« .‘yrl. ... A .124: 1051.103 'Y 3 '. n' r1 's'. mm .1 15.7‘v'1-3rur . .- .Pl..vvtt.e.ualll.'.¥:. {lttlfbAnirvlf 3.1.1.! J II.- \l?i:.§\i3‘VII-rl.vl.~ .. A... ‘I f. $0-...fl'fltv “‘1‘ l r . “14......2-78'2U». DCI‘OpOI'I'l'b l.- 9-. 0- :5 p A‘. v» I: 91... :LIIVIHWWI. . .Ilrr. .19 . 25.! :1. 0 3| ’1) .1.»- 4.. 01'..r uD'OIw .l I: ‘3. .. {P 4.... g :31 Ill... :5 .1, 0 .0... 11:.-- 1. . fl: 1...» 5 515003.)..an . l .nbli“-.3’. £‘h..f‘.|t t. .II ‘II‘ .,1..r.:... z {113253. 519%.... -3..- .313..- -. - ..-.I! a. 15!. up. .. .......:o.. 1...; i-.- 312...} . - - 12:35.91»... 2... .41 - . , . . .. . ..4b...u.~.wvvr..§gx. .F . [.73 L TATE NWERSITY LIBRAHIE Itit‘tl‘l’ilt mt illllllll Ii Ill 31293 00917 5518 El This is to certify that the dissertation entitled EVALUATION OF THE EFFECTIVENESS OF A CAMPING REFUND OFFER AND THE RELATIONSHIPS AMONG CAMPERS' CHARACTERISTICS presented by ICK-KEUN OH has been accepted towards fulfillment of the requirements for PH . D . degree inPARKS & RECREATION RESOURCES éhflflfi Major professor Date flbd 6)} ///fl MSU is an Affirmative Action/Equal Opportunity Institution 012771 fl—_ LIBRARY Michigan State University J "v“ PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or betore date due. 1 DATE DUE DATE DUE DATE DUE .____._=__J fir— MSU Is An Affirmative Action/Equal Opportunity Institution cmmme-m EVALUATION OF THE EFFECTIVENESS OF A CAMPING REFUND OFFER AND THE RELATIONSHIPS AMONG CAMPERS' CHARACTERISTICS BY Ick-Keun Oh A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirement for the degree of DOCTOR OF PHILOSOPHY Department of Parks and Recreation Resources 1990 J/Uag JV“ / ABSTRACT EVALUATION OF THE EFFECTIVENESS OF A CAMPING REFUND OFFER AND THE RELATIONSHIPS AMONG CAMPERS' CHARACTERISTICS BY Ick-Keun Oh There is a lack of research in evaluating sales promotions (e.g., coupons, refund offers) utilized by the recreation and tourism industry. Furthermore, no research could be identified which delineates the characteristics of persons who are prone to utilize recreation-related sales promotions. The primary objective of this study was to evaluate the effects of a refund offer on Campvention attendees' pre- and post-Campvention camping. The refund offer was $1 off per night of camping in Michigan state parks or commercial campgrounds. The second objective was to examine the relationships among intent to utilize the refund offer, income, education, and camping affiliation. A systematic random sample of the 4,729 who preregistered for the 1988 National Campers and Hikers Association's (NCHA) Campvention (held in Michigan), was selected (n = 1,575). The pretrip and the posttrip questionnaires were mailed out to the sample. Response Ick-Keun Oh rates for the pretrip and the posttrip questionnaires were 49.4% and 53.8%, respectively. A t-test was used to determine whether or not the $1 refund offer influenced campers to spend more nights in state parks or commercial campgrounds. The proportion and the number of nights in these campgrounds were used as dependent variables. The proportion was obtained by dividing the number of nights camped in "refund eligible" campgrounds by the total camping nights in Michigan. The awareness of the refund offer and intent to utilize it were independent variables. A log-linear analysis was employed to discover the relationships among the above-mentioned four categorical data. A backward elimination method was used to determine the best fitting model. The t-test results indicate that the refund offer was not enough to induce Campvention attendees to camp more nights in "refund eligible" campgrounds. Log-linear analysis suggests that Campvention attendees who were strongly affiliated with camping were more likely to utilize the refund offer than those who were not. There was no significant relationships between intent to utilize the refund offer and income or education. ACKNOWLEDGMENTS I am indebted to many people who helped me throughout the course of this dissertation. First of all, I'd like to express sincere appreciation to Dr. Edward Mahoney. He has been source of inspiration throughout my Ph.D. program. I am very grateful for his creativity, insight, and financial support. I thank Dr. Donald Holecek for his intellectual support at critical points and editorial criticism. I would like to thank Dr. Dale Wilson and Dr. Gordon Miracle for their constructive comments and encouragement. My special thanks go to Dr. Daniel Stynes for his valuable guidance with initial drafts of my dissertation. I am greatly indebted to Douglas Jester of the Michigan Department of Fisheries Division for his statistical guidance and criticism. Mr. Bill Noble of the Department of Statistics was so much helpful and patient whenever I needed consulting services. My brother-in-law, Dr. Jong-Won Park at the University of Illinois gave me constructive comments. My acknowledgments can not be completed without my thanks to my wife, Helen, my son, John, and my parents. They have been fountains of encouragement and love. They are the people who made me what I am today. iv TABLE OF CONTENTS Page LIST OF TABLES ......................................... viii CHAPTER I. INTRODUCTION ................................... 1 Definition and Use of Sales Promotions ....... 2 Statement of the Problem ..................... 5 The National Campers and Hikers Association's Michigan Campvention and Sales Promotion ..................... 7 Study Objectives and Hypotheses ............... 8 Objectives ................................ 9 Hypotheses ................................ 9 Organization of This Dissertation ............ 10 II. LITERATURE REVIEW .............................. 12 Sales Promotion Trends ....................... 12 Reasons for the Increased Use of Sales Promotions .................. 13 Recreation 8 Tourism Related Use of Sales Promotions .................. l6 Refund Offers: Use, Advantages, and.Disadvantages ....................... 18 Methods for Evaluating the Effectiveness of Sales Promotions ..................... 19 Sales Increase ............................ 20 Profitability ............................. 22 Redemption Rate ........................... 24 Characteristics and Behavior of Coupon-Prone Consumers .................. 26 Summary ...................................... 28 III. RESEARCH METHODS ............................... 30 Sampling Design .............................. 30 Data Collection Methods ...................... 32 Survey Design ............................. 32 CHAPTER Page Survey Administration ........................ 35 Response Rates ............................... 37 Data Preparation and File Management ......... 38 Assessment of Nonresponse Bias ............... 39 Method Used to Evaluate the Effect of the Refund Offer ... .................. 42 Statistical Methods .......................... 44 IV. FINDINGS AND RESULTS . .......................... 46 Characteristics of Persons/Parties Attending the Campvention ............... 46 Socioeconomic Characteristics ............. 47 Camping Behavior .......................... 49 Characteristics of Campvention Trip ....... 50 Experience with Coupons/ Refund Offers ........................ 52 Intent to Utilize the $1 Refund Offer ...................... 53 Comparisons of Parties Who Indicated that They Would, and Would Not-Apply for a Refund ............................ 55 Tests of the Effects of the $1 Refund Offer ............................ 64 Test Results .............................. 67 Summary ...................................... 70 V. LOG-LINEAR ANALYSIS AND PROBABILITY OF UTILIZING THE MONEY-BACK OFFER ............ 71 Log-linear Analysis .......................... 71 Advantages and Disadvantages of Log-linear Analysis .................. 72 Applications of Log-linear Analysis ....... 73 Comparisons of Log-linear Analysis with Alternative Statistical Methods ...... 74 Types of Log-linear Models ................ 76 Saturated vs Unsaturated Models ........ 76 Hierarchical vs Nonhierarchical Models ............................ 77 Independence Models .................... 78 Parameters in Log-linear Models ........... 78 Estimation of Parameters .................. 80 Model Building and Selection .............. 81 Examination of Relationships .................. 84 Hypothesis and Variables .................. 84 Selection of a Log-linear Model ........... 88 vi CHAPTER Page Models Generated Using the Backward Elimination Method .......... 89 Test of the Significance of the Effects in Models That Best Fit the Data ..... 91 Significance of the Effects in the Final Log-linear Model ............... 93 Residual Analysis ..... . ................ 95 Paramater Tests ........................... 97 Probabilities of Utilizing the $1 Refund Offer ..................... 100 Summary ...................................... 104 VI. SUMMARY, RESEARCH RECOMMENDATIONS, AND MARKETING IMPLICATIONS ................ 105 Summary of Findings .......................... 105 Limitations/Weaknesses ....................... 108 Research Recommendations ..................... 109 Marketing Implications ....................... 111 BIBLIOGRAPHY ........................................... 117 APPENDICES ............................................. 125 A. 1988 Michigan Campvention Pretrip Survey B. 1988 Michigan Campvention Posttrip Survey C. $1 Refund Offer Form D. 1988 Campvention Follow-up Survey vii 10. ll. 12. 13. LIST OF TABLES Response Rates to the Pretrip, Posttrip, and Nonresponse Questionnaires ...................... Comparisons of Characteristics of Respondents and Nonrespondents ............................... Format for Calculating the Refund Effect ......... Socioeconomic Characteristics of Respondents ...................................... Camping Behavior of Campvention Attendees ........ Camping Affiliation: Subscription to Camping Magazines, Membership in Camping Clubs, and Attendance at Camping/Outdoor Shows .............. The Characteristics of Michigan Campvention Trips and Sources of Camping Information ......... Experience with Coupons or Refund Offers for Camping, Other Recreational Opportunities, and Nonrecreational Products ...................... Awareness of and Intent to Utilize the $1 Refund Offer .............................. The Percent of Campvention Attendees Who Indicated that They Would Likely Have Utilized a Higher Refund Offer ................... A Comparison of Those Who Intended and Did Not Intend to Utilize the $1 Refund Offer ........ ANOVA and T-Tests for the Differences in Intention to Utilize the Refund Offer by Income and Education Levels ................... ANOVA and T-Tests for the Differences in Intention to Utilize the Refund Offer by the Number of Nights Camped in Michigan Before and/or After the Campvention .............. viii 51 53 63 Table 14. 15. 16. 17. 18. 19. 20. Tests of Homogeneity of Variance and Decisions on the Use of Different Types of T-test ....... T-Tests on the Proportion and Number of Camping Nights between Those Who Were Aware and Not Aware, and Those Who Intended and Did Not Intend to Seek a Refund ............... Results of Chi-Square Tests of Association between Intent to Utilize the $1 Refund Offer and Income, Education, and Camping Affiliation Models Fitted to Four-Way Crosstabulation of Education, Income, Camping Affiliation, and Intent to Utilize the $1 Refund Offer ......... Four-Dimensional Contingency Table with Observed, Expected, and Standardized Residuals for the [AC][EI] Model (N=459) ................ Estimates for Parameters Coefficients Relevant to the [AC] and [EI] Interaction Effects ...... The Percentage of Persons Who Indicated They Would Seek a Refund by the Different Characteristics of Campers .................... ix 86 CHAPTER I INTRODUCTION The use of various types of sales promotions is becoming more prevalent among tourism and recreation industries. Hotels, airlines, and restaurants provide their customers with free tickets, coupons, or price discounts in order to attract business. Recreation providers of activities such as skiing, camping, fishing, and golf similarly use these techniques. In 1990, for example, the Michigan Association of Private Campground Owners (MAPCO) offered two nights of camping for the price of one night during the off-season. In 1989 the Michigan State Parks Division offered a one-day, free admission coupon to state parks when the consumer purchased a six-pack of Pepsi Cola during the promotion period. The intent of these sales promotions was to increase the purchase of services or products and/or influence the timing of the purchase. As tourism and recreation businesses become more competitive, sales promotions are used as important marketing tools to promote products and services to prospective consumers. 1 2 Spending on sales promotions in the United States was approximately $100 billion in 1987, which accounted for 64.4% of the total expenditure on promotion, while advertising only accounted for an expenditure of 35.6% in the same year (Johnson, 1988). Because of the increased use and spending on sales promotions, more attention is being directed to sales promotion as marketing and communication techniques. Definition and Use of Sales Promotions Sales promotions, as defined by Kotler (1984), are "short-term incentives to encourage purchase or sale of a product or service" (p. 603). Strang (1976) further defined sales promotions as those activities which provide an incentive, additional to the basic benefits provided by a product or service, and which temporarily change the perceived price/value relationship of that product or service. Sales promotions consist of a wide variety of techniques designed to stimulate a sales response from consumers, wholesalers, and/or retailers. Some techniques are directed at the sales force. There are two primary types of sales promotions currently in use. Consumer-oriented sales promotions include free samples, coupons, refund offers, price discounts, premiums, contests, trading stamps, and product demonstrations. Trade-oriented sales promotions, on the 3 other hand, include free goods, merchandise allowances, cooperative advertising, dealer sales contests, buying allowances, and push money. Sales-force promotions, one of the trade-oriented sales promotions, include a variety of incentives, bonuses, contests, and sales rallies. A majority of sales promotions are directed toward frequently purchased consumer goods. However, as noted before, tourism and recreation businesses are increasing their use of sales promotions. Because of a keenly competitive market, the tourism and hospitality industries have adopted both consumer-oriented and trade-oriented sales promotion techniques. Consumer-oriented sales promotions offer the consumer various types of incentives for purchasing a product or service for the first time, purchasing more of a product or service at a time, or increasing the purchase frequency. For example, frequent traveler programs are used by most airlines as well as an increasing number of hotels (triple mileage offers by airlines; special rate for frequent users by hotels). Airlines also offer coupons, providing passengers with discount fares to specific destinations, with some restrictions. Restaurants, too, offer coupons or price discounts for food or beverages for a limited time. Trade-oriented promotions are, on the other hand, incentives to businesses and organizations, i.e., retailers or wholesalers comprising distribution channels. Tourism 4 providers, such as hotels, airlines, and tourist destinations, utilize a variety of trade-oriented sales promotion techniques. Travel agents and tour wholesalers are also given special rates at hotels and resorts and free trips to a given destination as prizes. Airlines reward travel agents who sell a certain number of tickets on their airlines. Although sales promotions can contribute to the accomplishment of a variety of different promotional objectives, the primary emphasis of most sales promotions is to increase short-term sales. A short-term sales increase can be achieved by encouraging current consumers to buy more of a product at one time than they normally would, or to increase purchase frequency for a certain period of time. Another common objective sales of promotions is to attract nonusers of a product (Schultz & Robinson, 1982a). Many sales promotions are designed to encourage first-time purchases (trials) by nonusers. For example, Michigan ski areas cooperate in sponsoring a "free skiing day" designed to attract nonskiers by providing them with free lift tickets, equipment rentals, and lessons. The Michigan Department of Natural Resources sponsors a "free fishing weekend," at which time people can fish without purchasing fishing licenses. Various sport fishing organizations offer a variety of fishing programs during that particular weekend. 5 A third objective of sales promotions is to encourage new customers to repeat their purchase(s) of a product or service. If the consumer is initially satisfied with the product or service, sales promotions can be used to generate repeat purchase(s). For example, some campgrounds and ski businesses offer customers price discounts or coupons for additional services/products to be used during their next visitf Finally, sales promotions are used to offset promotional activities of competitors. For example, if one ski area advertises on TV or radio, another ski area might use a counterbalancing sales promotion. Statement of the Problem Evaluating the effectiveness and efficiency of sales promotions is essential, given the increased spending and reliance on sales promotions. Although sales promotions are important in marketing products or services, many sales promotions are not evaluated. Studies of the effectiveness of sales promotions which have been performed, frequently have not been conducted in a scientific manner. Jolson, Wiener, and Rosecky (1987) assert that planning and evaluation of sales promotions have often been conducted by intuition and personal judgment rather than by scientific research methods. According to Reibstein and Traver (1982) 6 few firms are known to have conducted scientific evaluations of their sales promotion programs. In addition, most research on the results from sales promotions has been narrowly focused on the effectiveness of coupons, and more specifically, on the relationship between redemption rates and different coupon values. However, the redemption rate, itself, is not a good measure of the effectiveness of refunds or coupons. The major objective of sales promotions is to increase purchases by new and/or repeat customers, not to maximize the number of coupons redeemed or refunds given. Thus, the real measure of the effects of promotional offers should be incremental sales induced by the promotional offers. There is a deficiency of evaluation research on the effects of sales promotions for manufactured goods and even a greater deficiency in evaluating sales promotions utilized by recreation and tourism businesses, agencies and promotional organizations. A literature review (see Chapter II) identified approximately 30 evaluation studies of sales promotion techniques conducted since 1970. Only three dealt with the evaluation of sales promotions which were used in the hospitality industry and all three were restaurant- related promotions (Block, Brezen, 8 Schultz, 1986; Chapman, 1986; Varadarajan, 1984). Although sales promotion techniques are being used more by recreation and tourism businesses, organizations, and 7 agencies, relatively few studies have evaluated the effects of recreation and tourism-related sales promotions. Without this type of evaluation, it is difficult to assess the cost effectiveness of these sales promotions or to identify ways of increasing their cost effectiveness. Furthermore, no research could be identified which identifies the characteristics of persons who are prone to utilize recreation/tourism-related sales promotions. This information would assist in designing, targeting, and promoting sales promotions. While studies of nonrecreation or nontourism sales promotions provide some important insight, there is a need to undertake research to determine whether these findings carry over to recreation/tourism sales promotions. The National Campers and Hikers Association's Michigan Campvention and Sales Promotion The National Campers and Hikers Association (NCHA) is one of the largest and most active camping organizations in the country with more than 25,000 members. Every year NCHA holds a Campvention in a different state. The 1988 Campvention was held in Michigan during July 8-14 at Highland State Recreation Area. Approximately 4,000 parties attended this event. The Michigan Association of Private Campground Owners (MAPCO) and the Division of State Parks (Michigan Department 8 of Natural Resources) formed a promotional partnership to capitalize on the Campvention and the thousands of parties who would be attracted to Michigan. The partnership's 1 for each strategy included a $1 money-back (refund) offer night Campvention attendees camped in state parks or commercial (MAPCO) campgrounds before or after the Campvention (from June 24 to August 5). MAPCO and Division of State Parks jointly financed the promotion, including the refunds. The objective of the money-back offer was to encourage Campvention attendees to camp additional nights in state parks and/or commercial (MAPCO) campgrounds and, therefore, to increase their stay (spending) in Michigan. MAPCO and Division of State Parks requested that Michigan State University assist in evaluating the effectiveness of this promotion. Study Objectives and Hypotheses The primary objective of this study is to evaluate the effects of the $1 money-back offer on pre- and post- : A money-back (refund) offer is defined as "an offer by a manufacturer or a marketer to refund a certain amount of money when the product is purchased alone or in combination with some other products." (Schultz & Robinson, 1982b, p. 131). A refund offer is a small incentives of cash, checks, or coupons mailed back to consumer. The " terms, "refund offer" or money back offer" are used interchangeably in this study. 9 Campvention camping in Michigan state parks and/or commercial (MAPCO) campgrounds. Another objective of this study is to examine the relationships among intent to utilize the $1 refund offer, income, education, and camping affiliation (subscriptions to camping magazines and attendance at camping or outdoor shows). Log-linear modelling was the primary method used to examine these relationships. Objectives The following objectives were established to assist in guiding the study: Objective 1. To provide a marketing relevant descriptive profile of campers who intended to take advantage of the Campvention money-back offer. Objective 2. To assess the effects of the money-back offer on pre- and post-Campvention camping, (i.e., number of camping nights) in Michigan. Objective 3. To examine the relationships among intent to utilize the refund offer, income, education, and camping affiliation. Objective 4. To estimate the probability of utilizing the refund offer by different segments of Campvention attendees. Hypotheses A thorough review of theoretical and applied studies of sales promotions, especially coupon and money-back offer, provided the basis for three primary hypotheses. Hypothesis 1: Hypothesis 2: Hypothesis 3: 10 Parties who intended to utilize the $1 refund offer differ from those who did not intend to seek a refund in terms of their socioeconomic characteristics, camping behavior, camping affiliation, or past experience with coupons or refund offers. The $1 money-back offer influenced campers to stay more nights in state parks and/or commercial (MAPCO) campgrounds. a. Parties who were aware of the money-back offer camped (l) a greater proportion of their nights and (2) a greater number of nights in Michigan state parks and/or commercial (MAPCO) campgrounds than those who were not aware of the offer. b. Parties who intended to take advantage of the money-back offer camped (l) a greater proportion of their nights and (2) a greater number of nights in state parks and/or commercial (MAPCO) campgrounds than those who did not intend to utilize the offer. Intent to utilize the $1 refund offer is strongly related to levels of income, education, and camping affiliation. Organization of This Dissertation The purpose of this chapter was to introduce the study, define the primary research problem, and specify the objectives and hypotheses. In Chapter II, a review of the literature on sales promotion with special focus on the importance and the increased use of sales promotions and methods for evaluating the effects (effectiveness) of sales promotion techniques was presented. Research methods including the sampling method, data collection procedures, 11 and methods used to evaluate the effects of the Campvention refund offer are described in Chapter III. In Chapter IV, a comparison of those who did and did not intend to utilize the $1 refund offer and results of tests of hypothesis 1 and 2 was provided. The results of tests of hypothesis 3 using log-linear analysis were presented in Chapter V. Finally, in Chapter VI, the summary, research recommendations, and marketing implications were provided. CHAPTER II LITERATURE REVIEW This chapter begins with a review of literature on trends in the use of sales promotions and various types of sales promotions. It is followed by a section which focuses on literature concerning the use and role of sales promotion in an overall promotion strategy as well as the advantages and disadvantages of refund offers. In the third section, methods for evaluating the effectiveness of sales promotions (e.g., refund offers or coupons) were introduced and discussed. Literature pertaining to characteristics and behavior of coupon-prone consumers was reviewed in the final section. Sales Promotion Trends The most significant change in product/service marketing over the last two decades has been the transfer of promotion dollars from advertising to sales promotion activities (Rosenfeld, 1987). An increasing number of business managers realize the importance of sales promotion 12 13 as part of an effective marketing strategy. Even owners and managers of small businesses have increased their use of sales promotions (Varadarajan, 1984). Studies show that annual spending on promotion exceeded advertising expenditures every year since 1969 (Bowman, 1988). Donnelly Marketing's 1988 Annual Report shows that sales promotion's share of promotion budgets increased from 59% in 1980 to 69% in 1987, while the percent of promotional budgets spent on media advertising decreased from 41% to 31% during the same period (Bowman, 1988). Although trade-oriented sales promotions account for a large part of sales promotional dollars, the trend is for manufacturers to place greater emphasis on consumer-oriented sales promotions (Higgins, 1986). Consumer-oriented sales promotion programs accounted for 32% of the money spent on promotion and coupons accounted for more than 70% of consumer-oriented sales promotion (Higgins, 1986). Dun and Bradstreet Corporation (1988) reported that seven billion manufacturers' coupons were redeemed by 0.8. consumers out of the 200 billion coupons which were distributed in 1987. Reasons for the Increased Use of Sales Promotions Strang (1976) and Schultz and Robinson (1982a) discuss a number of reasons for the increasing use of sales promotions. First, consumers became more sensitive to the price of products and services during periods of inflation l4 and recession during the 19703 and early 19803. This made consumers more responsive to sales promotion programs. For example, when the oil crisis occurred in the 19705, almost two-thirds of the U.S. households used coupons in order to lower living costs. Second, the efficiency of advertising declined because of increasing media costs and media clutter. Third, competition among the existing brands in most product categories has intensified. Because of the increasing number of new brands and the promotional cost associated with launching new products, most manufacturers of consumer goods heavily rely on sales promotion to introduce their new products (Kotler, 1984). Finally, more top managers now consider sales promotions as an acceptable marketing activity. In the past, many felt that sales promotions (e.g., coupons, refund offers, contests) cheapened the brand. This attitude has changed with the success of sales promotion programs. Among various types of sales promotion techniques, coupons have become an important component in promoting frequently purchased consumer products. Coupons are defined by the American Association of Advertising Agencies (1978) as a certificate given to consumers which entitles them to an immediate price discount when they purchase the stated item. To get the benefit of coupons or refunds, users are required to make an effort to collect and redeem them. This 15 differs from other consumer-oriented sales promotion techniques such as bonus packs, contests, premiums, sweepstakes, or samplings. Coupons also have some advantages over simple price discounts. Studies have shown that coupons appear to generate a greater sales response than equivalent reductions in price (Cotton & Babb, 1978; Gardner & Strang, 1984; Schindler, 1984). Schindler (1984) discusses three psychological mechanisms which in part explain why coupons are more effective (in generating sales response) than simple price reductions. First, the agtention;§nd awareness mechanism explains why consumers are more likely to be aware of a coupon than a simple price discount. Consumers generally demonstrate a low level of price awareness. Second, the discount information mechanism explains that since consumers perceive price as an index of quality, a simple price reduction can lead consumers to believe that a product/service is worth only the reduced price. Conversely, a price decrease in the form of a coupon may make consumers feel that they are getting more (a product/service valued at the original price) for their money. Finally, the price choice mechanism demonstrates that when consumers obtain a discount through exercising their own judgment, they can take credit for the savings. And such feelings can increase the subjective value of a coupon . 16 Although coupons are still the most prevalent form of sales promotion, the wide-scale use of coupons has resulted in a decrease in redemption rate. As a result, there is a trend toward more selective coupons and sample mailings, multibrand (coupon) promotions, and refund offers instead of cents-off coupons (Schleier, 1985). Recreation 8 Tourism Related Use of Sales Promotions Sales promotions, as already mentioned, are also being used much more by hospitality businesses including restaurants, airlines, and hotels. Money-back offers, coupons, or premiums are used by restaurants primarily to generate short-term sales increases. Fast food restaurants are increasing sales promotions (e.g., give-away toys) aimed at the "young children" market. For example, McDonald's restaurants regularly offer different in-pack premiums (e.g., toys, games) as part of their effort to establish brand loyalty among children (Hume, 1989). The airline and hotel industries employ continuity 2 as long-term promotion tools to develop brand offers loyalty. Continuity offers are designed for long-term action by encouraging customers to purchase the product at more frequent intervals in order to acquire additional or 2 They includes stamp plans (collecting a certain number of stamps that may later be traded for cash, merchandise or a combination of the two. 17 complementary units of the product as a premium (Boddewyn & Leardi, 1989). Airlines use frequent-flyer programs and hotels employ frequent-traveler programs in order to induce brand loyalty. For example, frequent flyers can earn free flights after accumulating points for total mileages flown. Hotels, such as Sheraton, Inter-Continental, and Omni International, have frequent traveler programs which provide frequent (loyal) customers with free merchandise and/or discounts on airfare and car rentals (Higgins, 1986). Another trend is the increasing use of cooperative sales promotions3tw'tourism/hospitality businesses. Some hotel chains and car rental companies have established partnerships with airlines, enabling travelers to accumulate (travel/purchase) points and win awards (Higgins, 1986; Wright, 1989). For example, members of the Omni Select Guest Program can accumulate points (for nights spent in Omni hotels, for renting a car from Hertz, or buying a ticket on New York Air) which entitle them to discounts on airfare, car rental, and hotel room charges. 3 Cooperative promotions are also called tie-in or group promotions which involve two or more brands simultaneously. The consumer is offered an incentive to purchase all of the participating brands. This technique is usually linked to a common theme, and often uses other forms of sales promotion (Boddewyn & Leardi, 1989). 18 Refund Offers: Use, Advantages, and Disadvantages Refund (money-back) offers became widespread sales promotion techniques in the 19805 (Jolson et al., 1987). Refund offers now challenge the dominance of couponing as a sales promotion tool (Telzer, 1987). Telzer reported that households using refund offers increased from 27% to 51% between 1977 and 1984. Several factors contribute to the increasing use of refund offers. First, refund offers provide a relatively large value compared with coupons. Few refunds are less than 50 cents and many offer a substantial number of dollars, whereas most coupons usually offer less than $1 (A.C. Nielsen Company, 1983). Second, the redemption (refund) cost to the sponsor/promoter is relatively lower than for coupons because of a high slippage rate (the ratio of the number of consumers who purchase a product with the intent to claim the refund, but fail to do so, to the total number of consumers who purchase the product with intent to claim the refund). Usually only about 1% of refund offers distributed through media are redeemed (Schultz & Robinson, 1982b). Thus, the sponsor/promotor's average redemption cost per product unit is less than the refund value (Jolson et al., 1987). However, there are a number of disadvantages associated with refund offers. The primary disadvantage to the 19 business offering the refund is that refunds often benefit those who are already brand loyal or have already decided to purchase a product rather than generating sales by nonusers (Schultz 8 Robinson, 1982b). Studies have shown that sales promotions are unlikely to influence brand loyal consumers to switch to other products (Brown, 1974; Massy 8 Frank, 1965; Montgomery, 1971; Neslin 8 Clarke, 1987; Neslin, Henderson, 8 Quelch, 1985; Raju 8 Hastak, 1979; Webster, 1965). To brand loyal consumers, a refund offer sometimes means a simple price reduction rather than an incentive for inducing a purchase. However, this drawback is common to other sales promotion techniques as well as to refund offers. Another disadvantage of refund offers is that they include no immediate reward and require additional effort (inconvenience) from customers (Schultz 8 Robinson, 1982b). Unlike coupons or price reductions, consumers are often required to collect the proof(s) of purchase and/or receipts, mail them to a redemption center, and then wait several weeks or more for the refund. Methods for Evaluating the Effectiveness of Sales Promotions The effectiveness of sales promotions has been measured using a variety of methods and criteria including short-term sales increase, profitability, and redemption rate (Chapman, 20 1986; Gupta, 1988; Henderson, 1985; Irons, Little, 8 Klein, 1983; Klein, 1981; Neslin et al., 1985; Shoemaker 8 Tibrewala, 1985; Walters 8 McKenzie, 1988). Sales Increase As stated earlier, a primary objective of sales promotions is to generate short-term sales increases. The amount of increase in short-term sales is, therefore, widely used as a measure of the effectiveness of sales promotions. A sales increase is measured by comparing sales during the promotional period with sales during comparable periods prior to the sales promotions (Brown, 1974; Gupta, 1988). Sales increase results from purchase acceleration including larger quantity purchases, more frequent purchases, and brand switching during a promotion period (Blattberg, Eppen, 8 Lieberman, 1981; Neslin et al., 1985; Neslin 8 Shoemaker, 1983; Shoemaker, 1979). Neslin and Shoemaker assert that purchase acceleration could change market share, and consequently, impact the profitability of a sales promotion. Recently, scanner panel data has been used to assess sales increases. Scanners read the Uniform Product Code (UPC) symbol and keep track of how many of a specific (promoted) item have been sold. These scanners also keep track of all the purchases of specific individuals when they show shopping identification cards. Scanner panel data can, therefore, provide accurate records of individual consumer purchases during the promotion period. 21 Neslin et al. (1985) showed that coupon sales promotions increased purchase quantity during the sales promotion period but also lengthened the purchase interval between the last purchase during the promotion period and the next purchase after the promotional period. Using a scanner panel over 28 weeks, it was found that a sales increase for products such as bathroom tissue and coffee was more likely to come from larger quantity purchases than from reducing between purchase interval. Gupta (1988) examined sales increases resulting from three sales promotion techniques: promotional price cut, feature and display, and feature or display. Two-year's scanner panel data from 100 households was employed. The result showed that more than 84% of the total sales increase in ground coffee was accounted for by brand switching, 14% or less by shortened purchase interval, and less than 2% by larger quantity purchase. However, these two studies failed to determine whether the sales increase was a direct result of coupons. As Strang (1976) asserts, sales comparison between the promotion period and the preceding period or the same period a year ago is not sufficient because many other factors such as weather, competition, and price changes influence sales. Also, Neslin et al.'s study (1985) showed that temporary sales increases during the promotional period are sometimes at the expense of future sales. 22 Instead, incremental sales are recommended as a more valid measure of effectiveness (Irons et al., 1983; Klein, 1981). Incremental sales are sales that take place as a result of a sales promotion that would otherwise not have occurred (Klein, 1981). They are measured by monitoring sales using an experimental treatment. The sales rate for a brand in a market with a sales promotion is compared to the rate in a similar market without it. Klein measured the sales impact of nine different coupons for frequently purchased consumer goods by examining purchase differences between test and control markets. He used a supermarket scanner panel divided at random into three demographically equivalent cells and found statistically significant sales effects over a 4 to l6-week period. Profitabilitx The profitability of sales promotions is often measured in conjunction with incremental sales and costs associated with a promotion campaign (Bawa 8 Shoemaker, 1987b; Chapman, 1986; Irons et al., 1983; Neslin 8 Shoemaker, 1983; Shoemaker 8 Tibrewala, 1985; Walters 8 McKenzie, 1988). The profit associated with a sales promotion is the sales volume times the gross margin per unit minus promotion costs. Promotion costs include production (e.g., printing), communication/distribution of the promotional offer, and redemption costs. Redemption costs are a function of face 23 value, retailer and clearing house handling charges, as well as the number of coupons redeemed. Chapman (1986) analyzed the profitability of a $1-off coupon promotion for a take-out pizza restaurant. A sales response function was used to predict base and experimental sales over a six-week period. Profits from couponing were calculated by subtracting fixed costs (materials and labor costs) associated with conducting the couponing promotion from incremental profits. The predicted profitability of couponing was estimated to be approximately 5.2% of the normal gross margins derived from sales without a couponing effort. Walters and McKenzie (1988) tested hypotheses on the effects of loss leaders (products temporarily priced at or below retailer cost), of in-store price specials, and of double coupon promotions on overall store sales, profit, and traffic. They found that: (1) most of loss leader promotions had no effect on overall store profit, (2) some loss leader promotions affected profit by increasing store traffic, not by increasing sales of the promoted items, (3) double-priced coupon promotions affected profit by increasing sales of couponed products rather than increasing store traffic, and (4) in-store price specials had no effect on store profit, sales, or store traffic. 24 Redemption Rate Redemption rates are important to managers in budgeting and evaluating sales promotion programs, especially coupon offerings (Henderson, 1985). Coupon redemption rate, the ratio of the number of coupons redeemed to the number of coupons distributed, is the most frequently used criteria to evaluate the effectiveness of coupon offers (Bowman, 1980; Kuehn 8 Rohloff, 1967; Neslin 8 Clarke, 1987; Reibstein 8 Traver, 1982; Ward 8 Davis, 1978). Strang (1976) reported that redemption rates for coupon promotions range from 2% to 25% depending on the coupon face value and distribution methods. Neslin et al.'s study (1985) showed that coupon redemption rates for bathroom tissue ranged from 5.2% to 26.3%. A study by Nielsen Company (1983) disclosed an average coupon redemption rate of 4.5% for frequently purchased consumer goods. Numerous studies have focused on how redemption rates are influenced by: (1) coupon or refund values (Bearden, Teel 8 Williams, 1981; Irons et al., 1983; Keon 8 Bayer, 1986; Shoemaker 8 Tibrewala, 1985), (2) purchase behavior, e.g., brand loyalty (Bawa 8 Shoemaker, 1987a; Brown, 1974; Henderson, 1985; Lee 8 Brown, 1985), (3) household characteristics, e.g., income, education, household size (Bawa 8 Shoemaker, 1987b), and (4) methods of coupon distribution (Reibstein 8 Traver, 1982; Ward 8 Davis, 1978). Shoemaker and Tibrewala (1985) measured consumer's 25 self-reported likelihood of redeeming coupons with different face values. They found that higher face values produced a substantial increase in the likelihood of redeeming coupons among infrequent and nonusers of a brand, while increasing face value had relatively little effect on redemption rate among regular (loyal) buyers of the brand. Bearden et al. (1981) discovered that the amount of coupon value, reflecting percentage reduction of a product price, influenced the willingness to try a new brand. Between 20- 25% and 70-75% of the respondents were willing to try a new brand for the 5% and 40% price reduction, respectively. Other studies have examined the relationship of other factors and coupon redemption rates. Henderson (1985) found that frequent coupon users were more likely to redeem coupons than were infrequent users. Brown (1974) and Lee and Brown (1985) reported that brand loyal buyers were less likely to respond to sales promotions. A study by Ward and Davis (1978) determined that direct mail distribution of coupons was most effective for increasing redemption rates; on-package coupons was least effective. Although redemption rate is a commonly used measure of coupon effectiveness, some researchers strongly question the use of redemption rate as an actual measure of the effectiveness of sales promotions. Irons et a1. (1983) and Klein (1981) assert that redemption rate only measures the cost of a coupon promotion; it offers no information about 26 the magnitude of incremental sales resulting from the promotion. In reality, the higher the redemption rate, the greater the cost of a coupon promotion (Irons et al., 1983). Characteristics and Behavior of Coupon-Prone Consumers A number of studies have attempted to identify and profile the socioeconomic characteristics and/or shopping behavior of sales promotion prone consumers (Blattberg, Buesing, Peacock, 8 Sen, 1978; Dodson, Tybout, 8 Sternthal, 1978; Henderson, 1985; Kono, 1985; Montgomery, 1971; Neslin 8 Shoemaker, 1983; Teel, Williams, 8 Bearden, 1980; Ward 8 Davis, 1978; Webster, 1965). No published studies of the characteristics of persons who are prone to take advantage of recreation/tourism-related sales promotions were identified. However, the studies of sales promotion prone consumers which are reviewed provided useful insight and information to formulate a hypothesis for this study. Blattberg et al. (1978) examined the relationship between deal proneness and household characteristics. A cross-classification analysis was applied to five frequently purchased consumer products, such as detergent and tissue. On an average, upper-income households were found to be more deal prone than lower-income households. Of the households owning a car and a home, 34.4% were deal prone, while only 20.5% of the households that did not own either a car or a 27 home were deal prone. In addition, working women were less deal prone than were nonworking women. It appeared that a housewife's employment status and the presence of children under six years of age affected deal proneness for frequently purchased consumer goods, but not as strongly as did car and home ownership. Narasimhan (1984) observed that tendency to utilize coupons was higher for households with higher levels of education and with no children under 18, while it was lower for households with an employed wife. His study found that the number of coupon associated purchases increased to a point and then decreased as household income increased. Teel et al. (1980), on the other hand, discovered that households intending to try new grocery products with coupons were likely to have higher incomes, have larger families, and were younger than the households who did not utilize coupons to purchase new products. Finally, Bawa and Shoemaker (1987b) found that coupon-prone market segments tended to be somewhat younger, higher incomed, more educated, more likely to live in urban area, and less likely to have children than noncoupon-prone segments. A number of studies have concluded that deal-proneness is inversely related to brand loyalty (Dodson et al., 1978; Kono, 1985; Montgomery, 1971; Webster, 1965). Kono (1985) noted, however, that brand loyal consumers were reluctant to switch from their favorite brand to unknown private brands 28 or generics to take advantage of a coupon offer. This study raises questions concerning the ability of a coupon to induce brand trials or switching among consumers who already have a preferred brand. On the other hand, Hackleman and Ducker (1980) found that deal proneness was highly correlated with the volume of a product purchased or used, that is, the more of a product consumers purchase, the more deal-prone they are. Few studies have attempted to identify the characteristics of consumers who are prone to utilize refund offers (Jolson et al., 1987; Lincoln, 1978). In one study, Jolson et a1. examined shopping attitudes of persons who purchased home appliances and found that consumers who recognized obvious bargains were more likely to take advantage of refund offers. Their study also showed that responsiveness to the refund offer was inversely associated with the perceived effort (e.g., time and cost) required for redemption. Summary Spending on sales promotions is increasing and now exceeds advertising expenditures. Couponing is still the most frequently used sales promotion technique. However, refund offers have been gradually adopted by marketers because of the relatively low redemption cost and the great potential to appeal to their customers. The effectiveness 29 of coupon sales promotions is most commonly measured in terms of sales increase, profitability, and redemption rates. Many studies identified the characteristics and shopping behavior of deal-prone consumers for frequently purchased consumer goods. In general, income and education are positively related to the use of sales promotions. No published research on the sales increase or profitability of recreation and tourism sales promotions could be identified. CHAPTER III RESEARCH METHODS In this chapter, the methods employed to collect, prepare, and analyze the data gathered for this study were described. It is divided into eight sections dealing with: sampling design, data collection methods, survey administration, response rates, data preparation and file development, assessment of nonresponse bias, method used to evaluate the effect of $1 refund offer, and statistical methods. Sampling Design A computer file containing 4,729 NCHA (National Campers and Hikers Association) parties who preregistered for the 1988 Campvention on or before April 30, 1988, was used as a sampling frame. Parties who preregistered after April 30 or who registered at the Campvention site were excluded from the sampling frame. Fewer than 100 parties who did not preregister attended the Campvention. 3O 31 The NCHA staff responsible for Campvention registration were instructed to select every third party with a random start from their computer file of preregistrants. A total sample of 1,575 (33%) of the 4,729 preregistrants was selected. A systematic random sampling method was used because it was convenient for the NCHA personnel to implement. Since preregistrants were randomly arranged on the computer file, each party had an equal opportunity to be selected. The relatively large sample size was needed for the following reasons. 1. One of the study objectives was to develop a marketing-relevant profile of persons who intended to take advantage of the $1 money-back offer. A review of sales promotion literature showed that coupon/refund redemption rates for many products were less than 10%. It was assumed that, at the very most, 20% of the parties attending the Campvention would take advantage of the offer. This, coupled with expectations of a 50% response rate to the questionnaire, necessitated a large sample size. 2. Another objective was to identify and analyze different market segments, e.g., off-season campers, attributes sought segments, residents and nonresidents (see Mahoney, Oh, 8 Ou, 1989). A large sample size was deemed necessary to capture an adequate number of observations for each distinct market segment. 32 3. Finally, another objective of the overall study (see Mahoney et al., 1989) was to compare pre-Campvention trip and posttrip perceptions of Michigan campgrounds. This required a sufficient number of respondents for both the pretrip and posttrip questionnaires. The literature review indicated that an insufficient number of responses to successive rounds of pretrip and posttrip questionnaires is a common problem. Data Collection Methods Data needed to achieve the study objectives were collected using two different self-administered mail survey instruments administered before (pretrip questionnaire) and after (posttrip questionnaire) the Campvention. Survey Design Meetings were held with representatives from MAPCO, Michigan State Parks, and NCHA in March 1988 to determine their informational needs and Campvention logistics to be incorporated in the study design. Drafts of the questionnaire were sent to these representatives for comments and suggested changes. The final four-page pretrip questionnaire was completed in April, 1988 (see Appendix A). The pretrip questionnaire included a brief introduction which indicated the study's purpose, sponsors, and guaranteed confidentiality. It also indicated that 33 respondents would receive another questionnaire after they returned home from their Campvention trip. This prenotification had the purpose of improving the response rate. The following information was collected on the pretrip questionnaire. 1. Camping behavior: campground preference for public or private campgrounds (See Q16, Appendix A); number of nights camped annually (Q18); off-season--before Memorial Day (Q21), and after Labor Day (Q22)--camping. 2. Socioeconomic characteristics: age (Q4); zip code of permanent residence (Q26); gender (Q27); work status (Q28); marital status (Q29), and presence of children living at home (Q30). Other information was also collected on: attendance at the 1987 Iowa Campvention, pre-Campvention trip plans including number of nights, nights planned at the Campvention site, and additional nights of camping in Michigan, number of Michigan campgrounds already selected and reserved, likelihood that they would take advantage of the $1 per night money-back promotional offer, the importance assigned different attributes in selecting campgrounds, and pretrip perceptions of Michigan campgrounds. This information is reported in Mahoney et al. (1989). Some of the same data (e.g., camping equipment, party size, ages) collected on the pretrip questionnaire were also collected on the posttrip questionnaire. Overlapping data were needed to assess possible response biases including 34 difference between those who completed both the pretrip and the posttrip questionnaires, and those who did not complete either of the two questionnaires. Data collected on the posttrip questionnaire included: 1. Number of persons in travel party and the age of party members (Q3). 2. Length of their Campvention trip (Q7a), nights at the Campvention site (Q7b), additional nights in Michigan~~ at campgrounds, in hotels/motels, at friend/relative's home(s) (Q7c & d). 3. Distribution of the additional (pre- and post- Campvention) nights camped in Michigan-~state parks, local/county, state forests, national parks or forests, and commercial (MAPCO) campgrounds--across four regions in Michigan (Q8b). 4. Sources of information used to select Michigan campgrounds (Q12). 5. Whether or not respondents were aware of the $1 refund offer (Q16). 6. Whether or not they would likely take advantage of the $1 refund offer (Q17), amount of refund (number of nights) they intended to apply (Q17a), whether the refund offer influenced people to camp more nights in state parks or commercial (MAPCO) campgrounds (Q17b), and whether they would have utilized a refund with a differing amount of value (Q17c). 7. Whether or not they have utilized money-back offers or coupons for camping, other recreational activities, and nonrecreational goods (Q18-20). 8. Subscriptions to camping-related magazines, memberships in camping organizations other than the NCHA, and whether or not they attended camping or outdoor shows (Q22-24). 9. Annual household income in 1987, and education level (Q26-27). 35 The posttrip questionnaire also included questions with respect to trip spending, importance assigned different attributes in selecting a campground, perceptions of Michigan campgrounds, and evaluation of the overall Campvention and Campvention facilities (see Appendix B). Persons who indicated that they would likely not take advantage of the $1 money-back offer were asked whether or not they would likely have taken advantage of a higher (larger) money-back offer. Seven higher offers ranging from $2 to $8 were tested. The question (17c) reads, "If the refund had been 5 per night, would you have taken advantage of the offer?" Questionnaires with a refund value of $2, $3, $4, $5, and $7 inserted into the blank were mailed to 215 people/parties each. Questionnaires with refund value of $6 and $8 were sent to 250 people each. Persons asked about their likely response to different "higher" refunds were randomly chosen. Survey Administration A combined money-back offer promotion and refund application form (see Appendex C) was mailed, as part of the Campvention registration package, to all parties who preregistered. The refund offers were valid for nights camped in Michigan state parks and/or commercial (MAPCO) campgrounds from June 24 through the night of August 5, 1988. Parties attending the Campvention were requested to 36 return the application form and proof of purchase (camping permits/receipts) to MAPCO headquarters by September 1, 1988. NCHA provided three sets of mailing labels for each person/party who comprised the sample. Each label included name and mailing address as well as their unique Campvention registration number. Two labels were used for the pretrip questionnaire: one for mailing the questionnaire and postage-paid return envelope; the other was attached directly to the questionnaire. The label attached to the questionnaire was needed to match the pretrip and posttrip questionnaires completed by the same respondent. The remaining label was used to mail the posttrip questionnaire. The pretrip questionnaire and postage-paid return envelope were mailed by the NCHA on May 9, 1988, eight weeks before the Campvention, as part of the Campvention registration package. The registration package also included: (1) information about Campvention activities and programs, (2) a combined $1 money-back offer promotion and refund application form, (3) a form to be used to apply for refunds, and (4) Michigan Campground Directory. Because of the time constraints, no nonresponse follow-up was attempted for the pretrip questionnaire. One week after the Campvention (July 21), the same 1,575 parties who had received pretrip questionnaires were sent the four-page posttrip questionnaire and postage-paid 37 return envelopes by Michigan State University (see Appendix B). Instructions on the top of the questionnaire stressed that it was important that the same person who completed the pretrip questionnaire also complete this (the posttrip) questionnaire. Since more than half of the posttrip questionnaires had not been returned by August 15, follow-up letters were mailed on August 16 to the 350 persons/parties who had returned pretrip questionnaires but had failed to respond to the posttrip questionnaires. The primary reason for focusing on obtaining the posttrip questionnaires from respondents to the pretrip questionnaires was to acquire complete (pre and posttrip) information from as many respondents as possible. The budget for this study did not permit a comprehensive follow-up of all nonrespondents. In order to assess and adjust for nonresponse bias, on September 8, a reduced version of the posttrip questionnaire with a letter requesting cooperation was mailed to a random sample of 100 (20.8%) of 481 parties who had failed to return both pretrip and posttrip questionnaires. Response Rates The cut-off date for receiving the pretrip question- naires was July 7, one day prior to the start of the Campvention. A total of 794 (50.4%) were returned, of which 778 were usable (Table 1). 38 The response rate was somewhat higher (54.6%) for the posttrip questionnaire. This was due in part to the fact that reminders were sent to late respondents (see survey administration). Of the 860 posttrip questionnaires returned by the cut-off date (September 30), 847 were complete enough to be used in the analysis. Approximately 35% (560) of the 1,575 persons/parties comprising the sample completed and returned both the pretrip and posttrip questionnaires. Table 1. Response Rates to the Pretrip, Posttrip, and Non- Response Questionnaires. Sample Questionnaire Size Returns Response Rate a b Pretrip 1,575 794 (778) 50.4% (49.4%) Posttrip 1,575 860 (847) 54.6% (53.8%) Nonresponse 100 50 50% aNumber in parentheses indicates the number of usable questionnaires. 5Number in parentheses indicates response rate of usable questionnaires. Data Preparation and File Management All questionnaires were returned directly to Michigan State University. When received, the questionnaires were checked for completeness and edited in preparation for data entry. The SPSS PC+ Data Entry program made it possible to 39 enter the questionnaire data directly into the micro-- computer without precoding. The Data Entry program significantly reduced the number of data entry errors by controlling for out-of—range entries. It also facilitated file development (e.g., SPSS system file) and data analysis. Three different data files were developed and used for different analyses. Data from the pretrip and posttrip questionnaires formed two of the files. The matched pretrip and posttrip data from respondents who completed both questionnaires comprised the third file. A series of frequency distribution were run to identify possible data entry problems. Potential errors were checked against the original questionnaires and data entry problems were corrected. Assessment of Nonresponse Bias Fifty of the 100 nonrespondents returned the "nonresponse bias" questionnaire they were sent (Appendix D). Possible nonresponse bias was assessed by comparing characteristics of nonrespondents to both questionnaires with those of respondents who returned both questionnaires. Table 2 shows that there is little difference between respondents and nonrespondents on: their rating of the Campvention, size of Campvention party, length (number of nights) of their Campvention trips, likelihood of camping again in Michigan, work status, marital status, and presence 40 of children living at home. However, nonrespondents were less likely to have attended the Campvention. Of the non- respondents, 14% did not attend the Campvention, compared to less than 5% who responded to both of the questionnaires. This, in part, explains the discrepancy between the number of parties preregistered and the number of parties which actually attended the Campvention. However, this did not require a nonresponse bias correction since the actual attendance numbers were used as the expansion numbers (see Mahoney et al., 1989). Fewer nonrespondents were aware of (66%) or intended to take advantage (6.1%) of the $1 money-back offer. One reason for the difference in the intent to utilize the offer between respondents and nonrespondents was that the nonresponse questionnaires were mailed much later than the posttrip questionnaires, and much closer to the deadline for applying for a refund. By this time, most persons had, or had not, applied for a refund. Whereas, many persons who indicated they would seek a refund on the posttrip questionnaire evidently failed to follow through on their plans. This, in part, explains the difference between the percentage (14.8%) of those who indicated in the posttrip questionnaire that they intended to seek a refund and the percentage (6.3%) of Campvention attendees who actually applied for the refund. 41 Table 2. Comparisons of Characteristics of Respondents and Nonrespondents. Characteristics/Behavior Respondents Nonrespondents Participated in Campvention 95.3% 86.0% Average Campvention Party Size 2.6 persons 2.8 Rating of Campvention Facilitiesa 3.4 3.3 Rating of Campventionb 3.5 3.4 Total Campvention Trip Nights 12.9 nights 12.9 Nights at Campvention Site 5.8 nights 5.8 Additional Nights Camped in Michigan 4.5 nights 4.3 Aware of the $1 Offer 84.4% 66.0% Likely Would Utilize the Refund Offer 14.8% 6.1% Retired 60.7% 52.0% Married 94.6% 93.3% Have Children Living at Home 28.9% 24.4% 3On a 5 point scale (1=excellent, 5=poor). EOn a 5 point scale (1=much better, S=much worse). 42 Methods Used to Evaluate the Effect of the Refund Offer Two different potential effects of the refund offer were measured and evaluated: (1) refund effect and (2) redemption effect.4 The refund effect was measured by the proportion of nights camped in "refund eligible" campgrounds (state parks and/or commercial campgrounds) to the total number of nights camped in Michigan before and/or after the Campvention. If the refund offer had effectively induced campers to decide to stay more nights in state parks or commercial (MAPCO) campgrounds, persons who had been aware of the offer would have camped a greater proportion of their nights in state parks and/or commercial campgrounds (refund effect). Conversely, it would be expected that those who had not been aware of the offer would have camped fewer of their before and after Campvention nights in state parks and/or commercial campgrounds. The total number of camping nights in state parks and/or commercial campgrounds was also used to assess the refund effect. It was hypothesized that persons who were aware of the offer camped more nights in state parks and/or commercial campgrounds. 1 ‘2 It does not indicate actual redemption. It represents the intent to seek a refund. 43 Tests were also conducted to determine whether persons who indicated their intentions to seek a refund camped a greater proportion of their before and after Campvention nights in "refund eligible" campgrounds (redemption effect). It was hypothesized that persons who intended to seek a refund camped a greater proportion of nights in state parks and/or commercial campgrounds. The existence of refund effect means that the refund offer actually induced additional camping in these campgrounds. However, if there is a redemption effect but no refund effect, it means that persons who camped more nights in these campgrounds were more likely to seek a refund, but the offer did not influence their selection of "refund eligible" campgrounds. Table 3 shows the different proportions which were used to test for the refund effect. If A/(A + B) is Table 3. Format for Calculating the Refund Effect. Type of Campground Aware Not Aware Number of Nights Refund Eligible Campgrounds (State Parks, Commercial) A C Other Michigan Campgrounds (County/Local, State Forests, National Parks or National Forests) B D Total Nights Camped in MI A + B C + D 44 significantly greater than C/(C + D), a refund effect exists. The same formulation was used to test for the redemption effect. Statistical Methods As previously stated, the study had four objectives, with each requiring different statistical procedures. The first objective was to provide a descriptive profile of persons/parties who intended to take advantage of the $1 refund offer. Percentages and means were used to develop a general descriptive profile--socioeconomic characteristics, Campvention trip characteristics, camping behavior, camping affiliation, and experience with coupons or refund offers-- of persons/parties who attended the Campvention. The differences between those who indicated on the posttrip questionnaire that they did and did not plan to seek a refund was examined using t-tests. ANOVA tests were conducted to discover whether the percentage of respondents who intended to take advantage of the refund offer was different by income, education, and the length of camping nights in Michigan. The second objective was to assess the effects of the $1 refund offer on the amount of pre- and post-Campvention camping in Michigan. T-tests were used to determine differences in the number of camping nights in Michigan before and/or after the Campvention between persons who were 45 and were not aware of the offer, and between persons who intended and did not intend to seek a refund. The third objective was to test for relationships among income, education, camping affiliation, and intention to utilize the $1 refund offer. A log-linear analysis was used to discover the relationships among these multi-dimensional cross-classified data. A more in-depth discussion of the log-linear procedure is included in Chapter V. Finally, the fourth objective was to estimate the probability of utilizing the refund offer by different segments (e.g., income, education, and camping affiliation) of Campvention attendees. The percentage of persons who intended to utilize the refund offer was calculated for each segment. Findings relevant to Objective 1 and 2 are presented in Chapter IV. Chapter V exclusively deals with Objective 3 and 4. CHAPTER IV FINDINGS AND RESULTS This chapter presents (1) the characteristics of persons/parties who attended the Campvention, (2) a comparison of persons/parties who intended and who did not intend to take advantage of the $1 refund offer, and (3) results of tests of hypotheses regarding the effects of $1 refund offer on pre- and post-Campvention camping in Michigan. Additional information on the characteristics and behavior of Campvention attendees is presented in Mahoney et al. (1989). Characteristics of Persons/Parties Attending the Campvention This section describes the characteristics of persons who completed and returned both the pretrip and posttrip questionnaires. This includes: (1) socioeconomic characteristics, (2) camping behavior, (3) Campvention trip characteristics, (4) experience with coupons/refund offers, and (5) intent to utilize $1 camping refund offer. Only 46 47 persons/parties who completed both questionnaires were used for data analysis.5 Socioeconomic Characteristics The socioeconomic characteristics of Campvention attendees are shown in Table 4. Only about 20% of the attendees were Michigan residents. Of non-Michigan residents attending the Campvention, 37% had never camped in Michigan prior to the Campvention. The average age was 59. Approximately 70% of the attendees were 55 or older and the majority (60.7%) were retired. This was somewhat older than the average age of campers surveyed in other Michigan camping studies (Stynes & Mahoney, 1986; Fridgen et al., 1986). Almost all (95.6%) were married. Fewer than 30% had children living at home with the respondent, which was not surprising given the mean age. About 50% of the respondents had an annual household income between $20,000 and $39,999 in 1987. Fewer than half (43.3%) attended some colleges or had undergraduate or advanced degrees. The majority (57.3%) of respondents to the questionnaires were male. 5 Some questions about demographic characteristics (e.g., gender, marital status) and camping behavior were included in the pretrip questionnaire, whereas income, education, trip characteristics, experience with coupons/refund offers, and intent to utilize $1 refund offer were included in the posttrip questionnaire. To describe these characteristics of the sample and to meet one of overall study objectives (comparison of pretrip and posttrip perceptions of Michigan campgrounds), the use of these samples was indispensable. 48 Table 4. Socioeconomic Characteristics of Respondents. Characteristics Percent Michigan Resident 19.6 Nonresidents Camping for the First-time in Michigan 37 Average Age 59 years Less Than 55 29.8 55 or older 70.2 Married 95.6 Retired 60.7 Have Children Living at Home With Them 28.9 Income Under $20,000 22.8 $20,000-$39,999 49.7 $40,000 or more 27.5 Education High School Degree or Less 56.7 Some College or More 43.3 Male 57.33 a This indicates that a greater proportion of male members of camping families/couples completed the questionnaire. In part, this was due to the fact that male members were more likely to be the registrants of record. The questionnaires were mailed to the registrants of record. 49 ngping,Beh§yior Campers who attended the Campvention are, on the average, very active, high-volume campers. As shown in Table 5, they camped 52 nights during the previous year (1987). They are also very active off-season campers with a high percentage of the attendees camping before Memorial Day (87.9%) or after Labor Day (94.2%). About 86% camp both before Memorial Day and after Labor Day. More than half (56.1%) have no preference for either public or private (commercial) campgrounds. A quarter (24.6%) have a preference for private (commercial) campgrounds. Table 5. Camping Behavior of Campvention Attendees. Nights Percent a Camping Nights in 1987 52 nights Off-Season Camping Camp Before Memorial Day 87.9 Camp After Labor Day 94.2 Camp Both Before Memorial Day and After Labor Day 86.0 Campground Preference Public Campgrounds 19.2 Private (Commercial) 24.6 No Preference 56.1 3Does not include persons who lived in their camping equipment. Approximately 1% of respondents reported camping 365 days in 1987. They were excluded from the calculation. 50 The attendeeswere asked a series of questions to ascertain their degree of camping involvement. This information was also considered important and was used in assessing their predisposition to utilize camping-related sales promotions. Approximately three quarters (74.8%) of the respondents subscribed to a camping-related magazine(s) (Table 6). Almost half (48.4%) belonged to camping organizations in addition to NCHA. Around 70% attended at least one camping or outdoor show the previous year (1987). Table 6. Camping Affiliation: Subscription to Camping Magazines, Membership in Camping Clubs, and Attendance at Camping/Outdoor Shows. Camping Affiliation Percent Subscribe to Camping Magazines 74.8 Belong to Camping Organizations 48.4 Attend Camping/Outdoor Shows over the Previous Year (1987) 69.1 Characteristics of Campvention Trip Party size, and Campvention trip length and behavior are presented in Table 7. The average Campvention party consisted of 2.5 persons. The average party spent 14 (13.7) nights on their Campvention trip. Approximately half (52%) of the parties camped additional nights before and/or after the Campvention in Michigan. The average party spent nine (9.1) nights in Michigan on their Campvention trips. 51 Table 7. The Characteristics of Michigan Campvention Trips and Sources of Camping Information. Trip Characteristics Percent/Nights Camped in Michigan Before and/or After the Campvention 52.3% Party Size 2.5 persons 1 1.5% 2 70.0% 3 11.4% 4 10.8% 5 + 6.3% Total Trip Nights 13.7 nights Nights Spent in Michigan 9.1 nights At Campvention Site 5.9 nights At Campgrounds in Michigan 2.7 nightsa Hotels/Motels, Friends/Relatives 5 nights Pre- and Post-Campvention Camping Nights in Michigan 4.8 nightsh State Parks Campgrounds 1.0 nights Commercial (MAPCO) Campgrounds 3.2 nights Local/County, State Forests National Parks/Forests .6 nights Sources of Information Used to Select Michigan Campgroundsc Michigan Campground Directoryd 34.2% Woodalls Camping Directory 19.9% Recommend. from Friends/Campers 19.6% Trailer Life 17.1% Kampground of America (KOA) 13.9% Camping Experience in MI 12.1% aThe average camping nights of parties who attended the Campvention (Parties who did not camp before and/or after the Campvention are included). bThe average camping nights of parties who camped in Michigan before and/or after the Campvention (Parties who did not camp additional nights are excluded). cRespondents were allowed to indicate more than one source of information, so percent does not add to 100%. dAll preregistrants were direct mailed a copy of the Michigan Campground Directory. 52 Of these, six (5.9) nights were spent at the Campvention site. On the average, Campvention attendees camped three (2.7) additional nights in Michigan. Some parties spent a portion of their Michigan nights at hotels/motels or visiting friends/relatives. Campvention attendees who camped additional nights in Michigan camped an average of five (4.8) nights. On the average, three of these camping nights were spent in commercial campgrounds and one in state parks campgrounds. Campvention attendees who camped in Michigan before and/or after the Campvention relied on a variety of different informational sources to select the campgrounds where they stayed. The Michigan Campground Directory (which was direct-mailed to all preregistrants) was used by the greatest number of parties (34.2%), followed by Woodalls gamping Directory (19.9%) and recommendations from friends, relatives, or campers (19.6%). Trailer Life (17.1%) and figmpgrounds of America (13.9%) were the next frequently used informational sources. Experience with CouponsLRefund Offers Whether or not Campvention attendees have utilized coupons/refund offers for camping, other recreational opportunities, or nonrecreational products was presented in Table 8. Fewer than half (42.8%) of the attendees have utilized camping-related coupons or refund offers (Table 8). Most (67.7%) of these only occasionally utilized camping- 53 related coupons/refund offers, and only 13.6% often utilized coupons or refund offers related to camping. Nearly 25% have utilized coupons/refund offers to purchase other recreational opportunities, but 62% only occasionally utilized these offers. Almost all (90.9%) have used coupons or sought refunds in the purchase of other nonrecreational products. The majority (56%) indicated that they often utilized coupons/refund offers to purchase other nonrecreational products. Table 8. Experience with Coupons or Refund Offers for Camping, Other Recreational Opportunities, and Nonrecreational Products. Experience Utilizing Coupons/Refund Offers Percent For Camping 42.8 Seldom Utilize 18.7 Occasionally Utilize 67.7 Often Utilize 13.6 For Other Recreational Opportunities 24.9 Seldom Utilize 22.6 Occasionally Utilize 62.3 Often Utilize 15.1 For Nonrecreational Products 90.9 Seldom Utilize 5.4 Occasionally Utilize 38.8 Often Utilize 55.8 Intent to Utilize the $1 Refund Offer Although 90% of parties indicated that they were aware of the $1 Campvention money-back offer, only 14.8% indicated 54 on the posttrip questionnaire that they were planning to apply for a refund (Table 9). As previously stated, correction for nonresponse bias--a higher percent of nonrespondents did not intend to seek a refund-~reduced the estimate of the percent of Campvention attendees who intended to apply for a refund to 11.5%. In actuality, only 6.3% applied. Parties who intended to apply for a refund for one to three nights ($1 - $3) comprised 45% of the parties who indicated their intentions to apply for a refund. The Table 9. Awareness of and Intent to Utilize the $1 Refund Offer. Awareness/Intent Percent Were Aware of the Offer 89.6 Planned to Apply for a Refund 14.8(11.5)a Number of Nights Intended to Apply for a Refund 1 - 3 45.4 4 - 5 25.3 6 or more 29.3 Mean 5.1 nights The $1 Offer Influenced Them to Camp More Nights in State Parks and/or Commercial (MAPCO) Campgrounds 15.3 3Estimate of the percent of respondents who planned to apply for a refund after adjusting for nonresponse bias. average refund applied for was 5.1 nights ($5.10). Only 15.3% of those who intended to apply for a refund indicated 55 that the refund offer influenced them to camp more nights in state parks and/or commercial (MAPCO) campgrounds. As was discussed in the Chapter III, Campvention attendees who indicated that they would not likely seek a refund (of $1 per night) were asked whether or not they would likely have taken advantage of a higher refund offer ranging from $2 to $8. As was expected, the percentage of those who would have utilized a higher refund offer increased as the value of refund offer increased (Table 10). An increase in the refund value from $1 to $2 doubles the percent of persons who would have utilized a refund offer. Only 27.7% of respondents to $2 offer and 48.3% of respondents to $8 offer indicated they would have utilized the offer. On the average, 36% would have taken advantage of a higher refund offer ranging from $2 to $8. Comparisons of Parties Who Indicated that They Would, and Would Not Apply for a Refund This section describes and compares both parties who indicated on the posttrip questionnaire their intentions to apply for a refund and those who did not intend to apply. Again, it is important to emphasize that some of the persons who indicated they would apply for a refund did not actually follow through and apply. Since it was not possible to identify persons who did not actually apply for a refund, 56 Table 10. The Percent of Campvention Attendees Who Indicated that They Would Likely Have Utilized a Higher Refund Offer. Refund Percent That Would Have Amount .na Utilized a Refund Percent $1 527 14.8 $2 70 27.7 $3 65 29.3 $4 78 34.0 $5 77 35.3 $6 88 38.2 $7 78 42.3 $8 90 48.3 aThe number of persons who returned a posttrip questionnaire which asked whether they would have likely utilized a refund offer of this amount. 57 the intent to seek a refund was used instead of actual redemption. Persons who did not intend to utilize the refund offer included parties who were and were not aware of the offer. Parties who were not aware of the offer comprised only 10% of those who did not intend to seek a refund and statistical analyses showed that they did not differ significantly from those who were aware of the offer. One of this study's objectives was to provide a marketing- relevant descriptive profile of campers who intended to take advantage of the Campvention money-back offer. A null hypothesis was formulated to test for differences between those who intended and who did not intend to utilize the refund offer. Null Hypothesis 1: Parties who intended to utilize the $1 refund offer do not differ from those who did not intend to seek a refund in terms of their socioeconomic characteristics, camping behavior, camping affiliation, or past experience with coupons or refund offers. The results of t-tests for differences are reported in Table 11. Statistically significant differences were found between those who intended and did not intend to seek a refund with respect to the following: (1) state of residence, (2) length of Campvention trip, (3) nights camped in Michigan before and/or after the Campvention, (4) the use of the Michigan Campground 58 Directory to decide on where to camp in Michigan, (5) preference for public campgrounds, (6) subscription to camping magazines, and (7) membership in camping organizations. A higher percentage of out-of-state residents intended to seek a refund. In part, this is due to the fact that a higher proportion of nonresidents camped in Michigan before and/or after the Campvention and on the average they camped more nights in Michigan than Michigan residents. Persons who indicated they intended to seek a refund spent significantly more nights on their Campvention trips and spent more nights camping (6.2 nights) in Michigan before and/or after the Campvention than persons who did not intend to seek a refund. As hypothesized, persons who intended to seek a refund were more likely to subscribe to camping magazines (84%) and belong to camping organizations other than the NCHA (60%) than persons who did not. In other words, they were more actively involved campers. Persons who intended to apply for a refund were also more likely to have utilized the Michigan Campground Directory_when selecting Michigan campgrounds. There were no significant differences between those who did and did not intend to seek a refund with respect to their gender, work status (e.g., retired), family status (e.g., married, presence of children living at home), 59 Table 11. A Comparison of Those Who Intended and Did Not Intend to Utilize the $1 Refund Offer Did Not Intended Intend T-test Categories to Use to Use Probability Socioeconomic Characteristics Male 59% 58% .797 Age (mean) 59 years 59 years .699 Retired 56% 61% .403 Married 97% 94% .182 Out-of—State Resident 91% 78% .001 ** Have Children Living at Home 30% 29% .887 Education High School or Less 53% 57% .482 Some College or More 47% 43% .487 Income Under $20,000 23% 22% .837 $20,000-$39,999 46% 50% .539 $40,000 or More 31% 28% .621 Campvention Trip Nights Total Trip Length 17 nights 13 nights .001** Nights At the Campvention Site 5.7 6.0 .165 Nights Camped in Michigan 6.2 4.1 .003** Used MI Campground Directory to Plan Campvention Camping 48% 10% .001** (continued on next page) 60 Table 11 (cont'd.). Did Not Intended Intend T-test Categories to Use to Use Probability Camping Behavior Off-season Camping 87% 86% .966 Campground Preference Public 31% 16% .012* Private 22% 25% .551 Average Number of Nights Camped a Year 57 nights 52 nights .354 Camping Affiliation Subscribe to Camping Magazines 84% 73% .021* Belong to Other Camping Clubs 60% 46% .017* Attend Camping/Outdoor Shows 78% 68% .067 Experience with Coupons/Refund Offers Camping 37% 44% .244 Other Recreational Opportunities 29% 24% .325 Nonrecreational Products 94% 91% .384 * indicate a significance at p_< .05. ** indicates a significance at p < .01. 61 education, income, whether or not they camp in the off- season (both before Memorial Day and after Labor Day), campground preference, number of nights they camp a year, whether or not they attend camping/outdoor shows, and past experience with coupons/refund offers for camping, other recreational activities, and nonrecreational products. In particular, ANOVA and T-tests were performed to determine specifically whether or not intent to utilize the refund offer varied with income, education, or the number of nights camped in Michigan before and/or after the Campvention. Intent to utilize the $1 refund offer (which was expressed as a percentage) was not significantly different among persons with different income or education levels (Table 12). However, a statistically significant difference was found among persons reporting different number of nights camped in Michigan (Table 13). The t-test results indicated that Campvention attendees who camped eight or more nights in Michigan before and/or after the Campvention were more likely to take advantage of the offer than people who camped fewer nights. Although persons who camped 4 - 7 nights were more likely to utilize the refund offer than those who camped fewer than four nights, the difference was not statistically significant. 62 Table 12. ANOVA and T-Tests for the Differences in Intention to Utilize the Refund Offer by Income and Education Levels. . F ANOVA na Percent“ Probability Income: .819 Below $20,000 108 15 $20,000-$39,999 241 14 $40,000 or More 134 16 T-TEST Percent? T-Value D.F. Probability Education Level: High School Degree or Less 14 - .70 511 .482 College or More 16 aThe number of persons in the income category. bIndicates the percentage of parties who intended to utilize the $1 refund offer. 63 Table 13. ANOVA and T-Tests for the Differences in Intention to Utilize the Refund Offer by the Number of Nights Camped in Michigan Before and/or After the Campvention. ANOVA na Percentb Probability Number of Nights Camped in Michigan .009** 1 - 3 136 21 4 - 7 84 31 8 or More 41 51 T-TEST Percenth T-value D.F. Probability Number of Nights Camped in Michigan (1 - 3) x (4 - 7) 21 x 31 -1.60 218 .110 (4 - 7) x (8 + ) 31 x 51 -2.22 123 .028* aThe number of persons in the category of the number of nights camped in Michigan before and/or after the Campvention. 23Indicates the percent of parties who intended to utilize the refund offer. * indicates a significance at p_< .05. ** indicates a significance at p < .01. 64 Tests of the Effects of the $1 Refund Offer A principal objective of this study was to evaluate the effects of the $1 refund offer on before and/or after Campvention camping in Michigan. Both the proportion and the number of nights camped in Michigan state parks and/or commercial campgrounds were used to assess the refund and redemption effects. The proportion, the number of nights camped in state parks and/or commercial (MAPCO) campgrounds to the total number of nights camped in Michigan campgrounds (e.g., state parks, commercial, county, state forests, national parks, and national forests), was analyzed to determine whether the $1 money-back offer influenced pre- and post-Campvention camping behavior in Michigan. For obvious reasons, only parties who actually camped in Michigan before and/or after the Campvention were included in the analysis. Hypothesis 2: The $1 money-back offer influenced campers to stay more nights in state parks and/or commercial (MAPCO) campgrounds was formulated to test both the refund and redemption effect of the refund offer. It has two subhypotheses. Subhypothesis (2a) tests the refund effect and subhypothesis (2b) tests the redemption effect. Null hypothesis (2a): Parties who were aware of the money- back offer did not camp (1) a greater proportion of their nights and (2) a greater number of nights in Michigan state parks and/or commercial (MAPCO) campgrounds than those who were not aware of the offer. 65 Null hypothesis (2b): Parties who intended to take advantage g; the money-back offer did not camp (1) a greater proportion of their nights and (2) a greater number of nights in state parks and/or commercial (MAPCO) campgrounds than those who did not intend to utilize the offer. The independent variable for subhypothesis (2a) is the awareness of the refund offer. Whether or not the attendees were either aware or unaware of the refund offer was determined on the posttrip questionnaire. The independent variable for subhypothesis (2b) is intention to utilize the refund offer. Again, respondents were asked on the posttrip questionnaire whether or not they intended to seek a refund. The dependent variables for both subhypotheses are (1) the proportion of nights camped in state parks and/or commercial (MAPCO) campgrounds and (2) the number of camping nights in these campgrounds before and/or after the Campvention. The proportion of nights camped in state parks or commercial campgrounds is obtained by dividing the number of nights camped in "refund eligible" campgrounds by the total camping nights in Michigan. One-tailed t-tests were used to test subhypothesis. First, the assumption of equal variance was tested for each subhypothesis. The F value was explored to test the homogeneity of variance. Table 14 reports the results of 66 Table 14. Tests of Homogeneity of Variance and Decisions on the Use of Different Types of T-test. Decision on Groups F Value Probability T-test Average Proportiona Aware vs Not Aware of the Refund Offer 1.39 .323 Pooled-Variance Intended vs Did Not Intend to Seek a Refund 3.07 .001* Separate-Var. Average Number of Nights“ Aware vs Not Aware of the Refund Offer 2.30 .013* Separate-Var. Intended vs Did Not Intend to Seek a Refund 2.47 .001* Separate-Var. aRatio of the number of nights camped in "refund eligible" campgrounds to the total number of nights camped in Michigan before and/or after the Campvention. hNumber of nights camped in "refund eligible" (state parks and/or commercial) campgrounds. * indicates a significance at p.< .05. 67 homogeneity of variance tests and the decision on whether to use the pooled variance or the separate variance t-test.6 T_est ReJul L Null hypothesis (2a) was not rejected at 23.05 level of significance (Table 15). Those who were aware of the offer did not camp a significantly greater proportion (92%) of nights in state parks and/or commercial (MAPCO) campgrounds. In fact, those who were not aware of the offer camped a greater proportion (95%) of their Michigan camping nights in "refund eligible" campgrounds than those who were aware of the offer. Although those who were aware of the offer camped more nights (4.2) in state parks or commercial (MAPCO) than those who were not aware of it (3.5 nights), the difference was not statistically significant. Therefore, the awareness of the offer did not appear to significantly increase the proportion or the number of pre- and post-Campvention nights camped in Michigan state parks and/or commercial campgrounds. However, this may be in part due to the fact that all participants received the Michigan Campground Directory which gives most attention to 5 Norusis (1986) notes that "if the pooled-variance t test is used when the population variances are not equal, the probability level associated with the statistic may be in error. The amount of error depends on the inequality of the sample sizes and of the variances. However, using the separate-variance t value when the population variances are equal will usually result in an observed significance level somewhat larger than it should be" (p. 122). 68 Table 15. T-Tests of the Proportion and Number of Camping Nights between Those Who Were Aware and Not Aware, and Those Who Intended and Did Not Intend to Seek a Refund. Proportion/Nights Groupsa T-Value Probability Aware.s Not-Awareb Average Proportion .92 .95 - .53 .597 Average Nights 4.2 3.5 1.19 .241 3 Did Not Intended“ Intend “ Average Proportion .96 .91 2.06 .040 * Average Nights 6.0 3.4 4.07 .001 ** aParties who camped in Michigan before and/or after the Campvention during the Michigan Campvention trip. hn = 238, aware; n = 27, not aware; n = 78, intended; n = 187, did not intend. * indicates a significance at p,< .05. ** indicates a significance at p < .01. 69 commercial (MAPCO) campgrounds. Also commercial campgrounds and state parks provide the majority of Michigan's deve10ped (e.g., electricity, water) campsites which are desired by recreation vehicle campers.7 Null hypothesis (2b) was rejected. Parties who indicated their intentions to seek a refund spent a greater proportion (96%) of their pre- and post-Campvention camping nights, and a greater number of nights (6.0) in state parks and/or commercial campgrounds than parties who did not intend to seek a refund (91%, 3.4 nights). The test of subhypothesis (2a) indicates that $1 refund offer was not effective in inducing Campvention attendees to camp more nights in "refund eligible" campgrounds. A significant refund effect was not found. However, persons who intended to seek a refund camped significantly more nights in state parks and/or commercial campgrounds. Thus, there was a significant redemption effect. Therefore, the offer did not induce more camping in state parks or commercial (MAPCO) campgrounds, but persons who camped more nights in 'refund eligible" campgrounds were more likely to seek a refund. 7 The vast majority of full service campgrounds are listed in the Directory and most of these were "refund eligible". 70 Summary In this chapter, (1) a profile of persons/parties attending the 1988 Michigan Campvention, (2) comparisons of persons who intended and who did not intend to seek a refund, and (3) test results regarding the effect of the $1 refund offer on camping in Michigan state parks and commercial campgrounds were presented. Special emphasis was directed at laying a foundation for the log-linear analysis which is presented in the next chapter. Out-of-Michigan residents were more likely to utilize the refund offer. A higher percentage of persons who intended to utilize the $1 refund offer subscribed to camping magazines and belonged to camping organizations other than the NCHA. Intent to utilize the refund offer increased as the amount of pre- and post-Campvention camping in Michigan increased. Significantly, more of the persons who camped eight or more nights in Michigan indicated they would take advantage of the offer. No significant difference in the intent to utilize the offer was found across annual household income and education levels. The $1 refund offer does not appear to have had influenced persons/parties to camp more nights in state parks and/or commercial campgrounds (MAPCO). However, the analysis revealed a potential redemption effect. CHAPTER V LOG-LINEAR ANALYSIS AND PROBABILITY 0F UTILIZING THE MONEY-BACK OFFER Log-linear analysis was used to examine the relationships among intent to utilize the $1 refund offer, income, education, and camping affiliation. This chapter begins with a brief but necessary description of log-linear analysis, including its advantages and disadvantages. Log-Linear Analysis Log-linear models describe the structure of multi- dimensional contingency tables (Fienberg, 1977). The natural logarithms of expected cell frequencie584are predicted by log-linear models, using linear equations as in regression. This section describes advantages and disadvantages of log-linear analysis, applications of log-linear analysis, 8 Reynolds (1977) notes that whether a model is built for probabilities, expected frequencies, or the logarithm of expected cell frequencies is mainly a matter of taste and convenience because a model for one can easily be translated into models for the others. 71 72 comparisons of log-linear analysis with alternative statistical methods, types of log-linear models, parameters in log-linear models, estimation of parameters, and model building and selection. Advantages and Disadvantages of Log-linear Analysis As Fienberg notes, most researchers analyze multi- dimensional, cross-classified data using a series of chi- square tests. But chi-square tests of independence have limitations when analyzing multi-dimensional cross- classified data. Fienberg points out that chi-square tests do not allow for simultaneous examination of relationships among three or more variables. Therefore, it ignores the possibility of three-factor and higher-order interactions. Log-linear analysis overcomes this shortcoming. Log-linear models provide information about which variables are related and how they are related. Moreover, the absolute value (or coefficient) of interaction effect indicates the strength of the relationships. The disadvantage of log-linear analysis is that the numerical values (coefficients) of effects are difficult to interpret because there is no underlying physical scale (Reynolds, 1977). Furthermore, when each variable has several categories, a large number of effect coefficients should be calculated and interpreted. Finally, log-linear analysis deals with nominal or ordinal data. However, when 73 interval or ratio data are used, information about the data may be lost in the process of categorizing/ordering the data. Applicationsgof Log-linear Analysis The use of log-linear analysis has been increasing in social science research, marketing, and advertising. In sociology, log-linear models have been used to discover relationships among job attitude, education, time, and geographical region (Goodman, 1972); occupational relationships between father's occupation and son's occupation (Duncan, 1979); interaction between husband's behavior and wife's behavior (Allison & Liker, 1982); dyadic interaction between parents' political preferences and children's preferences (Feick & Novak, 1985); social relationship between sex (boy, girl) and frequency of toy offering (Iacobucci & Wasserman, 1987); and prediction of occupational aspirations as a function of socioeconomic status, residence, gender, and IQ (Elliott, 1988). In the fields of advertising and marketing, log-linear models have been used to assess promotional efficiency. They have been used to develop a magazine exposure distribution (ED) model (Boyd, 1985; Danaher, 1988); and to predict the adoption of innovations (Green, Carmone, & Wachspress, 1977). Green, Carmone, and Wachspress also 74 suggest that log-linear models can be used for analyzing brand switching. Categorical and/or ordinal data (e.g., demographic characteristics, presence or absence of a disease, and treatment outcome) have been used in biomedical studies. Log-linear models have been frequently used for survival study in biomedical research (Bishop, 1969). Psychologists have utilized log-linear models to analyze psychological behavior using cross-classified data (e.g., agree/disagree or pass/fail) (Bonnet 8 Bentler, 1983; Kriska & Milligan, 1982). Comparisons of Log-linear Analysis with Alternative StatisticglgMethods Other statistics, such as dummy-variable regression, ANOVA, discriminant analysis, or logit analysis, are often used as alternatives to log-linear analysis. But in this study, all variables used for examining the relationships are categorical. When the criterion variable is binary, the application of dummy-variable regression or ANOVA has some limitations. These techniques are used with the assumption that population are normally distributed with constant variance. Either dummy-variable regression or ANOVA requires interval or ratio data for the dependent (criterion) variable. Categorical data can not satisfy this critical assumption. 75 Discriminant analysis is used to classify respondents into different categories of the dependent variable using several metrically-scaled, independent variables (Hair, Anderson, Tatham, 8 Grablowsky, 1979). Discriminant analysis investigates relationships between a criterion variable and some predictor variables as in multiple regression. It does not identify interdependencies among a number of independent variables. Linear discriminant functions are often utilized in the case of dichotomous variables (Gilbert, 1968; Moore, 1973). However, Dillon and Goldstein (1984) noted that the linear discriminant function is not appropriate when the independent variables are all binary or a mixture of continuous and discrete variables. Logit models are used to examine the relationships between a dichotomous dependent variable and one or more 9 A log-linear model focuses on the independent variables. joint probabilities of a set of qualitative variables, whereas a logit model examines the conditional probability of a single qualitative variable, given a set of other variables (Aldrich & Nelson, 1984). Furthermore, interaction terms among independent variables in a log- linear model are not easily included in a logit model. Another disadvantage of a logit model is loss of information 9 In this respect, a logit model is closely analogous to ordinary regression. 76 on the relationships among the independent variables (Fienberg, 1977). In this study, one objective was to examine relationships among various categorical data (e.g. intent to utilize the $1 refund offer, income, education, and camping affiliation). The multivariate normality assumption that is necessary for regression, ANOVA, and discriminant analysis, is not required for log-linear analysis. Logit models are not appropriate for analyzing the relationships of more than three variables. Therefore, log-linear analysis was determined to be the most appropriate method to analyze the relationships among multi-dimensional categorical data. Types of Logrlinear Models Log-linear models are categorized as saturated vs. unsaturated, hierarchical vs. nonhierarchical, and independence models. Saturated vs Unsaturated Models: A saturated log- linear model contains all possible parameters (or effects).10 It has as many independent parameters as there are cells in the table. Saturated models always fit the observed data, but they are not parsimonious and have little value in and of themselves. They serve as an excellent H Parameters indicate effects which represent the increments or decrements from the base value (u) for particular combinations of values of variables. 77 starting point (model) for exploring other, more parsimonious, log-linear models that fit the data. When an independent parameter equals zero (i.e., no effect), a log-linear model does not contain that parameter. A log-linear model is unsaturated when the model has fewer parameters than the number of cells in the contingency table (Reynolds, 1977). Hierarchical vs Nonhierarchical Models: The hierarchical principle is that the inclusion of a higher order interaction effect requires the inclusion of all lower order interaction effects as well as main effects. For example, if an ABC interaction effect is in the model, AB, BC, and AC interaction effects and A, B, and C effects should be included in the model (a generating class [ABC] includes [A], [B], [C], [AB], [BC], and [AC]). Alternatively, if A is not included in the model, then neither are AB, AC, and ABC interaction terms. Nonhierarchical models do not follow this rule. Most, but not all (Knoke & Burke, 1980) log-linear models are hierarchical. The restriction to hierarchical models is a characteristic of the iterative proportional fitting algorithm for estimating the expected cell frequencies in the log-linear models. The primary reason for avoiding nonhierarchical models is that hierarchical 78 models are easier to interpret than are nonhierarchical models (Fienberg, 1977).11 Independence Modela: The model representing independence among variables contains main effects as well as a grand mean. In independence models, there is no interaction among the variables. Parameters in Log-linear Models To help understand parameters in a log-linear model, a two-factor interaction model is presented below. The log- linear model is L" = LD(F]') = U + 113(1) + uh”) + u 1] 1 35(13} where, Ffi = the expected cell frequency for the (in)m ' cell of variable in the model (i: 1, ....... I, Ln (PM) = the natural logarithm of the expected cell frequency for the (i,j)t cell in the table. u = grand mean which is defined as the average of the logarithms of all expected cell frequencies. n ufiU = main effect of the i category of variable A. n main effect of the j category of variable B. “MD thuifi = interaction effect fqg the ith category of variable A and the j category of variable B. H Refer to Knoke and Burke (1980, p. 74) for the reason why hierarchical models make sense. 79 Therefore, u = 242 = 1 . IJ IJ 2 Li: Li+ L” u :\ = - u = '- a ‘. (’ 3 J J IJ 2 Li] L+j LH u ~ - - u = ' D} ' ' 1 I I IJ uab{ij) = L1: ' u ‘ “am ‘ “om Lg Lg L++ z LH ‘ ' ” J I IJ L++ is the sum of the logs of frequencies in the all cells. The "+" signs denote summation of frequencies over the appropriate subscript. “MM is the average log of the frequencies in the J categories of variable B at the ith level of variable A minus the grand mean u, while uNfi is the average log of the frequencies in the I categories of variable A at the jth level of variable B minus the grand mean. udfl or “NH reflects the extent to which the number of cases in the ith or j” category of variable A or B is different from the average across all categories of A or B. The larger the absolute value of the coefficient, the more the distribution of A or B deviates from the average across all categories of A or B. 80 The value for the interaction effect, utflfi)' is estimated by subtracting the grand mean and main effects of A and B from the log of the expected frequency in the (i,j)th cell. u w indicates that the marginal distribution auin of A depends on levels of B and vice versa. In other words, the distribution of A for a given level of B differs from the distribution of A across all levels of B. If there is no association between the two variables, the interaction parameter would be zero. Since udfi and “ND represent deviations from the grand mean of the logarithms of the all expected cell frequencies, = 0, also 2 u;:n = 0 and 2,u;:;\= 0. ~ 3~(-.I 1 a“(*ll V ‘ a The u terms (or effects)12 must sum to zero across the categories of a variable. For each variable, similar constraints are imposed on the interaction terms. Estimation of Parameters There are two basic approaches to estimating the parameters in a log-linear model: Maximum Likelihood Estimate (MLE) and Weighted Least Square (WLS). MLE is based on individual observations and is obtained by either the iterative proportional fitting algorithm or the Newton- M The u-terms are generally referred tou as effects. u is called a zero-order effect or grand mean, :) and u are first-order or main effects, and u j) is secoun -order éffect . . . an (i or first-order interaction. 81 Raphson algorithm. The iterative proportional fitting algorithm estimates the expected cell frequencies for a hierarchical model, and the Newton-Raphson algorithm yields a parameter vector.13 On the other hand, WLS is based on the group observation (Flath 8 Leonard, 1979; Malhotra, 1984). WLS is concerned with deriving parameter estimates that yield the smallest sum of squared errors in the fit between the model and data. A WLS estimation proceeds in a similar way as Ordinary Least Square (OLS) approach, except that it uses the weighted sum of squared errors (Flath 8 Leonard, 1979). Because MLE procedure produces consistent and efficient statistical estimates (Knoke 8 Burke, 1980), most studies (Benedetti 8 Brown, 1978; Bishop et al., 1977; Bonett 8 Bentler, 1983; Elliott, 1988; Fienberg, 1977; Green et al., 1977; Knoke 8 Burke, 1980; Norusis, 1986) employ MLE with an iterative proportional fitting algorithm Model Building,and Selection A general strategy for the analysis of cross-classified categorical data involves tests of several models, including not only an independence model, but also models containing various interaction terms among the variables. The process of model building begins with the selection of a base model, B Bishop, Fienberg, and Holland (1977, pp. 83-102) and Haberman (1978, pp. 64-69) explain the procedures of calculating these algorithms. 82 followed by addition of effects to the base model (forward procedure) or deletion of effects (backward procedure). Each model generates expected frequencies, and then the expected frequencies are compared with observed frequencies. Models where expected frequencies fit the observed frequencies are compared on different criteria--goodness-of— fit, parsimony--to identify the most appropriate model for the analysis. There are two methods to evaluate goodness-of-fit of a model, which indicates how well the model fits the observed data. One is to use either the likelihood-ratio chi-square or the Pearson chi-square statistic, and the other is to examine the standardized residuals (Green, 1988; Kriska 8 Milligan, 1982; Norusis, 1986). The likelihood ratio chi- square statistic (L2) is defined as L2 = 2 z fij 1n (fij/Fij) where, f” indicates the observed cell frequency and Ffi represents the expected cell frequency. The larger the L3 relative to the number of degrees of freedom, the more the cell expected frequencies depart from the actual cell frequencies. Degrees of freedom to be associated with various goodness-of-fit test statistics are determined by subtracting the number of independent parameters fitted for the model from the total number of cells in the table. d.f. = # cells - # parameters fitted. 83 The total number of independent parameters for a specific model is 1 (for the grand mean), plus the number of independent parameters pertaining to the u terms included in the model. For the I X J interaction model, the number of degrees of freedom for uflU’ “ND’ and u ) are I-l, J-1, aMij and (I-1)(J-1), respectively. For large sample sizes“ the likelihood-ratio chi- square and the Pearson chi-square statistics are usually very close (Norusis, 1986). However, the likelihood ratio chi-square is preferable to Pearson's chi-square because (1) the expected cell frequencies are estimated by maximum likelihood methods, and (2) I} can be partitioned for more powerful tests of conditional independence in multiway tables (Knoke 8 Burke, 1980). I} follows the chi-square distribution with degrees of freedom equal to the number of independent parameters which have no effect on the expected cell frequencies. The other criterion for evaluating log-linear models is parsimony (Benedetti 8 Brown, 1978; Knoke 8 Burke, 1980; Reynolds, 1977). A parsimonious model refers to a model that contains as few parameters as possible but still fits the data. H As a result of thumb, a large sample size refers to at least ten times the number of cells in the table (Fienberg, 1976, p. 37). 84 Examination of Relationships The second objective of this study was to discover the relationships among Campvention attendees' intentions to utilize the $1 refund offer, income, education, and camping affiliation. Hypothesisaand Variables Null hypothesis 3 was formulated to test the relationships as follows: Null hypothesis 3: Intent to utilize the $1 refund offer is not strongly related to levels of income, education, and camping affiliation. The variables, education, income, and camping affiliation were included in the log-linear analysis based on past studies and statistical tests. Income and education were included in the model even though univariate statistical analyses showed no significant relationship between intent to utilize the refund offer and either variable for two reasons. First, the literature review of sales promotion studies revealed that income and education were almost always used for describing the characteristics of coupon-prone consumers. Income and education were often found to be strongly (e.g. positively or negatively) related to coupon redemption rate for manufactured goods. People with higher income were more likely to utilize coupons than people with 85 lower income (Bawa 8 Shoemaker, 1987b; Blattberg et al., 1978; Teel et al., 1980). However, the coupon redemption rate was not always positively related to income and education (Narasimhan, 1984). Although no studies have examined whether or not income and education are strongly related to the use of recreation/tourism/ camping-related sales promotions, a decision was made to include these variables given the results of studies of non-recreation sales promotions. Second, the univariate statistical tests in Chapter IV did not examine the relationships between different levels of income/education and intent to utilize the refund offer. These relationships can be determined using log-linear analysis. The inclusion of camping affiliation as one of the log- linear variables was made on the basis of previous camping studies (Fridgen et al., 1986; Mahoney et al., 1989; Stynes 8 Mahoney, 1986). Also, subscription to camping related magazines was statistically related to intention to use the refund offer (see Chapter IV). There was no significant difference between those who intended and who did not intend to utilize the refund offer with respect to whether they attended camping or outdoor shows. However, since attending these shows is considered an information-seeking effort, this variable was included with subscriptions to camping magazines, as a measure of camping affiliation. It was 86 assumed that persons who were inclined to seek information about camping would be more likely to utilize coupons or refund offers because they would have more opportunities to find these offers than those who are less inclined to seek this information. Also, these information-seeking campers are generally more active campers. Camping affiliation-- magazine subscriptions and attendance at camping/outdoor shows also provides a means of reaching deal-prone campers. The final stage was to test association between stated intentions to utilize the refund offer and the three variables. The results are shown in Table 16. Table 16. Results of Chi-Square Tests of Association Between Intent to Utilize the $1 Refund Offer and Income, Education, and Camping Affiliation. Variables Chi-Square D.F. Significance Income ' .401 2 .818 Education .336 1 .562 Camping Affiliation .633 1 .012 * * indicates a significance at p_< .05. The chi-square tests showed that only camping affiliation was significantly related to intent to utilize the refund offer (p_= .012). Income and education were not significantly associated with intent to utilize the refund offer. 87 Although there was no statistically significant association between income or education and intent to utilize the refund offer, they were included in the model because it was expected that there would be some significant interactions when all four variables were incorporated in the model. Also, since relationships of multi-dimensional cross-classified data cannot be discovered with a chi-square test, chi-square test results were not sufficient reason to delete income and education variables for log-linear analysis. It was expected that some relationships would exist among these four variables. Each of the three variables was reported by respondents in two or three categories: Annual household income was expressed in three categories:15 1. Less than $20,000 (low) 2. $20,000 to $39,999 (middle) 3. $40,000 or more (high) Education background was expressed in two categories:” H There were 12 income categories. Median of income was $25,000 - $29,999. Persons with low and income consisted of 22.8% and 27.5%, respectively. M There were seven education categories. Median of education was high school degree. 88 1. Low : high school degree or less 2. High: college or more Camping affiliation was also expressed in two categories: 1. Strong: those who both subscribed to a camping related magazine(s) and attended camping or outdoor shows. 2. Weak : those who subscribed to a camping related magazine(s) 9; attended a camping/outdoor show 9; did neither. The intent to utilize the $1 refund offer was expressed in two categories:17 1. Yes: those who intended to take advantage of the $1 refund offer. 2. No : those who did not intend to take advantage of the $1 refund offer. Selection ofaa Log-linear Model STATGRAPHICS microcomputer software was used to develop the log-linear model. A two-step procedure was employed. First, a backward elimination method was employed starting with a saturated model containing a four-factor interaction among intent to utilize the $1 refund offer, income, education, and camping affiliation. The least significant three-factor u-terms were first eliminated and then the least significant two-factor u terms were eliminated until only significant effects (p_< .05) remained. Second, the :7 In this study, the intent to utilize the $1 refund offer was used instead of actual redemption. 89 effects in the model selected by the backward elimination method were tested to confirm whether or not they were substantially significant. Models Generated Using the Backward Elimination Method Table 17 reports the likelihood ratio chi-square statistic, the associated degrees of freedom, and the significance level for the goodness-of-fit test for 13 different log-linear models, including independence and saturated models. These models were some of the models generated during the backward elimination process. They were developed in order to explain the goodnees-of-fit, to test significance of effects, and finally to select a parsimonious model. A saturated four-factor interaction model, Model 1: [EICA]”, served as the starting point. A saturated model always fits the data with the significance level of 1.0 and zero degrees of freedom. However, as stated previously, saturated models do not provide meaningful information since all, even insignificant effects, are included. The backward elimination process identified Model 8: [AC][EI]19 as a model which best fits the data. Four other models with N E: education, I: household income, C: camping affiliation, A: intent to utilize the $1 off per night refund offer. M [AC][EI] model indicates two 2-factor interactions between intent to utilize the refund offer and camping affiliation and between income and education. 90 Table 17. Models Fitted to Four-Way Crosstabulation of Education, Income, Camping Affiliation, and Intent to Utilize the $1 Refund Offer. Likelihood Ratio Degrees of Models Chi-Square Freedom Significance 1. [EICA] 0.000 0 1.000 2. [ACI][EI] 9.839 9 .364* 3. [AEI][AC] 16.566 10 .085* 4. [ACI] 81.787 12 .000 5. [AEI] 26.584 12 .009 6. [AC][AI][EI] 17.030 13 .198* 7. [AC][CI][EI] 15.094 13 .301* 8. [AC][EI] 17.755 15 .276* 9. [EI][A] 27.774 17 .048 10. EI][C] 269.694 17 .000 11. [AC][E] 136.520 19 .000 12. [AC][I] 89.703 18 .000 13. [E][I][C][A] 90.249 18 .000 Abbreviations: Education E I: C: A Annual household income Camping affiliation Intent to utilize the refund offer * indicates that the model fits the data well at p_< .05. 91 significant likelihood ratio chi-square statistics were also identified (Model 2, 3, 6, and 7). Hypotheses that these models fit the data were not rejected at p=.05 level of significance. The size of Type I error is one criterion for selecting a model that fits the data. If Type I error is small, effects of the model which exist in the population are likely to be omitted. However, increasing the chances of Type I error may result in effects in the model which are not true in the population (Bishop et al., 1977). Knoke and Burke (1980) contend that if the probability of a Type I error lies between about .10 and .35, the model fits the data well. The probability of Type I error for Model 2 (p=.364) and Model 3 (p=.085) almost meets this criterion. However, Models 6, 7, and 8 were between .10 and .35, and simpler and easier to interpret than Model 2 and 3. Therefore, Model 2 and 3 were eliminated for consideration as possible final models. Test of the Significance of the Effects in Models that Best Fit the Data The next step in the model selection process was to test the significance of the effects in each of the remaining three models (Model 6, 7, and 8). To test the significance of each effect, the value of the likelihood ratio chi-square of one model including an effect (or term) was compared with that of the other model that does not 92 include that effect. The significance of an effect is determined by the difference in the likelihood ratio chi- square value with the difference in degrees of freedom. The difference in L2 approximately follows the chi-square distribution with the difference in degrees of freedom between the two models (Knoke 8 Burke, 1980). The analysis revealed that even though Model 6 and 7 fit the data, both include an effect which is not significant. Model 8: [AC][EI] was compared with Model 6: [AC][AI][EI] and Model 7: [AC][CI][EI]. The difference in 1.2 between Model 6 and 8 is .725 (17.755—17.030), and the difference in the number of degrees of freedom is 2 (15-13). The chi-square value with 2 d.f. is 5.99 at 2;.05 level. The hypothesis that [AI] effect equals zero was not rejected. Model 7: [AC][CI][EI] was compared with Model 8. The difference in.13 between the models is 2.661 (17.755 - 15.094) and the difference in the number of degrees of freedom is 2. Thus, the hypothesis that [CI] effect equals zero was not rejected. Even though these models fit the data, they have insignificant effects. In addition, they are not simpler than Model 8. Simpler models having fewer generating classes (or fewer combinations of main and interaction effects), were then formulated and compared with Model 8 because sometimes models developed using the backward elimination method 93 contain effects that are not significant. A stepwise method does not automatically provide the best fitting model that is parsimonious.m Model 9, 10, 11, and 12 (see Table 17 for details) were formulated to determine whether or not any simpler models than Model 8 were available. However, the hypotheses that the models fit the data were rejected for each of the four models. The independence model, Model 13: [A][C][E][I], does not fit the data either. Therefore, no simpler models were identified. Significance of the Effects in the Final Logélinear Model The [AC][EI] model consists of two interaction effects: interaction between intent to utilize the refund offer [A] and camping affiliation [C], and interaction between income [I] and education [E]. The hypothesis was that: the [AC][EI] model fits the data well. Respectively, there are 2, 2, 2, and 3 categories for the variables: intent to utilize the refund offer, camping affiliation, education, and income. Therefore, degrees of freedom for this model, are equal to 2x2x2x3-[1+(2-1)+(2-1)+(2-1)+(3-1)+(2-1)(2-1) +(2-1)(3-1)]=15. The hypothesis was not rejected at p_<.05 (p=.276). Thus, this model fits the data. H Fienberg asserts that "various stepwise methods should not be thought of as automatic devices for deciding up on appropriate log-linear models. At best they can be of aid in limiting attention to a reduced set of models that give a reasonable fit to the data" (1977, p.68). 94 The significance of [AC] effect was tested by comparing Model 5: [AEI] with Model 3: [AEI][AC]. Both models have the [AEI] effect in common, and the difference in the likelihood ratio chi-square between the two models is due to the [AC] effect. The difference in the likelihood ratio chi-square between the models is 10.018 (26.584 - 16.566), and the difference in degrees of freedom is 2 (12 - 10). The critical value of chi-square with 2 degrees of freedom at p; .05 level is 5.99. Therefore, the hypothesis that the [AC] effect equals zero was rejected. Thus, the interaction effect between intent to utilize the refund offer and camping affiliation was statistically significant. The significance of [EI] effect was tested by comparing the significance of the difference in likelihood chi-square ratios of Model 2: [ACI][EI] and Models 4 [ACI]. The hypothesis that [EI] effect is zero was also rejected; the interaction effect between income and education was also statistically significant. Model 8 was selected as the best model because: (1) the likelihood chi-square ratio is significant; (2) the probability of Type I error is .276 (between .10 and .35); (3) the two interaction effects are statistically significant; and (4) it is also parsimonious and easy to interpret. 95 Residual Analysis Residuals were analyzed to determine how well Model 8: [AC][EI] fits the data. In residual analysis, the difference between the observed and expected cell frequency in the model is examined. The standardized residual is calculated using the following formula. (Observed - Expected) frequency tandardized residual = Expected frequency Standard residuals greater than 1.96 (p_<.05) in absolute value suggest important differences between the observed and expected frequencies in the model which indicates the model does not fit the data. If the model fits the data, the standardized residuals are approximately normally distributed with a mean of zero and standard deviation of one (Norusis, 1986). Table 18 reports the results of residual analysis on Model 8: [AC][EI]. All but one of the standardized residuals are smaller than 1.96 in absolute value, indicating that [AC][EI] model fits the data well and, the mean of standardized residuals is .027, which is not significantly different from a mean of zero. 96 Table 18. Four-Dimensional Contingency Table with Observed, Expected, and Standardized Residuals for the [AC][EI] Model (N=459). Intent to Camping Obs. Exp. Std. Utilize Education Affil. Income Count Resid. Yes Low Strong Low 10 9.5 .16 Middle 12 12.5 -.15 High 6 4.9 .49 Weak Low 4 4.4 -.17 Middle 4 5.8 -.73 High 1 2.3 -.83 High Strong Low 1 1.5 -.38 Middle 13 11.3 .51 High 6 8.3 -.79 Weak Low 1 .7 .40 Middle 3 5.2 -.96 High 9 3.8 2.68* No Low Strong Low 35 39.3 -.68 Middle 48 51.8 -.52 High 23 20.3 .61 Weak Low 42 37.9 .67 Middle 56 49.9 .86 High 17 19.6 -.58 High Strong Low 3 6.0 -1.24 Middle 53 46.6 .94 High 36 34.1 .33 Weak Low 9 5.8 1.32 Middle 39 44.9 -.89 High 28 32.9 -.85 * indicates a significant difference between the observed and expected frequency (Z-value > 1.96). 97 Parameter Tests Parameter tests were performed to examine which and to what degree the independent parameters were related. Table 19 presents the parameter coefficients and standardized parameter coefficients for each effect in the final model, [AC][EI]. The ratio of the coefficient to its standard error, which is the standardized coefficient or Z-value, was used to test the null hypothesis that the effect is not significant. Standardized coefficients parameters greater than 1.96 in absolute value are considered significant at p<.05. The interaction between intent to utilize the refund offer and camping affiliation was significant with the standardized coefficient of 1.975 in absolute value. The standardized coefficient for the interaction of strong camping affiliation and intention to seek a refund was 1.975, indicating a significant interaction effect. Therefore, persons who exhibited a strong affiliation with camping were more likely to utilize the $1 refund offer than were persons who exhibited a weak camping affiliation. The interaction between education and income was also significant with a Z-value of 4.473 and 3.793 for low and high income levels. That is, as would be expected, people with less education are more likely to have lower income (Z—value =4.473). People with a high education levels are more likely to have a high income (Z = 3.793). Standardized 98 Table 19. Estimates for Parameters Coefficients Relevant to the [AC] and [EI] Interaction Effects. Standard Standardized Parameter Coefficients Error Coefficient Likelihood of Utilizing Offer(A) Yes -.894 .096 -9.340 * No .894 .096 9.340 * Education (E) Low .243 .096 2.537 * High -.243 .096 -2.537 * Camping Affiliation (C) Strong .204 .096 2.130 * Weak -.204 .096 -2.l30 * Income (I) Low -.565 .155 -3.646 * Middle .595 .116 5.124 * High -.030 .133 -.227 A*C Yes Strong .189 .096 1.975 * Yes Weak -.189 .096 -1.975 * No Strong -.189 .096 -1.975 * No Weak .189 .096 1.975 * E*I Low Low .693 .155 4.473 * Low Middle -.190 .116 -1.639 Low High -.503 .133 -3.793 * High Low -.693 .155 -4.473 * High Middle .190 .116 1.639 High High .503 .133 3.793 * * indicate a significance with a Z-value greater than 1.96. 99 coefficients for education, income, camping affiliation, and intent to utilize the refund offer were also significant except for the high income category. When the frequency in a particular category is larger than the average frequency across all categories, the parameter coefficient appears positive. In case of intent to utilize the refund offer, the parameters for "Yes" and "No" categories were -.894 and .894, respectively. The standardized parameters were -9.340 for "Yes" and 9.340 for "No", showing that the effect of intent to utilize the refund offer was significant. The negative coefficient indicated that the number of people who answered "Yes" was fewer than the average number of two categories. In actuality, 70 and 389 Campvention attendees answered "Yes" and "No", respectively. The coefficient for strong camping affiliation was .204 with the standardized coefficient of 2.130. The positive coefficient indicated that there was a significant difference in the number of respondents who showed strong affiliation and the average number of two categories (strong and weak). That is, the number of respondents who subscribed to camping magazines aad_attended camping or outdoor shows (strong affiliation) was much greater than the number (213) of respondents who either subscribed to camping magazine gr attended camping/outdoor shows or neither. 100 The number (258) of respondents with a high school degree or less was greater than the average number of two categories. The number of respondents with some college education or advanced degrees was 201. The parameter for the lower level of education was .243, with the Z-value of 2.537. The number (105) of respondents with an annual household income less than $20,000 was far fewer than the average number of respondents across the three categories of income. The coefficient was -.565 and Z-value was -3.646. The coefficient for the middle income group between $20,000 and $39,999 was .595 with Z-value of 5.124. A relatively large number of respondents (228) belonged to the middle group. Fewer than the average number of campers (126) had an annual household income of $40,000 or more. No relationships between intent to utilize the refund offer and income or education were found. Probabilities of Utilizing the $1 Refund Offer The findings of the log-linear analysis provided a basis for estimating the probabilities that different segments would take advantage of the refund offer. The probabilities of utilizing the offer were estimated for segments formulated using different segmentation bases, including different levels of camping affiliation, 101 education, and income, and the combinations of these three variables. These variables were all included in the final log-linear model. Table 20 shows the percentage of segments who indicated their intention to seek a refund. Some segments are more refund-prone. Segments exhibiting strong camping affiliation and low education are more prone to utilize the refund offer. Persons with weak camping affiliation and low education are unlikely to seek a refund. These differences across segments suggest that organizations should target sales promotions at segments which show a relatively higher likelihood of utilizing the $1 refund offer. 102 Table 20. Characteristics of Campers. The Percentage of Persons Who Indicated They Would Seek a Refund by the Different Characteristics/Segments Percent Camping Affiliation Strong 19.5'(246)h Weak 10.3 (213) Education Low (High School or Less) 14.3 (258) High (College or Advanced Degrees) 16.4 (201) Income Low (Below $20,000) 15.2 (105) Middle ($20,000-$39,999) 14.0 (228) High ($40,000 or More) 17.5 (126) Education * Income Low Low 15.4 (91) Middle 13.3 (120) High 14.9 (47) High Low 14.3 (14) Middle 14.8 (108) High 19.0 (79) Camping Affiliation * Education Strong Low 20.9 (134) High 17.9 (112) Weak Low 7.3 (124) High 14.6 (89) (continued on next page) 103 Table 20 (cont'd.). Characteristics/Segments Percent Camping Affiliation * Income Strong Low 22.4 (49) Middle 19.8 (126) High 16.9 (71) Weak Low 8.9 (56) ' Middle 6.9 (102) High 18.2 (55) Camping Affiliation * Education * Income Strong Low Low 22.2 (45) Middle 20.0 (60) High 20.7 (29) High Low c Middle 19.7 (66) High 14.3 (42) Weak Low Low 8.7 (46) Middle 6.7 (60) High 5.6 (18) High Low 10.0 (10) Middle 7.1 (42) High 8.1 (37) aThe table entry is the percentage of parties/persons who indicated their intentions to seek a refund. bNumbers in parentheses are total within-cell sample sizes on which each percentage is based. :Only four persons are included in this segment. 104 Summary The variables used in the log-linear analysis were selected using a multi-step procedure. Income and education were incorporated because they were almost always used in previous studies describing the characteristics of coupon- prone segments for frequently purchased consumer goods. Statistical tests showed a significant relationship between camping affiliation and whether or not Campvention attendees intended to seek a refund. The backward elimination method resulted in a log- linear model with two statistically significant interaction effects between intent to utilize the refund offer and camping affiliation, and between income and education. People who were strongly affiliated with information-seeking efforts for camping were more likely to take advantage of the refund offer. Although some interaction effects among income, education, camping affiliation, and intent to utilize the refund offer were expected, no statistically significant higher-order interaction effects (e.g., 3-factor interaction effects) were found. Thus, the intent to utilize the refund offer was not likely to be influenced by the combined levels of income and education, camping affiliation and education, or camping affiliation and income. CHAPTER VI SUMMARY, RESEARCH RECOMMENDATIONS: AND MARKETING IMPLICATIONS This study had four objectives: (1) to provide a marketing-relevant profile of people who intended to utilize the $1 money-back offer, (2) to assess the effects of the money-back offer on pre- and post-Campvention camping in Michigan, (3) to examine the relationships among the intent to utilize the refund offer, income, education, and camping affiliation, and (4) to estimate the probabilities of utilizing the $1 refund offer by different segments of Campvention attendees. In this chapter, (1) summary of the major findings, (2) limitations/weaknesses of the study and recommendations focusing on the use of log-linear analysis, and (3) marketing implications were provided and discussed. Summary of Findings Persons atttending the 1988 NCHA's Michigan Campvention were very active, heavy-volume campers. They camp an average of 52 nights per year. Approximately 86% usually camp before Memorial Day and after Labor Day, while 82% camp each year in states other than where they reside. Almost 105 106 three quarters subscribed to camping magazines; about half belonged to other camping organizations; and 70% attended camping or outdoor shows. More than half of Campvention attendees camped an average of 4.8 nights in Michigan before and/or after the Campvention. On the average, they spent one night in state parks and three nights in commercial campgrounds. Approximately 90% of those who attended the Campvention were aware of the $1 refund offer. However, only 11.5% of the attendees intended to take advantage of the offer. Out-of- staters were more inclined to utilize the refund offer than were Michigan residents. In part, this was due to the fact that out-of-Michigan residents camped more nights before and/or after the Campvention, and the results showed that persons who camped more nights were more likely to apply for a refund. A higher percentage (51%) of parties who camped eight or more nights in Michigan other than at the Campvention site had intentions to take advantage of the $1 refund offer than those who camped four to seven nights (31%) and one to three nights (21%). Persons who indicated their intentions to seek a refund were more likely to subscribe to camping magazines and belong to camping organizations other than the NCHA. However, unlike a number of other sales promotion studies which were reviewed, there was no statistically significant 107 relationship between income or education levels and intentions to utilize the money-back offer. The results indicate that the $1 refund offer was not enough to induce Campvention attendees to camp more nights in state parks or commercial campgrounds (no refund effect). Only 11.5% of the parties who camped additional nights in Michigan before and/or after the Campvention intended to seek a refund for 4.5 nights ($4.50). In actuality, only 6.3% of the parties who camped additional nights in Michigan applied for the refund, and the average refund applied for was 5.3 nights ($5.30). There was a significant discrepancy (slippage) between those who indicated their intentions to seek a refund and those who actually applied. Log-linear analysis was used to determine the relationship among intent to utilize the refund offer, income, education, and camping affiliation (e.g., subscriptions to camping magazines and attendance at camping or outdoor shows). A backward elimination method resulted in a two-factor interaction model. Significant interactions were found between intent to utilize the refund offer and camping affiliation. Campvention attendees who were strongly affiliated with camping were more likely to utilize the offer than those who were not. There were no significant relationships between intent to utilize the refund offer and income or education. The findings of the log-linear analysis were used as the base to estimate the 108 probabilities of utilizing the refund offer by different segments of Campvention attendees. Limitations/Weaknesses The major weakness of this study is that intent to utilize the refund offer was assumed to be equal to their actual redemptions. The intention to utilize promotional offers is not a perfect indicator of actual redemption due to the slippage effect. There are always some discrepancies between actual behavior and intention to take advantage of promotional offers. However, the results of this study are still useful because self-reported intentions have been correlated with actual behavior (Kalwani 8 Silk, 1982; Morrison, 1979). Although this measurement issue may not limit the results of this study, future studies should use actual redemption instead of intentions to obtain accurate results. A second and much more minor weakness resulted from the sampling method. The sample was selected only from preregistrants for the 1988 Campvention. NCHA members who did not preregister for the Campvention were not included in the sample. Persons/parties who did not preregister but participated in the Campvention were not included in the sample. So these results cannot be generalized to all NCHA members. NCHA members who preregistered but did not attend the Michigan Campvention were included in the sample, but 109 the questionnaire identified them as nonattendees. These sampling problems are not viewed as significant since the primary purposes of this study were to profile Campvention attendees and to evaluate the effectiveness of the refund in increasing pre- and post-Campvention camping. Since no previous studies were identified which evaluated camping-related sales promotions, the questionnaires, hypotheses, and variables included in the log-linear analysis were based on studies of sales promotions for nonrecreational products. Although these studies provided information on the characteristics and behavior of sales promotion prone consumers, there were no studies which confirmed that the same characteristics were relevant to the use of recreation/tourism-related sales promotions. Therefore, this study was for the most part exploratory in nature. Other variables (e.g., differing amount of refund) should have been examined and included in the log-linear model. Research Recommendations Log-linear analysis an appropriate statistical technique when simultaneously examining the relationships among multi-dimensional cross-classified data. A series of chi-square tests can not be used to investigate the relationships between a number of categorical variables. 110 In order to find a linear association between two categorical variables, similar future research should incorporate the ordering of the categories into log-linear analysis. The extention of log-linear models (e.g., the linear row-effects, linear column-effects, or linear by linear association models) can be used to test a linear association between two variables that are ordinal. Although log-linear parameters are relatively easy to analyze in four and higher-dimensional contigency tables, researchers who conduct similar analyses should also consider using odds ratio (cross-product ratio) to interpret the types and degree of association in 2x2 tables (e.g., the relationship between intent to utilize the refund offer and camping affiliation). An odds is the ratio of the frequencies that an event occurs and does not occur. However, the intuitive appeal of an odds ratio declines when a higher order interaction (more than three-way interaction) exists because the odds ratio must be calculated for each level of the other variable(s) involved in the interaction (Elliott, 1988; Gilbert, 1981). As the literature shows, coupon-prone segments for one product are inclined to display similar tendencies for other products. According to Bawa and Shoemaker (1987), households were found to be consistent in coupon usage across product classes. This study focused on camping only. Therefore, the relationships found in this study could be 111 tested for other recreational activities or tourism services (e.g., fishing, boating, skiing, foods, or lodging). Marketing Implications This study has provided significant marketing information. It provided the first comprehensive marketing- oriented profile of persons who attended the National Campers and Hikers Association's (NCHA) annual Campvention. Not only can this information be used by NCHA more effectively to market Campventions to its members, it can also be used by states, where Campventions are held. For example, the findings (e.g., trip length, trip planning, trip information) can be used to design marketing strategies to increase the amount of before and/or after Campvention camping. In fact, about half of the persons who attended the Campvention had not camped in Michigan previously. Thus, the Campvention offers host states a unique Opportunity to market recreation and tourism, especially camping opportunities to first-time visitors, and by doing so promote repeat visits. The findings provided current marketing relevant insight about a "heavy-volume" segment of the camping market. This segment is identifiable, reachable and exploitable, especially for sales promotion campaigns. The information generated by this study can be used by states (e.g., travel bureaus, agencies) and campground businesses, 112 to design strategies to attract and better serve heavy- volume campers. The findings indicate that the $1 money-back offer did not significantly affect the proportion of camping nights or the amount of pre- and post-Campvention camping in Michigan state parks or commercial campgrounds. In part, this is due to the relatively low refund compared to the overall cost of a night of camping, and the limited time Campvention attendees had available to camp additional nights in Michigan (beyond the seven nights at the Campvention site). It is likely that other factors (e.g., time available, trip itinerary, overall trip costs) constrained the number of nights which were available to camp in Michigan. It is important to recognize that the cost of campsite rental is only a small portion of the overall costs of a camping trip. A recent study reported that the average camping party spent $163 per trip, not including campsite rental costs (Mahoney 8 Yu, 1988). The findings also show that the refund value would have to have been substantially greater to affect the number of nights camped in Michigan. However, even if the refund had been $8, fewer than half would have taken advantage of the refund. Whether or not a higher refund would have generated sufficient, additional camping to offset the refund costs is doubtful. The percentage of persons who would likely have taken advantage of a higher refund offer during their Campvention 113 trips increased as the refund value increased. However, as Gupta (1988) points out, unless the promotional offerings are beyond a threshold level, it is not likely to affect choice probabilities. The implication is that money-back offers, such as this one, must be only one component of an overall promotional effort, including other types of sales promotions (e.g., contests, premiums). It is likely that the money-back offer would have been more effective in generating camping nights before and/or after the Campvention if Campvention attendees had received information about the offer and recreational opportunities in Michigan more in advance of the Campvention and before they had decided on the length of their trip and trip itinerary. A number of persons volunteered that they would not utilize the offer because they had already planned to stay at some other campgrounds (e.g., KOA) or at campgrounds in other states prior to receiving information about the offer. Therefore, recreation/tourism sales promotions such as this refund offer should commence before the targeted customers have decided on trip itinerary. Only 30% of persons who camped additional nights in state parks and/or commercial campgrounds had intentions (at least at the time when they received the posttrip questionnaire) to utilize the $1 refund offer. Fewer than a half of those who intended to seek a refund actually applied. Although it was estimated that persons who 114 attended the Campvention camped approximately 9,000 nights in Michigan state parks and commercial campgrounds before and/or after the Campvention, only $1,300 was refunded. The 6.3% redemption rate is in line with response rate to similar sales promotions. However, it was lower than the Division of State Parks or MAPCO (the sponsors) expected. In part, the lower than expected redemption rate was due to the cost and time required to apply for a refund relative to the refund amount. Undoubtedly many parties, especially those who camped only a few nights in state parks and/or commercial campgrounds, decided that it was not worthwhile to apply for a refund. Others may have misplaced the application form or the necessary proof of purchase. Some failed to return the required materials prior to the refund cut-off date. The relatively short time period for applying for a refund may also have discouraged some busy people from applying for the refund. Each sales promotion technique has strengths and weaknesses. For example, coupons offer the convenience of payment at purchase time, whereas refunds are not so immediate. A more customer-convenient refund procedure (e.g., coupon books with immediate refunds) at the participating/eligible campgrounds, would likely have increased the number of persons who sought a refund. However, the small refund value ($1 off per night) and camping parties' already-fixed schedule before Campvention 115 would tend to indicate that it is unlikely that a more convenient refunding procedure would have significantly increased the number of nights camped in Michigan state parks or commercial campgrounds. The result of log-linear analysis indicates that campers who both subscribed to camping-related magazines and attended camping/outdoor shows were more likely to utilize the refund offer. Those who did only one of the preceding activities, or neither of the activities were less likely to utilize the refund offer. The relationship between intent to utilize the refund offer and camping affiliation has two important marketing implications and uses. First, recreation marketers should consider degree of affiliation/involvement as a base for segmenting markets for the purpose of designing and targeting sales promotions. This information also suggests a way of reaching sales promotion prone campers. Second, it is equally important to consider sales promotions as only one component of the promotion aimed at achieving a specific objective. The insignificant refund effect indicate that a package of complementary sales promotions would have likely been more effective. Although the findings indicate that the $1 refund offer did not significantly increase before and/or after Campvention-related camping in Michigan, it was still a cost effective promotion. The offer generated approximately 305 116 additional nights of camping in Michigan state parks or commercial campgrounds. The average campsite fee in Michigan was approximately $8.50 and the average spending per night/party was $78.70, including campsite fee. So the money-back offer generated an additional $24,000 in revenues or spending that would not have accrued without the refund offer. The total cost associated with the refund offer was almost $4,000. Approximately $1,300 was paid out in the form of refunds. Printing, mailing (refunds to the applicants), and handling costs were around $2,700. MAPCO paid all refunds associated with nights camped in commercial campgrounds and the Division of State Parks covered refunds for nights camped in state parks campgrounds. In addition, the offer sent the message that Michigan wanted the attendees to camp and spend additional time in Michigan. This is important since Campvention attendees are high-volume campers which Michigan needs to target and attract. B I BL I OGRAPHY B I BL I OGRAPHY Aldrich, J. H., 8 Nelson, F. D. (1984). Lineargprobability, logit, and probit models. Beverly Hills, CA: Sage Publications. Allison, P. D., 8 Liker, J. K. (1982). Analyzing sequential categorical data on dyadic interaction: A comment on Gottman. Psychological Bulletin, 21(2), 393-403. American Association of Advertising Agencies (1978). Sales promotion techniques: A basic guidebook. New York. Bawa, K., 8 Shoemaker, R. W. (1987a). The effects of a direct mail coupon on brand choice behavior. Journal of Marketing Research, 24(4), 370-376. (1987b). The coupon-prone consumer: Some findings based on purchase behavior across product classes. Journal of Marketing, 51(4), 99-110. Bearden, W. O., Teel, J. E., 8 Williams, R. A. (1981). Consumer response to 'cents-off' coupons. In K. Bernhardt et a1 (Ed.), The changing marketing environment: New theoriesaand applications. Chicago: American Marketing Association. Benedetti, J. K. 8 Brown, M. B. (1978). Strategies for the selection of log-linear models. Biometrics, 33(December), 149-166. Bishop, Y. M. (1969). Full contingency tables, logits, and split contingency tables. Biomatrics, 25(2), 383-399. , Fienberg, S. E., 8 Holland, P. W. (1977). Discrete multivariate analysis: Theoryaand practice. Cambridge, MA: MIT Press. Blattberg, R., Buesing, T., Peacock, P., 8 Sen, S. (1978). Identifying the deal prone segments. Journal of Marketing Research, 15(3), 369-377. 117 118 Blattberg, R., Eppen, G. D., 8 Lieberman, J. (1981). A Theoretical and empiricall evaluation of price deals for consumer nondurables. Journal of Marketing. i§(1), 116-129. Block, M, Brezen, T., 8 Schultz, D. (1986). The impact of sales promotion in fast food restaurants. In E. Larkin (Ed.), Proceedings of the 1986 Conference of the Emerigaa Academy of Advertising (pp. R17-R22). Boddewyn, J. J. 8 Leardi, M. (1989). Sales Promotions: Practice, regulation and self-regulation around the world. International Journal of Advertising, £14), 363-374. L7 Bonett, D. G. 8 Bentler, P. M. (1983). Goodness-of-fit procedures for the evaluation and selection of log-linear models. Psychological Bulletin, 91(1), 149-166. Bowman, R. (1980). Couponing1and rebates. profit on the dotted line. New York: Lebhar-Friedman Books. (1988). 1987: The year in promotion. Marketing 8 Mediaapecisiona,.;;(7), 150-154. Boyd, M. M. (1985). The Dirichlet multivariate multinomial distribution as a magazine exposure model. Unpublished Ph.D. dissertation. University of Florida. Brown, R. G. (1974). Sales response to promotions and advertising. Journal of Advertising Research, 11(4), 33-39. Chapman, R. G. (1986). Assessing the profitability of retailer couponing with a low-cost field experiment. Journal of Retailing, filll), 19-40. Cotton, B. C., 8 Babb, E. M. (1978). Consumer response to promotional deals. Journal of Marketing, 32(July), 109-113. Danaher, P. J. (1988). A log-linear modeel for predicting magazine audiences. Journal of Marketing Research, 25(4), 356-362. Dillon, W. R., 8 Goldstein, M. (1984). Multiyariate analysis : methods and_applications. New York: Wiley. Dodson, J. A., Tybout, A. M., 8 Sternthal, B. (1978). Impact of deals and deal retraction on brand switching. Journal of Marketing Research, 11(1), 72-81. 119 Dun 8 Bradstreet Corporation (1988). Study finds slight rise in 1987 coupon redemption. News (July 12). Duncan, 0. D. (1979). How destination depends on origin in the occupational mobility table. American Journa} of Sociology, 84(4), 793-803. Elliott, G. C. (1988). Interpreting higher order interactions in log-linear analysis. Psychologica1 Bulletin, 103(1), 121-130. Feick, L. F., 8 Novak, J. A. (1985). Analyzing sequential categorical data on dyadic interaction: Log-linear models exploiting the order in variables. Psychological Bulletin, 2§(3), 600-611. Fienberg, S. E. (1977). The analysis of cross-classified categorical data. Cambridge, MA: MIT Press. Flath, D., 8 Leonard, E. W. (1979). A comparison of two logit models in the analysis of qualitative marketing data. Journal of Marketing Research, 16(4), 533-538. Fridgen, J. D., Mahoney, E. M., Nelson, C. M., 8 Holecek, D. (1986). Michigan state park atudy: Campers and day usera. East Lansing: Michian State University. Gardner, M. P., 8 Strang, R. A. (1984). Consumer response to promotion: Some new perspectives. In K. C. Kinnear (Ed.), Advances in Consumer Research, 11J 745-748. Gilbert, E. S. (1968). On discrimination using qualitative variables. Journal of the American Statistical Association. 63:1399-1412. Gilbert, G. N. (1981). Modelling society: An introduction to loglinear analysis for §ocia1 resaarchers. London, UK: George Allen 8 Unwin. Goodman, L. A. (1972). A general model for the analysis of surveys. Amerigan Journal of SociolggyJ 11(6), 1035-1086. Green, J. A. (1988). Loglinear analysis of cross-classified ordinal data: Applications in developmental research. Child Development, égj 1-25. Green, P. E., Carmone, F. J, 8 Wachspress, D. P. (1977). On the analysis of qualitative data in marketing research. Journal of Marketing Resaarch, 14(1), 52-59. 120 Gupta, S. (1988). Impact of sales promotions on when, what, and how much to buy. Journal of Marketing Research, 25(4). 342-355. Haberman, S. J. (1978). Loglinear fit for contingency tables Applied Statistics, 21, 218-225. Hackleman, E. C., 8 Ducker, J. M. (1980). Deal proneness and heavy usage: Merging two market segmentation criteria. Journa} of the Academy of Marketing Science, 5(Fall), 332-344. Hair, J. F., Anderson, R. E., Tatham, R. L., 8 Grablowsky, B. J. (1979). Multivariate data analysis. Tulsa, Oklahoma: Petroleum Publishing Company. Henderson, C. M. (1985). Modeling the coupon redemption decision. E.C. Hirschman 8 M. B. Holbrook (Eds.), Advances in Consumer Research, 12. Provo, Utah: Association for Consumer Research. Higgins, K. T. (1986). Sales promotion spending closing in on advertising. Marketing News, 25(14): 8, 11. Hume, S. (1989). Fast food: Panic creeps in amid more sluggish sales. Advertising Age, 55(39), S4. Iacobucci, D., 8 Wasserman, S. (1987). Dyadic social interactions. ggychological Bulletin, 102(2), 293-306. Irons, K. W., Little, J. D., 8 Klein, R. L. (1983). Determinants of coupon effectiveness. In F. S. Zufryder (Ed.), Adyancea_andgpractices of marketingaacience. Proceedings of the 1983 ORSA/TIMS Marketing Science Conference (pp. 157-164). University of Southern California. Johnson, W. C. (1988). Sales promotion: It's coming down to 'Push Marketing'. Marketing News, 22(5), 8-9. Jolson, M. A., Wiener, J.L., 8 Rosecky, R. B. (1987). Correlates of rebate proneness. Journal of Advertising Research, 21(1), 33-43. Kalwani, M. V., 8 Silk, A. J. (1982). On the reliability and predictive validity of purchase intention measures. Marketing Science, 1(2), 243-286. Keon, J. W. 8 Bayer, J. (1986). An expert approach to sales promotion management. Journal of Advertising Research, 25(3), 19-26. 121 Klein, R. L. (1981). Using supermarket scanner panels to measure the effectiveness of coupon promotions. In J. W. Keon (Ed.), Proceedings: Third ORSAlTIMS special interest conference on market meaaurementgand analysis (pp.118 - 124). Providence, RI:The Institute of Management Science. Knoke, P. 8 Burke, D. (1980). Log-linear models. Beverly Hills, CA: Sage Publications. Kono, K. (1985). Are generics buyers desl-prone? on a relationship between generics purchase and deal-proneness Jouraal of the Academygof Marketing Science, 15(1),62-74. Kotler, P. (1984). Marketing management: analysisy_planning. and control. Englewood Cliffs, N.J.: Prentice-Hall, Inc. Kriska, S. D. 8 Milligan, G. W. (1982). Multiple regression analysis for categorical data with an illustrative application in personnel selection. Psychological Bulletin, 52(1),193-202. Kuehn, A. A., 8 Rohloff, A. C. (1967). Consumer response to promotions. In P.J. Robinson (Ed.), Promotional deciaions using mathematica1 models (pp.45-148). BostonzAllyn and Bacon, Inc. Lee, J .Y., 8 Brown, M. G. (1985, August). Coupon redemption and demand for frozen concentrated orange juice: a switching analysis. American Journa1 of Agricultura1 Economics, 647-653. Lincoln, D. J. (1978). Consumer rebates: A suggested decision framework for marketers and some consumer view- points. In R. S. Franz et al (Eds.), Proceedings of the Southern Marketing Association. Mahoney, E. M., Oh, I., 8 Ou, S. (1989). A study of the National Campersaaad Hikers Association'a 1988 Michigan gampvention. East Lansing: Michigan State University, Department of Park and Recreation Resources. 8 Yu, Y. (1989). Michigan Commercia1_ Campground Study. East Lansing: Michigan State University, Department of Park and Recreation Resources. Malhotra, N. K. (1984). The use of linear logit models in marketing research. Journal of Marketing Research, 21(1), 20-31. 122 Massy, W. F., 8 Frank, R. E. (1965). Short term price and dealing effects in selected market segments. Journal of Marketing Research, 2(2), 171-185. Montgomery, D. B. (1971). Consumer characteristics asso- ciated with dealing: An empirical example. Journal of Marketing Research, 15(1), 118-120. Moore, D. H. (1973). Evaluation of five discrimination procedures for binary variables. Journa1 of the American Statistical Association, 55:399. Morrison, D. G. (1979). Purchase intentions and purchase behavior. Journal of Marketing, 55(2), 65-74. Narasimhan, C. (1984). A price discrimination theory of coupons. Marketing Science, 1(Spring). 128-147. Neslin, S. A., 8 Clarke, D. G. (1987). Relating the brand use profile of coupon redeemers to brand and coupon characteristics. Journal of Advertiaing Research, 25(1), 23-32. Neslin, S. A., Henderson, C., 8 Quelch, J. (1985). Consumer promotions and the acceleration of product purchases. Marketing Science, 5(Spring), 147-165. Neslin, S. A., 8 Shoemaker, R. W. (1983). A model for evaluatig the profitability of coupon promotions. Marketing Science, 2(Fall), 361-388. Nielsen Company (1983). Promoting with consumer refunds. A special report from the Nielsen Clearing House. Chicago: Nielsen Company. Norusis, M. J. (1986). SPSS PC+: Advanced statistics. Chicago: SPSS, Inc. (1988). SPSS PC+ V3.0 update manual. Chicago: SPSS , Inc . Raju, P. S., 8 Hastak, M. (1979). Consumer response to deals: A discussion of theoretical perspectives. In J. C. Olsen (Ed.), Advances in Consumer Research, 1J 296-301. Reibstein, D. J., 8 Traver, P. A. (1982). Factors affecting coupon redemption rates. Journa1 of Marketing, 15(4), 102-113. Reynolds, H. T. (1977). Analysis of nominal data. Beverly Hills, CA: Sage Publications. 123 Rosenfeld, J. (1987). Eighth annual advertising 8 sales promotion report. Marketing Communicationa, 12(8), 7-16. Schindler, R. (1984). How cents-off coupons motivate the consumer. In K. E. Jocz et al (Eds.), Research onaaalaa Promotion (pp.47-62). Cambridge, Massachusetts: Marketing Science Institute. Schleier, C. (1985). Marketing image plots turnaround. Advertising Age, 55(63), 15-16. Schultz, D. E., 8 Robinson, W. A. (1982a). Sales promotion management. Chicago: Crain books. (1982b). Sales promotion essentials. Chicago: Crain Books. Shoemaker, R. W. (1979). An analysis of consumer reactions to product promotions. Educators' Conference Proceedings (pp. 244-248). Chicago: American Marketing Association. , 8 Tibrewala, V. (1985). Relating coupon redemption rates to past purchasing of the brand. Journal of Advertising Research, 25(5), 40-47. Strang, R. A. (1976, July-August). Sales promotion-fast growth, faulty management. Harvard Businesa'Review, 5gJ 115-124. Stynes, D. J., 8 Mahoney, E. M. (1986). 1984 Michigaa commercial campground marketing study. East Lansing: Michigan State University, Department of Park and Recreation Resources. Teel, J. E., Williams, R. H., 8 Bearden, W. (1980). Correlates of consumer susceptibility to coupons in new grocery product introductions. Journal of Advertising, 5(3), 31-46. Telzer, R. (1987). Sales promotion: rebates challenge coupons' redeeming values. Advertiaing Ages, 55(12), 518-520. Varadarajan, P. R. (1984). Consumer responses to small business coupon sales promotions. American Journal of Small Business. 5(2), 17-26. Walters, R. C., 8 McKenzie, S. B. (1988). A structural equations analysis of the impact of price promotions on store performance. Journal of Marketing Research, 25(1), 51-63. 124 Ward, R. W., 8 Davis, J. E. (1978). Coupon redemption. Journal of Advertising Research, 15(4), 51-58. Webster, F. E. (1965). The 'deal-prone' consumer. Journal of Marketing Research, 2(2), 186-189. Wright, T. (1989). Frequent-flier programs spawn host of imitators. Marketing News, 25(10), 2 8 8. APPENDICES ___F— ...—.1 Yes (the 1988 will be my FIRST CAHPVEMTlON) (GO TO QUESTION 4) I A 125 APPENDIX A 1988 wicntggu CAMPVEMTIOM sruov Michigan State University, Michigan Division of State Parks, Michigan Association of Private Caapground Owners,and the Rational Canpers and hikers Association are corducting a cmprehensive study of persons who attend the 1988 MICMIGAN CAMPVEMTIOM being held at highland Recreation Area. The study will provide information which will be useful in decisions regarding future canpventions. we will also be sending you another brief questionnaire after you return home from your trip to gather information on your satisfaction with the 1988 Caapvention and carping in Michigan. If you are planning to attend the 1988 Canpvention PLEASE COMPLETE the following questiomaire and RETURN it to us in the attached postage paid envelope. PLEASE take the time to couple-ta the questionnaire. without your help the study will not be Successful. We guarantee that your response will remain strictly confidential. l. DATE 7w COMPLETED this QUESTIONNAIRE / I (MONTH/DAY/YEAR) 2. Hill the 1988 Michigan CAMPVEHTIOM be the FIRST National Caupers and Hikers CAMPVEMTIOM you have attended? ____ "0 -—€> Did you attend the 1957 IOWA CAMPVEMTIOM? Yes no (60 TO oussrtou 4) 3. ON YogRLIOS? ICUA CAMPVEMTTOM TRIP how many nights did you spend on: 3a) On your entire CAMPVENTIOM TRIP (This includes nights at the Canpvention, nights in Iowa before and after the Campvention, and nights in other states traveling to and from the Caapvention) umber of total nights away from home 3b) At the Iowa CAMPVEMTIOM SITE: Mather of nights 3:) At campgrounds in Iowa (OTHER THAN THE CAMPVENTION SITE): Member of nights at other carpgrotrds 3d) At campgromds OJTSIOE Iowa: Murber of nights '" (3a should equal the SUM of 3b, 3c, and Id) ”' y ......OQOOQOOOOQOOOOOOQO ‘9” "Icnlmu WPVENYIO“ QUESTIONS "..“Q'm.”;....'. \ 4. At the I988 MICHIGAN CAMPVEMTIOM how many persons includiggyourself will STAY OM YOUR SIT§ with you 7 La) what are the AGES of the persons who will gray on your gite? YGJRSELF , Person 2 , Person 3 , Person 6 , Person 5 , Person 6 , Person 7 . 5. On your MICHIGAN CAMPVEMTIOM TRIP what type of camping equipment will you utilize? __ rent ’53 ~ __ Caaping Trailer m __ Travel Trailer Motor Home Van/Bus Conversion 5th wheeler Other 6. On your MICHIGAN CAMPVENTION TRIP how many TOTAL MIQflTS AUAY FROM HOME will you spend? This includes: nights at the Campvention, nights in Michigan before and after the Campwention and, nights in other states traveling to and from the Camvention. Total 1988 CAMPVEMTION TRIP nights 126 7. How many nights are you planning to cap at the {Michigan} mflgHTIOH SIT§ located in Highland Recreation Area? (The CAMPVEHTIDH will last 7 nights) Huber of nights at the Michigan QAMPEHTIOH SITE 8. Other than the nights at the CAMPVEHTIOH SIT§ are you planning to cars: Eitional nights in MICHIGAN either before or after the Camvention? [ Mo (to T0 outsnou 15) Yes -> How many ADDITIONAL nights (not countim nights at the CAMPVEHTIOH SIT§) are you planning to can: in Michigan 7 Huber of additional nights (GO TO WESTIOH 9) 9. will you likely SEL§CT TH: cawpcnouuogsz. or have you alrem gelgted the catipgrmndu) (OTHER THAN CAMPVEHTIDM SITE) you will stay at in Michigan DEFDR§ LEAVING HCM§ on the trip? ,_...____.l -_—i No (GO TO QUESTION 11)] Yes (60 TO QUESTION 9a) 9a) Have you already selected the canogromdu) (OTHER THAN CAMPVEHTIO! SITE) you will stay at in Michigan ? _ Yes -> How many Michigan camgrotnds have ygg alregy 14.-[5th ? runner of camgrolnds V 10. HILL you make, or have you already made, reservations at these canpgrotnds (OTHER THAN CAMPVEHTIDH SITE) before leaving houie on the trip? MO Yes 9 We) Have you ALREADY mace reservations at caaogromds in Michigan ? No Yes I I ’11. In your registration package there is an offer for a $1.00 OFF REFUND for each night you spend carping at a Michigan State Park or cmround which i; a merber of Michigan Assgiagion of Private cm Mg (MAPCO). The refund offer will not apply to other Michigan canngrotnos O_3_ nights at the cmtim site. WILL YOU LIKELY TAKE ADVANTAGE 0F ””5 OFFER? No why not? Yes ‘ 12. what PRIMARY swaggsz informtion will you rely on most to select the caapgrotndls) OTHER THAN TH§ QMPVEHTIOH SITE you will stay at in Michigan? (Please check all that apply) Rana McHally Canning Directory Caspgromd brochures woodalls Canning Directory Recmtions from other camera at CAMPVEHTIDH Michigan Caaogrotnd Directory Recoaaendations from camera you meet in Michigan camgrounds Trailer Life Recounendations of friends 8 relatives Highway signs Past canning experience in Michigan Other (specify) 127 13. CIRCLE THE NUME§RS (1°51 on the map at the right TO SHOW TM§ REGIONS of 0 Michigan YOU PLAN TO CAMP IN while on YOUR 1988 CAMPVEMTIOM TRIP. CIRCLE the numbers of ALL REGIONS you are planning to camp in. i‘ONLY CIRCLE REGION 1 If YOU PLAN TO CAMP AT CAMPGROUNDS (OTHER THAN CAMPVENTION SITE) in this region. 14. Have you already written or called, or do you plan to write or call, for additional Michigan travel/recreational information? ___ Yesi€a>14a) which Organization(s) have you written or called, or will you write or call for more information? ___ Michigan Travel Bureau ____ Nest Michigan Tourism Association ___ Michigan Dept. of Natural Resources ____ Southwest Michigan Tourism Association ___ East Michigan Tourism Organization ____ Upper Peninsula Tourism Association 21_, ‘___ Southeast Michigan Tourism Organization ____ Other (Specify) ‘L IS. Please rate the IMPORTANCE of the following CAMPGROUND ATTRIBUTES AND/OR FACILITIES UHEN SELECTING A CAMPGROUND? CAMPGROUNO ATTRIBUTES Crucial Very Important Important Somewhat Important Not Important Large sites Shaded Sites Cleanliness Quietness Site Privacy Security Hospitality of campground staff Low Price Flush toilets Electricity Showers Laundromat Campground store water hookups Sewer hookups Natural surrounding Situated on a lake/stream Hiking trails Pool 16. Do you USUALLY prefer to coup in public or private (coaliercial) calpgrou'lds? Public campground Private (commercial) campground No preference 1?. who is USUALLY MOST INFLUENTIAL in deciding which campgrounds you stay at? Myself My spouse Children Family (Group) decision Other 128 18. Approximately how many nights did you camp LAST YEAR (1987)? (If you didn't camp, write “0“ on the line) 19. How many of these nights were OUTSIDE TH§ STATE gflgngYOU LIVE? (If none, write “0“ on the line) 20. How many states (not includigg your hogs state) did you camp in during 1987? (If no other states write "0“) 21. Do you USUALLY camp BEFORE Memorial Day ? No Yes 22. Do you USUALLY camp AFTER Labor Day? No Yes 23. Have you EVER camped in Michigan ? No ___Yes-€E> when was the last year you camped in MICHIGAN? I9 24. Based on your impressions, experience, information from others, or travel/camping literature, please complete the following perception of MICHIGAN CAMPQROUNDS which include public and private campgrounds. Strongly Strongly No Michigan campgrounds: Agree Agree Disagree Disagree Impression are very large (nurber of canpsites) are inexpensive are crowded have hospitable campground staff offer many (in-campground) recreation facilities provide large campsites are clean are quiet are family oriented offer modern hookups (electric,sewer,water) are secluded provide modern restroom/shower facilities are safe/secure are well maintained 25. m.- you a RESIDENT of MICHIGAN? Yes _ No 925.) Have you gvta LIVED in Michigan 7 _ch __No 25b) Do you have family/friends LIVING in Michigan? ___Yes __No 25c) will you VISIT them on Your Campvention trip? ___Yes __No I 26. what is the gig code of YOUR PERMANENT RESIDENCE? 27. Are YOU male or female? Female Male 28. Are YOU retired? No Yes 29. Are you currently: Single Divorced/widowed Separated Married -€E> Is your spouse retired? Yes ___ No 30. Do you have children LIVING AT HOME VITH YOU ? No Yesawhat are their ages ? Child 1 Child 2 Child 3 Child 4 Child 5 129 APPENDIX B 1988 MICHIGAN CAMPVENTIOM POSTTRIP SURVEY Michigan State University, Michigan Division of State Parks, Michigan Association of Private Canpgrou'ids (MAPCO) and NCHA are conducting a study of the I988 Campvention. Now that the 1988 Carpvention is over we are interested in obtaining information about where and how long you caaped, and your evaluation of the Canpvention and carping in Michigan. If someone in your household DID NOT conplete the questionnaire we sent you BEFORE the Canpvention. Please carplete this one. He desperately need your assistance. If someone in your household COMPLETED AND RETURNED the questionnaire we sent BEFORE the Canpvention, it is important that the PERSON UHO COMPLETED THAT QUESTIONNAIRE also COMPLETE THIS ONE. 1. Did you participate in the 1988 Caupvention at Highland Recreation Area in Michigan? Yes No (Go to Question 22) 2. when did you leave home for the Michigan Caapvention? / (Month/Day e.g., July / IO) 3. At the 1988 MICHIGAN CAMPVENTION how many persons including yourself STAYED OM YwR SITE? persons (If you were alone, write "I" on the line) 3a) Vhat are the AGES of the persons who stayed on your site? Yourself, Person 2, Person 3 Person 4, Person 5, Person 6, Person 7 4. what type of carping equipment did you utilize on your CAMPVENTION TRIP? A . . Tent n Carrping trailer [$1 Travel trailer Motor home .1 Van/bus conversion 5th wheeler. Other 5. How would you rate the facilities at the Canpvention site in Highland Recreation? Circle the appropriate number. Excellent 1 2 3 4 5 Poor 6. How would you rate the 1988 Caapvention conpared to other Canpventions you have attended? __ Much better ___Setter __ About the same __worse _ Much worse __ Not attended others 7. On your MICHIGAN CAMPVENTION TRIP how many NIGHTS AUAY FRO! M046 did you spend on: 7a) Your ENTIRE Michigan CAMPVEHTION TRIP (This includes nights at the Caapvention, nights in Michigan before and after the Canpvention, and nights in other state traveling to and from the Canpvention) , Huber of nights 7b) At the MICHIGAN CAMPVENTION SITE: Nutber Of nights 7:) Cauping in MICHIGAN (OTHER THAN THE CAMPVENTION SITE): (If none, write "0") Huber of nights 7d) Hotels/motels, second home, friend/relative's home(s) in MICHIGAM:(If none, write "0“) Huber of nights 7e) Cauping in OTHER STATES (Including in your home state): (If none, write '0“) Huber of nights 7f) Hotels/motels, second home, friend/relative's home(s) in OTHER STATES (If none, write 0) Huber of nights """(7a should aggal the total of Thy 7c. 7dy 7e and 7f) """ 130 8. Did you camp at MICHIGAN CAMPGROUNDS other than the Caggyention Site? Yes (Go to Ouestion 8a) No (Go to Ouestion 13) 8a) First CIRCLE THE NUMBERS (1-6) ON THE Mag at the right to show the regions you canped in. Only circle Region I if you canped at canpgrounds in this region other than the Canpvention site. I S -. Qr 8b) Then in the TABLE BELOU indicate How MANY NIGHTS you camped at each type of campground ‘T " (NOT INCLUDING CAMPVENTION SITE) in each region you circled on the map. 3 4 ......O... "W‘- 0f night! ......fl.... Region 1 Region 2 Region 3 Region 4 Region 5 Region 6 2 1 Michigan State Parks Commercial (private) campgrounds Couity/Local Campgrounds State Forest Canpgrounds National Park or National forest Illll lllll lllll lllll ||||| lllll "'"“' Do not include the nights you canped at the Caapvention site ““““ 9. HOV MANY CAMPGRQJHDS in Michigan (not comting the Caupvention Site) did you carp at on your CAMPVENTION trip? lumber of Caapgrounds ID. How many of these canpgrolnds had you §§LECT§D BEFOR§ leaving home on your trip? Number of caepgrouids 11. How many of these caupgromds did you RESERE a site in 3.5% you left home? Huber of canpgrotnds 12. which PRIMARY szCEgsz of information did you rely on MOST TO S§L§CT the Michigan caupgromds (other than the camvention sits) you canped at? __ Michigan Caapgrou'ad Directory __ Highway signs __ woodalls Caaping Directory _ Canpgrotnd brochures __ Trailer Life __ Recounendations from other calipers at the Michigan Caepvention __ Rand McNally Caaping Directory __ Recomendatlons from caliper you met in Michigan canpgrouids __ Kanpgrouid of America (RDA) __ Recoumendations from friends 8 relatives _ Coast to Coast _ Past Caeping experience in Michigan _ AAA __ Cauping Organization/Cub _ Highway Mapls) _ Other (specify) 13. was your CAMPVENTION TRIP the first time that you'canped in Michigan? Yes No 11.. For your CAMPVENTION TRIP did you write or call for additional Michigan travel/recreational information? No __ Yes —-P which organization(s) did you write or call for more information? _ Michigan Travel Bureau __ Vest Michigan Tourism Association _ Michigan Dept. of Natural Resources _ Southwest Michigan Tourism Association __ East Michigan Tourism Organization __ Upper Peninsula Tourism Association Southeast Michigan Tourism Organization Other (specify) 131 15. Please indicate your perception(s) of Michigan campgrounds which include public and private campgrounds. O Strongly i Strongly N Michigan campgrounds Agree Agree Disagree Disagree Impression are very large (number of campsites) are inexpensive are crowded have hospitable staff offer many (in-campground) recreation facilities provide large campsites are clean are quiet are family oriented offer modern hookups (electric, sewer, water) are secluded provide modern restroom/shower facilities are safe/secure are well maintained 16. were you aware of the SPECIAL CAMPVENTION OFFER of S1.00 OFF per night of camping in Michigan State Parks or camp- grounds which are member of Michigan Association of Private Campground Owners (MAPCO) ? Yes No -e would you have taken advantage of the S 1.00 off per night offer if you had been aware of it? Yes No 17. Did you take advantage of the S 1.00 Off REFUND OFFER ? ;4_ Yes ->17a) For how many nights will(have) you apply for a refu'id? Nuvber of nights 17b) Did the S1.00 per night refund influence you to camp MORE NIGHTS In Michigan State Parks or MAPCO campgrounds than you otherwise would have? Yes No NO -—-017c) If the REFUND had been S PER NIGHT, would you have taken advantage of the offer? No --> why not? Yes --> On your campvention trip would you have camped more nights in Michigan State Parks or MAPCO caapgroinds if you had received S off per night? Yes No why not? 18. Have you ever utilized money back offers or coupons for camping? No Yes --> How often? Seldom Occasionally Often I9. Have you utilized money back offers or coupons for other recreational ggtivities? r Yes -v19a) For what type of recreation activities have you taken advantage of a money back offer or coupon? 19b) How often did you utilize these offers? Seldom Occasionally Often NO 132 20. Do you or your spouse utilize money back offers or coupons when purchasing other products? No Yes -—h How often? Seldom Occasionally Often 21. we'd like to know how much your PARTY spent on your CAMPVENTION TRIP. It's important that you provide us with PARTY SPENDING, so please consult other persons who stayed with you at your caggvention sit . If your party did not spend anything in a particular category, write "0“ on that line. 00 not include cagggention REGISTRATION FEE. 21a) Total PARTY spending on your ENTIRE CAMPVENTION trip: ... ....... . ..... ........................ S 21b) PARTY spending while camping AT THE CAMPVENTION site: ..... ....... . ...... ............. ....... . S 21c) PARTY Spending in MICHIGAN (Not including spending while camping at the campvention site): ... S 21d) PARTY spending in OTHER STATES except Michigan (Include spending at home preparing for this trip): ......OOOOOOOOOO0.03.00.00.00... ..... 0.0.0.000.........OIOOOOOOO ..... O ......... s """"""'(21a should equal the total of 21b, 21c, and 21d)""""""' 22. Are you likely to camp in Michigan in the future? Yes No why not? 23. Do you currently subscribe any camping related magazines? No Yes -—~> Hame(s) of magazine(s): 24. Do you currently belong to any camping organization other than National Camper and Hiker Association (NCHA)? No Yes --o which one(s)? 25. Did you attend any camping or outdoor shows in 1987? Yes No THE REMAINING QUESTIONS ON YOURSELF ARE NEEDED SO THAT HE CAN GENERALIZE OUR FINDINGS TO ALL CAHPVENTION PARTICIPANTS. AGAIN, 8E ASSURED THAT THE INFORMATION UILL REMAIN STRICTLY CONFIDENTIAL. 26. In 1987, what was your household's annual gross income (before tax)? Under S 10,000 $20,000 to $26,999 $35,000 to $39,999 $50,000 to $59,999 $10,000 to $16,999 $25,000 to $29,999 $60,000 to $4£,999 $60,000 to $69,999 $15,000 to $19,999 $30,000 to $34,999 $45,000 to $49,999 $70,000'or more 27. What is the highest level of education you have completed? ____ Elementary (1-8) Some college ____ Advanced degree (M.S, Ph.D., M.D., 0.0.8. J.D) Some high school College degree High school diploma Some graduate school 133 APPENDIX C ..mcuauos mucuon monsoon updsoo no >000 ~30» suouu<. mason: oum>uua 00mdEmu ocwp3aozw 00: .m um303< 0cm vm 0:30 coozuun comical: cm puasmu hogan: Houoe ”30> OB quH .mm - sz:m mmmmoo< a ZOHBaEmo <=uz 020 am we .uocscu pouuumwuou. oEom 020 on um3E mququ moasmu manoeoopmu Ham :0 pco 000300 quu :0 05m: 029 u .vanniwoamv H: .uonu4 cc< .vmnn xom .0.m .oum pco mw3200u0 mwzu cu3uum .muaEuom mandamo u30> no >000 u30> 0002 i A.Uouquumn uoc mum muc3OUme 0H03oo .xuoa m 00 000600 30> con: muc300mdo cu>Hm 350:0 mum: 30> uuucz magma: new pa0> ma uc3oumap maze .00aucu>dEmu 020 um macaw: acaQEou mp3aocw no: moon. .mmma .m um303< uo anon: on» £030uzu vm 0:35 EOuu 0:30uOQEmu comacuH: oum>fiud mdqummwOAuumd uo xumm oumum cmoflcuwz >co ca QEmU u .mmu4 coflummuumm pcmacod: 0:... am .cojcmcasmo <=uz mama ecu pcmuu< u "am3z 30> muuqezo 09 .chAuuonm u30> mon 0H0; 00 >u0uuuuwa 0:30uUQEmu cmmwcowz oum 0H3mcoo .mcoaum>u0mou u3o> Eufiucou 0a 000m muflocu u30> mo mxuod 0:0 uumucoo .xumm spam on. meow—masons uo >u33m3m>m 05 0» uuonn3m m.“ .3qu «1:. .cho mmmq .m umzm:d m0 ucmnc 0.3 29.0.2.3 3.. 0:3... 50.5 000.50 mucmfic u0u 03cm ma uouuo maze .c0«uc0>05co any Ecuu 0:0 00 qm>ouu 30> mm pcoauopcoz.uoumz H3umu3moc many ducaaxu Cu >uwc3uu0020 many mxme .coquco>05mu 0:0 uo uumm o no mama ca comacuA: acquamw> u0~ 30> xzmzu on. >uac3uuommo menu umuuo 03 .pc30uOQEmu oum<2 oum>qu0 0004:0q: m:qummnufluumm >cm cm uo xumm mumum :mquuu: >3 5 059.50 >2 9:02. SE oo.3» u>> .0.0.n_.<.S_ .w mxm