of casino development: namely, social exchange theory and social carrying capacity theory. 2.1. Social exchange theory Ap (1992) argued that the existing research lacks a theoretical framework for explaining residents attitudes towards the impact of tourism. He advocated a social exchange theory as an appropriate framework for explaining residents perceptions of the impact of tourism. His research ndings had revealed that residents would evaluate tourism development in terms of expected benets or costs obtained in return for their services, that is, social exchange. In other words, residents who perceived personal benet from tourism development expressed positive attitudes toward it. It is a behavioral theory that attempts to understand and predict the reactions of individuals in an interactive situation (Ap, 1990). Pizam (1978) tested the relationship between residents attitudes towards tourism and their economic dependency. The results of the study indicated that residents who were employed in the tourism industry

ARTICLE IN PRESS
468 C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480

expressed the most positive attitudes toward tourism. On the other hand, residents who were not employed in the tourism industry expressed the most negative attitudes toward tourism. The ndings implied that the social exchange theory was supported. However, the results of Pizams study (1978) appeared to be counter to those from the study by Liu and Var (1986), which revealed no signicant difference between residents in tourism businesses and those in non-tourism businesses. Milman and Pizam (1988) also conrmed the social exchange theory in that residents who were employed in the tourism industry tended to be highly supportive as compared to those not employed in the industry. The study by King, Pizam, and Milman (1993) supported the social exchange theory in that residents who received economic benets from tourism were more likely to support tourism development. The social exchange theory was also observed in the study by Caneday and Zeiger (1991). Their ndings revealed that residents employed in tourism businesses were likely to express a more positive attitude toward the impact of tourism than those employed in nontourism businesses. The social exchange theory was supported by the study on gambling by Perdue et al. (1995). The results of their research indicated that personal benets were strongly correlated with support for gambling and the positive impacts of gambling, such as jobs and recreation opportunities. Jurowski, Uysal, and Williams (1997) examined residents reactions to tourism based on the social exchange theory. Their ndings indicated that potential for economic gain, use of tourism resources, eco-centric attitude, and community attachment affected resident perceptions and modied resident support for tourism. The study by Perdue et al. (1999) supported the social exchange theory in that residents who perceived personal benets from gaming were more likely to be positive in assessing their quality of life. Roehls study (1999) also supported the concept of social exchange theory in that residents who perceived that the casino generated more jobs gave higher scores on a quality of life scale. Additionally, Andereck and Vogt (2000) conrmed the social exchange theory in that those residents who perceived tourism positively supported most specic types of tourism development. Lee, Kim, and Kang (2003) also supported the social exchange theory in that variables of personal benets were positively correlated with positive economic and social impacts. The results indicate that residents who perceived personal benets from casino development were likely to express economic and social impacts more positively. The social exchange theory is methodologically easy to design since tourism impact can be evaluated by residents in terms of perceived benets or costs obtained in return for their services. The social exchange theory was

suggested by Ap (1992) as an appropriate framework for explaining residents attitudes towards the impact of tourism. In this respect, this study employs the social exchange theory in exploring the underlying relationship among perceived impacts, benet and support by residents in a longitudinal pattern, as listed below: H1a: H1b: H2a: H2b: H3a: H3b: H 4: H5a: H5b: H6a: H6b: H7a: H7b: Positive economic impact has a positive association with perceived benet. Negative economic impact has a negative association with perceived benet. Positive social impact has a positive association with perceived benet. Negative social impact has a negative association with perceived benet. Positive environmental impact has a positive association with perceived benet. Negative environmental impact has a negative association with perceived benet. Perceived benet has a positive association with support. Positive economic impact has a positive association with support. Negative economic impact has a negative association with support. Positive social impact has a positive association with support. Negative social impact has a negative association with support. Positive environmental impact has a positive association with support. Negative environmental impact has a negative association with support.

2.2. Social carrying capacity theory Allen, Long, Perdue, and Kieselbach (1988) asserted that each community has a certain capacity to absorb tourism development. Development beyond the level of carrying capacity will result in negative social and environmental impacts as well as negative economic impact. A social carrying capacity theory postulates that residents express positive attitudes at an initial stage of tourism development followed by negative attitudes after the community reaches a certain level of acceptable change, or carrying capacity (Perdue et al., 1999). Allen et al. (1988) explored the relationship between levels of tourism development and residents satisfaction with community life. They measured the level of tourism development based on a percentage of gross retail sales attributable to tourism (accommodations and restaurants) in each community. The results of their study conrmed the carrying capacity theory in that residents beneted highly with lower to moderate levels of tourism development, while their perceptions took a

ARTICLE IN PRESS
C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480 469

downward trend with increase in tourism development at certain level. Long, Perdue, and Allen (1990) investigated differences in resident perceptions and attitudes across communities on the basis of the communitys existing level of tourism development. In 1985, survey data were collected in 28 small rural communities in Colorado and then grouped into ve levels of tourism development. Researchers measured the level of tourism by considering the percentage of retail sales derived from tourism and the total retail sales of each community. The results of the study revealed that residents support for additional tourism development initially increased with increasing levels of tourism, but became less favorable when a carrying capacity level was reached. Carmichael et al. (1996) investigated changing residents attitudes toward tourism over 3 years. The results of their study indicated that residents perceived the casino town to be a more desirable place to live in 1992, but perceived it to be gradually less so over the 19931995 period, implying support for the carrying capacity theory. A direct application of this theory to this study raises one issue. Since it is difcult to measure the levels or stages of tourism development (Allen et al., 1988), the maximum carrying capacity of the casino communities was not clear. However, it is hypothesized in this study that the level of perceived benet and support will be decreased eventually, since the residents perception reaches the maximum carrying capacity as follows: H 8: Perceived benet will be signicantly decreased over time after the certain level of carrying capacity. Support will be signicantly decreased over time after the certain level of carrying capacity.

N. Social

N. Env.

Benefit

Support

N. Econ. P. Social

P. Env.

P. Econ.

Fig. 1. Proposed model of residents perceptions of casino development.

H 9:

3. Conceptual model According to the social exchange theory, the researchers have developed a conceptual model. This model postulates that exogenous variables have both direct and indirect effects on benet and support. Specically, the theoretical model to be tested, as shown in Fig. 1, involves eight constructs: positive economic, negative economic, positive social, negative social, positive environmental, and negative environmental, as well as benet and support.

4. Methodology 4.1. Data collection Two casino communities, designated by a special law as run-down mining areas, were chosen for survey

research. The data for this study were collected in four different time frames, using one pre-survey and three post-surveys. The pre-survey was conducted 6 months prior to the casinos opening. The number of samples was proportionately allocated based on occupation, using the ofcial statistics of Chongsun County and Taeback City, Kangwon Province. A self-administered questionnaire was given to those who preferred to complete the questionnaire by themselves. Otherwise, the eld researchers completed the questionnaire via personal interview. Respondents were at least 18 years of age; one person was chosen for the sample in cases where there was a group of people. To increase the response rate, small gifts (a set of memory-pens) were given to respondents who participated in the survey and completed the questionnaire. Since the researchers intended to conduct a post-survey, they informed respondents that they would return and asked for their name and phone number so that the same respondent could be identied in the post-survey. A total of 517 usable questionnaires were nally collected during the pre-survey. The Post 1-survey was administered to those who had responded to the pre-survey after the Kangwon Land Casino had been open for 6 months. Each researcher was posted to the same survey site as in the pre-survey and was given information on respondents name, phone number, and workplace as collected in the pre-survey so that they could be easily identied. The researchers asked respondents whether they had participated in the pre-survey. If they conrmed previous participation, they proceeded with the post-survey. During the Post 1survey, a total of 404 usable questionnaires were collected and used for analysis. The Post 2-data were collected one and a half years later than the previous data, in summer 2002, which was equivalent to almost 2 years post-opening. A total of 415 usable questionnaires were collected and used for analysis. Among these respondents, about half (205) participated in the previous two surveys because some of them were out of town, had moved away, quit their jobs, or closed businesses.

ARTICLE IN PRESS
470 C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480

The Post 3-survey was conducted 1 year after the previous data were collected, in summer 2003. A total of 567 usable questionnaires were collected in the Post 3data collection effort and used for analysis. Since there was a signicant time gap between the Pre- and the Post 3-data, 187 of 567 respondents did not participate in this entire longitudinal study because they were out of town, had moved away, quit their jobs, or closed businesses. The authors intended to use panel data over the whole periods of research. However, due to the mortality effect of sampling, a signicant number of respondents from the pre-data were dropped for several different reasons, including relocation of residency, death, and closed businesses. The cohort sample size of 187 is not large enough to conduct structural equation modeling with a great number of variables. So the authors partially changed the original sampling method by sampling additional participants. 4.2. Measurements Measurement items were selected after (1) an extensive literature reviewa preliminary list of measurement items was initially generated from a review of tourism literature pertaining to residents perceptions of tourism and casino impacts (Pizam, 1978; Pizam & Pokela, 1985; Liu & Var, 1986; King et al., 1993; McCool & Martin, 1994; Perdue, Long, & Allen 1990, Perdue et al., 1995, 1999; Carmichael, et al., 1996; Long, 1996; Jurowski et al., 1997; Lindberg & Johnson, 1997); (2) interviews with tourism academics in the eld of tourism impacts; and (3) interviews with community leaders in the casino town. All items were pilot-tested and respondents were asked to evaluate the appropriateness of the measuring instruments. 4.2.1. Exogenous variables A ve-point Likert-type scale was used, with potential responses ranging from 1 being strongly disagree to 5 being strongly agree to assess each variable. The positive economic factor was composed of six items: investment and business, tourist spending, standard of living, tax revenue, employment opportunity, and personal income. The negative economic factor was comprised of two items: cost of living and increased tax burden. The positive social factor was comprised of ve items: quality of life, improvement of educational environment, consolidation of community spirit, pride of local residents, and conservation of local culture. The negative social factor was comprised of seven items: occurrences of gambling addicts, bankruptcy, destruction of family, prostitution, divorce, alcoholism, and crime. The positive environmental factor was made up of two items: preservation of historic sites and natural beauty.

4.2.2. Endogenous variables Two variables were used in the study: residents perception of the benet of casino development and their support level. These residents perceptions were measured on a ve-point Likert-type scale: 1 strongly disagree, 3 neutral, and 5 strongly agree. Respondents were asked to rate how much they agreed with each item on the scale. The benet factor was comprised of four items: casino development provides benets to myself, casino development provides benets to local community, casino provides various job opportunities, and casino provides casino related job opportunities. The support factor was comprised of ve items: I believe the future of our city is bright due to the casino industry; I am proud that I live in this city, the casino industry makes this city a better place to live, I support development of the casino, and the development of casino is the right choice for this city. Table 1 provides descriptive statisticsincluding means, standard deviations, and intercorrelationsfor the measurement scales. Due to the measurement and structural discussions centered on each sample, analyses were performed for the four data samples to provide a comprehensive assessment of the measures. Considering mean values in Table 1, residents perceived that positive economic, social, and environmental impacts as well as benets and support lessened 6 months after the casino opening as compared to the rst days after the casino opening. However, these positive impacts slightly improved 2 years after the casino opened. Residents also perceived that negative economic impacts had reduced gradually, whereas negative social impacts had increased as time went by due to gambling addictions and usury. Negative environmental impacts lessened 6 months after the casino opening as compared to time immediately surrounding the casino opening, but gradually increased due to expansion of casino facilities and construction of the theme park and road. 4.2.3. Measurement modelconrmatory factor analysis The measurement scales utilized in the study are presented in Table 2. The psychometric properties of the eight constructs were evaluated by employing the method of conrmatory factor analysis using LISREL 8.54 (Jo reskog & So rbom, 2001). Each measurement item was only allowed to load on one factor and could not cross-load on any other factors. The specic items were evaluated based on the items error variance, modication index, and residual covariation (Anderson & Gerbing, 1988; Fornell & Larcker, 1981). The model t was evaluated using the CFI, NNFI, and RMSEA t indices, which are recommended due to their relative stability and insensitivity to sample size. Since the w2 is almost always signicant with large sample sizes, the primary evaluation of t appropriate for this study is the RMSEA, CFI, and

ARTICLE IN PRESS
C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480 Table 1 Summary statistics Variable Pre data (n 517) Pos. econ. (1) Neg. econ. (2) Pos. social (3) Neg. social (4) Pos. env. (5) Neg. env. (6) Benet (7) Support (8) Mean Std. Dev. (1) (2) (3) (4) (5) (6) (7) (8) 471

3.70 3.31 2.81 3.49 3.31 4.04 2.68 3.20

0.61 0.87 0.71 1.14 0.87 0.88 0.74 0.93 0.77 0.82 0.73 0.72 0.76 0.76 0.91 0.93 0.73 0.86 0.72 0.73 0.78 0.72 0.97 0.93 0.81 0.85 0.76 0.76 0.81 0.71 0.98 0.96

1.00 0.45 0.19 0.24 0.03 0.21 0.26 0.44 1.00 0.14 0.70 0.24 0.47 0.02 0.66 0.63 1.00 0.11 0.65 0.05 0.48 0.12 0.55 0.52 1.00 0.34 0.73 0.10 0.57 0.02 0.64 0.64

1.00 0.20 0.28 0.03 0.24 0.05 0.19

1.00 0.12 0.42 0.02 0.15 0.43

1.00 0.17 0.45 0.06 0.02

1.00 0.02 0.03 0.29

1.00 0.07 0.01

1.00 0.42

1.00

Post 1-data (n 404) Pos. econ. (1) 2.86 Neg. econ. (2) 3.04 Pos. social (3) 2.39 Neg. social (4) 3.39 Pos. env. (5) 2.78 Neg. env. (6) 3.38 Benet (7) 2.41 Support (8) 2.80 Post 2-data(n 415) Pos. econ. (1) Neg. econ. (2) Pos. social (3) Neg. social (4) Pos. env. (5) Neg. env. (6) Benet (7) Support (8) 3.10 3.19 2.45 3.54 2.90 3.46 2.64 3.00

1.00 0.16 0.29 0.16 0.32 0.06 0.03

1.00 .20 0.77 0.05 0.61 0.64

1.00 0.12 0.49 0.27 0.35

1.00 0.03 0.46 0.41

1.00 0.06 0.11

1.00 0.72

1.00

1.00 0.06 0.12 0.01 0.19 0.01 0.08

1.00 0.21 0.59 0.02 0.48 0.51

1.00 0.04 0.42 0.24 0.27

1.00 0.00 0.44 0.43

1.00 0.02 0.02

1.00 0.85

1.00

Post 3-data (n 567) Pos. econ. (1) 3.00 Neg. econ. (2) 3.02 Pos. social (3) 2.40 Neg. social (4) 3.63 Pos. env. (5) 2.92 Neg. env. (6) 3.57 Benet (7) 2.58 Support (8) 2.87

1.00 0.25 0.23 0.01 0.23 0.08 0.03

1.00 0.12 0.64 0.02 0.61 0.66

1.00 0.02 0.51 0.16 0.24

1.00 0.05 0.56 0.63

1.00 0.01 0.02

1.00 0.88

1.00

NNFI. Models are generally considered to have good ts if RMSEA is lower than 0.08 and CFI and NNFI are greater than 0.90, although models with indices lower than this may be acceptable in some circumstances (Bentler & Bonett, 1980; Browne & Cudek, 1993). The coefcient alpha estimates for the multi-item scales used in this study are presented in Table 2. Most of the alpha coefcients for the data exceed or are close to the minimum standard for reliability of 0.7 recommended by Nunnally and Bernstein (1994) except for the negative economic construct in the second post-data appeared to be marginal. 5. Results 5.1. Structural model The hypothesized relationships between the factors shown in Fig. 1 were based on the social exchange

theory. By using LISREL 8.54 (Jo reskog & So rbom, 2001), the goodness of t of the various models was testable and the relative t of particular pairs of models could be assessed. Three approaches to ascertain consistency of factors were taken using multi-group analysis: (1) investigating the signicant relationships among the variables; (2) testing signicant differences among Pre-, Post 1-, 2-, and 3-data, to examine whether the residents perception of the casino development changed over the period; and (3) identifying specic differences in the attributes in the residents perceptions as revealed in the longitudinal study. This model was then compared with the competing model, which excluded the direct paths from each exogenous variable to support. The w2 was signicant (po:001), while the CFI and NNFI were about .91 and .90, respectively; whereas RMSEA was .062.067. In comparing this model with the competing model, it was clear that the theoretical model t better than the

ARTICLE IN PRESS
472 C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480 Table 2 Results of conrmatory factor analysisPre, Post 1, Post 2, and Post 3-dataa Construct Positive economic (PE) Item Parameter estimates: Investment and business Tourist spending Standard of living Tax revenue Employment opportunity Personal income Coefcient alpha: Parameter estimates: Costs of living Tax burden Coefcient alpha: Parameter estimates: Quality of life Educational environment Community spirit Pride of local residents Conserve local culture Coefcient alpha: Parameter estimates: Gambling addicts Bankruptcy Destruction of family Prostitution Divorce Alcoholism Crime Coefcient alpha: Parameter estimates: Preservation of historic sites Preservation of natural beauty Coefcient alpha: Parameter estimates: Crowding Trafc congestion Quantity of litter Noise level Water pollution Destruction of natural environ. Coefcient alpha: Parametric estimates: Various job opportunities Casino-related job opportunities Personal benet Community benet Coefcient alpha: Parameter estimates: Bright future Pride Better place to live Support Right choice for the city Coefcient alpha: Pre Post 1 Post 2 Post 3

0.73 0.66 0.48 0.61 0.68 0.67 0.82 0.80 0.72 0.76 0.53 0.80 0.64 0.50 0.58 0.75 0.85 0.87 0.87 0.87 0.83 0.91 0.93 0.96 0.73 0.56 0.68 0.57 0.77 0.91 0.92 0.84 0.82 0.92 0.77 0.72 0.67 0.88 0.77 0.81 0.77 0.43 0.82 0.72 0.87

0.70 0.67 0.43 0.70 0.65 0.70 0.83 0.78 0.74 0.74 0.65 0.85 0.73 0.69 0.56 0.83 0.78 0.81 0.82 0.85 0.86 0.81 0.85 0.91 0.74 0.50 0.67 0.52 0.70 0.84 0.85 0.84 0.75 0.89 0.75 0.72 0.71 0.85 0.83 0.83 0.80 0.72 0.85 0.81 0.88

0.73 0.71 0.83 0.64 0.66 0.79 0.85 0.71 0.66 0.60 0.71 0.64 0.73 0.84 0.61 0.83 0.64 0.72 0.77 0.67 0.75 0.74 0.66 0.88 0.59 0.66 0.65 0.40 0.65 0.68 0.74 0.68 0.65 0.82 0.86 0.82 0.99 0.98 0.89 0.93 0.83 0.94 0.94 0.92 0.92

0.64 0.63 0.80 0.55 0.58 0.77 0.83 0.70 0.67 0.64 0.71 0.72 0.70 0.79 0.69 0.85 0.61 0.67 0.73 0.66 0.76 0.77 0.70 0.87 0.70 0.64 0.66 0.33 0.62 0.76 0.80 0.78 0.68 0.83 0.77 0.71 0.85 0.80 0.86 0.80 0.78 0.85 0.75 0.80 0.90

Negative economic (NE)

Positive social (PS)

Negative social (NS)

Positive environmental (PEv)

Negative environmental (NEv)

Benets (Ben)

Supports (Sup)

a Pre: CFI 0.91, NNFI 0.90, RMSEA 0.08; Post 1: CFI 0.91, NNFI 0.90, RMSEA 0.08; Post 2: CFI 0.94, NNFI 0.93, RMSEA 0.07; Post 3: CFI 0.95, NNFI 0.95, RMSEA 0.07.

ARTICLE IN PRESS
C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480 473

competing model (Dw2 160:78; Ddf 12; po:001), so the authors decided to use the theoretical model as the base model for further testing the invariance between pre- and post-data.

residents perceptions had changed toward casino development in a specic time period.

5.1.1. Entire group analysis: Pre, Post 1, Post 2, and Post 3-data Table 3 summarizes the model comparison for group analysis. First, two models were compared to test similarity among the data in the entire data set. In Model 1, the four groups were not restricted by any of the non-xed parameters so that they had the same value across groups. In Model 2, the factor loadings were constrained to be equal across the four groups. By comparing the goodness of t between Models 2 and 1, it could be decided whether the four groups had the same loadings or coefcients from the measurement model. Table 3 shows that there was a signicant difference between the models (Dw2 406:78; Ddf 87; po:001). The result was not unexpected, since the data were collected over a 4-year period with a heterogeneous sample. In order to investigate the relationships more specically, partial group analyses were conducted by comparing Pre- and Post 1-data, Post 1- and Post 2data, and Post 2- and Post 3-data. By following this procedure, the researchers could detect whether the
Table 3 Model comparisons for group analysis Model For full group analysis: Model 1 (Free) Model 2 (LXY IN) Model 2Model 1 (Dw2 =Ddf ) For partial group analyses: Pre vs Post 1 Model 3.1 (Free) Model 3.2 (LXY IN) Model 3.2Model 3.1 (Dw2 =Ddf ) Model 3.3 (LXY IN; BE IN; GA IN) Model 3.3Model 3.2 (Dw2 =Ddf ) Post 1 vs Post 2 Model 4.1 (Free) Model 4.2 (LXY IN) Model 4.2Model 4.1 (Dw2 =Ddf ) Post 2 vs Post 3 Model 5.1 (Free) Model 5.2 (L(X)Y IN) Model 5.2Model 5.1 (Dw2 =Ddf ) Model 5.3 (L(X)Y IN, BE IN, GA IN) Model 5.3Model 5.2 (Dw2 =Ddf ) Model 5.4 (L(X)Y IN, BE IN, GA IN, TE(D) IN, PS IN) Model 5.4Model 5.3 (Dw2 =Ddf )

5.1.2. Partial group analyses: Pre- and Post 1-data The Pre- and Post 1-data were compared without any restrictions to the parameters in Model 3.1. Consequently, the authors constrained all factor loadings to be equal in these two data sets as developed in Model 3.2. The w2 difference test was conducted between Models 3.1 and 3.2, resulting in no signicant difference (Dw2 41:06; Ddf 29; p4:001). Since Model 3.2 had equality constraints in factor loadings for Pre- and Post 1-data, the results showed that two groups had an invariant structure of manifest variables for indicating latent variables. As Bollen (1989) suggested, the next step was to assess the invariance of all regression coefcients across groups by adding equality constraints in Model 3.3. This restricted hierarchical model, Model 3.3, helps to assess invariance for the general structural equation model with both latent and manifest variables (Bollen, 1989). Although Model 3.3 appeared to be a good t, the w2 difference test and practical indicator difference test showed that the t of Model 3.3 was signicantly worse than that of Model 3.2 (Dw2 121:70; Ddf 13; po:001). Thus, the hierarchical structural equation modeling step stopped in Model 3.2, suggesting that

w2

df

RMSEA

CFI

NNFI

8791.42 9198.20 406.78

2400 2487 87

0.071 0.073

0.94 0.94

0.93 0.93

4307.94 4349.00 41.06 4471.11 121.7 4536.52 4718.19 181.67 3841.16 3884.29 43.13 3900.40 16.11 4043.56 143.16

1200 1229 29 1242 13 1200 1229 29 1200 1229 29 1242 13 1303 61

0.071 0.072 0.074

0.94 0.93 0.93

0.93 0.93 0.93

0.074 0.075

0.93 0.93

0.92 0.92

0.071 0.072 0.069 0.068

0.95 0.95 0.94 0.94

0.94 0.94 0.94 0.94

ARTICLE IN PRESS
474 C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480

the residents perceptions changed before and six months after the casino opened. Table 4 presents the standardized parameter estimates for both Pre- and Post 1-data. In both data sets, support was directly signicantly predicted by positive (g 0:29 for Pre, g 0:26 for Post 1) and negative social factors (g 0:16 for Pre, g 0:13 for Post 1) as well as benet (b 0:25 for Pre, b 0:66 for Post 1) at pso0.01. Also, benet was directly signicantly affected by positive (g 0:26 for Pre, g 0:47 for Post 1) and negative economic factor (g 0:23 for Pre, g 0:08 for Post 1) at pso.01. Neither the positive nor negative social and environmental factors had signicantly affected benet in both Pre- and Post 1-data. In the Pre-data, both positive economic and environmental factors had a signicant relationship with support, but the relationship was no longer signicant in the Post 1data. 5.1.3. Partial group analyses: Post 1- and Post 2 data The Post 1- and Post 2-data were compared without any restrictions in the parameters in Model 4.1 (see Table 3). Like the previous procedure, Model 4.2 was developed by constraining all factor loadings to be equal in these two data sets. Then, the w2 difference test was conducted between Models 4.1 and 4.2, resulting in signicant difference (Dw2 181:67; Ddf 29; po:001). The hierarchical structural equation modeling step stopped in Model 4.2. The results indicated that these two groups had different structures of manifest variables
Table 4 Results of comprehensive model testing Estimates Paths PE-Ben NE-Ben PS-Ben NS-Ben PEv-Ben NEv-Ben PE-Sup NE-Sup PS-Sup NS-Sup PEv-Sup NEv-Sup Ben-Sup R2 Ben R2 Sup w2 (p-value) df CFI NNFI RMSEA Pre 0.26(4.04)a 0.23(3.22)a 0.10(1.41) 0.08(1.22) 0.01(0.15) 0.03(0.51) 0.33(5.87)a 0.02(0.34) 0.29(4.46)a 0.16(3.01)a 0.26(3.88)a 0.01(0.11) 0.25(5.06)a 0.12 0.44 2178.49(.00) 600 0.93 0.92 0.077

for indicating latent variables. The main reason for this result may be the heterogeneity of the sample over the period. Post 2-data were collected one and a half years later than the Post 1- survey. Thus, there were signicant changes in respondents in that some were excluded from and some were added to the survey. Table 4 shows that the benet factor had a signicant positive effect on support (b 0:77 for Post 2, t 12:23) at po:01: The strength of the relationship dramatically increased from the Post 1-data. The negative economic factor was no longer signicant with regard to benet, whereas the positive economic factor had a signicant positive effect on benet (g 0:38 for Post 2, t 4:90) at po:01: Unlike the previous data, both negative social and positive environmental factors had signicant impacts on benet in Post 2-data. It is interesting to note that the positive social factor no longer had a signicant direct effect on support, whereas the negative social factor remained directly signicant on support. In addition, the positive environmental factor had a signicant direct effect on support. 5.1.4. Partial group analyses: Post 2- and Post 3-data The Post 2- and Post 3-data were compared without any restrictions in the parameters in Model 5.1 (see Table 3). Consequently, constraining the all factor loadings to be equal in these two data sets was developed as Model 5.2. The w2 difference test was conducted between Models 5.1 and 5.2, resulting in no signicant difference (Dw2 43:13; Ddf 29; p4:001).

Post 1 0.47(6.12)a 0.08(2.53)b 0.11(0.86) 0.19(1.29) 0.15(1.25) 0.02(0.27) 0.04(0.57) 0.01(0.22) 0.26(3.76)a 0.13(2.67)a 0.14(1.45) 0.05(1.10) 0.66(8.87)a 0.63 2129.45(.00) 600 0.94 0.93 0.076

Post 2 0.38(4.90)a 0.03(0.86) 0.08(2.01)b 0.26(4.42)a 0.19(2.04)b 0.16(0.79) 0.03(0.51) 0.09(0.64) 0.11(1.95) 0.12(2.31)b 0.10(2.04)b 0.01(0.23) 0.77(12.23)a 0.41 0.75 1749.48(.00) 600 0.94 0.93 0.071

Post 3 0.35(4.95)a 0.07(1.32) 0.15(2.15)b 0.13(2.66)b 0.23(3.07)a 0.02(0.36) 0.05(0.88) 0.16(0.63) 0.12(2.09)b 0.10(2.66)a 0.17(3.02)a 0.03(1.11) 0.71(13.91)a 0.50 0.84 2091.68(.00) 600 0.95 0.95 0.069

R2 Fit

Note: PE: positive economic; NE: negative economic; PS: positive social; NS: negative social; PEv: positive environmental; NEv: negative environmental; Ben: benet; Sup: support. a,b indicate statistical signicance at pp.001, pp.01, respectively; t-values are in parentheses.

ARTICLE IN PRESS
C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480 475
Negative Social

Since Model 5.2 had equality constraints in factor loadings for Post 2- and Post 3-data, the results showed that the two groups had invariant structures of manifest variables for indicating latent variables. Like the comparison between Pre- and Post 1-data, the next step was to assess the invariance of all regression coefcients across groups by adding equality constraints in Model 5.3. This restricted hierarchical model, Model 5.3, helped to assess invariance for the general structural equation model with both latent and manifest variables. Although Model 5.3 appeared to be a good t, the w2 difference test and practical indicator difference test showed that the t of Model 5.3 was not signicantly different from that of Model 5.2 (Dw2 16:11; Ddf 13; p4:001). Thus, the hierarchical structural equation modeling proceeded to the next step in which the covariance matrix of error variance was xed to be equal. Model 5.4 was developed to make all lambda loadings, regression coefcients, and error variances equal across Post 2- and Post 3-data. The t of Model 5.4 was signicantly worse than that of Model 5.3 (Dw2 143:16; Ddf 61; po:001). Thus, the modeling step stopped in Model 5.3, suggesting that the residents perceptions were not much changed 2 and 3 years after the casino opened. As Table 4 presents, both Post 2- and Post 3-data revealed signicant effects from the positive economic, social, and environmental factors on benet, as well as a signicant effect from the negative social factor on benet. Also, both data sets showed signicant direct effects of negative social and positive environmental factors on support. In addition, support was directly predicted by benet in both Post 2- and Post 3-data. Fig. 2 visually illustrates Table 4 by presenting the completely standardized parameter estimates for comparison in the entire data set. Specically, common metric completely standardized regression coefcients for Post 2- and Post 3-data are presented since both data sets have invariant latent and manifest structures. 5.1.5. Hypotheses testing H1a: Positive economic impact has a positive association with perceived benet. As hypothesized, the positive economic impact had a positive signicant effect on the perceived benet throughout the entire longitudinal data of 4 year period (see Table 4). It is therefore concluded that H1a is supported. This nding indicates that residents who perceived personal benet from the casino development expressed positive economic impact. Thus, the empirical results were found to support the social exchange theory. Negative economic impact has a negative association with perceived benet. The empirical results afford partial support for this hypothesis

Negative Eviron.

-0.10(2.88) a -0.13 (2.67) a a - 0.16 (3.01)

-0.18(4.80) a -0.19 (1.29) -0.08 (1.22)

Negative Econ.

0.04(1.13) a -0.08 (2.53) b -0.23 (3.22) 0.13(2.15) b 0.11 (0.86) 0.10 (1.41)

0.73(12.57) 0.66 (8.87) a a 0.25 (5.06)

Support

Benefit

0.10(2.40) b 0.26 (3.76) a 0.29 (4.46) a

0.11 (2.11) 0.14 (1.45) a 0.26(3.88)

0.23(3.03) a 0.15(1.25) -0.01 (0.15)

Positive Social

0.05(1.25) 0.04 (0.54) 0.33 (5.87) a

Positive Environ.

0.47 (6.12) a 0.26 (4.04) a

0.37(4.95) a

Significant for all data Significant in post-2 & 3 Not significant in post-2 & 3

Positive Econ.

Fig. 2. Standardized maximum-likelihood parameter estimates for the hierarchical model: comparison among Pre-, Post 1-, Post 2- and Post 3- data. a,b indicate statistical signicance at pp.001, pp.01, respectively; t-values are in parentheses; italic gure denotes pre, non-italic gure denotes Post 1 and underlined italic gure denotes standardized common metric coefcient estimates of Post 2 & 3. Dashed lines indicate paths that are not signicant at .05 or better; plane lines indicate paths that are only signicant for Post 2 & 3-data; and bold lines indicate paths that are signicant for all the data.

H2a:

H2b:

H1b:

(see Table 4). That is, the negative economic impact had a negative signicant effect on the perceived benet in Pre- and Post 1- data, whereas it had an insignicant effect on benet in Post 2- and Post 3-data. Positive social impact has a positive association with perceived benet. The empirical results afford partial support for this hypothesis (see Table 4). The positive social impact had a positive signicant effect on the perceived benet in Post 2- and Post 3-data, whereas it had an insignicant effect on benet in Pre- and Post 1-data. This nding implies that the positive social impact of casino development on benet does not appear in the early stage of development. Negative social impact has a negative association with perceived benet. The empirical results afford partial support for this hypothesis (see Table 4). The negative social impact had a negative signicant effect on the perceived benet in Post 2- and Post 3-data, whereas it had an insignicant effect on benet in Pre- and Post 1-data. This nding also implies that the

ARTICLE IN PRESS
476 C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480

H3a:

H3b:

H 4:

H5a:

H5b:

H6a:

negative social impact of casino development on benet does not appear in the early stage of development. Positive environmental impact has a positive association with perceived benet. The empirical results afford partial support for this hypothesis (see Table 4). The positive environmental impact had a positive signicant effect on the perceived benet in Post 2- and Post 3-data, whereas it had an insignicant effect on benet in Pre- and Post 1-data. Negative environmental impact has a negative association with perceived benet. The empirical results indicated that the negative environmental impact did not have a signicant effect on the perceived benet throughout the entire longitudinal data (see Table 4). So, H3b is not supported. Perceived benet has a positive association with support. As hypothesized, the perceived benet had a positive signicant effect on support throughout the entire longitudinal data (see Table 4), thus supporting H4. This nding indicates that residents who perceived personal benet from casino development expressed their support for its development. Thus, the empirical results were found to support the social exchange theory. This nding also indicates that benet perceived by residents is the most inuencing variable for supporting the casino development. Positive economic impact has a positive association with support. The empirical results indicated that the positive economic impact generally did not have a signicant effect on support throughout the entire longitudinal data except Pre-data (see Table 4). So, H5a is not supported. The results indicated that the positive economic impact inuenced support at the pre-opening stage of casinos but it had no more inuence on support at the post-opening stages. This might be attributed to the fact that the residents perceived a positive economic impact in the post-opening periods that was not greater than they expected in the early stage. Negative economic impact has a negative association with support. The empirical results indicated that the negative economic impact did not have a signicant effect on support throughout the entire longitudinal data (see Table 4). So, H5b is not supported. This nding implies that the residents do not perceive the negative economic impact as an important variable for support. Positive social impact has a positive association with support. The empirical results indicated

H6b:

H7a:

H7b:

that the positive social impact generally had a signicant effect on support throughout the entire longitudinal data except Post 2-data (see Table 4). So, H6a is generally supported. Negative social impact has a negative association with support. As hypothesized, the negative social impact had a negative signicant effect on support throughout the entire longitudinal data (see Table 4). It is therefore concluded that H6a is supported. This nding indicates that those residents who perceived negative social impact from the casino development did not support its development. Thus, the empirical results were found to support the social exchange theory. This nding also indicates that negative social impact (e.g., gambling addicts, bankruptcy, divorce, crime) as perceived by residents is the most inuential variable in determining support for casino development. Positive environmental impact has a positive association with support. The empirical results indicated that positive environmental impact generally had a signicant effect on support throughout the entire longitudinal data except Post 1-data (see Table 4). So, H6a is generally supported. Negative environmental impact has a negative association with support. The empirical results indicated that the negative environmental impact did not have a signicant effect on support throughout the entire longitudinal data (see Table 4). So, H7b is not supported. This nding implies that the residents do not perceive the negative environmental impact as an important variable for support.

Table 5 represents the result of analysis of variance for testing Hypotheses 8 and 9: H 8: Perceived benet will be signicantly decreased over time after the certain level of carrying capacity. Support will be signicantly decreased over time after the certain level of carrying capacity.

H 9:

Both perceived benets and support level signicantly dropped from pre-data to Post 2-data. However, they did not change signicantly from Post 2-data to Post 3data. In other words, residents had great perceptions about potential benets of casino development before it opened and signicantly lowered their perceptions 6 month later. Then, they slightly increased the perceived benet level as well as support level after being exposed to the casino for 2 years. Then, during the third year,

ARTICLE IN PRESS
C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480 477

Pillai Trace 1.53(po0.000); Wilks Ramba 0.12 (po0.000);HotellingLawley Trace 3.12(po0.000); and Roys Greatest Root 1.17(po0.000).

residents felt consistently about the perceived benets and support. Thus, Hypotheses 8 and 9 were somewhat supported up to Post 2-data, but not for Post 3-data. The residents seemed to have constant attitudes toward the casino development after 2 years of operation. The researchers could conclude that this was a sign of social carrying capacity level and may predict drops in the near future.

IIIIV

0.07 0.14

6. Conclusion The results of the longitudinal analysis reported here shed light on our understanding of resident perceptions of casino development, which has not been explored in past research. First, residents revealed signicantly different perceptions about the casino before and after its development. According to the pre-data, residents support was predicted by positive economic, positive and negative social, and positive environmental factors as well as by benet. As Table 4 (see also Fig. 2) indicates, respondents supported the casino when they perceived receiving personal and community benets (b 0:25; t 5:06) as based on social exchange theory. Among those signicant factors in addition to benet, the positive economic factor had the strongest direct effect on support (g 0:33; t 5:87). In other words, when residents perceived a positive economic factor, such as enhancing investment and business, employment opportunities, income, and tourist spending, their support level increased dramatically. However, when they were concerned about negative social factors, such as the occurrence of gambling addiction, bankruptcies, and alcoholism, their support level decreased (g 0:16; t 3:01). Positive and negative economic factors also had indirectly signicant effects on support mediated by benet. Second, 6 months after the casino opened, Post 1-data showed some minor changes in residents perceptions about support. Positive economic and environmental factors had no signicant direct effects on support. Specically, the positive economic factor had a signicant effect on support only when mediated by benet (g 0:47; t 6:12; b 0:66; t 8:87). This result can be interpreted as residents non-exposure to the casino their perceptions of the positive economic factor did not cause them to support it. Rather, they tended to become more supportive when they actually received economic benets. In addition, the positive effect of benet on support dramatically increased from 0.25 to 0.66. The positive impact of receiving actual personal and community benets became even stronger after the casino opened. However, positive and negative social factors were less likely to predict residents level of support for the casino 6 months after its opening. Before the casino opened, people were much concerned about the social

IIIII

0.24 0.20

Table 5 ANOVA test of benet and support over 4-year period

Benet 2.68a Support 3.20

po.05.

Mean values were computed on the basis of 5-point Likert scale(1: strongly disagree, 3:neutral, 5: strongly agree).

Scheffe multiple range tests

Pre (I) Post 1 (II) Post 2 (III) Post 3 (IV) F-value

III

2.41 2.80

2.64 3.00

2.58 2.87

8.50 17.84

0.27 0.40

ARTICLE IN PRESS
478 C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480

impacts of a casino on their community. As a result, they might have exaggerated both the positive and negative social impacts on their community. When they actually experienced the casino development, their perceptions of social impacts less signicantly affected their support level. Third, residents changed their perceptions dramatically 2 years after the casino opened. A comparison of Post 1- and Post 2-data revealed signicant changes in residents perceptions. The positive relationship between benet and support was further strengthened (b 0:77; t 12:23). Both the negative social and positive environmental factors had signicant direct effects on support (g 0:12; t 2:31; g 0:10; t 2:04). Residents might be still seriously concerned about gambling addiction due to the increasing number of problem gamblers within the community who might experience bankruptcies, destruction of families, and suicides. Also, residents were more concerned about positive environmental factors. As far as benets were concerned, residents seriously cared about positive economic, social, and environmental factors as well as negative social factors. Residents believed that a positive social factor, such as enhancements to quality of life, educational environment, and community spirit, signicantly affected benet. Once they actually received their benet, then they would support the casino (b 0:77; t 12:23). This could be applied in the same manner to the positive economic factors. In addition, residents perceived a signicant negative social impact on benet (g 0:26; t 4:42). Once they experienced the casino, they started to believe that social problems might discourage or decrease the number of job opportunities, personal, and community benets. It is also interesting to note that residents no longer showed a signicant effect of negative economic factors on benets. This might be attributed to the fact that prior to the opening of a casino, residents had not seriously experienced negative economic factors, such as increases in cost of living and tax burdens, as experienced after the casino opened. Last, Post 3-data were collected 1 year after the collection of Post 2-data. Interestingly, the structural equations were not statistically different between the two sets of data. In addition, the mean scores of perceived benets and support level did not change signicantly based on the ANOVA results. In other words, residents perceptions did not signicantly change 2 years after the casinos opening and thereafter. The results suggested that residents had stable attitudes toward casino development after a certain point. Benet was still the strongest factor in determining support. Positive economic, social, and environmental factors showed strong direct effects on benet, whereas negative social and positive environmental factors had both signicant direct and indirect effects on support.

To summarize, the social exchange model ts the entire data set very well in explaining residents attitudes toward casino development. The results showed that residents had relatively stable perceptions of the casino and its impacts 3 years after it opened in their community. The positive economic impact was most signicant in determining the benet. Also, respondents perceived signicant effects of negative social impact on both benet and support. Furthermore, benet was the most important determinant in residents support for casino development according to the entire data set. Based on the results, both social exchange theory and social carrying capacity theory were generally conrmed.

7. Implication The results of this longitudinal study can be used in evaluating changes to residents perceptions. That is, this longitudinal study might be useful in evaluating whether the impacts of casino operations proceed as desirably as the residents rst expected before the casino development. Otherwise, policy-makers can take appropriate actions to make the community a pleasant place to live and to improve the quality of life for residents. Thus, this study has specic implications for tourism and social development as follows. First, policy-makers should identify how to provide benets to local residents. Due to the important role of the benet factor in this longitudinal study, local government could increase job opportunities and income levels by developing a number of tourism projects, including building lodging facilities and restaurants, hosting seasonal cultural events, and constructing a mining themed-park. Specically, developing the mining communities as themed villages would be an additional attraction to the project and might favor both residents and casino customers in terms of creating more jobs and more leisure activities, respectively. Second, casino operators and policy-makers should make efforts to minimize the negative social factors. Since the results of this study showed the signicant effect of negative social factors, casino operators and policy-makers should develop an effective strategic planning for prevention program of problem gambling in order to minimize the negative social impacts of casino on the residents. Currently, the Kangwon Land Casino is operating its own problem gambling center. However, the center is facing many difculties with funds and personnel. Since two or three ofcials of the Ministry of Culture and Tourism supervise casinos, it is strongly recommended that the Korean government organize the gaming control board so that the casino industry may be operated in a more effective way to promote responsible gambling. In addition, by abiding

ARTICLE IN PRESS
C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480 479

by strict regulations and supervision, the casino industry will come up with more creative ideas and effective gambling clinics. Also, like the Alberta Alcohol Drug Abuse Commission (AADAC) in Canada, the Korean government could operate a problem gambling center at the state level or establish network systems with mental clinics to help problem gamblers effectively. Third, in order to maximize the positive social impacts to the community, policy-makers should consider creating a positive educational environment. By enhancing the educational environment, local residents children will stay in the community and receive an equivalent level of education to that found in metropolitan areas. Consequently, the size of the community will not decrease, but rather will be increased because those educated residents will decide to stay when they have more job opportunities in their communities. Furthermore, local government in partnership with the Kangwon Land Casino should consider developing a community college or education center to train local residents to work in the industry (i.e., jobs for casino dealers, hotel and restaurants, ski slopes, golf courses). This study is limited in that the sample was not all panel data. Hence, the results should be interpreted with caution. If possible, future research should use the panel data over the whole period of the research in order to more accurately examine changes in residents perceptions in the longitudinal pattern. In particular, by evaluating residents perceptions on a continual basis, the researcher could clearly identify the maximum carrying capacity level of the communities from the resident perspective and check whether the phenomenon becomes positive or negative.

Acknowledgement We wish to acknowledge the nancial support of the 55th Kyung Hee University Anniversary Research Promotion Fund.

References
Allen, L. R., Long, P. T., Perdue, R. R., & Kieselbach, S. (1988). The impact of tourism development on residents perceptions of community life. Journal of Travel Research, 27(1), 1621. Andereck, K. L., & Vogt, C. A. (2000). The relationship between residents attitudes toward tourism and tourism development options. Journal of Travel Research, 39(1), 2736. Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: a review and recommendation two-step approach. Psychological Bulletin, 103(3), 411423. Ap, J. (1990). Residents perceptions research on the social impacts of tourism. Annals of Tourism Research, 17(4), 610616. Ap, J. (1992). Residents perceptions on tourism impacts. Annals of Tourism Research, 19(4), 665690.

Bentler, P., & Bonett, D. (1980). Signicance tests and goodness of t in the analysis of covariance structures. Psychological Bulletin, 88(3), 588606. Bollen, K. (1989). Structural equations with latent variables. New York: Wiley. Browne, M., & Cudeck, R. (1993). Alternative ways of assessing model t. In K. A. Bollen, & J. S. Long (Eds.), Testing structural equation models (pp. 445455). Newbury Park, CA: Sage. day, L., & Zeiger, J. (1991). The social, economic, and Cane environmental costs of tourism to a gambling community as perceived by its residents. Journal of Travel Research, 30(Fall), 4548. Carmichael, B., Peppard, D., & Boudreau, F. (1996). Mega-resort on my doorstep: local resident attitudes toward Foxwood casino and casino gambling on nearby Indian Reservation Land. Journal of Travel research, 34(3), 916. Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18(1), 3950. Getz, D. (1994). Residents attitudes towards tourism: a longitudinal study in Spey Valley, Scotland. Tourism Management, 15(4), 247258. Giacopassi, D., Nichols, M., & Stitt, B. (1999). Assessing the impacts of casino gambling on crime in Mississippi. American Journal of Criminal Justice, 18(1), 117131. Hsu, C. (2000). Residents support for legalized gaming and perceived impacts of riverboat casinos: changes in ve years. Journal of Travel Research, 38(4), 390395. Jo reskog, K., & So rbom, D. (2001). LISREL 8: Users reference guide. Chicago, IL: Scientic Software International. Jurowski, C., Uysal, M., & Williams, D. R. (1997). A theoretical analysis of host community resident reactions to tourism. Journal of Travel Research, 36(2), 311. Kangwon Land Casino (2001). Casino visitors and revenues (internal report). Kangwon Province: Kangwon Land Casino. Kangwon Land Casino (2004). Casino visitors and revenues (internal report). Kangwon Province: Kangwon Land Casino. Kangwon Province (2000). Evaluation report on initial operation of Kangwon Land Casino. Kangwon Province: Internal government material. King, B., Pizam, A., & Milman, A. (1993). Social impacts of tourism: host perceptions. Annals of Tourism Research, 20(4), 650665. Lee, C. K., & Back, K. J. (2003). Pre- and post-casino impact of residents perception. Annals of Tourism Research, 30(4), 865885. Lee, C. K., Kim, S. S., & Kang, S. Y. (2003). Perceptions of casino impactsa Korean longitudinal study. Tourism Management, 24(1), 4555. Lindberg, K., & Johnson, R. L. (1997). Modeling resident attitudes toward tourism. Annals of Tourism Research, 24(2), 402424. Liu, J., & Var, T. (1986). Resident attitudes toward tourism impacts in Hawaii. Annals of Tourism Research, 13(2), 193214. Long, P. (1996). Early impacts of limited stakes casino gambling on rural community life. Tourism Management, 17(5), 341353. Long, P., Perdue, R., & Allen, L. (1990). Rural resident tourism perceptions and attitudes by community level of tourism. Journal of Travel Research, 28(3), 39. Mason, P., & Cheyne, J. (2000). Residents attitudes to proposed tourism development. Annals of Tourism Research, 27(2), 391411. McCool, S., & Martin, S. (1994). Community attachment and attitudes towards tourism. Journal of Travel Research, 32(3), 311. Milman, A., & Pizam, A. (1988). Social impacts of tourism on central Florida. Annals of Tourism Research, 15(2), 191204. Nunnally, J., & Bernstein, I. (1994). Psychometric theory (3rd ed.). New York: McGraw-Hill.

ARTICLE IN PRESS
480 C.-K. Lee, K.-J. Back / Tourism Management 27 (2006) 466480 Pizam, A. (1978). Tourisms impacts: the social costs to the destination community as perceived by its residents. Journal of Travel Research, 16(4), 812. Pizam, A., & Pokela, J. (1985). The perceived impacts of casino gambling on a community. Annals of Tourism Research, 12(2), 147165. Roehl, W. (1999). Quality of life issues in a casino destination. Journal of Business Research, 44(3), 223229. Perdue, R., Long, T., & Allen, L. (1990). Resident support for tourism development. Annals of Tourism Research, 17(4), 586599. Perdue, R., Long, T., & Kang, Y. (1995). Resident support for gambling as a development strategy. Journal of Travel Research, 34(2), 311. Perdue, R., Long, T., & Kang, Y. S. (1999). Boomtown tourism and resident quality of life: the marketing of gaming to host community residents. Journal of Business Research, 44(3), 165177.