Predicting Attendance At southeastern U.s. Amusement Parks Using
Transcription
Predicting Attendance At southeastern U.s. Amusement Parks Using
Predicting Attendance At Southeastern U.S. Amusement Parks Using Accessibility- and Amenity-Based Models James Andrew McCall Miami-Dade County Park & Recreation Department Miami, Florida 33128 Joe Weber* Department of Geography University of Alabama Tuscaloosa, AL 35487 E-mail: [email protected] * Corresponding author The Geographical Bulletin 47: 45-59 ©2006 by Gamma Theta Upsilon ABSTRACT This research examines the extent to which traditional geographic concepts of proximity can explain attendance to amusement parks in the Southeastern United States. Accessibility models were calculated for nineteen parks using both miles and driving times as measures of distance from population centers. Results show only weak relationships between attendance and distance from population. To test whether other factors are more important predictors of attendance, regression analysis was carried out using variables describing amenities and characteristics of the parks. The results show that a high level of explanation for attendance to parks is provided by ticket price, parking cost, the number of years the park has been open, and centrality within the Southeast. Geographic proximity to population does not appear to be important to park attendance. Key Words: Amusement Parks, Accessibility, Distance, Attendance Introduction Tourism in the Southeast has been associated with all manner of colorful and wellknown attractions and locales, including not only Disney World but the beach resorts of the Florida panhandle (Hollis 2004), Mammoth Cave in Kentucky (Algeo 2004), the Smoky Mountains (Tooman 1997), assorted roadside and urban attractions (Jakle 1985; Hollis 1999; Newman 1999; Starnes 2003), and amusement parks. The latter can be defined as entertainment-oriented businesses, usually requiring admission fees, containing attractions, rides, food, and shopping, and other entertainment (Adams, 1991). These amusement parks are enclosed and separated from the tasks of everyday life, and range in size from small family-owned business to vast corporate parks. Theme parks involve the application of an overall theme to a particular amusement park. In the early twentieth century amusement park development was strongly tied to local urban populations, but since the 1950s the 45 James Andrew McCall and Joe Weber rebirth of the park industry has been tied more to automobile travel and regional or even international tourist travel (Adams 1991). Increasing ease of long distance travel, and the growth of tourist complexes centered around beaches, national parks, or large cities has made any expectation that local populations will make up a large component of total attendance appear doubtful. Despite this, proximity to population remains a concern for amusement park operators (Adams 1991). This research will examine accessibility of theme parks to the surrounding population in the Southeast U.S. and relate this to attendance, with the goal of understanding the importance of population proximity to park attendance patterns. THEME PARK DEVELOPMENT IN THE SOUTHEASTERN U.S. The origins of amusement parks have been traced back to medieval trade fairs that became popular for their entertain- ments, to European pleasure gardens from the eighteenth and nineteenth centuries, and to the Chicago Columbian Exposition of 1893 (Adams 1991; Younge 2002). This Exposition included the first Midway with food and themed entertainment as well as an overarching theme for the entire event. The first enclosed amusement parks with rides, games, food, and other entertainment appeared on Coney Island in the 1890s, and these soon spread throughout the country (Adams 1991). These early parks were located within urban areas at locations well served by streetcar or subway lines (and in fact were often created by streetcar lines to generate passengers), and benefited from shorter working weeks and rising incomes during the early 20th century. By 1920, there were as many as 2000 amusement parks in the U.S., but they began a steady decline through the automobile era, Great Depression, the Second World War, and the television era (Adams 1991). Figure 1: Amusement Parks in the Southeastern United States 46 Predicting Attendance At Southeastern U.S. Amusement Parks Using Accessibility- and Amenity-Based Models The return of the amusement parks to importance, or, in the terminology of the destination life-cycle model, their rejuvenation (Butler 1980; Tooman 1997), is due to Disneyland and the success of theming, with their attention to a safe, clean environment, and a mix of rides and attractions for all family members (Adams 1991). New parks, often controlled by large corporations, soon followed, with the 1970s seeing the construction of most of the postwar theme parks now operating. Many of these parks cater to vacationers from distant locations rather than local residents, and are typically located in a suburban location, adjacent to freeways. Since the opening of Disneyland, no new theme park has succeeded in a traditional urban location (Adams 1991). Within the Southeast United States a number of amusement parks exist (Fig. 1). There is great diversity among these parks, from small family-owned facilities to large corporate structures. Six Flags Over Georgia opened in 1967 outside of Atlanta and was the first modern theme park in the Southeast (Adams 1991). Parks in this area range in size from the five-acre Family Kingdom Amusement Park in Myrtle Beach, South Carolina, to Paramount’s colossal 400-acre King’s Dominion amusement park in Richmond, Virginia. There is also great variation in the number of rides and attractions offered at each park (Table 1). Several of the smaller parks contain only one roller coaster and a few adult rides. Large corporate parks owned by Six Flags and Paramount have as many as twelve roller coasters and as many as thirty adult rides and attractions. In addition, at least eighty Southeastern amusement parks have closed their gates in the twentieth century (Samuelson and Yegoiants 2001; Styer 2005). Even with close proximity to urban populations, at least five amusement parks built in Atlanta failed, while Nashville also experienced at least five closings. Baltimore has been historically the worst place in the Southeast to build amusement parks, with at least twelve defunct parks. In addition, there have been failed attempts to create new parks, as when the Walt Disney Company announced in 1993 plans to build Disney America in Virginia, but abandoned the idea after strong public opposition (Moe and Wilkie 1997). GEOGRAPHIC ANALYSIS OF THEME PARKS Tourist oriented areas, including amusement and theme parks, have been examined from a range of perspectives in recent decades (Jakle 1985; Adams 1991; Marling 1997; Young 2002). Theme parks, especially Disneyland and Disney World, have attracted a great deal of attention for how they have been designed, their impact on guests, and for their economic and political impact on surrounding communities (Zukin 1991; Findlay 1992; Sorkin 1992; Francaviglia 1996; Warren, 1996; Archer 1997; Marling 1997; Foglesong 2001; Mannheim 2002; Young and Riley 2002). Critiques of downtown commercial-entertainment districts in large cities have used the metaphor of theme parks to describe these areas and the way they structure activities (Warren 1994; Hannigan 1998; Young 2002; Bryman 2004). Theming is not limited to amusement parks or downtowns, and can be applied to entire cities or small towns, such as the UFO-oriented theme developed for Roswell, New Mexico (Paradis 2002). Rather than examining the cultural meaning and design of theme parks, the goal here is to better understand attendance and their relationship to local population patterns. There is a limited amount of geographical research and analysis of amusement park attendance patterns in the U.S., and no recent studies have been published on amusement park attendance. This research addresses this shortage by examining the extent to which traditional geographic concepts of proximity are useful for explaining yearly attendance to amusement parks in twelve states of the Southeastern U.S. Given the increasing scale of amusement parks, the greater the distance they attempt to attract visitors from, and the longer time these visitors will remain, it is tempting to ask whether traditional approaches to accessibility will remain relevant to them. An analysis of attendance at amusement parks is also useful because it provides 47 48 Nashville, TN Charlotte, NC Doswell, VA Upper Marlboro, MD Louisville, KY New Orleans, LA Austell, GA Bessemer, AL Gulf Shores, AL Valdosta, GA Opryland USA Paramount’s Carowinds Paramount’s King’s Dominion Six Flags America Six Flags Kentucky Kingdom Six Flags New Orleans Six Flags Over Georgia Visionland Waterville USA Wild Adventures Theme Park 170 20 70 245 140 59 345 400 105 70 140 Hot Springs, AR 67 5 Magic Springs Maggie Valley, NC Ghost Town in the Sky 88 Myrtle Beach, SC Family Kingdom Amusement Park 100 150 Pigeon Forge, TN Dollywood 26 360 Memphis, TN Huntington, WV Libertyland Williamsburg, VA Camden Park 368 1997 1987 1998 1967 2000 1990 1982 1975 1973 1972 1978 1976 1925 1961 1992 1961 1903 1975 1898 304 150 105 151 122 145 135 130 120 145 60 65 80 135 200 215 93 167 95 Park Size (Acres) Year Opened Days Open Lake Winnepesaukah Amusement Park Rossville, GA Bowling Green, KY Busch Gardens Williamsburg Location Beech Bend Park Amusement Park Table 1. Characteristics of Southeastern Amusement Parks 1,250,000 175,000 400,000 2,250,000 1,095,000 1,200,000 1,550,000 2,090,500 1,850,000 2,000,000 280,000 250,000 375,000 140,000 80,000 2,300,000 105,000 2,600,000 400,000 Annual Attendance 9 1 2 9 6 7 9 12 11 6 5 3 3 2 2 3 3 6 1 Roller Coasters 26 6 11 21 17 19 20 18 11 12 12 7 16 6 12 11 15 8 12 Adult Rides yes yes yes yes no yes yes yes yes no yes no no no yes yes no yes no Nearby Water Park yes no yes yes yes yes yes yes yes yes yes yes yes yes no yes yes yes no Shows 19 2 5 21 16 15 24 7 30 17 0 7 0 1 0 8 1 6 0 Restaurants 6 0 4 19 12 16 10 7 2 3 1 1 0 1 0 9 1 9 0 Shops 2002 2002 2001 2002 2000 2001 2002 2002 2002 1997 2002 2002 2002 2002 2002 2002 2002 2002 2002 Year Data Available James Andrew McCall and Joe Weber Predicting Attendance At Southeastern U.S. Amusement Parks Using Accessibility- and Amenity-Based Models a perspective on the aspects and amenities at parks that are most attractive to visitors. DISTANCE, POPULATION, AND ACCESSIBILITY TO THEME PARKS Proximity to population is of fundamental concern to theme park developers and operators, with close populations given greater weight than more distant potential visitors. For example, the local population might be counted “within radiuses of 25, 50, 100, and 150 miles of a potential site” (Adams 1991, p. 109). A market penetration model refines this into the primary, secondary, and tertiary markets (Miami Metrozoo 2003). The size of each market area is based on the percentage of residents who will visit an attraction. The primary market for a theme park might be a radius of 25 miles, while the secondary market would include the population between one and two hours driving time, and the tertiary market consists of populations (especially in larger cities) located between three and four hours driving time away. The tertiary market is however quite variable, with little or none for a small attraction, and the entire world for Disney World. However, the use of discrete distance intervals can be problematic, for it has been shown in a wide variety of situations that interaction falls off at a non-linear rate, an effect commonly known as distance decay. This idea has long been a part of recreation planning (Clawson and Knetsch 1966; Hanink and White 1999). One study concluded that, “The average number of trips made to Six Flags Over Georgia in 1968 decreased according to an exponential function as the travel time to Six Flags Over Georgia increased” (Dyer 1970, p. 33). In their study of the effects of distance on travelers’ attendance to the Great Smoky Mountains National Park, Cole and Mitchell (1969, p. 14) conclude that attendance is “a negative function of distance.” However, in another study it was found that distance decay had only limited power to explain attendance, in this case to a Saudi Arabian national park, due perhaps to the uneven local population distribution and importance of distant urban populations (Paul and Rimmawi 1992). More recently, Meyer-Arendt (1997) used a version of the traditional gravity model to explain attendance at casinos in Mississippi. This strongly suggests that amusement parks can be examined in the same fashion, with parks being patronized primarily by local populations. Both travel costs (measured linearly with distance) and area and amenities of water recreation sites were found to be important (Siderelis and Moore 1998). Travel times have been used to predict attendance at recreation sites with some success (Brainard, Lovett, and Bateman 1997). This study used estimates of highway driving time and the origin locations of visitors to a park. It can be expected that incorporating information about the attractiveness of recreation destinations as well as the presence of competing destinations will improve such an approach. However, there are increasing reasons to suspect that traditional proximity-based approaches might no longer be valid in the case of theme parks. The growth of commercial air travel and the construction of the Interstate Highway System have shrunk travel times between cities, resulting in space-time convergence (Janelle 1968). This has been mapped for the Southeast U.S., showing greatly reduced driving times between 1950 and 1975 (Carstensen 1981). In response to faster speeds of travel (resulting in greater accessibility to major cities), amusement parks in larger cities may experience increased patronage due to a wider area from which to draw visitors. Similar issues apply to tourist attractions that increasingly function on a global scale (Williams 1998). An important issue with tourism is that travel can be an important and valued part of the vacation experience, and so cannot necessarily be expected to show a distance decay effect for all forms of travel (Hanink and White 1999). In the case of theme parks, which may be infrequently visited or are part of a larger vacation, this appears to be a strong possibility. While this is very likely the case with Disney World and adjacent parks, it may also be the case for many other Southeastern parks. A shrinking importance for distance could also be the result of the 49 James Andrew McCall and Joe Weber increasing role of new and larger attractions at parks. Theme parks increasingly compete by building new rides, especially thrill rides such as roller coasters (Newman 2004). These new rides are essential for attracting repeat customers (Braun and Soskin 1999), whose loyalties may not be based on proximity. These conflicting possibilities for the importance of proximity to population will be tested with data about amusement park attendance in the Southeast. STUDY AREA AND DATA This study examines amusement parks located in the Southeastern United States, excluding those in Florida. Amusement parks were identified as entertainment-oriented attractions that charged admission and had at least one roller coaster. There are nineteen amusement parks distributed throughout the twelve states of this area, with the majority of the parks located near large cities (Table 1). This is not surprising given the emphasis Figure 2: Attendance at Amusement Parks Figure 2: Attendance at Amusement Parks 50 on proximity to population. Two parks are located adjacent to the Great Smoky Mountains National Park, which is the most visited National Park in the country and clearly an ideal place to attract tourists. Since Florida has many large amusement parks, it may seem unusual that it is not included. However, several of Florida’s amusement parks, especially Walt Disney World in Orlando, are attractive globally, whereas most parks in the study are attractive only regionally. Consequently, the inclusion of amusement parks in Florida would greatly skew the results. In addition, three parks were excluded because they offer free admission and are located along public piers that also contain restaurants, clubs, and shopping attractions. Yearly attendance estimates for these piers were reported in local papers, but it is impossible to know how many of the people that visited the pier actually visited the amusement park. Finally, one additional park was excluded because no attendance es- Predicting Attendance At Southeastern U.S. Amusement Parks Using Accessibility- and Amenity-Based Models timates were found and park officials refused to release attendance data. While Opryland is no longer open, it represents a modern and recently operating theme park and so was included in the study. For each park, the attendance data used in the analysis were for the most recent year available. Figure 2 shows attendance at parks in the study area. Tourist attendance is volatile over time, and related to many factors that extend beyond the attributes of the destination (Frechtling 2001). This represents the effective demand, and not the potential demand for these destinations, represented by the total population interested in attending such a park (Hall and Page 1999). While there are several variables that could be used to represent attendance, here the number of visitors per year is used. Finding attendance data for amusement parks can be quite troublesome, as “many amusement parks by policy do not reveal or report attendance figures even to industry associations . . .” (Adams 1991, 168). The following sources were used to compile attendance data: Amusement Business (ten parks), articles in local papers furnished by local visitors’ centers and tourism development councils (seven parks), and phone interviews with public relations personnel (two parks). ACCESSIBILITY OF SOUTHEASTERN PARKS Accessibility is a fundamental concept in geography, which in its traditional proximitybased form relates a set of origin locations to one or more destination locations (Pirie 1979; Kwan and Weber 2003). A number of different accessibility measures exist, depending on how they take into account distance and whether they incorporate the size of the destination. The Shimbel or network accessibility measure, which was first introduced to geography by a pioneering analysis of accessibility in the Southeast (Garrison 1960), measures the highway distance from each city to all other cities. Lower total mileage values represent greater accessibility to other cities. This has been a common means for measuring accessibility among cities (Gauthier 1968; Murayama 1994; and Spence and Linneker 1994). Another set of accessibility models have been termed population potential or Hansen measures, first developed in the 1940s (Pooler 1987) and suggested as an accessibility measure by Hansen (1959). These measures take into account not only the distance from the origin to destinations, but the size of destinations. In this case, the closer a park is to more and larger population centers, the higher its accessibility. The population potential model can be written as follows: Ai = k Pj D ij b where accessibility of park i is equal to the sum of the population of each center (Pj) divided by distance from the park to that center (Dij b). k represents a calibration constant. Distance is raised to a power represented by b, which allows for non-linear distance decay. This is similar to the formulation of distance decay in gravity models for spatial interaction, with higher b values meaning that accessibility will fall off more rapidly with distance. This population potential accessibility measure has been used in many recreation applications, including accessibility to golf courses (Mitchelson and Lazaro 2004), and the accessibility of national parks to population (Hanink and White 1999; Hanink and Stutts 2002). It has been suggested for evaluating the accessibility of wilderness areas to population, although because support for wilderness areas tends to be greatest among those living farthest away, multiplying the attractiveness of the wilderness by distance could actually give a more suitable measure for the value of wilderness to society (Hanink 1995). The population potential model was calculated using ArcView 3.2 Geographic Information System (GIS). County centroids were used to represent population centers, using population values from the 2000 Census. A short script was written to allow ArcView to calculate population potential models between each amusement park and every county in the study area. To account for visitors to parks from outside the study 51 James Andrew McCall and Joe Weber area, a 200-mile buffer was created around the entire study area, and counties inside this buffer added to those in the Southeast. Both highway mileage and driving time were used to represent distance. Doing so allows some indication of how households choose parks. While travel times to parks can be important given long trips, mileage may be of more importance to households on long vacations. In both cases the 2003 National Transportation Atlas Database (Bureau of Transport Statistics 2003), which includes GIS representation of all major highways in the Southeast, was used for the highway network. Because there is no highway driving time data present in this dataset, travel times were estimated for each highway link. To do this, average driving speeds were estimated for different classes of roads, a common approach at the interurban level (Carstensen 1981; Brainard, Lovett, and Bateman 1997). Average speeds of 62.3 mph were computed for Interstates and 52.1 mph for other major highways based on average values for these highways throughout the Southeast (Rand McNally 2004). These average speeds were converted to driving times and assigned to all links in the street network database. In addition to calculating population potential using travel times and miles, accessibility was measured using different distance decay values. In the case of gravity Table 2. Results of Correlation Analysis Measure of Distance Beta Value Correlation Coefficient Miles 1 0.485a Miles 1.5 0.457a Miles 2 0.397 Miles 2.5 0.362 Miles 3 0.347 Travel Time 1 0.488a Travel Time 1.5 0.456a Travel Time 2 0.397 a Statistically Significant at p <=.05 52 models there has been considerable discussion about the proper b value (Pooler 1987; Mikkonen and Luoma 1999). Because there is no expectation about exactly at what rate proximity to population will decrease as a potential influence on park attendance, b parameters ranging from 1 to 3 were used for mileage, and 1 to 2 with travel time. If distance is cubed instead of being squared in the model, lower predictions of attendance should result. EXPLAINING AMUSEMENT PARK ATTENDANCE The accessibility results were tested to see whether variations in attendance were strongly related. Pearson correlation analysis was used to test the accuracy of the population potential accessibility model predictions with the actual yearly attendance data for each park. Network distances and travel times were used, with several beta values (Table 2). The best results were obtained with linear distance, which shows very clearly that distance decay is not as great a factor for amusement park attendance as it is for other kinds of trips (Hanson and Schwab 1995). However, because amusement park trips are likely to be rare for most households, it should not be surprising that other considerations are far more important. For travel times a beta exponent of 1 again provided the most accurate prediction of attendance to amusement parks in the Southeast. The fact that travel time provides a higher level of explanation than distance must be due to the unevenness of travel times on the highway system, as expected from space-time convergence (Carstensen 1981). Although proximity to population provides a moderate level of explanation for attendance, a variety of other factors can also be expected to contribute to attendance at amusement parks. These could include an amusement park’s location near physically attractive features like beaches or mountains; the possibility of attending an adjacent water park, restaurants, or shopping center; elaborate marketing schemes by parks and local tourism developers; incentives to bring large institutionalized tourist groups to the park; Predicting Attendance At Southeastern U.S. Amusement Parks Using Accessibility- and Amenity-Based Models cheaper admission rates for senior citizens and for stays longer than one day; and the appeal of amusement parks as a perfect family vacation destination. These are a more restricted set of variables than might be used for more general tourism studies, but include destination attractiveness, distance, and price (Frechtling 2001). And like national parks, amusement parks include a wide range of features that appeal to different segments of the tourist market, making it difficult to select variables (Hanink and White 1999). Although most of these parks do not reach the level of publicity of Florida’s large theme parks, it may be that they draw on populations well beyond those found locally. Strong loyalties or preferences to particular parks might also account for attendance patterns that are not based on distance. These possibilities were explored using stepwise multiple regression analysis. Although tourist attendance is commonly modeled and predicted using time series analysis (Frechtling 2001), a regression analysis with a range of geographical data is used here, since the goal is not to predict attendance based on past patterns, but instead to relate attendance to cross-sectional explanatory variables. As the attributes of visitors, or their origin locations, are not available, the analysis necessarily focuses on destination attributes (Table 1). Yearly attendance at each of the nineteen amusement parks in the study area was the dependent variable, with 34 independent variables describing amenities and attractive characteristics of the parks (Table 3). Independent variables were obtained from the websites and brochures for each individual park. The 34 independent variables for each park fit into five general groups: 1) General park information; 2) Park admission price information; 3) Accessibility; 4) Park ride information; and 5) Park attractions and amenities. The original continuous data were also recoded to create seventeen categorical variables to better isolate parks with similar features. As with national parks (Hanink and White 1999), it is expected that older theme parks will be more widely known. Larger parks should offer more attractions, while those open for longer periods during the year should attract more tourists. Ownership by a larger corporation such as Six Flags would likely be associated with a park being better known and as a indicator of quality and satisfaction. As all of these parks are seasonal, the number of days each operated per year was included, with the expectation that longer seasons should be related to higher attendance. Price information includes adult, child, senior admissions, season pass, and daily parking fees. The lower these prices, the greater the expected attendance. Variables were also included describing attractions present in the parks. The number of roller coasters, rides for adults and children, presence of a nearby water park, shows, restaurants, and shops were included. Adding new rides and attractions is essential to allowing a park to stand out from the competition, and make the park more appealing as a family destination and as a repeat destination (Newman 2004; Cobb 2005). It can be expected that the presence of more of these amenities and features will have a positive effect on attendance. The population potential accessibility measures discussed earlier were included in the regression model. Shimbel accessibility was also included, representing the total distance from each park to all county centroids. A high accessibility value for this measure indicates a park is less central, as it has a higher total mileage from the park to all county centroids, and vice versa. Stepwise regression produced a model that used five variables to explain attendance to amusement parks in the Southeastern U.S. (Table 4). These variables are the adult-one day admission price for each park, daily parking fees, year the park opened, a coding for the year parks opened, and the network accessibility of each park. Although the limited number of parks in the Southeast allows for a small number of observations, the overall model was significant and the model has a very high level of explanation for park attendance (R2 = 0.953). However, interpretation of the explanatory variables is not straightforward. Surprisingly, variables describing the attractions and amenities offered in the park 53 James Andrew McCall and Joe Weber Table 3. Regression Variables Variable Park Size Description Acres CODE Park Size Year Opened CODE Year Opened Length of Operating Season CODE Operating Season CODE Owned by Corporation CODE Owned by Six Flags Adult 1-Day Admission <100 acres=1, 101-199=2, 200-299=3, ≥ 300=4 Year Before 1945=1, 1946-1977=2, 1978-2003=3 Days of Operation per Year 1-99 days=1, 100-150=2, 151-199=3, ≥ 200=4 NO=0, YES=1 NO=0, YES=1 $ CODE Adult 1-Day Admission Child 1-Day Admission CODE Child 1-Day Admission Senior 1-Day Admission CODE Senior 1-Day Admission Season Pass Price CODE Season Pass Parking Fees/Day CODE Parking Fees/Day Population Potential Accessibility Network Accessibility Number of Roller Coasters CODE # of Roller Coasters Number of Adult Rides 0$=0, 1-19=1, 20-29=2, 30-39=3, 40-49=4, ≥ 50=5 $ 0$=0, 1-16=1, 17-26=2, 27-34=3, ≥ 35=4 $ 0$=0, 1-16=1, 17-26=2, 27-34=3, ≥ 35=4 $ 0$=0, 1-49=1, 50-79=2, 80-105=3, ≥ 106=4 $ 0$=0, 1-4=1, 5-8=2, 9-12=3 CODE # of Adult Rides Number of Child Rides CODE # of Child Rides CODE Nearby Water Park Shows CODE Shows Restaurants CODE Restaurants Shops CODE Shops CODE Conventions, Catering 1-5=1, 6-10=2,11-15=3, 16-20=4, 21-25=5, ≥ 26=6 1=1, 2-5=2, 6-8=3, ≥ 9=4 1-5=1, 6-10=2,11-15=3, 16-20=4, 21-25=5, ≥ 26=6 NO=0, YES=1, OCEANFRONT=2 Number of shows in park NO=0, YES=1 Number of restuarants and food stands in park 0=0, 1-5=1, 6-10=2, 11-15=3, ≥ 16=4 Number of shops in park 0=0, 1-5=1, 6-10=2, ≥ 11=3 0=NO, YES=1 do not show up. Instead, the most important variable in the regression analysis is the adult one-day admission charge. Adult one-day admission price is directly related with yearly attendance at each park. This must indicate 54 a relationship between willingness to pay a high fee and satisfaction from previous visits (or expectations prior to a visit). Higher priced parks therefore reflect high demand based on their attractiveness. This is surpris- Predicting Attendance At Southeastern U.S. Amusement Parks Using Accessibility- and Amenity-Based Models Table 4. Regression Results for Amusement Parks Dependent Variable Attendance Independent Variable Constant Adult Admission CODE Parking Year Opened CODE Year opened Accessibility ing because the size of parks or the number of rides does not appear in the final model, and a categorical variable representing whether parks are part of large corporate chains also does not seem to be related to attendance. Whatever park characteristics represent their attractiveness to visitors is therefore not represented in the variables. A related variable is the daily parking fee charged by parks, which again show that the prices charged by parks are directly related to its yearly attendance. This variable must also be a reflection of past demand, and not a predictor of future attendance. Contradictory results are found for the year that parks are opened. Newer parks are associated with higher attendance values. This may reflect novelty or publicity given to newer parks, or perhaps the fact that newer parks are larger. However, when parks are coded by their opening date, an inverse relationship is found between attendance and age. This must reflect the grouping of parks into the three age categories (opened before 1945, 1946-1977, and after 1977) and large attendances at parks opened in the 1960s and 1970s. Patterns exist here that have clearly not been fully identified. Finally, the results for Shimbel accessibility show that parks with higher centrality have higher attendance. Several amusement parks near the center of the study area (including Six Flags Over Georgia and Dollywood) tend to have high yearly attendances, though clearly parks in Virginia and Maryland also have substantial numbers of visitors. Standardized Coefficient -23626393 0.658 0.352 0.419 -0.548 -0.185 Significance 0 0.002 0.018 0.004 0.032 R2 0.953 CONCLUSIONS This study has attempted to explain yearly attendance at nineteen amusement parks in the Southeastern U.S. From a geographical perspective, use of accessibility models is appealing for explaining attendance to parks based on their distances from residents of the study area. The use of distance as an influence on attendance or travel behavior is an important concept in geography, and is supported by work on national parks, which has shown that attendance varies with distance from the park (Cole and Mitchell 1969; Hanink and Stutts 2002). Yet this is not necessarily the case, as unlike other results with national parks, attendance at amusement parks does not strongly decrease as a negative function of distance. Parks that are closer to larger populations do not necessarily have higher attendance than those in more remote locations. It is undoubtedly the case that attendance levels at amusement parks have likely been augmented by space-time convergence within the Southeastern U.S. resulting from improvements in transportation systems, and centrality within the region is also associated with higher numbers of visitors. However, it does not stem fully from proximity to population. Neither does the size or number of rides in the parks explain attendance, though newer parks do tend to have larger crowds. Instead, the adult admission price and parking fees are the primary factors found in this research for explaining attendance at Southeastern amusement parks. 55 James Andrew McCall and Joe Weber These findings are interesting because they must be a reaction to other factors that have previously created high attendance. This has been shown for amusement parks in central Florida, where parks steadily increased their admissions through the 1980s, when they saw steady attendance increases (Braun and Soskin 1999). During this period Disney World’s ticket price rose faster than other parks as it was seen as the best park in the area and so could charge a premium. Because families may choose park destinations based on attractions and amenities, and the total cost of travel to parks may make the admission cost relatively unimportant, amusement parks have no incentive to reduce cost to encourage larger crowds (Braun, Soskin, and Cernicky 1992). This study does not take into account the potential for competition between amusement parks for the same customers. Competition does not have any significant effects on national park attendance (Hanink and White 1999), but does have some for national battlefields (Hanink and Stutts 2002). It is possible that there is significant competition among parks, so that the population potential measure is not appropriately specified. While accessibility in the sense of proximity of parks to population does not seem to be important, proximity of parks to nearby parks (or other recreation activities) could be of greater significance. The crucial issue is the extent to which parks offer a similar range of attractions. If parks differ considerably, then competition among them should be taken into consideration (Hanink and Stutts 2002). However, it may also be that people perceive each park as a unique destination, in which case competition is not an issue. This research also did not test for the importance of large tourist attractions (such as found at Myrtle Beach or the Smoky Mountains) adjacent to amusement parks, except for a dummy variable that describes whether there is a nearby water park. The high level of explanation produced by the final regression models suggests that agglomeration effects are not important, or have otherwise been exploited by the parks in their admission 56 and parking prices. Also, while Dollywood, which is adjacent to Great Smoky Mountains National Park and the tourist attractions of Gatlinburg and Pigeon Forge, does have high attendance, the Shimbel accessibility measure relates this attendance to centrality within the Southeast rather than proximity to those other tourist attractions. It would be interesting to analyze all amusement parks in the U.S. to determine if these findings remain consistent. Population potential predictions of attendance for Florida’s amusement parks would likely be less accurate because of the enormous actual attendance figures generated by colossal resorts like Walt Disney World, Universal Studios Orlando, and Busch Gardens Tampa Bay, which draw visitors from around the world. However, a regression analysis of variables most related to attendance for amusement parks in Florida may have similar results. Walt Disney World, Universal Studios Orlando, and Busch Gardens Tampa Bay all charge high adult admission prices and high parking fees, while all ranking within the top eleven in yearly attendance figures for North American amusement parks. Although the results of this study suggest that traditional concepts of proximity may not explain attendance to amusement parks, the results for national parks shows this need not be the case with other tourist-oriented facilities. It would be interesting to test the population potential model with attendance to beaches, ski resorts, concerts, and other tourist destinations to see if results similar to these for amusement parks hold true. It would also be useful to examine more sophisticated interaction models to better understand the importance attached to distance. Identifying home locations of visitors would be essential, as assessments of the accuracy of beta parameter exponents for distance in the model can potentially be improved. ACKNOWLEDGEMENTS The authors would like to thank Gerald Webster, James Salem, and several anonymous reviewers for their helpful comments. Predicting Attendance At Southeastern U.S. Amusement Parks Using Accessibility- and Amenity-Based Models LITERATURE CITED Adams, J. A. 1991. The American Amusement Park Industry: A History of Technology and Thrills. Boston: Twayne Publishers. Algeo, K. 2004. Mammoth Cave and the Making of Place. Southeastern Geographer 44: 27-47. Archer, K. 1997. The Limits to the Imagineered City: Sociospatial Polarization in Orlando. Economic Geography 73: 322336. Brainard, J.S., Lovett, A. A. and Bateman, I.J. 1997. Using Isochrone Surfaces in Travel-Cost Models. Journal of Transport Geography 5:117-126. Braun, B. M. and Soskin, M. D. 1999. Theme Park Competitive Strategies. Annals of Tourism Research 26: 439-443. Braun, B. M., Soskin, M. D. and Cernicky, M. 1992. Central Florida Theme Park Pricing: Following the Mouse. Annals of Tourism Research 19: 131-136. Bryman, A. 2004. The Disneyization of Society. London: Sage. Bureau of Transportation Statistics. 2003. National Transportation Atlas Database. http://www.bts.gov. Butler, R.W. 1980. The Concept of a Tourist Area Cycle of Evolution: Implications for Management of Resources. Canadian Geographer 24: 5-12. Carstensen, L.W., Jr. 1981. Time-Distance Mapping and Time-Space Convergence: The Southern United States, 1950-1975. Southeastern Geographer 21: 67-83. Clawson, M. and Knetsch, J.L. 1966. Economics of Outdoor Recreation. Baltimore: Johns Hopkins University Press. Cobb, M.H. 2005. For Your Amusement: After Bankruptcy and Disappointing Attendance, is Visionland Bouncing Back? The Tuscaloosa News, June 12, pgs 1E and 8E. Cole, L. S., Jr. and Mitchell, L.S. 1969. Attendance as a Negative Function of Distance, Great Smoky Mountains National Park Campgrounds. Southeastern Geographer 9: 13-23. Dyer, C.D. 1970. An Investigation of the Effect of Traveltime on Trips Attributed to a Major Recreational Area. Master’s Thesis, School of Civil Engineering, Georgia Institute of Technology. Findlay, J. M. 1992. Magic Lands: Western Cityscapes and American Culture after 1940. Berkeley: University of California Press. Foglesong, R.E. 2001. Married to the Mouse: Walt Disney World and Orlando. New Haven: Yale University Press. Francaviglia, R.V. 1996. Main Street Revisited: Time, Space, and Image Building in Small-Town America. Iowa City: University of Iowa Press. Frechtling, D. C. 2001. Forecasting Tourism Demand: Methods and Strategies. Oxford: Butterworth Heinemann. Garrison, W. L. 1960. Connectivity of the Interstate Highway System. Papers and Proceedings of the Regional Science Association 6: 121-137. Gauthier, H. L. 1968. Transportation and the Growth of the Sao Paulo Economy. Journal of Regional Science 8: 77-94. Hall, C.M. and Page, S.J. 1999. The Geography of Tourism and Recreation: Environment, Place, and Space. New York: Routledge. Hanink, D. M. 1995. Evaluation of Wilderness in a Spatial Context. Growth and Change 26: 425-441. Hanink, D. M. and Stutts, M. 2002. Spatial Demand for National Battlefield Parks. Annals of Tourism Research 29: 707-719. Hanink, D.M. and White, K. 1999. Distance Effects in the Demand for Wildland Recreational Services: the Case of National Parks in the United States. Environment and Planning A 31: 477-492. Hannigan, J. 1998. Fantasy City: Pleasure and Profit in the Postmodern Metropolis. New York: Routledge. Hansen, W. G. 1959. How Accessibility Shapes Land Use. Journal of the American Institute of Planners 25: 73-76. Hanson, S. and Schwab, M. 1995. Describing Disaggregate Flows: Individual and Household Activity Patterns. In: Hanson, S. (ed.) The Geography of Urban Transportation, 2nd Edition. New York: Guilford Press, pp. 166-187. 57 James Andrew McCall and Joe Weber Hollis, T. 2004. Florida’s Miracle Strip: From Redneck Riviera to Emerald Coast. Jackson: University of Mississippi Press. Hollis, T. 1999. Dixie Before Disney: 100 Years of Roadside Fun. Jackson: University Press of Mississippi. Jakle, J. A. 1985. The Tourist: Travel in Twentieth Century North America. Lincoln: University of Nebraska Press. Janelle, D. G. 1968. Central Place Development in a Time-Space Framework. The Professional Geographer 20: 5-10. Johnston, R. J. 1973. On Frictions of Distance and Regression Coefficients. Area 5: 187-191. Kwan, M. and Weber, J. 2003. Individual Accessibility Revisited: Implications for Geographical Analysis in the Twenty-First Century. Geographical Analysis 35: 341353. Mannheim, S. 2002. Walt Disney and the Quest for Community. Aldershot: Ashgate. Marling, K. A., ed. 1997. Designing Disney’s Theme Parks: The Architecture of Reassurance. Paris: Flammarion. Meyer-Arendt, K.J. 1997. Mississippi Casinos and Geographic Concepts. Mississippi Journal for the Social Studies 8: 1-12. Miami Metrozoo. 2003. Miami Metrozoo Master Plan & Further Developments. Miami: Miami-Dade County Park & Recreation Department. Mikkonen, K. and Luoma, M. 1999. The Parameters of the Gravity Model are Changing-How and Why? Journal of Transport Geography 7: 277-283. Mitchelson, R. L. and Lazaro, M. T. 2004. The Face of the Game: African Americans’ Spatial Accessibility to Golf. Southeastern Geographer 44: 48-73. Moe, R. and Wilkie, C. 1997. Changing Places: Rebuilding Community in the Age of Sprawl. New York: Henry Holt and Company. Murayama, Y. 1994. The Impact of Railways on Accessibility in the Japanese Urban System. Journal of Transport Geography 2: 87-100. 58 Newman, A. 2004. Higher! Faster! Wetter! Racing to be Summer’s Hot Ride. New York Times June 4. [http://www.nytimes. com]. Retrieved June 6, 2004. Newman, H. K. 1999. Southern Hospitality: Tourism and the Growth of Atlanta. Tuscaloosa: University of Alabama Press. Paradis, T. W. 2002. The Political Economy of Theme Development in Small Urban Places: the Case of Roswell, New Mexico. Tourism Geographies 4: 22-43. Paul, B. K. and Rimmawi, H. S. 1992. Tourism in Saudi Arabia: Asir National Park. Annals of Tourism Research 19: 501-515. Pirie, G. H. 1979. Measuring Accessibility: a Review and a Proposal. Environment and Planning A 11: 299‑312. Pooler, J. 1987. Measuring Geographical Accessibility: a Review of Current Approaches and Problems in the Use of Population Potentials. Geoforum 18: 269-289. Rand McNally. 2004. Road Atlas. Skokie, IL: Rand McNally. Samuelson, D. and Yegoiants, W. 2001. The American Amusement Park. St. Paul, MN: MBI Publishing Company. Siderelis, C. and Moore, R. L. 1998. Recreation Demand and the Influence of Site Preference Variables. Journal of Leisure Research 30: 301-318. Sorkin, M. 1992. See you in Disneyland. In: Sorkin, M. (ed.) Variations on a Theme Park: The New American City and the End of Public Space. New York: Hill and Wang. Spence, N. and Linneker, B. 1994. Evolution of the Motorway Network and Changing Levels of Accessibility in Great Britain. Journal of Transport Geography 2: 247264. Starnes, R.D., ed. 2003. Southern Journeys: Tourism, History, and Culture in the Modern South. Tuscaloosa: University of Alabama Press. Styer, J. 2005. Defunct Amusement Parks. [http://www.defunctparks.com/] Retrieved May 5, 2005. Tooman, L. A. 1997. Applications of the Life-Cycle Model in Tourism. Annals of Tourism Research 24: 214-234. Predicting Attendance At Southeastern U.S. Amusement Parks Using Accessibility- and Amenity-Based Models Warren, S. 1996. Popular Cultural Practices in the “Postmodern City”. Urban Geography 17: 545-567. Warren, S. 1994. Disneyfication of the Metropolis: Popular Resistance in Seattle. Journal of Urban Affairs 16: 89-107. Williams, S. 1998. Tourism Geography. London: Routledge. Young, T. 2002. Grounding the Myth- Theme Park Landscapes in an Era of Commerce and Nationalism. In: Young, T. and R. Riley (eds.) Theme Park Landscapes: Antecedents and Variations. Washington, DC: Dumbarton Oaks Research Library. Young, T. and Riley, R. eds. 2002. Theme Park Landscapes: Antecedents and Variations. Washington, DC: Dumbarton Oaks Research Library. Zukin, S. 1991. Landscapes of Power: from Detroit to Disney World. Berkeley: University of California Press. 59