A Comparison of Economic Impact Measurement Techniques for a Tourism Special Event

Abstract

The study’s purpose was to examine and update the research methods and measurement issues associated with assessing the economic impact (EI) of an extended duration, regional tourism event. Specifically examined were the issues with the implementation of EI studies in regard to further testing and re-verifying the methods used in (1) measuring a representative sample from an information-seeking/sharing web site and social media registration and engagement lists; (2) comparing locals and nonlocals within a region using different definition techniques; (3) examining the impact of VFRs within the local markets; and (4) comparing the estimation of purchasing within versus outside an impact area. Recommendations for further analysis and implementation for EI studies and the management of tourism special events are presented.

Keywords

economic impact measurement techniques special events regional fair

Introduction

One of the most extensively researched areas of special events is their economic impact (EI) (Raybould and Fredline 2012), and an accurate estimate of the event-related expenditures within the impact area obtained through an attendee expenditure survey is critically important in EI assessments (Rogers 2007). To date, there has been relatively little research on the quality of the measurement data entered into the EI models, and surprisingly few experimental investigations of measurement issues as noted by Raybould and Fredline (2012). Recently, Warnick, Bojanic, and Xu (2015) brought into question some of the initial measurement techniques used to construct EI measures. EI studies have long been the focus of tourism destination areas and, in the last decade, there has been more focus on them for short-term special events that last one to two days, such as festivals (Damonte, Marcis, and Rella 2013) and sports events (Li and Jago 2013). Periodically, there have been concerns expressed about overestimation and improper measurement and application techniques (Jeong, Crompton, and Dudensing 2016; Damonte, Marcis, and Rella 2013; Crompton, 2006, 2010;Crompton, Lee, and Shuster 2001; Stanley et al. 2000). For example, Rogers (2007) stated that event and festival organizers are often called upon to make EI estimates and are not prepared to do so, resulting in estimates that are not credible.

All too often, the estimates are inaccurate at least in part because the manner in which information is collected and projected does not meet acceptable research standards. (7)

Tensions appear to exist between academics, consultants, and event managers over the need to balance academic rigor and competency with everyday time and cost constraints in EI applications (Davies, Ramchandani, and Coleman 2010). Furthermore, these researchers have noted that the process to complete an acceptable research-based EI study is costly because it requires professionals who are knowledgeable in the techniques of sampling, survey data collection, weighting of initial measurements and application of the adjusted measures into a projection model. While there is a wealth of literature regarding the application of the appropriate EI models and methods to utilize in EI research (e.g., Jeong, Crompton, and Dudensing 2016; Abelson 2011; Crompton 2010; Tyrrell and Johnston 2006; and Crompton and Lee 2000), there is substantial evidence to suggest more research on the methods related to the input measurements of EI techniques is still needed (e.g., Davies, Ramchandani, and Coleman 2010; Raybould and Fredline 2012; Ramchandani and Coleman 2012).

John Crompton (2006, 2010) has been one of the most vocal advocates of the need to examine the EI techniques used over the years. He has also been an innovator in reviewing and conducting studies on the various aspects of EI study analysis and coined the concept of economic significance (ES). Prior to Crompton’s work, Dan Stynes (1999) was also a strong advocate of conducting EI studies in tourism settings, and his techniques formed some of the fundamental EI principles and procedures for applications in tourism settings. Crompton, Lee, and Shuster (2001) followed this lead by systematically improving the techniques for input measurements and they created a framework for undertaking an EI study in the Springfest case. Rogers (2007) also contributed to improving the process by providing guidelines for both gated and ungated special events through a series of research directives to address a number of event complexities and to standardize the measurements in conducting EI studies across Canada. Rogers (2007) stressed the need to obtain representative samples, while also noting how costly it is to obtain large, representative samples.

Others have also worked to improve measurement techniques in EI studies. Recently, Warnick, Bojanic, and Xu (2015) indicated that a Trade Market Analysis (TMA) technique could help define locals more precisely in EI studies, while Vogelsong (Hans Vogelsong, personal interview, April 7, 2013) indicated that the easiest way to classify locals and nonlocals in the EI process is to simply have respondents self-identify their local versus nonlocal status. Others (Backer 2008, 2012; Young, Corsum, and Baloglu 2007) have suggested that various subgroups of event attendees such as visiting friends and families (VFR) groups are discounted or not adjusted for in EI studies. Nevertheless, more research is needed to verify if these new techniques tend to hold up over time.

The gap in the literature also suggests that more verification and testing of measurement techniques applied to special events and, in particular, extended special events is still needed (Warnick, Bojanic, and Xu 2015; Wynen 2013; Ramchandani and Coleman 2012; Raybould and Fredline 2012: Backer 2008; and Rogers 2007). This article examines the measurement issues associated with the implementation of EI studies:

to determine if sampling from the various attendee and social media registration and engagement databases is representative;

to compare the various approaches for defining “locals” and “nonlocals” within a region using three different techniques—the Springfest technique used by Crompton, Lee, and Shuster (2001); a trade market analysis (TMA) technique first used by Warnick, Bojanic, and Cartier (2013); and a simplified self-described technique used in coastal zone management EI studies called the HV technique named after its creator, Hans Vogelsong (Personal interview, April 7, 2013 ) (2013).

to consider and examine the size and impact of visiting friends and relatives (VFRs) expenditures within the local markets; and

to compare a one-column expenditure collection method recommended by Stynes (1997, 1999) to a two-column method recommended by Crompton, Lee, and Shuster (2001).

These research measurement issues are addressed by conducting an initial EI measurement analysis through a quasi-experimental design method of a large, special event fair located in Western Massachusetts. This event, a regional six-state fair called the Big E, runs for 17 days each year beginning in September and is hosted at the Eastern States Exposition grounds in West Springfield, Massachusetts.

Review of the Literature

Crompton and associated researchers have illustrated the conceptual rationale(s) for undertaking EI studies. They suggested that EI studies can be used to examine direct and indirect economic gains (Harnick and Crompton 2014); can be used to measure economic significance (ES, the overall expenditures by all attendees) (Crompton, Lee, and Shuster (2001); and have expounded on the basic principles of EI studies (Crompton and Lee 2000; Crompton, Lee, and Shuster 2001). For instance, they emphasized the need to differentiate between expenditures made locally in the impact community and outside the impact area. However, as demonstrated in more recent research, Warnick, Bojanic, and Xu (2015) revealed that additional factors can arise, including the method to classify locals and nonlocals, the presence of visiting friends and relatives (VFR) markets, and how to improve the accuracy of the measurements before estimating both the EI and ES.

In addition, Warnick, Bojanic, and Cartier (2013) suggested that the nature, increased popularity, and the timing of large regional and short-term special events requires additional inquiries and measurement techniques for the purposes of benchmarking and comparison. Abelson (2011) indicated that methods used to research the economic impact of major events remained contested, and considerable debate has continued to exist over the most appropriate EI methods and measures to evaluate events. However, most of the issues have revolved around some of the initial basic techniques advocated by Crompton (2006), Crompton, Lee, and Shuster (2001), and Stynes (1997, 1999).

Rogers (2007) of Research Resolutions and Consulting in Canada agreed with this need by creating a guidebook titled Guidelines: Survey Procedures for Tourism Economic Impact Assessments of Gated Events and Festivals, and a companion publication for ungated events. These works provided very specific procedures to generate EI measures that are credible and standardized, and that were proven to create opportunities to benchmark EI measures, as well as determine when an EI study is not appropriate. Furthermore, Ramchandani and Coleman (2012) concluded that a genuine gap exists in the academic research literature regarding EI estimate measures and discussed the underlying factors that contribute to the variances.

Quality of EI Input and Sampling Data

Researchers have also suggested that the modeling of EI studies was inextricably linked to the quality of the information collected. Ramchandani and Coleman (2012) examined differences in ex ante and ex post estimates of EI in determining where the accuracies of forecasted estimates could be improved. Their research found that (1) direct expenditure measurements were the most difficult to predict among event attendees; (2) reliable forecasts of overall attendance were crucial for credible estimates of anticipated EIs; and (3) the accuracy of forecasting models were intrinsically linked to the quality of the information, especially the expenditure measures, that were used as input for the EI models. The process to identify the expenditures for the EI models is improved by using attendee surveys rather than secondary data, as documented by Ramchandani and Coleman (2012). Similarly, Damonte, Marcis, and Rella (2013) suggested that direct tourism spending examination can bias the economic impact.

Sampling and the differentiation of segment spending are critical to EI assessments, and have been the topics of previous EI research (see Taks et al. 2013 for an example). Now through the professional marketing and planning of special events, the opportunity exists to explore other sampling frameworks such as the utilization of registration lists through various pre-event inquires and social media engagement opportunities (Hudson and Hudson 2013). According to Ramchandani and Coleman (2012), Rogers (2007), and Crompton, Lee, and Shuster (2001), the administration of an EI study requires intercept and detailed sampling techniques that are both costly and complicated. An EI study requires a carefully sampled set of “stints” (Rogers 2007) and trained personnel to intercept attendees on-site throughout the event. However, researchers have also found substantial resistance to detailed on-site intercepts and noted the shortcomings of projecting expenditures before the event has concluded (e.g., Warnick, Bojanic, and Cartier 2013). Other researchers have noted that extensive survey methods (i.e., diary surveys and personal on-site intercept interviews) and extensive expenditure categories create additional burdens and data collection expenses (Raybould and Fredline 2012; Stynes and White 2006; and Howard, Lankford, and Havitz 1991). Thus, it would be useful to investigate more cost-effective data collection methods. Even Crompton, Lee, and Shuster (2001) recommended a periodic evaluation of the process and not an annual EI event study, and Rogers (2007) recommended a five-year plan because of high implementation and evaluation costs.

The opportunities to reach attendees through more widespread audience engagement techniques such as various social media and targeted marketing registration strategies that have evolved may also lead to the further refinement of EI measurement techniques and data collection techniques. However, do these sample frameworks provide truly representative samples? This is still unknown and not fully tested. These new mediums offer opportunities to gain valuable consumer insights, interactions, and rich data faster before, during, and after the event (Hudson and Hudson 2013). These same researchers also indicated that the academic literature is still in its infancy and most event marketing managers have had very little guidance into incorporating social media into their communications or audience engagement strategies (Mangold and Faulds 2009). Others state that investing in social media marketing strategies produces a very strong return on investment for a wide variety of event assessments and marketing-based research opportunities (Dholakia and Durham 2010; Cruz and Mendelsohn 2010; Xiang and Gretzel 2010).

Measuring Expenditures for Locals and Nonlocals

EI studies usually start with estimating visitor attendance and expenditures. This requires a direct participant (self-reported) estimate of expenditures either during or immediately after the event (Rogers 2007; Crompton, Lee, and Shuster 2001). To project the impacts, attendee estimates and how the visitor or attendee expenditure estimates convert to changes in local income and employment are accomplished by applying these measures and estimates to a projection model (i.e., input–output model, computable general equilibrium [CGE] model, etc.). However, initial basic measurements are critical and the formatted EI questions need to be asked in the proper sequence (Rogers 2007; Crompton, Lee, and Shuster 2001; Stynes 1997, 1999). While the definition of locals and exclusion of them has been strongly advocated by Crompton (2006) and Crompton, Lee, and Shuster (2001), a number of critically important preliminary measurement steps are needed. They include (1) the techniques of measuring who is within and outside the local economy, (2) the definition of the local economy area and how locals shop in the area, (3) how and when the basic expenditure data are collected, (4) the problems associated with recall and projection issues, (5) how to include both locals and nonlocals and to further segment these groups, and (6) methods to obtain more accurate data from both locals and nonlocals.

Locals and nonlocals were defined using the trade market analysis (TMA) technique advocated by the Extension Service at the University of Wisconsin (2012) in an air show study conducted by Warnick, Bojanic, and Xu (2015). Another technique presented at a research conference was a simple, but straightforward, approach called the HV technique named after the researcher Hans Vogelsong (Personal interview, April 7, 2013) who suggested that a simple query was an easy way to differentiate between locals and nonlocals in coastal zone management EI studies. Crompton, Lee, and Shuster (2001) segmented locals and nonlocals by defining the locals as only those who reside in the host community zip codes and all others were classified as nonlocals. Defining locals is critical in the EI process as they should not be included in further EI modeling because they do not circulate new money within the local economy (Crompton 2006; Crompton, Lee, and Shuster 2001). In addition, other attendee profiles can, and should, be used for further EI analysis such as day-trip and overnight attendee groups (Wynen 2013).

The VFR Market Impact in EI Studies

Another subgroup of local attendees that has largely been overlooked in EI measurements is the group of people visiting friends and relatives (Backer 2008, 2012; Young, Corsum, and Baloglu 2007). Backer (2012) also indicated that the attempts to measure VFR markets are typically well underestimated. She found that the VFR market is not valued in the role and context of commercial public accommodations, and that attempts to measure this market must identify at least three different VFR segments (pure VFRs, commercial accommodating VFRs, and exploiting VFRs). In addition, most EI studies have focused either on permanent tourism destinations or relatively short-term special events lasting only a few days. Few have examined larger, annual events such as regional fairs that may be held over a period of a week to as much as three weeks in duration. However, the viability of a VFR market becomes more prominent when there are downturns in the economy that make overnight stays costly and when people take shorter trips that are closer to home. Consequently, the main purpose of this study is the verification and testing of ways to improve EI measurements.

Methodology

The Big E began in 1916 and is one of the largest regional fairs in the country. The fair represents the regional economic and agricultural achievements and initiatives of the six New England states and is held annually at the Eastern States Exposition Grounds in West Springfield, Massachusetts. This event was chosen to investigate the issues surrounding the measurement techniques of EI studies because of its long duration, diverse attendee population, and regional attendance. Furthermore, the Big E has invested in a variety of social media platforms to engage a broad cross section of potential attendees and is an ideal location for the demand for fair-based research (Lillywhite, Simonsen, and Wilson 2013).

The first step in the study was to develop a detailed data collection instrument following the guidelines of Crompton, Lee, and Shuster (2001) and Stynes (1999). Then, an online implementation process following the guidelines of a modified Dillman (2000) approach was employed and attendees were contacted through event registration and social media contact e-mail addresses, engagement, and contact invitations. Public reminder notices were posted on social media platforms (i.e., Facebook, Twitter, etc.) and the fair web site, but because of the high response rates, no reminder e-mail notices were sent as recommended by Dillman (2000). The online Qualtrics™ survey application platform served as the source of the survey data collection, and the data collection period lasted from September 13 to 30, 2013. For those groups with no e-mail addresses, mailed surveys were available.

The online registration processes were obtained through web site requests for fair registration information and through social media registration platforms (e.g., Facebook, Twitter, etc.). The purposes for obtaining the e-mail address contacts and to drive these potential and actual visitors to registrations sites were (1) to convey important event planning information, (2) to sell event/concert/entertainment tickets, (3) to reserve seats for particular fair events/shows, (4) to participate in various judging and fair show competitions, and (5) to participate in incentives to be obtained at the Big E. The registration processes were enhanced through targeted advertisements, informational brochures/postings, publicity in the area media, and on the Big E web site. Also, an intercept process was used to distribute invitation cards to obtain additional e-mail addresses at assigned times, and approximately 5,000 cards were randomly distributed periodically by randomly assigned days and times of the event based on previous attendance patterns (Rogers 2007).

To examine the various measurement and sampling techniques, a quasi-experimental design was implemented. An extensive five-part survey was developed, and pretested, during a similar event prior to the Big E. One individual was chosen to represent each polled group or travel party as recommended by Rogers (2007), Crompton, Lee, and Shuster (2001), and Stynes (1999), and incentives for seasonal passes to the 2014 Big E drawing were offered. Each individual was then randomly assigned to one of five intrastudy surveys during the actual online survey administration that covered the following areas: (1) the tourism economy experience, (2) the branded experience, (3) the memory experience dimensions, (4) a single-column expenditure EI method sample, called the Stynes method, and (5) a two-column expenditure EI method sample, called the Crompton method. Only the respondents to the intrasurveys 4 and 5 noted above were included in this analysis.

The Big E attracted 1.48 million individuals over 17 days in 2013, and this survey collected information from individuals representing 3,274 groups. It was conducted in September 2013 from a pool of 5,126 registration groups that were randomly sampled from the web site and social media platform registration and engagement databases. An overall response rate of 63.8% was obtained (3,274 responding groups ÷ 5,126 polled groups). The overall sample was drawn from a total registration population of 152,760 household groups. These contacts were obtained from randomly sampling 146,274 web page and Facebook registrations; 5,736 Twitter accounts; 419 InstaGram accounts; and 331 Pinterest accounts.

The two subsamples involving the EI studies included 665 (20.4% of the total) groups for the Stynes method and 674 groups (20.7% of total) for the Crompton method samples. The same 11 expenditure categories were used for both methods, but the Crompton method included two expenditure columns—one for collecting expenditure inside and one for outside of the local area, compared to the Stynes method that focused on only one expenditure column requesting expenses from inside the area. Sample representativeness, additional techniques for defining locals (Springfest vs. TMA vs. HV), and the measurement of the VFR segment were examined for both samples.

Verification of Sample Representativeness

To verify the representativeness of each sample, the actual distribution of respondents was compared to a zip code analysis of the local and nonlocal attendees as prepared and estimated by The Big and Eastern States Exposition staff. Comparisons of demographic attributes (e.g., education status, professional employment status, and household income) and daily attendance for each of the EIS (economic impact study) samples were also examined, and it was determined that there were no significant differences between the two samples (note: the detailed results are not included here because of space limitations).

Methods for Defining Locals and Nonlocals

Three techniques were used to differentiate locals and nonlocals: (1) the Springfest technique used by Crompton, Lee, and Shuster (2001); (2) the trade market analysis (TMA) technique advocated by the University of Wisconsin Extension Service and used by Warnick, Bojanic, and Cartier (2013) and Warnick, Bojanic, and Xu (2015); and (3) the simplified, self-designated HV technique introduced by Hans Vogelsong (Personal Interview, April 7, 2013) in coastal zone management EI studies. The Springfest technique was developed by Crompton, Lee, and Shuster (2001) and defined locals as those who resided within the immediate local community zip codes and nonlocals as those who lived outside of the immediate local community zip code areas. It was slightly modified in this study because of the large fairgrounds and included the West Springfield zip codes and those zip codes bordering on the fairgrounds in the West Springfield area.

The TMA technique was adapted from the Warnick, Bojanic, and Xu (2015) study and used to identify four different markets: (1) a local convenience market consisting of attendees who shop locally within a 10-mile driving radius that included 17 zip codes, (2) a local destination market consisting of attendees who shop for major shopping items within a 10- to 20-mile driving radius that included 27 zip codes, (3) a market of day-trip attendees who live beyond a 20-mile driving radius, but less than 100 miles, and (4) an overnight tourist market consisting of attendees who traveled 100 miles or more and stayed overnight while attending the event. The driving radius was determined by using an online zip code calculator that estimated the one-way driving distance between the zip codes of the respondents’ home residences and the Big E location in West Springfield, Massachusetts.

The HV technique included two basic questions: (1) whether or not the respondents considered their households to be local residents of the West Springfield area, and (2) if they regularly shop in the area. Those who answered no to both questions were identified as nonlocals. The three methods (Springfest, TMA, and HV) were assessed by comparing the expenditure patterns of each technique’s classification of locals and nonlocals to determine if the expenditures were different, and to determine if the actual locally defined groups were different across the comparisons of the three different techniques.

Measuring the VFR Markets Method

To examine the VFR markets, a set of survey questions was posed to the travel group representatives to determine if friends or relatives were part of their group’s purposes for attending the fair. These questions included identification of the number of adults and children who were visiting them locally from beyond a 50-mile radius of the West Springfield area. In previous studies, the percentages of local groups that contained VFRs were found to vary from 10% to nearly 28% (Warnick, Bojanic, and Xu 2015; Warnick, Bojanic, and Cartier 2013).

Group Expenditures within versus outside Impact Area Methods

The two different EI subsample surveys were administered to collect the data using slightly different approaches. The first subsample followed a one-column approach seeking expenditure information in 11 different expenditure areas for the overall event experience—the basic approach recommended by Stynes (1999), called the Stynes method. The second subsample followed a two-column approach seeking expenditure information of the same 11 expenditure categories but designating between money spent within, and outside, the local community as recommended by Crompton, Lee, and Shuster (2001) and hereafter referred to as the Crompton method.

Coding and Data Analysis Methods

The coding and adjustments to the primary EI measures (such as mileage calculations and the recoding of locals/nonlocals, day trippers, and overnight attendees) were made in Microsoft Excel. The Qualtrics online survey platform was used to determine the random intrasurvey assignments and provide an expenditure total as respondents input their individual categories to allow them to check their information for accuracy. Analysis of variance tests were applied in examining the expenditures between and within different groups and the significance was tested at the .05 level. Statistical analyses were computed using the statistical platform StatPlus.

Selected Findings and Results

The results for the basic distribution of respondents by category for the two samples (Stynes and Crompton) are presented in Table 1. For the 645 groups in the Stynes sample, 100 were identified as convenience locals (15.5%), 97 as regional locals (14.7%), 369 as day-trip attendees (57.5%), and 79 as overnight attendees or tourists (12.1%). For the 658 groups in the Crompton sample, the convenience locals numbered 103 (15.7%), regional locals numbered 97 (14.7%), day-trip attendees numbered 378 (57.5%), and overnight tourists numbered 80 (12.2%).

Table 1.

The Respondents and Distribution of Group Types in the Big E Study.

Case	Location	Time	Sample Size	Resp. Rate (%)
The Big E Regional Fair	West Springfield, MA	September 13–30	5,126	N = 3,253 (63.4%)
Respondents Study	Five subsamples		Sample size	Percentage of sample
Experience economy	Subsample 1		645	19.8%
Branded experience	Subsample 2		642	19.7%
Memory experience	Subsample 3		629	19.3%
EIS Stynes method	Subsample 4		663^a	20.4%
EIS Crompton method	Subsample 5		674^a	20.7%
Total sampled	5 groups		3,253	100.0%
Group Type	N, Groups	Percentage	Average Group Size	Total Polled Individuals
Stynes sample (one-column) TMA technique
Locals (overall)	197	30.2%	4.73	932
Locals (conv.)	100	15.5%	4.71	471
Regions (dest.)	97	14.7%	4.75	461
Nonlocals	448	69.8%	4.96	2,222
Day-trippers	369	57.5%	4.72	1,742
Overnight	79	12.1%	6.08	480
Total	645^a	100.0%	4.89	3,154
Stynes method HV technique
Locals	168	26.2%	4.89	822
Nonlocals	471	73.8%	4.92	2,317
Totals	639^a	100.0%	4.91	3,139
Crompton sample (two-column) TMA technique
Locals (overall)	200	30.4%	4.51	901
Locals (conv.)	103	15.7%	4.67	481
Regions (dest.)	97	14.7%	4.33	420
Nonlocals	458	69.6%	4.89	2,241
Day-trippers	378	57.4%	4.81	1,818
Overnight	80	12.2%	5.29	423
Total	658^a	100.0%	4.78	3,142
Crompton method HV technique
Locals	179	27.0%	4.44	795
Nonlocals	484	73.0%	4.90	2,372
Totals	663^a	100.0%	4.78	3,167

Note: EIS = economic impact study; TMA = trade market analysis; HV = Hans Vogelsong; exp. = experience; conv. = convenience; dest. = destination.

EI one-column Stynes method sample had 18 incomplete cases. EI two-column Crompton method sample had 16 incomplete cases. These cases were not included in the distribution. Some data were not complete enough, which yielded slightly smaller samples; n = 639 in the Stynes method sample (one-column EI approach) and n = 663 in the Crompton method sample (two-column EI approach). The Springfest technique, counting only the immediate West Springfield zip codes and adjacent Big E destination zip codes, yielded 71 locals (11%) and 570 nonlocals (89%) in Stynes method sample, and 81 locals (12.3%) and 577 (87.7%) in the Crompton method sample for comparison purposes.

The HV technique of self-identification of local and nonlocal status was examined for both the Stynes and Crompton samples. In the Stynes sample (the one-column EI measurement technique), 168 (26.2%) of the respondents were classified as local attendees and 471 as nonlocal attendees (73.8%). In the Crompton sample, using the HV self-identification (two-column EI measurement technique), the locals numbered 179 (27.0%) and the nonlocals numbered 484 (73.0%). Group size was also reported for each segment of locals and nonlocals, and the overall estimate of the number of individuals reached through the survey is indicated in Table 1.

Representative Samples Compared to Actual Attendance Distribution

Relatively small variations were found when the samples were compared to the actual attendance day distribution. The percentage of variance for under- and overrepresentations by day were small, ranging between +2.7% and −2.4% within both samples, and the combined over- and under representation was only 1/10th of 1% (0.1%). (Note: Because of space limitations, the full day-by-day tabular analysis is not provided here, but it was made available to the reviewers.) The zip code analysis indicated that actual attendees compiled by the Big E staff classified 26.2% as locals and 73.8% as nonlocals. Using the same zip code analysis technique, the Stynes method sample (one-column sample) classified survey participates as 26.3% locals and 73.7% as nonlocals, while the Crompton method sample (two-column sample) classified 27.0% as local and 73.0% as nonlocal groups. Therefore, the data analyses suggested that both samples were highly representative of the actual visitation analysis conducted by the fair staff (see Table 2).

Table 2.

Zip Code Analysis of the Local and Nonlocal Sampling for the Stynes and Crompton Methods.

	Actual^a Percentage	Stynes Method Count^b	Stynes Method Percentage	Crompton Method Count^b	Crompton Method Percentage	Combined Count	Percentage
Local	26.2%	168	26.3%	179	27.0%	347	26.7%
Nonlocal	73.8%	471	73.7%	484	73.0%	955	73.3%
Total	100.0%	639	100.0%	663	100.0%	1,302	100.0%

Actual percentage based on zip code analysis of attendees upon entrance to fair by Eastern State Exposition staff.

Stynes method sample represented a one-column expenditures data collection method with emphasis on expenditures within the local market area and the Crompton method sample represented a two-column expenditures data collection method with emphasis on expenditures within the local market and expenditures outside the local market during the event. (Note: Zip codes were not available for all survey respondents.)

Comparison of Techniques for Identifying Locals and Nonlocals

In the Warnick, Bojanic, and Xu (2015) study, the TMA technique was used to improve the estimate of locals by defining groups through different local shopping areas that best described how the local economy worked. This approach demonstrated an improvement over Crompton’s original research utilizing a simple local zip code analysis and defining locals by the most immediate local zip codes (i.e., in their Springfest study, only the two most local zip codes were used for locals). The TMA, HV, and Springfest techniques were compared to each other to determine if significant differences existed in expenditure patterns and to determine if one technique should be preferred over the others.

The results with the EI Stynes method (one-column sample) indicated that the distributions for the three techniques were not significantly different (at p <.05) in terms of average overall expenditures (see Table 3a). However, the Springfest technique (Crompton, Lee, and Shuster 2001) identified the fewest locals. The TMA identified 197 (30.5%) as locals, the HV technique identified 168 (26.3%) as locals, and the original Springfest technique advocated by Crompton, Lee, and Shuster (2001) identified 71 (11.0%) as locals. The comparison of HV locals (mean = $364.93) to TMA locals (mean = $363.62) was not significant (F[1, 364] = 0.00037, p = .98464); TMA locals (mean = $363.62) to Springfest locals (mean = $328.80) was not significant (F[1, 267] = 0.05184, p = .82006); and the HV locals (mean = $364.93) to Springfest locals (mean = $328.80) was not significant (F[1, 238] = 0.05436, p = .8158). The same was true for the nonlocals’ overall expenditures. The comparison of HV nonlocals (mean = $354.71) to TMA nonlocals (mean = $354.62) was not significant (F[1, 924] = 0.00001, p = .99718); TMA nonlocals (mean = $354.62) to Springfest nonlocals (mean = $358.86) was not significant (F[1, 1021] = 0.02457, p = .87548); and the HV nonlocals (mean = $354.71) to Springfest nonlocals (mean = $358.86) was not significant (F[1, 1049] = 0.02433, p = .87607).

Table 3a.

Average Expenditures by Locals and Nonlocals for the Stynes Method Sample.

Category: One-Column Stynes Method/Sample	TMA^a Nonlocals	HV^b Nonlocals	Springfest^c Nonlocals	TMA^a Locals	HV^b Locals	Springfest^c Locals
Groups (n)	448	471	570	197	168	71
Average group size	4.96	4.92	4.88	4.73	4.89	5.14
% total (within category)	69.5%	73.7%	89.0%	30.5%	26.3%	11.0%
Parking	$14.31	$14.34	$14.70	$16.23	$16.48	$15.21
Admission	53.99	54.20	56.00	63.54	4.61	60.92
Average states purchases	60.58	59.27	61.23	61.57	65.46	55.56
Refreshments	69.23	67.49	71.32	80.94	87.91	80.82
Food and drinks before and after event	15.47	15.03	17.37	18.41	20.18	8.34
Souvenirs or gifts	60.78	62.04	62.70	61.66	58.26	45.80
Clothing or accessories	23.95	28.42	26.71	39.79	29.75	43.10
Transportation	32.81	30.73	27.88	9.90	11.82	8.11
Local attractions	3.01	3.00	3.75	6.98	7.68	8.03
Overnight accommodations	15.45	14.51	12.05	.89	1.05	2.46
Other	6.53	7.00	6.40	4.19	2.44	1.06
Total	$354.62	$354.71	$358.86	$363.62	$364.93	$328.80
Average expenditure per person	$71.50	$72.10	$73.53	$76.88	$74.63	$63.97

Note: TMA = trade market analysis; HV = Hans Vogelsong.

TMA locals here include both types of locals (locals–convenience and regionals–destination TMA groups) and nonlocals included both day-trip and overnight visitors.

HV (Vogelsong, personal interview, April 7, 2013) technique asks respondents to self-identify as local or nonlocal.

Springfest defined locals as those from the most immediate destination zip codes, in this case West Springfield zip codes 01089 and 01090 and the immediate adjacent or bordering zip code communities to the Big E; all others were nonlocals. All nonlocal comparsions—TMA nonlocals to HV nonlocals, HV nonlocals to Springfest nonlocals, and TMA nonlocals to Springfest nonlocals—were nonsignificant at the .05 level. All local comparisons—TMA locals to HV locals, HV locals to Springfest locals, and TMA locals to Springfest locals—were nonsignificant at the .05 level.

The results with the EI two-column sample (the Crompton method) indicated that the distributions for the three techniques were not substantially different (at p <.05) in terms of average overall expenditures between the locals; however, the Springfest technique (Crompton, Lee, and Shuster 2001) identified the fewest locals as expected (see Table 3b). The TMA identified 200 (30.4%) as locals, the HV technique identified 179 (27.0%) as locals, and the original Springfest technique identified 81 (12.3%) as locals. While there were differences in the actual numbers, the overall expenditures between each of the local groups yielded no significant differences (at p <.05). The comparison of HV locals (mean = $289.42) to TMA locals (mean = $284.76) was not significant (F[1, 259] = 1.02483, p = 0.31233); TMA locals (mean = $284.76) to Springfest locals (mean = $247.87) was not significant (F[1, 278] = 0.62387, p = .43029); and the HV locals (mean = $289.42) to Springfest locals (mean = $247.87) was not significant (F[1, 280] = 0.067894, p = .41066). However, the comparison between the techniques among the nonlocals revealed a significant difference (at p <.05). The comparison of HV nonlocals (mean = $353.39) to TMA nonlocals (mean = $360.44) was significant (F[1, 858] = 3.95758, p = .04698); TMA nonlocals (mean = $360.44) to Springfest nonlocals (mean = $341.59) was not significant (F[1, 1034] = 0.16371, p = .68585); and the HV nonlocals (mean = $353.39) to Springfest nonlocals (mean = $341.59) was not significant (F[1, 1057] = 0.02552, p = .8731). While one may note that most of these differences were not significant and might conclude that any of these techniques would work in defining locals and nonlocals, the additional analysis of the TMA segmentation of both of these groups revealed additional, and significantly important, insights concealed in the broader classifications of both locals and nonlocals using the other techniques.

Table 3b.

Average Expenditures by Locals and Nonlocals for the Crompton Method Sample.

Category: Two-Column Crompton Method/Sample	TMA Nonlocals	HV Nonlocals	Springfest Nonlocals	TMA Locals	HV Locals	Springfest Locals
Groups (n)	458	484	577	200	179	81
Average group size	4.89	4.90	4.94	4.51	4.44	4.75
% total (within category)	69.6%	73.9%	87.7%	30.4%	27.0%	12.3%
Parking	$13.97	$13.79	$14.52	$16.31	$17.23	$13.69
Admission	55.53	54.16	44.08	53.33	57.04	55.92
Average states purchases	49.94	49.24	47.72	42.18	42.78	41.68
Refreshments	57.08	56.06	55.58	55.46	57.46	60.63
Food and drinks before and after event	19.30	18.50	16.07	11.41	12.27	13.49
Souvenirs or gifts	69.63	64.82	66.22	48.36	59.18	38.55
Clothing or accessories	19.99	20.36	19.08	17.44	16.40	18.55
Transportation	41.67	37.70	37.57	18.49	26.59	8.02
Local attractions	1.05	1.00	1.28	1.81	1.95	1.44
Overnight accommodations	23.01	22.11	16.67	3.15	2.39	0.00
Other	10.64	16.91	14.58	16.82	0.45	0.00
Total	$360.44^a	$353.39^a	$341.59	$284.76	$289.42	$247.87
Average expenditure per person	$73.71	$72.12	$69.15	$63.14	$65.18	$52.18

Note: TMA = trade market analysis; HV = Hans Vogelsong.

Nonlocal comparsions: TMA nonlocals to HV nonlocals were significant at the .05 level. HV nonlocals to Springfest nonlocals, TMA nonlocals to Springfest nonlocals were not significant at the .05 level. All local comparisons—TMA locals to HV locals, HV locals to Springfest locals, and TMA locals to Springfest locals—were not significant at the .05 level.

Differences within the Nonlocals and Locals by the TMA Technique

Within nonlocal groups’ expenditures using the TMA classification in the Stynes method one-column sample, differences were found between the day-trip and overnight tourist attendee groups. On average, the overnight respondents spent significantly more per group ($443.73) than the day-trip groups) spent ($335.54) on the overall expenditures as would be expected (F[1, 448] = 5.64, p = 0.0179). These differences compare to an overall average of $354.62 per group for all nonlocals. For the locals, although there was a noticeable gap between the reported averages, the difference was not significant (F[1, 193] = 0.576, p = 0.449). The convenience trade locals on average spent $395.06 per group, while regional trade market locals spent $331.21 per group. The overall average was $363.62 per group for all locals and the overall average for all attendees was $357.37 per group in this sample. What is interesting about these expenditure patterns are the within category differences. For example, the nonlocals spent considerably more on the transportation and travel items (i.e., transportation, overnight accommodations, and other travel expenses) as would be expected, while the locals spent substantially more on other items (i.e., clothing and accessories, food/drink before and after fair, and the refreshment expenditures at the fair), with locals outspending both day-trip groups and regional local groups on average on these same items.

When these expenditures were compared in the Crompton method sample for nonlocal groups using the TMA classification, differences were found between the day-trip and overnight tourist attendee groups. On average, the day-trip attendees spent $299.63 per group, while the overnight tourists spent $647.77 per group, which was significantly different at the p < .05 level (F[1, 457] = 40.856, p = 0.0001). However, the difference was not significant for the locals (F[1, 199] = 0.06581, p = 0.7978). The convenience trade market locals spent $275.62 on average per group, while the regional trade market locals spent $287.99 per group, compared to an overall average of $335.44 per group for all attendees in this sample.

Visiting Friends and Relatives in the Local Markets

The impact of the visiting friends and relatives (VFR) market was found to be important in this study. If these VFR groups are excluded because they are labeled as locals, then the additional expenditures are incorrectly excluded when the full EI modeling is implemented. The presence of VFR groups was identified and their expenditure patterns were examined both within the Stynes method and the Crompton method samples. When locals were compared to the local VFRs, differences were found, not only in the expected group size, but also in expenditures overall in the one-column Stynes method (see Table 5). On average, it was found that the VFR groups comprised 21% of the convenience trade market and spent $912.19 per group ($123.27 per person) on average, with an average group size of 7.40 persons. This was significantly more than the average total local convenience trade market expenditure per group of $395.06, or $83.88 per person (F[1, 119] = 4.357, p = 0.039).

The VFR groups comprised 22.1% of the regional destination trade market of locals and spent $447.50 on average per group ($77.19 per person), with an average group size of 5.80 persons. The average overall destination trade market expenditures per group were $331.21, or $67.73 per person. However, this within-group difference was not significant at the p <.05 level (F[1, 113] = 2.336, p = 0.129). Although not significant, it is clear that the VFR market spent more per group than the regular locals, and they also exceeded the spending levels of the overall grouping of nonlocals in the TMA technique. In addition, the VFR market spent more on average than the day-trip market ($335.54 per group) and the overnight market ($443.73 per group; see Table 4 for this figure). These patterns suggested that this event might bring out substantially different local convenience market spending patterns and group configurations than would be expected and further justify collecting expenditure data on locals.

Table 4.

Average Expenditures by Category for Locals and Nonlocals for the TMA Technique Classification.

One-Column Stynes Sample	Nonlocals			Locals			Overall Averages
One-Column Stynes Sample	All	Day Trip	Overnight	All	Local (CTM)	Reg (DTM)	Overall Averages
Groups (n)	448	369	79	197	100	97	645
Average group size	4.96	4.73	6.08	4.73	4.71	4.75	4.89
Expenditure categories
Parking	14.31	12.58	22.42	16.23	19.02	13.35	$14.90
Admission	53.99	53.27	57.34	63.54	74.91	51.82	$56.92
Average states expenditure	60.58	60.07	62.96	61.57	67.22	55.74	$60.88
Refreshments	69.23	70.48	63.44	80.94	86.75	74.96	$72.82
Food/drinks before/after event	15.47	14.27	21.01	18.41	26.10	10.57	$16.37
Souvenirs or gifts	60.78	61.11	59.23	61.66	54.71	68.84	$61.05
Clothing or accessories	23.95	23.96	23.87	39.79	47.78	31.56	$28.76
Transportation	32.81	26.60	61.63	9.90	9.31	10.52	$25.79
Local attractions	3.01	2.83	3.86	6.98	4.75	9.28	$4.22
Overnight accommodations	15.45	7.46	53.19	.89	1.75	0.00	$10.99
Other	6.53	4.55	15.72	4.19	3.50	4.90	$5.81
Total	$354.62	$335.54^a	$443.73^a	$363.62	$395.06 (n.s.)	$331.21 (n.s.)	$357.37
Average expenditure per person	$71.50	$70.94	$72.98	$76.88	$83.88	$69.73	$73.08
Two-Column Crompton Sample	Nonlocals			Locals			Overall Averages
Two-Column Crompton Sample	All	Day Trip	Overnight	All	Local (CTM)	Reg (DTM)	Overall Averages
Groups (n)	458	378	80	200	103	97	658
Average group size	4.89	4.81	5.29	4.51	4.67	4.33	4.78
Expenditure Categories
Parking	13.97	13.50	16.23	16.31	18.19	14.39	$14.67
Admission	55.53	53.01	67.39	53.33	59.16	47.51	$54.77
Average states expenditure	49.94	44.67	76.73	42.18	48.02	36.30	$47.24
Refreshments	57.08	52.41	77.32	55.46	62.00	48.82	$56.51
Food and drinks before and after event	19.30	11.96	53.96	11.41	14.79	7.94	$16.80
Souvenirs or gifts	69.63	63.02	100.73	48.36	48.05	48.67	$63.18
Clothing or accessories	19.99	17.91	29.78	17.44	16.57	18.36	$19.25
Transportation	41.67	31.89	88.21	18.49	7.76	29.91	$34.60
Local attractions	1.05	.80	2.26	1.81	1.47	2.16	$1.27
Overnight accommodations	23.01	7.00	99.63	3.15	0.98	5.44	$16.87
Other	10.64	4.60	39.43	16.82	1.96	32.60	$12.41
Total	$360.44	$299.63^a	$647.77^a	$284.76	$275.62 (n.s.)	$287.99 (n.s.)	$335.44
Average expenditure per person	$73.71	$62.29	$122.45	$63.14	$59.02	$66.51	$70.18

Note: TMA = trade market analysis; CTM = convenience trade market; DTM = destination trade market; n.s. = not significant at the .05 level.

Significant at the.05 level.

Table 5.

Visiting Friends and Relatives Expenditures and Group Size Effects in the Stynes Method Sample.

Stynes (One-Column) EIS Group Sample	Nonlocal	Local (CTM)	Regional (DTM)	All
Number or groups	448	100	95	645
Expenditures per group: typical average expenditures per group	$354.62	$395.06^a	$331.21	$357.14
Average expenditures per group with VFR	NA	$912.19^a	$447.70	NA
Added group income from VFR market/group	NA	$517.13	$121.94	NA
Group Size, Expenditures and %	Nonlocal	Local	Regional	All
Number of VFR attendees in sample	NA	21	21	42
Typical group size per type	4.96	4.71	4.75	4.92
Average group size of VFR markets	NA	7.40	5.80	NA
Average group expenditure per person	$71.50	$83.88	$67.73	$72.59
Average expenditure per group person within VFR	NA	$123.27	$77.19	NA
Percentage of VFR attendees	NA	21.0%	22.1%	6.5%

Note: Two groups were excluded in the regional destination trade market as incomplete in this analysis. EIS = economic impact study; CTM = convenience trade market; DTM = destination trade market; NA = not available; VFR = visiting friends and relatives.

Significant at the .05 level.

When locals were compared to the local VFRs in the Crompton method (two-column approach), differences were found in the expected group size, as well as in expenditures (see Table 6). The VFR group represented 19% of the overall convenience trade market and spent an average $409.37 per group ($64.98 per person) with an average group size of 6.30 persons. This compared to the average local convenience market expenditures per group of $275.62 or $59.02 per person. This difference was significant within the groups at the p <.05 level (F[1, 117] = 3.975, p = 0.049). The VFR group comprised 22.7% of the overall destination trade market and spent an average $332.93 per group ($59.45 per person) with an average group size of 5.60 persons. This compared to the average destination market expenditures per group of $287.99 or $66.51 per person. However, there was no significant difference at the p <.05 level (F[1, 115] = 0.169, p = 0.682) within this group for VFR expenditures. Furthermore, there was no significant difference found between the local convenience attendees and the local destination or regional attendees with VFRs present in their travel groups in the two-column, Crompton method sample at the p <.05 level (F[1, 39] = 0.999, p = 0.324).

Table 6.

Visiting Friends and Relatives Expenditures and Group Size Effects in the Crompton Method Sample.

Crompton (Two-Column) EIS Group Sample	Nonlocal	Local (CTM)	Regional (DTM)	All
Number or groups	458	100	97	658
Expenditures per group: typical average expenditures per group	$360.04	$275.62^a	$287.99	$335.44
Average expenditures per group with VFR	NA	$409.37^a	$332.93	NA
Added group income from VFR market/group	NA	$133.75	$44.94	NA
Group Size, Expenditures and %	Nonlocal	Local	Regional	All
Number of VFR attendees in sample	NA	19	22	41
Typical group size per type	4.89	4.67	4.33	4.78
Average group size of VFR markets	NA	6.30	5.60	NA
Average group expenditure per person	$73.71	$59.02	$66.51	$70.18
Average expenditure per group person within VFR	NA	$64.98	$59.45	NA
Percentage of VFR attendees	NA	19.0%	22.7%	6.2%

Note: EIS = economic impact study; CTM = convenience trade market; DTM = destination trade market; VFR = visiting friends and relatives; NA = not available.

Significant at the .05 level.

While the management implications here suggested that the VFR markets were sizable and expenditure patterns were different from the smaller, average group size locals, it should be noted that these samples of locals with VFRs present were still relatively small overall. More oversampling of locals is likely necessary to more fully uncover and to further document this visitation trend. From both of these samples, the number of VFR groups totaled only 83 (20.9%) of the overall 397 local groups. The results also suggested mixed findings here, and the results may be affected both by the small sample sizes and/or unique or extraordinary/outlier group spending patterns in each sample. Nevertheless, these findings appeared to give rise to the need to begin to target and further analyze locals who do attract VFR markets to the area.

Inside and Outside Market Area Purchases

Crompton, Lee, and Shuster (2001) recommended that all EI expenditure data be collected using a two-column approach. This study compared two different methods—a Stynes method (one-column approach) with emphasis on requiring the respondent to only include expenditures spent locally and the Crompton method (two-column approach) with an emphasis on separating expenditures data into money spent inside and outside of the local area. Overall average expenditure data were collected for the both samples for comparative purposes (see Table 7). Each survey sample found that the overall expenditures were in the $300-$360 per group range, with the Stynes method (one-column approach) standing at an average $357.37 per group ($73.08 per person), and the Crompton method at an average $337.54 per group ($70.62 per person).

Table 7.

Average Expenditures by Category for the Stynes and Crompton Samples.

Category	Number of Groups and Average Group Size
Category	Stynes Method Overall	Crompton Method Overall	Inside West Springfield	Outside West Springfield
Groups (n) responding	645	658	658	658
Average group size	4.89	4.78	4.78	4.78
Expenditure categories
Parking	$14.90	$14.67	$13.30	$1.37
Admission	56.92	54.77	52.84	1.93
Refreshments (includes average states spending)	133.70	103.75	100.75	3.00
Food and drinks before and after the event	16.37	16.80	11.99	4.81
Souvenirs or gifts	61.05	63.18	62.18	1.00
Clothing or accessories	28.76	19.25	18.53	0.72
Transportation	25.79	34.60	27.78	6.82
Local attractions	4.22	1.27	0.72	0.55
Overnight accommodations	10.99	16.87	12.15	4.72
Other	5.81	12.41	12.41	0.00
Total	$357.37^a,b	$337.54^a	$312.65^b	$24.92
Average expenditure per person	$73.08	$70.62	$65.41	$5.21

Nonsignificant at the .05 level.

Significant at the .05 level.

There were no significant differences between the Stynes method and the Crompton method at the p <.05 level when the total overall expenditures were examined (F[1, 1307] = 0.772, p = 0.379). However, when the Stynes method was compared to the Crompton method inside the local economy area only (i.e., the average expenditures from within the West Springfield area), the difference was very close to being significant at the p <.05 level (F[1, 1307] = 3,717, p = 0.054, and critical F value of 3.8486). Also, it appears that even in the one-column Stynes method, individuals associate the cost of attendance as a total trip and likely do not differentiate between inside and outside the local area. The Crompton method identified expenditures at $312.65 per group inside the West Springfield area and $24.92 outside the Springfield area.

Conclusions and Recommendations

This study was undertaken to further examine and verify previous research measurement issues associated with constructing the preliminary expenditure data that eventually goes into a fully developed EI study. This study was unique in that a quasi-experimental design was developed to test the various measurement techniques and the data were collected over the full 17-day period of the event. Specifically, the issues examined were within the context of the implementation of EI studies to improve the efficiency in collecting large data sets that are representative of the full event populations. Second, this study sought to determine if different definitional techniques (TMA, HV, or Springfest) of classifying locals and nonlocals would provide similar results and if one technique might be preferred when applied in EI studies. Two samples were randomly generated through a unique assignment feature in the online Qualtrics platform to test the differences in these techniques, where one sample followed the one-column Stynes EI method and the other sample followed the two-column Crompton EI method. Within each sample, the techniques for defining locals and nonlocals were examined and compared, as well as the further review of the TMA technique to determine if more detailed segmentation of locals into convenience and regional markets, and nonlocals into day-trip and overnight guests/tourists, would provide different results. Third, both samples were examined to determine if VFR groups existed in the local markets and if spending patterns were different from the typical local attendees. Finally, both samples were examined to determine if the EI methods for estimating overall expenditures inside and outside of the local host community area provided different results using the Stynes method or the Crompton method. The following recommendations were made based on the results of the analyses.

Social Media Use and Registration Database Recommendation

First, database registration and the use of social media tools to assist in sampling for EI studies and initial expenditure measurements appear to yield representative samples. The post-event application resulted in high response rates, and the samples were representative of the actual visitation by both distribution of attendees’ profiles and by attendance days. This process of sampling proves to be cost effective, and it eliminates the problems associated with on-site intercepts (e.g., high refusal rates leading to less accurate projections). This sampling method can suffer from recall problems, but the similarity in the two samples in this study suggests that this is less of a problem than previously thought. The use of a post-event online survey also reduces the mailing costs and the costs of training and deploying interviewers, and it allows the assessment of nonattendee respondents to determine what prevented them from attending the event.

Techniques to Define Locals and Nonlocals Recommendation

Initially it appears that there is little difference between the techniques (TMA, HV or Springfest) in defining locals, as there were no significant differences found in their respective spending patterns at the same event in either of the samples. Each seems to be assessing the same types of locals. However, there did appear to be differences in the nonlocals in the Crompton method sample, possibly because the Crompton method forces respondents to think more carefully about where the expenditures occurred, and the different approach to segmentation of nonlocals by the TMA technique. Furthermore, although not statistically significant, the TMA technique does improve the classification of nonlocals in comparison to the Springfest technique, which tends to assign too many locals into the nonlocals classification. Part of this is due to the presence of day-trip attendees versus overnight guests or tourists. There were significant differences in both samples between day-trip and overnight attendees. While the HV shows promise, it is the conclusion here that the TMA technique provides the more detailed segmentation and discrete estimates, and likely provides more marketing opportunities and should be the preferred method for defining locals and nonlocals.

TMA Technique Recommendation

The application of a TMA technique improved the analysis and definition of locals and nonlocals as found in the Warnick, Bojanic, and Xu (2015) study. It was confirmed that it was still better to think of the local economy as a trade market area where various types of shopping and buying occur. This should also appeal to sponsors and vendors who wish to support the special events, including a regional fair. The HV technique is an easy method for classifying locals and nonlocals, but it does result in an estimated 10% overall error rate in the self-identification of locals and nonlocals when examined by zip codes. While the HV technique does simplify the process for management and reduces the coding and classifying of respondents, the TMA technique still appears to measure spending patterns more precisely and improved the differentiation of attendees over the Springfest and HV techniques that segment attendees into broader groups of locals and nonlocals. The exclusion of the additional TMA locals results in more accurate estimates of the “new money” coming into the local economy. This is especially important in a long-term event and might help management gain new insights into specific spending patterns. Furthermore, the marketing communication strategies in reaching these distinctly different markets may also engage more local/regional support and corporate sponsorships. The weakness and limitation of the TMA technique is that it requires more coding and understanding of the local market economy.

It should also be noted that measuring the expenditure patterns of locals is still important. For example, it was found that locals spent at higher levels when further segmented into convenience and regional markets than would be expected as found in other short-term events (Crompton, Lee, and Shuster 2001). However, with substantially different, and broader types and mixes of attendees, vendors may have created different promotional activities that actually enhanced local incentive purchases/expenditures. It was also possible that local groups may have contained a number of interrelated event groups (i.e., VFR groups) that exhibit these different expenditure patterns. This may be a weakness in this study and should be explored further in the future.

Unlike previous short-term studies of attendee spending (e.g., Crompton, Lee, and Shuster 2001) where locals typically account for the lowest levels of overall expenditures, this study found locals, especially the very local convenience type shoppers, spending at levels higher than day-trip attendees. Locals appeared to visit the Big E more frequently within a given year and possessed more fair shopping knowledge than their counterparts, the nonlocal attendees. These differences were significant and especially important in understanding how the event length and configuration impacts expenditure patterns over time, both between and within selected market groups, and further emphasizes the preferred technique of choice—the TMA technique for measurement application in EI studies.

Inclusion of VFRs in Local Market and Recommendation

The identification of VFRs continued to be an important research challenge. Even with the larger sample sizes in this study, the number of VFR groups counted was relatively small, but still a distinct proportion of the local subsamples at nearly 20%. However, more oversampling of locals appears to be needed to continue to verify the existence of the distinct VFR groups and their respective expenditure patterns across different events. It is recommended that managers of large attendance special events recognize this segment might be substantial, especially if there is a shortage of viable local accommodations or if an event has the potential of attracting VFRs (e.g., music festivals and concerts). VFR group sizes were larger than the other local groups and the nonlocal groups, and they spent more than a comparable local group. This research also served to support the works of Backer (2008, 2012) that VFR markets were largely underestimated. It is recommended that questions in any EI study be asked about the attendance of VFRs. Questions regarding who pays (host or visitor) for event attendance or admission, how long they stay in the area, and the nature and purpose of the travel to the area should be asked in order for the VFR groups to be correctly counted.

Inside and Outside Area Expenditure Estimates and Recommendation

Findings from this study suggested that the Crompton method (two-column EI method) was the best method for estimating expenditures inside the local area and adjusting them appropriately. It was clear in this study that respondents estimated their expenditures based on the whole trip and did not necessarily consider the definition of the local area even when they were asked to do so. Therefore, the Crompton method should be the preferred survey construction method for estimating on-site expenditures and should be included with instructions to carefully differentiate the expenditures inside and outside of the local area while attending a special event as suggested by Crompton, Lee, and Shuster (2001) and Rogers (2007). Furthermore, adjustments to the overall local EI would be needed to eliminate the overestimating of the outside the area expenditures that are incorrectly associated with inside the local area spending. In particular, the large number of day-trip attendees need to be classified correctly.

Need for Future Research

Examining these measurement principles in a longer, extended time special event context, such as a large regional fair, provided new evolving differences not found in previous studies. An additional management implication was the expenditure pattern of locals at this event. Attendees appeared to approach the longer-running special event differently, especially locals, and spent at higher levels when the event runs for a longer time period. Careful attention to both locals’ and nonlocals’ expenditure patterns to annual large-scale special events such as the Big E is recommended even though local expenditures are to be excluded in an EI study (Crompton 2006; Crompton, Lee, and Shuster 2001).

Further research to continue to improve and to test the TMA technique and other techniques in a variety of different special events is recommended. The TMA technique continued to show promise for application in longer-running events and might find promise in other economic impact applications such as tourist destinations and attractions. Specifically, more research is needed when the attendance of VFR groups at special events is expected to be large and that a significant portion of the attendees might stay with locals. Additionally, more research is recommended on the purpose(s) of the VFR market travel when they plan to attend longer-term special events. In conclusion, this study showed promise in (1) enhancing the basic measurement techniques applied in the preliminary measurements for EI studies; (2) further refining and addressing critical implications to longer-term special events; and (3) resolving some of the need to bridge the gap between practitioners seeking realistic EI measures and academics seeking to improve the accuracy of EI measurement and estimation before completing EI modeling estimates.

Footnotes

Acknowledgements

The authors acknowledge the assistance and the opportunity to conduct this study as provided by the Big E and the staff of the Eastern States Exposition Center in West Springfield, MA, and the support and assistance provided by the Greater Springfield Convention and Visitors Bureau and the Department of Hospitality and Tourism Management in the Isenberg School of Management at the University of Massachusetts at Amherst.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Rodney B. Warnick, PhD, is a professor in the Hospitality and Tourism Management Department at the University of Massachusetts Amherst. His primary research area is the monitoring and analysis of recreation, sport, and tourism trends and related consumer behavior and marketing topics. He teaches courses in club management and hospitality and tourism marketing, advertising, and promotion. He also is active in preparing grants and is a consultant to the tourism and resort management firms and organizations.

David C. Bojanic is the Anheuser-Busch Foundation Professor of Tourism in the Department of Marketing at the University of Texas at San Antonio. He has published two books and numerous research articles on tourism and hospitality marketing.

Elizabeth Cartier is a PhD candidate in the Department of Hospitality and Tourism Management of the Isenberg School of Management at the University of Massachusetts Amherst. Her primary research interests are in the areas of community based tourism, place attachment and human resource management in hospitality.

References

Abelson

Peter

. 2011. “Evaluating Major Events and Avoiding the Mercantilist Fallacy” Economic Papers 30 (1): 48–59.

Backer

Elisa

. 2008. “VFR Travellers—Visiting the Destination or Visiting the Hosts?” Asian Journal of Tourism and Hospitlaity Research 2 (1): 60–70.

Backer

Elisa

. 2012. “VFR Travel: It Is Underestimated.” Tourism Management 33 (1): 74–79.

Crompton

John L.

2006. “Economic Impact Studies: Instruments for Political Shenanigans?” Journal of Travel Research 45 (1): 67–82.

Crompton

John L.

2010. Measuring the Economic Impact of Park and Recreation Services. Ashburn, VA: National Recreation and Park Association, Research Series.

Crompton

John L.

Lee

Seokho

. 2000. “The Economic Impact of 30 Sports Tournaments, Festivals, and Spectator Events in Seven U.S. Cities.” Journal of Park & Recreation Administration 18 (2): 107–26.

Crompton

John L.

Lee

Seokho

Shuster

Thomas J.

2001. “A Guide for Undertaking Economic Impact Studies: The Springfest Example.” Journal of Travel Research 40 (1): 79–87.

Cruz

Brant

Mendelsohn

Josh

. 2010. “Why Social Media Matters to Your Business.” Chadwick, Martin, Bailey Research Report. http://www.cmbinfo.com/cmb-cms/wp-content/uploads/2010/04/Why_Social_Media_Matters_2010.

Damonte

L. Taylor

Marcis

John G.

Rella

Thomas

. 2013. “Methodology to Reduce Bias in Tourism-Driven Economic Impact Studies.” Atlantic Economic Journal 41:451–52.

10.

Davies

Larissa

Ramchandani

Girish M.

Coleman

Richard J.

2010. “Measuring Attendance: Issues and Implications for Estimating the Impact of Free-to-View Sports Events.” International Journal of Sports Marketing & Sponsorship 12 (1): 11–23.

11.

Dholakia

Utpal M.

Durham

Emily

. 2010. “One Café Chain’s Facebook Experiment.” Harvard Business Review 88 (3): 26.

12.

Dillman

Don A.

2000. Mail and Internet Surveys: The Tailored Design Method. New York: John Wiley.

13.

Harnick

Peter

Crompton

John L.

2014. “Measuring the Total Economic Value of a Park System to a Community.” Managing Leisure 19 (3): 188–211.

14.

Howard

Dennis R.

Lankford

Samuel V.

Havitz

Mark E.

1991. “A Method for Authenticating Pleasure Travel Expenditures.” Journal of Travel Research 29 (4): 19–23.

15.

Hudson

Simon

Hudson

Rupert

. 2013. “Engaging with Consumers Using Social Media: A Case Study of Music Festivals.” International Journal of Event and Festival Management 4 (3): 206–23.

16.

Jeong

Ji Y.

Crompton

John L.

Dudensing

Rebekka M.

2016. “The Potential Influence of Researchers’ ‘Hidden’ Procedure Decisions on Estimates of Visitor Spending and Economic Impact.” Journal of Travel Research 55 (7): 874–888. doi:10.1177/0047287515605932.

17.

ShiNa

Jago

Leo

. 2013. “Evaluating Economic Impacts of Major Sports Events—A Meta Analysis of Key Trends.” Current Issues in Tourism 16 (6): 591–611.

18.

Lillywhite

Jay M.

Simonsen

Jennifer E.

Wilson

B. J. H.

2013. “A Portrait of the US Fair Sector.” Event Management 17 (3): 241–55.

19.

Mangold

W. Glynn

Faulds

David J.

2009. “Social Media: The New Hybrid Element of the Promotion Mix.” Business Horizons 52 (4): 357–65.

20.

Ramchandani

Girish M.

Coleman

Richard J.

2012. “Testing the Accuracy of Event Economic Impact Forecasts.” International Journal of Event and Festival Management 3 (2): 188–200.

21.

Raybould

Mike

Fredline

Liz

. 2012. “An Investigation of Measurement Error in Visitor Expenditure Surveys.” International Journal of Event and Festival Management 3 (2): 201–11.

22.

Rogers

Judy

. 2007. Guidelines: Survey Procedures for Tourism Economic Impact Assessments of Gated Events and Festivals. Canadian Tourism Commission: Research Resolutions & Consulting, Ltd. http://www.mtc.gov.on.ca/en/research/resources/Guidelines_Econ_Impact_Gated_Events_2007.pdf

23.

Stanley

Dick

Rogers

Judy

Smeltzer

Sandra

Perron

Luc

. 2000. “Win, Place or Show: Gauging the Economic Success of the Renoir and Barnes Art Exhibits.” Journal of Cultural Economics 24 (3): 243–55.

24.

Stynes

Daniel J.

1997. Economic Impacts of Tourism. www.msu.edu/course/prr/840/econimpact/pdf/ecimpvol1.pdf.

25.

Stynes

Daniel J.

1999. Guidelines for Measuring Visitor Spending. www.msu.edu/course/prr/840/econimpact/pdf/ecimpvol3.pdf.

26.

Stynes

Daniel J.

White

Eric M.

2006. “Reflections on Measuring Recreation and Travel Spending. Journal of Travel Research 45 (1): 8–16.

27.

Taks

Marijke, B.

Green

Christine

Chalip

Laurence

Kesenne

Stefan

Martyn

Scott

. 2013. “Visitor Composition and Event-Related Spending.” International Journal of Culture, Tourism, and Hospitality Research 7 (2): 132–47.

28.

Tyrrell

Timothy J.

Johnston

Robert J.

2006. “The Economic Impacts of Tourism: A Special Issue.” Journal of Travel Research 45 (1): 3–7.

29.

University of Wisconsin. 2012. “Downtown Market Analysis/Trade Area Analysis.” http://fyi.uwex.edu/downtown-market-analysis/understanding-the-market/trade-area-analysis/.

30.

Warnick

Rodney B.

Bojanic

David C.

Cartier

Elizabeth A.

2013. “Economic Impact and Significance Studies: Research Measures.” Paper presented at the 2013 Northeast Recreation Research Conference Proceedings, USDA, US Forest Service Northeast Research Station. Cooperstown, NY, April.

31.

Warnick

Rodney B.

Bojanic

David C.

Feng

. 2015. “Using a Trade Market Analysis Technique to Refine Measurements for Economic Impact Analysis of Special Events.” Journal of Travel Research 54 (1): 52–65.

32.

Wynen

Jan

. 2013. “An Estimation of the Determinants of the Same-Day Visitor Expenditures in Belgium.” Tourism Economics 19 (18): 161–72.

33.

Young

Cheri A.

Corsum

David L.

Baloglu

Seyhmus

. 2007. “A Taxonomy of Hosts Visiting Friends and Relatives.” Annals of Tourism Research 34 (2): 497–516.

34.

Xiang

Zheng

Gretzel

Ulrike

. 2010. “Role of Social Media in Online Travel Information Search.” Tourism Management 31 (2): 179–88.