.-..,.. 7.: H . «5 M ‘~'o.‘v __,.... . n. ,1 ash. 4: ‘fl 'p.‘-‘ :L‘... A "vb-‘4 .4 . ' KP). L4. ‘4'"?1'“ l - , ”l"-V _, 1' 'u" J‘pl w 0'." u» «'1’ .. ~ a4~ ‘ .14 .‘ .- v ug‘wflI-J .ll.pb ' a hbkhur. ‘_ q J, 3'. P. ' .1 hagq}- in“? .- 5dr. ",_ 115:? e": if . . ‘ «gr .0 L..I’ n97 ‘ 1; “$5,.- 7* pm ' . a" Lu—‘I : '5' ' MM'CHKS/AN/ S?€4‘/f7/UNW$R/S/IIY L/ngRARIES . lllllllllll t A 4—--. LIBRARY Michigaasmte University I \ -'———— Tllis is to certify that the thesis entitled ECONOMIC IMPACTS OF SPORT FISHING IN OTTAWA COUNTY: A STUDY OF THE LAKE MICHIGAN FISHERIES FROM OCTOBER 1981 TO OCTOBER 1982 presented by Scott William Jordan has been accepted towards fulfillment of the requirements for Master nf__Sci_ence_degree in Eisheriemnd Wildlife 3&9 974/1; ‘ Major professor Daniel Telhelm Date May 16, 1985 0-7639 MSU is an Affirmative Action/Equal Opportunity Institution 'P.,. RE rum HQ. rem: we: )V1531.} Place in book drop to LIBRARJES remove this Checkout from JI-IIQIUIIL. your record. [ENES will be charged if hook is returned after the date stamped below. ' “/ V.--.v.»~v~\a - l ‘ waft” 1' 39- 0139 “1:! ”WW; I 5:02 V 1... ECONOMIC IMPACTS OF SPORT FISHING IN OTTAWA COUNTY: A STUDY OF THE LAKE MICHIGAN FISHERIES FROM OCTOBER 1981 TO OCTOBER 1982 By Scott William Jordan A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Fisheries and Wildlife 1985 Dedicated to my wife Robin ii ACKNOWLEDGEMENTS This research was funded by the Federal Coastal Zone Management Program (administered by the Land Resource Programs Division of Michigan's Department of Natural Re— sources), the Board of Commissioners of Ottawa County, the Ottawa County Department of Social Services, the North West Ottawa County Chamber of Commerce, the Grand Haven Charter Boat Association, the West Michigan Marine Association, the Holland Fish and Game Club, the Holland Area Steelheaders, and the Holland Area Chamber of Commerce. Also cooperating were the Michigan State University Agricultural Experiment Station, the MSU Cooperative Extension Service, the Ottawa County Department of Social Services and the Michigan Sea Grant College Program. The following people were of particular assistance during the project. In the Holland area: Terry Hofmeyer, Holland City Manager; Ross Giles and Louis Hallacy, Holland Chamber of Commerce; John Dumez, Holland Fire Chief; and interviewers Mike Leisher and Ken Talkie. In the Grand Haven area: Greg Buckley, Grand Haven Assistant City Manager; Ed Lystra and Jack Smant, Grand Haven Chamber of Commerce; Ken Whitney, Michigan Charter Boat Association president; and interviewers Larry Smith and Pete Tullis. iii I am especially grateful to my graduate advisor and research supervisor, Dr. Daniel R. Talhelm. I have the highest regard for him, and his unfailing support of both me and my work has always been a source of encouragement for me. I would also like to thank both Dr. Donald Holecek and Dr. Niles Kevern for serving as members of my graduate committee. Special appreciation is extended to Charles Pistis, Sea Grant District Extension Marine Agent, who acted as a liai— son between Michigan State University and all local inter- ested parties. His efforts and support were instrumental in initiating and implementating this study. Most importantly, much of the credit for the completion of this thesis goes to my wife, Robin. I cannot even begin to express the measure in which she has supported, encour- aged and sacrificially given of herself so that I could accomplish this important milestone in my life. iv TABLE OF CONTENTS Page LIST OF TABLES........................................ vii LIST OF FIGURES....................................... xi CHAPTER I INTRODUCTION.................................. 1 II RECREATION ECONOMICS THEORY AND LITERATURE REVIEWOOOOOIOIOOOOO0.000.000...OOOOOOOOCOOOOOO 9 III METHODS...OOOOOOOOIIOOOOOOOOOOCOOO00.0.0000... 26 surveYSOOOOOOIOOCOOOOOOIOOOOOCOOQOOOO 33 Ice, pier and shore fishing inter- views and estimates of use........... 35 Boat fishing use..................... 41 Charter boat fishing................. 44 BUSiHeSS surveYOOOOOOOOO0.0.0.0000... 44 Iv ICE FISHINGOCOIOI00.0.00...OOOOOOOOOOOOOOOOOOO 47 HOIIand.IOOOOOIIOOOOOIIOOOOOOOOI0.... 49 Grand Haveno.OOOOOOOOOIOOOOOOOOOOOOOO 55 v PIER FISHINGOOOOOOOOOOOOOOOOOOOOO0.00.00.00.00 63 HellandOOOOOOIOOOOOIOIOOOOOOOOOOOOOOO 64 Grand HavenOOOOOOOOOOO0.000.000.0000. 73 VI BOAT FISHINGOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO 84 H011and0000000000OOOOOOOCOOOOOIOOOOO. 85 Grand HavenOOOOCOOOOOOOO0.0.0.0...0.. 92 VII GRAND HAVEN SHORE FISHING..................... 101 VIII GRAND HAVEN BAYOU BOAT FISHING................ 108 IX CHARTER FISHING............................... 117 Total expenditures calculations...... 120 X SECONDARY IMPACTS............................. 120 XI SUMMARY, DISCUSSION AND RECOMMENDATIONS....... 125 Page APPENDIX A SURVEY QUESTIONNAIRESOOOOOOOOOOOOOOOOOOOOOOIOO 138 LIST OF REFERENCESOOOOCOOOOOOOOOOOOOOOOOOOCOOOOOOOOOOO 143 vi 11. 12. 13. 14. 15. LIST OF TABLES Hypothetical estimate of daily use at a fiShing Site .0.IOOOOOOCOOOOOOOOOOOOOO0...... Holland ice anglers' average daily expenditures made at home, en route, and in Ottawa County. County expenditure statistics for Holland ice angling. Sample size = 46 (non-resident = 4). Holland ice angler comments.................... Holland ice anglers' household incomes......... Grand Haven ice anglers' average daily expenditures made at home, en route, and in Ottawa countYOOOOIO00......DOOOOOOOOOOOOOOIOO County expenditure statistics for Grand Haven ice angling. Sample size = 48 (non-resident a 11)........................................ Grand Haven ice angler comments................ Grand Haven ice anglers' household incomes..... Holland pier anglers' average daily expenditures made at home, en route, and in Ottawa countYOOIOOOOOOOOOOOOOOOOOOOIOOOOOOOO. County expenditure statistics for Holland pier angling. Sample size = 193 (non-resident 9)........................................ Holland pier angler comments................... Holland non-resident pier angler origins....... Holland non-resident pier angler accommOdationSOOIOOOI...OOOOOOIOOOI...0...... Holland non—resident pier angler family aetiVitieSOOOCOOIOOOOOOOOOCOOIOOOOIOOO0...... vii Page 40 55 57 58 59 62 66 67 69 7O 7O 7O Table Page 16. Means by which Holland pier anglers learned about fishing in the Holland area............ 72 17. Holland pier anglers' household incomes........ 73 18. Grand Haven pier anglers' average daily expenditures made at home, en route, and in Ottawa countYOOOOOOCOOOOOOOOOOOOOIOOOOOOOOOOO 75 19. County expenditure statistics for Grand Haven pier angling. Sample size = 681, (non-reSident=347)....00000000000.0.0.0.... 76 20. Grand Haven pier angler comments............... 77 21. Grand Haven non—resident pier angler origins... 79 22. Grand Haven non—resident pier angler accommOdationSOOOOIOOIOOOOOOOOOOOOOOOOOOOOOO. 79 23. Grand Haven non-resident pier angler family actiVitieSOOOOOOIOCCOOOOOOIOOOOOOOOOOOOOOO0.0 81 24. Means by which Grand Haven pier anglers learned about fishing in the Grand Haven area........ 82 25. Species Grand Haven pier anglers primarily fiShed forOOOOOOOOOOOOOICOOOCOOOOIOOOOOOO 82 26. Grand Haven pier anglers' household incomes.... 83 27. Holland boat anglers' average daily expenditures made at home, en route, and in Ottawa countYOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO 86 28. County expenditure statistics for Holland boat angling. Sample size = 217 (non-resident =48)....00000000CO...IOOOIOOOOCOOOOOOOOOIOO. 87 29. Holland boat angler comments................... 89 30. Holland non-resident boat angler origins....... 89 31. Holland boat anglers' household incomes........ 92 32. Grand Haven boat anglers' average daily expenditures made at home, en route, and in Ottawa County..................C...’........0 94 33. County expenditure statistics for Grand Haven boat angling. Sample size =184, (non-resident =100)..IOOOOOOOOOOOOOOOOOIOOOOOOIOOOOOOOOOOO 95 viii Table 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. Grand Haven boat angler comments............... Grand Haven non-resident boat angler origins... Grand Haven non-resident boat angler accommOdationSOOOOOOOOOOOOOOOOOOOOOOOOO0.0... Means by which Grand Haven boat anglers learned about fishing in the Grand Haven area........ Grand Haven boat anglers' household incomes.... Grand Haven shore anglers' average daily expenditures made at home, en route, and in Ottawa COUfltYsoooo0.00.0.00000000000000.0000. County expenditure statistics for Grand Haven shore angling. Sample size = 210, (non-resident = 69).......................... Grand Haven shore angler comments.............. Species shore anglers primarily fished for..... Means by which non—resident shore anglers learned about fishing in the Grand Haven area Grand Haven bayou boat anglers' average daily expenditures made at home, en route, and in Ottawa countYOOOOOOIOOOOOOOOIOOOOOOOCOOOICOOO County expenditure statistics for Grand Haven bayou boat angling. Sample size = 137, (non—resident = 87).......................... Grand Haven bayou boat anglers' comments....... Grand Haven non-resident bayou boat angler originSOOOOOOOOOOIOOOOOOOIOOOIOOOOOOOOOC.0... Grand Haven non—resident bayou boat angler accommOdationSOOOOOOOIOOOIIOOOOOOOOOOOO0.0... Species bayou boat anglers primarily fiShed forOOOOO0..OOOOOOOOOOOOOOOOOOOOO0.0.0. Means by which non-resident bayou boat anglers learned about fishing in the Grand Haven area Grand Haven non-resident charter anglers' average daily expenditures in Ottawa County.. ix Page 96 97 99 100 100 102 103 104 106 107 110 111 112 113 113 115 116 120 Table 52. Adjusted gross expenditures and direct net income from non—resident angler expenditures in Ottawa county.OOOOIOOIOOOOOOOOOIOO0.0.0... 53. Summary of angler use (angler days) and expenditures for all angling for Great Lakes fish, and related angling, in Ottawa County in 1981—82................................... Page 124 126 LIST OF FIGURES Figure Page 1. A hypothetical demand function with a supply curve for angling (BD) and an income— compensated demand curve (CF) assuming compensation begins at point C............. 15 2. Holland ice angler major in—state origins.... 53 3. Grand Haven ice angler major in-state originSOOCOOC0.0......0000000000COCOCOOOO0.0. 60 4. Holland pier angler major in—state origins... 71 5. Grand Haven pier angler major in-state originSOOOOOOOOOOOIOOOOOOOOOOOOO0.0.0.0...0.. 80 6. Holland boat angler major in~state origins... 90 7. Grand Haven boat angler major in-state origins...OOOIOOOOCOOOOOO...0.0.0...0.0.0.... 98 8. Grand Haven shore angler major in-state originSOOOOOIIOOOOOOOOOOOOOOOOOO0.0.0.0000... 105 9. Grand Haven bayou boat angler major in—state Origins.0.0.000000000IOOOCOODOIIOI00.0.0.0... 1]-4 10. Grand Haven charter angler major in-state OriginSOOOOOOO0.00000COOOOOOOOOO0.00.00.00.00 119 xi ABSTRACT ECONOMIC IMPACTS OF SPORT FISHING IN OTTAWA COUNTY: A STUDY OF THE LAKE MICHIGAN FISHERIES FROM OCTOBER 1981 TO OCTOBER 1982 By Scott William Jordan Great Lakes anglers in Ottawa County were personally interviewed to estimate the angling effort, associated spending and related economic and marketing information for ice, shore, pier, boat and charter fishing. Anglers spent an estimated 237,796 days and almost $4.6 million in Ottawa County angling for Great Lakes and nearby fish. Of this, non-resident anglers spent an estimated 101,931 days and almost $2.5 million, generating total Ottawa County sales of over $6 million. Shore and ice anglers were primarily lower-income local people, whose major concern was adequate access to the fish- ing. Pier and boat anglers were a composite of all income levels, and were concerned about facility conditions, unres— tricted commercial fishing, fish plantings and the lack of conveniently located bait and tackle stores. The results showed Ottawa County enjoys a high level of Great Lakes anglers' use and expenditures, and that there is potential for future improvements and growth. INTRODUCTION At the time of this study Michigan's manufacturing—based economy was in the throes of an economic recession, and the economic contribution of recreation—tourism industries in Michigan took on increasing significance for many locali- ties. While recreation and tourism dollars will probably never replace all the manufacturing jobs and income lost throughout the state during that recession, the economic problems of those years has continued to focus the attention of public officials and private citizens on the present and potential future contribution of Michigan's tourism resources. Great Lakes sport fishing has for many years been one of Michigan's major recreational pursuits and tourist attractions. All coastal counties offer attractive fishing. Anglers' expenditures vary, but the economies of many coastal communities depend heavily on this spending. In a prior study of the economic impacts of Great Lakes sport fishing in Alcona County, Michigan (Jordan and Talhelm, 1982), it was found that the local economy was substantially impacted by angler expenditures. Alcona County (population 10,000) is located on Lake Huron in the northeast corner of Michigan's Lower Peninsula. In that rural area the economic base was limited and fishing pressure was great. In Ottawa County and other subsequent studies in the more populous and industrialized areas of Muskegon, Bay and Macomb counties (Jordan and Talhelm, 1983, 1984a, 1984b), it was found that, whereas the total dollar impacts in some instances were several times greater than they were in Alcona County, they comprised a smaller percentage of the much larger overall economies found in those counties. The Alcona County study was initiated when local bus- inesses became concerned that local residents and government officials incorrectly perceived that Great Lakes sport fishing was of no benefit to Alcona's economy. The results of that study showed Great Lakes anglers spent over $1.3 million per year in Alcona County, and that those dollars were distributed over a wide spectrum of the local business community. From the results of the Alcona study, communi- ties there were able to document and address those issues and problems which were of particular concern both to area residents and the anglers themselves. Prior to the Alcona study in 1980-81 several research- ers conducted investigations of Great Lakes fishing expen— diture impacts in Michigan. One study by Fox (1970) was a statewide analysis of salmon and trout anglers in 1967, the second year (there was some fishing for immature "jack" sal- mon in 1966) that anglers were able to fish for the newly introduced coho salmon Onchorhynchus kisutch. Fox's sample was almost entirely of boat anglers fishing for coho salmon, 3 and he estimated they had mean expenditures of $931.00 for durable equipment purchased at home (boats, fishing and camping equipment) and mean trip expenditures of $13.00 per day. He made no attempt to estimate angler use, and there— fore did not calculate either statewide or regional gross expenditures. Talhelm (1973a) and Ellefson (1973) estimated that licensed Michigan residents spent $20 million fishing for salmon and steelhead in 1970. Those estimates were updated for inflation and included in the 1979 Status Report of Great Lakes Fishery Values for the Great Lakes Fishery Commission. Also, the 1975 National Hunting and Fishing Survey estimated angler expenditures, and to some extent are applicable on a regional basis to Michigan. Another regional Great Lakes study by Kapetsky and Ryckman (1973) investigated the economic impacts of the trout and salmon fishery from 1969-1972 on Grand Traverse Bay. They estimated that anglers spent close to $500,000 in the four counties around Grand Traverse Bay in 1972. Since their investigation in 1972, commercial gillnetting appar- ently has almost completely eliminated the lake trout fishery in Grand Traverse Bay, which in 1972 was the main- stay of the fishery there. In a study of the Grand Traverse Bay sport fisheries (Jordan and Talhelm, 1984c), current impacts were estimated at only $56,000, or one-tenth of the 1972 levels! If the lake trout fishery had not been decimated, gross expenditures would probably have been on the order of what was found in similar studies done in the adjacent counties of Benzie and Manistee (Jordan and Talhelm, 1984d, 1984e), where in 1983, nonresident impacts were from 40-50 times greater than those found in Grand Traverse County! Almost ten years had elapsed between the Grand Traverse and Alcona regional studies. During that time, the salmonid fisheries were providing excellent fishing opportunities all over Michigan. Therefore, it was not surprising that soon after the Alcona report spread throughout the state, other counties realized their need for similar information about their own Great Lakes fishing opportunities. When Muskegon and Ottawa counties expressed interest in studying their Great Lakes fisheries, it presented an excellent opportunity to analyze an area of the state much different from Alcona County. Ottawa County has a varied economy with many light to heavy manufacturing industries, 3 large farming community, and a well established tourism trade based on a variety of natural resource and cultural attractions. The county has been a leader in the state for promoting and encouraging the use of its Lake Michigan fisheries resources, and has a well developed infrastructure for handling the needs of tourists. While the two major communities, Holland and Grand Haven, in Ottawa County are fairly similar in the structure of their economies and general demographics, this study analyzed angling in Holland and Grand Haven separately because interest groups from each wanted results specific to their city. The fishing opportunities available in those two cities are much more varied than the stictly open—water salmonid fishery available in Alcona County. A winter ice fishery offers a variety of gamefish (walleye, Stizostedion vitreum; northern pike, Esox lucius; yellow perch, Perca flavescens; crappie, Pomoxis spp.; and bluegill, Lepomis macrochirus) on Lake Macatawa, the Pigeon River, and the Grand River bayous. Those same waters, all of which are connected to Lake Michigan, also offer warm-weather fishing opportunities for those same species and largemouth bass, Micropterus salmoides; smallmouth bass, Micropterus dolomieui; and catfish, Ictalurus spp.. On Lake Michigan, anglers fish for salmon, Oncorhynchus spp.; lake trout, Salvelinus namaycush; steelhead, Salmo gairdneri; brown trout, Salmo trutta; menominee, Prosogium cylindraceum; and yellow perch from boats, piers, and the shore. The primary goals of this investigation were to: 1) estimate the total number of angler days (an angler day is one person fishing any part of one day) spent fishing by anglers in each of the Great Lakes-associated fisheries in Ottawa County, 2) estimate the average daily expenditures and totals by both county resident and county non-resident anglers for each of the above fisheries in Holland and Grand Haven, and 3) solicit subjective responses from anglers as to their perceptions of the adequacy of both public and privately offered goods and services in the county, along with their overall impressions of the Great Lakes fishing opportunities available in Ottawa County. A one year study always presents the risk of sampling a time period which does not represent the norm. From conver- sations with local people and from actual experience through the interviewing process, it appears that overall, fishing success was below normal in the 1981—1982 fishing year. Ice fishing was not as good as expected in both Holland and Grand Haven, with the fishing being particularly abysmal in the Grand River bayous. In the past the Grand River bayous have provided outstanding winter fishing (Richey, 1978), which according to local reports drew thousands of anglers to the area. However, during the course of the 1981-1982 winter season only about 15% of the interviewed anglers came from outside Ottawa County, and even the major— ity of those came from adjacent counties. The fishing success in the Bayous simply never reached a sustained level that season which was attractive enough to draw many non- resident anglers. In the Holland area, anglers fishing on Lake Macatawa at times had good catches, but success was not consistent and the fish were generally small. However, at Port Sheldon, catches of yellow perch on the average were better than those on Lake Macatawa, both in size and number. On Lake Michigan that year the catch of spring steel- head and brown trout from the piers in both Holland and Grand Haven was very low, and as the summer progressed, the usually good perch fishing on the piers never materialized. Offshore salmonid fishing was fair in May, terrible in June, not quite fair in July and August, and because of an unex— plainable delay in the salmon run, was only fair in Septem- ber and the first part of October. The fall pier fishing for salmonids in Holland was particularly dismal because of the late runs. Not until late October did anglers began to consistently catch fish. Although the salmonid fisheries were not generally consistent, at least in Grand Haven the overall catch rate for the season was a respectable one fish per day for pier anglers and almost two fish per day for boat anglers. In addition to angler success being below par in the study year and having its effects on angler participation (especially non-residents), the fact that the study year was during the time of one of Michigan's worst recessions could also explain the low angler participation rates. In a study of changes in leisure activities among the general popula— tion of Greater New Orleans, Louisiana during the 1974 re- cession, Wagner and Donahue (1976) found that significantly less time was spent in leisure activities away from the home than in previous years for households earning less than $13,000. It could be that for many anglers who used to travel to Ottawa County to fish, that either inflation or job lay-offs reduced their income to the point where extra— local fishing trips were unaffordable. Despite the below normal catch and use rates, Ottawa County enjoyed one of the highest levels of use and expendi— tures among the Michigan counties recently investigated (Jordan and Talhelm, 1982, 1983, 1984a, 1984b, 1984c, 1984d, and 1984e). For the entire study period, Great Lakes and associated anglers spent an estimated 238,000 days fishing and $4.5 million in Ottawa County, of which 102,000 days and $2.4 million was attributable to non-resident anglers. Those estimates are apportioned by fishery and city in the different sections of this report. RECREATION ECONOMICS THEORY AND LITERATURE REVIEW The focus of this investigation was to assess the gross expenditures of Great Lakes and related angling in Ottawa County. The focus was determined by the coastal communi- ties' (who in part funded this investigation) desire to know the impacts anglers' expenditures have in terms of county income and jobs. Their viewpoint was extremely pragmatic in that they wanted to know what the Great Lakes fisheries rneant to them, and they had no real concern for the value of tflie fishing resources t0“SOCIety as a whole. Therefore, the ruesults of this investigation do not reflect any analysis of tflie valué of the Great Lake fishery resource in Ottawa (hiunty, but only the impacts of anglers' expenditures in Pursuing that Resource. Although few recent studies have focused on angler ex- Peruiitures, past investigations have explored some of the in‘Plications of recreationists' expenditures on local econo— “1198. In a study of cottage developments, institutional (““nPS and public parks at Pigeon Lake, Alberta, Canada, BOhlin and Ironside (1976) measured the spatial distribution 0f Capital and trip expenditures by recreationists. Their investigation centered around three hypotheses: 1) that the 10 major part of the economic impacts attributable to a recreation resource will accrue to the regions where recrea- tionists originate and not to the destination area, 2) that there are major differences in money flows with respect to different recreational pursuits and 3) that the direct effects of recreational expenditures on the economy of the destination area will be negligible in terms of the area's total economy, if any. It was interesting to note that the first and third hypotheses were the general impressions residents and officials in Alcona County had expressed in the study there (Jordan and Talhelm, 1982). The results of the Pigeon Lake study showed that for: l) cottage users, 26% of their capital expenditures and 42% of their trip expenditures were made in the local economy, 2) park users, 45% of their trip expenditures were made in the local economy and 3) camp users, 37% of their capital expenditures and 35% of their trip expenditures were made in the local economy. The authors' accepted their hypotheses because in all instances less than half of recreationists' expenditures were made in the destination area. In a study of a state park in New Hampshire, Frick and Ching (1970) found that the local income generated by 125,000 park users was equivalent to that which would be expected from 12 permanent resident families, and also concluded that local income generated by at least park user expenditures is nominal. Other studies have estimated employment impacts ranging from 50 jobs for a state park in Tennessee (Dean, 11 et. al., 1978) to 350 jobs for a TVA reservoir in Tennessee (Garrison, 1974), while also stressing the levels were nominal. Although those studies showed that perhaps in many in- stances the larger share of recreational expenditure impacts go to origin rather than destination economies, the fact still remains that the portion which destination economies do receive may be vital to their economic well—being despite the smaller share and leakages to the outside. Garrison (1974) estimated that direct, indirect and induced employ- ment resulting from recreation expenditures accounted for 5% of the total private nonagricultural employment in the Norris Lake area of Tennessee. While he thought that was insignificant, one could imagine the response if a major city like Detroit were faced with the loss of 5% of its employment base. Therefore, it was not surprising to hear of a change in attitude among businesses and residents in Alcona County after release of that study's results; that sport fishing expenditures are important. The loss of even 12 local jobs would be of vital concern to a community the size of Harrisville in Alcona County, and cities like Grand Haven and Holland in Ottawa County would certainly not be apathetic about the loss of as many as 350 jobs. Although the value of the resource was not a focus of this investigation, a review of some of the theory in the literature is appropriate. The reason is there is a notion held by many non-economists, and especially many of the 12 parties encountered in the course of the Ottawa investiga- tion, that gross expenditures is an acceptable measure of the value of a recreational resource. They would argue that expenditures, in this case made by anglers, must represent at least the minimum value anglers place on the activity of fishing, or anglers would not have made them. In other words, they propose that the value of an angling day is worth at least what an angler spends per day for the experience of fishing. Gross expenditures and analyses of the associated multipliers may be useful from the standpoint of indicating the levels of income and job impacts to a community or region. They also determine the cost of "production" of an angling day. However, they are not a good determinant of the worth or benefit of angling. The reason they are not is because they represent the cost of "producing" the fishing experience. Expenditures made in "producing" the angling experience are not a payment for either the resource or for angling rights to the resource. At least that is the case for most North American fisheries. Many European fisheries, however, are a good example of where a payment is made for the angling rights to the resource. In those fisheries, anglers not only have to pay the costs of "producing" the angling experience, but they must in addition pay a fee for the right to fish on someone's property. That fee is a partial measure (because owners usually charge the same fee to everyone for a particular stretch of water, and cannot or 13 do not price-discriminate among users) of the value of the resource. Angler expenditures (costs) should be viewed as the supply function for the angling experience. The supply of angling is the schedule of the minimum prices (expenditures and opportunity cost of time) at which each given quantity of angling is available. The minimum price for any partic- ular angler will always be constant, or perfectly elastic (horizontal supply schedule) because the angler can con- ceiveably "produce" as little or as many visits to a fishing site as he wants all at the same average cost (Talhelm, 1984). However, because anglers both produce (supply) and con- sume (demand) the fishing experience, and because their ex— penditures do represent a part of their total demand or willingness to pay for the angling experience, it is under- standable that many people make the mistake of using gross expenditures as a measure of the value of the experience or of the resource. Another problem of viewing gross expendi- tures as the value of the resource, is it then implies that the farther recreation areas are located from population centers, the greater the benefits (Smith and Kavanagh, 1969). Such an implication is completely juxtaposed to what would truly maximize benefits - having the resource closer to population centers. Economists define the actual value of the resource as either the marginal net willingness to pay for or the 14 marginal net willingness to sell the recreational resource. If anglers have no ownership or rights to the resource, then willingness to pay, or what anglers would give to use the resource, would be the proper definition. If anglers do have ownership or rights to the resource, then willingness to sell, or what anglers would have to be compensated to give up use of the resource, would be the proper perspec- tive. Most empirical work in recreation economics has used the concept of willingness to pay to estimate recreation demand. In Figure 1 total willingness to pay or the total value of angling would be defined by the area (ACDE) under the demand curve for a recreational resource at some level of use (AE units). That value includes users' expenditures, ABDE, which again represents the cost of "producing" AE number of angler days. The value of the angling resource equals the total value of the angling (ACDE) minus the cost of angling (ABDE), or area BCD (Talhelm, 1984). Area BCD is anglers' marginal willingness to pay over and above their actual expenditures for AE days of fishing, and is the ap- propriate measure of the value of the resource. Dwyer, et. al., (1977) points out that area BCD is actually an approximation of the net willingness to pay, for if the full price for each unit demanded could be collected from each consumer (assuming we start at point C or with the first consumer who enters the market), there would be some effect on consumer incomes which would cause demand to shift PRICE 15 Demand Function Figure 1. USE A hypothetical demand function with a supply curve for angling (BD) and an income—compen- sated demand curve (CF) assuming compensation begins at point C. 16 to the left. This is known as the "income effect" and CF would be defined as the income—compensated (or Hicksian- compensated after Hicks, 1943) demand curve, making BCG the better approximation of the value of the resource. However, for almost all instances of evaluating recreation opportuni- ties, it is assumed that collecting the full willingness to pay for each unit will not raise expenditures (reduce in- come) enough to cause a shift in the demand curve. In fact, Bowes and Loomis (1980) have theoretically shown there is an exact relationship between the consumer surplus estimated using the travel-cost methodology (TCM) and the Hicksian compensated demand function for a site, or in other words, consumer surplus measured by the TCM is equivalent to that measured using entry prices. Two methodologies, the travel-cost method (TCM) and the survey method or contingent-valuation method (CVM), have been widely used to assess recreationists' willingness to pay. The TCM estimates recreationists' demand for a re— source by observing their actual behavior (travel) in util— izing the resource, and the CVM estimates recreationists' demand for a resource through a carefully constructed bidding-game survey. Both methodologies have their respec— tive strengths and weaknesses. The TCM was first suggested by Hotelling (1949) and later popularized by Clawson and Knetsch (1966). The TCM has had widespread empirical application by a host of l7 researchers, and has undergone perhaps the most scrutiny and modification. The TCM was originally developed for anal— yzing single site resources (Clawson, 1959; Knetsch, 1964; Clawson and Knetsch, 1966; Merewitz, 1966; Weithman and Haas, 1982; and Palm and Malvestuto, 1983). However, much of the work with the TCM has been to adapt it to multi-site analyses and site quality—change analyses (Brown et.al., 1964; Stevens, 1966; Burt and Brewer, 1971; Talhelm, 19733, 1976; Cicchetti et.al., 1976; Sutherland, 1982; and Vaughan and Russell, 1982). There have been many criticisms of the TCM and many modifications and improvements have been made to rectify the model's shortcomings. As was mentioned, the simple TCM estimates the value of a site based on the demand for the site as a function of travel costs and selected demographic variables (e.g., income). Some of the assumptions generally used in the TCM are: 1) recreationists travel solely for the pleasure of traveling and the only purpose of the trip is to visit the specified site, which means their travel expendi- tures accrue completely to the destination site, and 2) the prices of substitute recreational opportunities to the site are independent of the travel costs to the site, and that the availability of alternative recreational sites does not affect the demand for the site under investigation. Some researchers (Knetsch and Cesario, 1976; Mendelsohn and Brown, 1983) include only the vehicular costs of travel and the opportunity cost of travel time in their TCM l8 analyses, arguing that costs for food, lodging and equipment either enroute or on—site should not be included because the use of those items may have utility values to the recrea- tionist separate from that of the resource. They would include those costs only if there is reason to believe that the marginal utility of those inputs is zero. They also argue that costs incurred at the site (including time) should not be included as travel costs, because they are not related to the marginal cost of obtaining the resource - providing the resource is the focus of the valuation and not specific activities. The point they make is that the 1313; of either enroute or on-site expenditures both in money and time for items such as food and accommodations are inputs in the production of either satisfying meals or a relaxing environment. Therefore, they believe the value of those activities is separate from that of the resource, and that only resource-specific prices of non-travel inputs should play a role in the demand for a resource. The problem with that approach is that it is difficult to determine specifi— cally the prices for all the non-travel substitutes which could be used in estimating the demand for a resource. Talhelm (1984), on the other hand, avoids dfferences in specifying all such substitute prices by including all those enroute and on-site expenditures in the cost of "producing" the angling experience. He argues that if they are not included as part of the price in estimating the demand for the resource under study, then they should also not be 19 included as part of the prices of any alternatives considered. In studies that estimate the demand for one site, the problem arises in that as distance traveled increases for a recreationist, there is the likelihood that other sites are being utilized en route. If multiple purposes for an indi- vidual's trip can be specified, then travel costs should be allocated accordingly (Talhelm, 1973a, 1981, 1984; Haspel and Johnston, 1982). This will help avoid an overestimate of the value of the site under study. There is also the problem of determining at what rate to cost travel time. If traveling time accounted for as much disutility as working time, then travel time should be valued at the origin-specific net wage rate. However, Nelson (1977) has shown that even commuting travel time has some utility. Assuming that recreational travel time has even more utility, it seems reasonable that travel time should be costed at less than the wage rate, which Nelson (1977) suggests it should be at about one—third the wage rate. Researchers have critcized the TCM for: 1) not taking into account substitutes, 2) being an all—or—none evaluation incapable of estimating marginal changes in the resource, and 3) not being able to assess changes in the quality of a site. Burt and Brewer (1971) and Talhelm (1972, 1973a, 1973b) made some of the initial modifications of the TCM by 20 developing systems of demand schedules which recognized the substitution effects among heterogeneous recreation sites. In a study of three proposed lakes to be developed by the Army Corps of Engineers in Missouri, Burt and Brewer (1971) developed six classes of water-based recreation sites and estimated demand curves for each class using prices for the other classes as substitute prices in each demand equa- tion. Although they addressed the problem of substitutes, they did not make clear qualitative distinctions between their six classes of sites. Basically there were two unique lake groups (Table Rock and Lake of the Ozarks), a "typical" Corps lakes group, an Ozark Mountain's rivers group and all other lakes in two size categories. One recent study which addresses site quality or char- acteristics was done by Vaughan and Russell (1982) of fee- fishing sites in the U. S. using a varying parameter model (Maddala, 1977). Their hypothesis was that anglers value some species more than others, and therefore they used major species class as the most distinctive site attribute, breaking out demand into two separate equations - one for trout and one for catfish/roughfish. Within each species equation, they then used as many as 40 site characteristics (some specified from empirical work done by Holman and Bennett, 1973), including catch rates, as explanatory vari- ables. They did include substitute sites in their analysis, but not directly, as Burt and Brewer did. They included them by proxy as dummy variables, reflecting the extent of 21 competition from other sites perceived by the fee-site owner. Talhelm (1972, 1973a, 1973b, and 1976) has developed an approach which takes into account both site quality and substitutive effects between different qualities of recrea— tion. He and others using his techniques (Stanford, et al., 1982; Victor, et al., 1983; and Korson, 1979) use a behav- ioral model to partition sites or geographical units (coun- ties) into having one of a variety of recreational experi- ences, defined as "products", according to exhibited site characteristics. Each "product" of recreation represents a different quality of recreation analagous to the levels of quality found with any marketable commodity, such as compu- ters or clothes. Talhelm's behavioral model uses a discrim- nant analysis to select the "best" set of products, the "best" set defining how recreational participants most likely perceive quality differences within a recreational category. Having selected a set of recreational products, Talhelm's methodology then uses the TCM to derive a demand function for each product, using alternative products and some other closely related recreational categories as sub— stitutes in any particular demand function. One additional variation of the TCM is the hedonic method (Brown and Mendelsohn, 1980; and Mendelsohn and Roberts, 1982). The hedonic approach estimates the demand for different site characteristics, where each site is 22 analyzed as a set of characteristics. Characteristics refer to such things as size of stream, density of fish, vegeta- tive cover types, etc. In the first of two stages, the hedonic approach regresses travel cost as a function of a set of characteristics for each origin. In a second stage calculation, it then uses the estimated price for a marginal unit of a specified characteristic from the first stage calculation as an independent variable, along with a vector of demand shift variables to estimate the demand for that characteristic. The management implication is that the most valuable characteristics of a site or group of sites can be determined, and those characteristics either be preserved or developed accordingly. The other major approach in determining willingness to pay is that of the survey or contingent valuation method (CVM). The CVM is often refered to as the "bidding game" approach. In it a respondent, through a survey instrument, is placed in a set of hypothetical situations and asked to respond to a series of questions that elicits bids from him to either: I) obtain the hypothetical situation if it repre- sents an improvement in his utility level, or 2) prevent the hypothetical situation if it represents a deterioration in his utility level. The bidding game contains three elements: 1) an instru- ment by which a respondent's bid is placed in a realistic institutional context of payment; 2) a starting point at which the bidding process is begun; and 3) a set of 23 information that establishes the context of the hypothetical situation in which the respondent is to formulate his bids (Brookshire et al., 1976, 1978). For an individual respon- dent, the bids solicited for a particular public good simply represent that individual's indifference curve for that public good, the indifference curve being a locus of points for a given income level where for varying combinations of goods the individual's total level of utility remains the same. The measure of consumer surplus for the public good is then the aggregate bid curve obtained by algebraically summming the individual bids of the relevant population after subtracting all the expenses associated with the public good experience (Randall et.al, 1975). One critcism of the CVM is the possibility for respon- dents to get into a gaming strategy with the interviewer. While many researchers believe the problem exists, particu- larly in the case of public goods, at least one study by Bohm (1972) of the demand for public television broadcasting found little respondent gaming bias, suggesting, perhaps, that the problem is not as significant as believed. The concern with gaming-strategy bias arises from the assumption that respondents may understate their preference for the good, in hopes that they may escape being charged as much as they are actually willing to pay for the amount of the good they actually desire. Conversely, researchers also assume respondents may bid up their apparent willingness to pay if they feel it may help preserve the good in its present 24 state. Some researchers feel the challenge then is to phrase questions so that the respondent is not put in a position of considering his opinions about the propriety of charging for the use of the good (Knetsch and Davis, 1966). Some suggestions by Miller et al., (1977) to help prevent respondent gaming strategy are: 1. The less hypothetical the question, the more stable and reliable the response. 2. Questions should be asked while the respondent is engaged in the activity under investigation, to prevent him from having to project himself into hypothetical situations. 3. Consider only one change in conditions at a time. 4. Formulate questions so as to remove opinion. 5. Use test items in the instrument similar to those in the actual situation. 6. Make situations concrete rather than symbolic. A recent study by Bishop et al., (1983) has helped to show that many of the problems with the CVM stems from its artificial context - people answering CVM questions do not have well developed ideas about how they would actually act in a real market for the good under investigation. BishOp et al., (1983) evaluated Wisconsin permit goose hunting with several contingent valuation mechanisms along with a stan- dard TCM and a "simulated market", where permit holders were offered real money not to hunt (willingness-to-accept compensation or willingness-to-sell (WTS), as opposed to willingness—to-pay (WTP)). Their results showed that WTP 25 estimated by CVM could be in error (underestimate) by 50 percent or more. Their results also showed some other interesting relationships. First, as Dwyer et al., (1977) pointed out, estimates of WTS are always greater than estimates of WTP. The question is, where does the actual level of consumer surplus lie between the two measures. Although their "simulated market" was a WTS proposition, they found the average value taken as compensation for a permit ($63) was almost equally spaced between the high CVM-WTS value ($101) and the lower TCM-WTP value ($32). The CVM-WTP value they estimated was even lower ($21) than their estimated TCM-WTP. Again the authors felt the low CVM-WTP is due primarily to the artificiality of the CVM framework, because when people are unclear as to how to act in a hypothetical market, oftentimes their response is no response, or they express zero WTP. The authors noted this was borne out experien- tially in their study, where they observed respondents giving much more careful consideration in responding to actual money offers in the "simulated market" than to hypothetical offers in the CVM. Non-market goods evaluation has certainly progressed since the first TCM studies, but even as Bishop et al.'s (1983) study shows, there is still considerable room for minimizing the errors associated with all methodologies. METHODS As previously mentioned, the major goals of this study were to estimate Great Lake angler use and expenditures in Ottawa County in order to derive the total gross expendi- tures in the county by those anglers. In deriving total gross expenditures, however, some considerations must be made in describing or defining the population to be sampled. The findings of a recreation economic investigation would be suspect if: 1) it fails to either define or adequately de— scribe the population being sampled, and 2) the sampling scheme chosen has not adequately considered the bounds of the population under study. The defined population for this study was all angler use within a year's time at all designated fishing sites within Ottawa County. Therefore, by definition the popu- lation encompassed all anglers (men, women, children, li- censed and unlicensed), and their use had to be at specific sites and within a specific time frame. With the population so defined, there were implications to consider in selecting a sampling technique. In general, there are three techniques which have been used in socioeconomic research: personal interviews, mail surveys and telephone interviews. In determining the 26 27 appropriate technique, the primary factors to consider are: 1) the spatial distribution of the defined population, 2) the size and representation of the pOpulation and 3) the cost of obtaining the sample. If the spatial distribution of the defined population is small and the population is well represented within the geographic area under study, then personal interviews would be the prefered technique. Personal interviews provide the most reliable contact with a respondent because: 1) there is a greater control over the response in terms of identifying and eliminating respondent biases, 2) there is a higher response rate and 3) a more in-depth interview can be given (Kerlinger, 1973, Sellitz, et a1, 1962). Another factor to consider in choosing the personal interview is whether they will be conducted in households or on—site. Household personal interviews can be very costly. They can cost as much as ten times more per response than a mail interview and five times more than a telephone inter- view (Wellman et al., 1980). Lansing and Morgan (1971) noted that household personal surveys can range upwards from $25 per completed interview. On-site personal interviews are usually less costly, simply because the sample popula- tion is concentrated. By definition the population for this study was spa— tially confined and was well represented; therefore, on-site personal interviews were feasible and was the survey instrument chosen. Cost was a consideration in this study. 28 The average cost per personal interview turned out to be less than $10. However, the chief reason for this was that more than one-half of the interviewer hours for the study came from local people who either volunteered their time or were involved in Michigan's Workfare Program. As the spatial distribution of the population increases and/or the density of the population decreases, then mail or telephone surveys become the instruments of choice, primar- ily because of their cost and time efficencies. While the mail survey is oftentimes the survey instrument of choice in fisheries economics research because of population spatial and density considerations (it is the instrument regularly used by MDNR's Fisheries Division for evaluations of use and impacts), there are many reasons why, aside from the fact it was practical to do personal interviews on-site, it was felt a mail survey was not appropriate for this study. They are also reasons which should make angler use and expenditure estimates based on mail surveys suspect, especially when the estimates are based on a sample from a large geographic re— gion and are then applied to specific locations. Mail survey questionnaires are generally sent to li— censed anglers, which in Michigan eliminates most women and all anglers under age 16 from being included in the sample. Also not included are illegal nonlicensed anglers. This could severely bias a sample, especially if nonlicensed anglers or their use comprise a large proportion of the population under study. 29 Dunning and Hadley (1978) found that in Erie County, New York two-thirds of the angling population was comprised of nonlicensed anglers. Anglers 9 to 15 years of age made up approximately 42% of the angling population and accounted for over 55% of the total angling days. Illegal nonlicensed anglers accounted for 25% of the angling population. Martin (1977) estimated from U.S. Fish and Wildlife Service data that on the average about 40% of each state's anglers were nonlicensed in fiscal year 1975. Jamsen (1985) reports that the MDNR Fisheries Division investigated the percentage of non-licensed anglers in Michigan in 1973, 1976 and 1983. The Fisheries Division found that almost 50% of all anglers in Michigan are non- licensed. Non-licensed spouses account for 40% of the non-licensed total and anglers under 17 years of age for 60%. However, they did find that approximately 90% of anglers fishing Lake Michigan were licensed. Jamsen (1985) did say the 90% figure was probably biased towards repre- senting salmonid anglers, and likely did not take into ac- count much of the summer perch fishing from Lake Michigan piers. The point is that if mail surveys sent to licensed Michigan anglers do not sample most women, all children under age 17 and illegal nonlicensed anglers, then all those nonlicensed anglers should not be viewed as part of the population under study. However, when angler use is estimated by direct observation, with no distinction made 30 between licensed and nonlicensed anglers, the possible result is an overestimate of gross expenditures. That is because the anglers who generally spend the least (women, children and those who feel they cannot afford a license) are not sent a mail survey and consequently are excluded from the average expenditures calculations. However, if the average expenditures of licensed anglers are then expanded by estimates of angler use which included the unlicensed anglers, then the result may be an overestimate of impacts. Other problems with mail surveys are l) respondents' inability to recall details of past fishing trips and 2) their desire to be "helpful" in the answering process. Most mail surveys of angler expenditures ask the respondent to give an accounting of expenditures on the angler's most recent fishing trip. However, as the length of time since the last trip increases, the accuracy of the accounting suffers. In addition, there is the tendency to recall the most successful, memorable or expensive trip, which may not have been the most recent trip. With the personal on—site interview it was relatively easy to maintain the bounds of the defined population by asking for expenses incured on the current trip. With mail surveys, it is very easy for individual respondents to violate the bounds of the intended population defined by the investigator. One problem with the personal interview at this point, is that if the interview is conducted prior to the conclusion of the angler's trip, there is the possibility 31 the angler may not accurately assess what expenses he may have after the interview. The possible bias could go in either direction, and it is assumed it cancels out. All anglers in some measure feel that fishing is a worthwhile pursuit, and there is a general bias in their thinking that fishing is positively "valuable" to society. The prevalence of that idea was constantly encountered in conducting interviews for this study, where anglers were always commenting that they spend large amounts of money on their fishing. What they often meant was that they have spent what seems to them a large sum of money on fishing, and that over a number of trips. Even in the personal interview setting, anglers' would: 1) try to include the value of all their equipment in current trip expenditures, 2) have the interviewer document some past expensive trip, or 3) try to give an average per trip for their fishing expenses over a year's time. The point is, that if many of them tried to manipulate the question in the presence of an interviewer who repeatedly had to bring their responses into the constraints of the question, one can imagine the liberty many anglers take in answering such questions in the privacy of their homes. When held within the constraints of prepar- ing for and participating in the trip they were on at the time of the interview, interviewers found anglers usually spent considerably less than what, and with good intentions, they would like to have convinced the interviewer they had spent. 32 Telephone interviews present some advantages over mail surveys while at the same time having their own sampling problems. Telephone interviews allow for more control over respondent biases and for more in-depth interviews than mail surveys, although still not to the degree attainable with a personal interview. They also have a higher completion or return rate than do mail surveys, although while one-mailing surveys have a rather low average response rate of 48% (Heberlein and Baumgartner, 1978), many mail surveys of recreationists have attained response rates from 70% to 96% using intensive follow-up mailings (e.g., Burch and Wenger, 1967; Lucas and Oltman, 1971; Kennedy, 1974 and Kanuk and Berenson, 1975). Some of the problems associated with telephone inter- views are: 1) the inability to contact respondents who have unlisted numbers or privacy listings and potential respon- dents without phones, 2) a female response bias due in part to the timing of calls, 3) an inability to maintain a sample representative of the population under study and 4) an inability to gain respondent rapport. It should be noted that in a study done by Field (1973) a telephone sampling design and instrument was developed which alleviated many of those problems and which should be considered as an alternative to a mail survey when a sample is desired from a large heterogenous population within a large geographic region. 33 Again, the personal interview was chosen because: 1) the spatial distribution of the defined population was small, 2) the defined population was well represented in the geographic area of study, 3) it was at least as cost-effec- tive as either a mail or telephone survey would have been, 4) it afforded the greatest control over respondent biases and 5) was not selective of particular subgroups within the defined population. Surveys A total of 2,059 angler interviews were made at Lake Michigan and connecting waterways fishing access points within Ottawa County. Access points were either: 1) ob— served to have angling usage or 2) were pointed out by local people to be areas of fishing activity. In the Holland area angler use was sampled on: (1) Lake Macatawa, (2) the Lake Michigan north pier, (3) Lake Michi— gan (offshore salmonids) off the entrance to Lake Macatawa, (4) the Pigeon River near Port Sheldon, and (5) Lake Michi- gan (offshore salmonids) off the mouth of the Pigeon River. In the Grand Haven area angler use was sampled on: (1) the Grand River bayous, (2) the Lake Michigan piers, and (3) Lake Michigan (offshore salmonid) off the mouth of the Grand River. Interviews were conducted by me, undergraduate students hired at MSU, Holland Fire Department personnel, City of 34 Grand Haven personnel and other Ottawa County residents who were either volunteers or assigned to the project through the Michigan Department of Social Services' "Workfare" Pro- gram. All of the interviewers had at least a two-hour "classroom" training session in the proper administration of the questionnaire, along with semi-monthly or monthly checks in the field to assure that the proper regimen and delivery of the interview was maintained. In the personal interviews, anglers were questioned about their trip expenditures, their fishing success, where they were from, their length of stay, where they were staying, their impressions of the fishing in that sample area, whether they had reasons for their trip other than fishing and personal information. An example of the questionnaire is in Appendix A. Anglers' mean expenditures for various categories of goods and services are tabulated in the results for each fishery. Following the table of means in each fishery is a table of sample statistics. In those tables statistics for all anglers are listed first, followed by non-resident angler statistics in parenthesis. Listed are the sample mean, the standard deviation, the standard error of the mean, the 95% confidence interval of the mean and the mea- sure of skewness of the distribution. Since a large proportion of anglers in any particular fishery do not purchase a specific good or service within the time constraints of one trip, most categories of goods 35 and services in the samples have many observations of zero expenditures. This causes strongly positive estimates of skewness, meaning the frequency curve for most categories of expenditures is asymmetric to the right. The data can be transformed (for a large number of observations of zero, a log transformation is usually appropriate) to reduce the skewness. However, transformtions were not attempted nor were any nonparametric tests performed, as the interest was not in approximating a normal distribution or explaining variable variance, but it was in determining the actual sample means. Therefore, most of the sample statistics show large measures of skewness and variability in anglers' expenditures. Ice, pier and shore fishing interviews and estimates of use Ice, pier, and shore fishing use was estimated using a survey method developed by Talhelm (1972). This method is similar to that of other studies using stratified random sampling and roving creel surveys (Hayne, 1966 and 1972; and Malvestuto, Davies and Shelton, 1978). The technique con- sisted of systematic traverses of either: (1) sections of shoreline, (2) a pier, or (3) a concentration of ice anglers. In using Talhelm's method the following assumptions are made: 1) All anglers either along a stream or lake shore segment or a concentration of ice anglers could be 36 interviewed at any selected point in time on a sample day. 2. All anglers know how long they would fish in the sample area on the sample day. 3. The composition of the angler population in any sample area did not change significantly, in terms of many anglers coming and going, over the time period of one traverse by an interviewer of a stream or lakeshore segment or concentration of ice anglers. 4. Anglers are distributed throughout the day in a random pattern and arrive and depart from the sample area at random. A sample area (stream or lakeshore segment or concen- tration of ice anglers) was sampled at several points in time on a sample day. For each point in time, an inter- viewer would walk the length of the sample area and inter- view all anglers encountered. That was defined as a tra— verse. Traverses were never more than two hours in dura- tion. If there were more anglers in the sample area than could be interviewed in a traverse, the interviewer would do an "instantaneous" count, divide the count by ten and use the resulting quotient to determine how many anglers to skip between interviews. The interviewer would not attempt to interview anglers who left the area "ahead" of him or anglers who would come in "behind" him as he progressed with 37 his traverse. A sample area would have two or three traverses done on any sample day, and if the interviewer en- countered anglers more than once within an area on a sample day, they were not re—interviewed on subsequent traverses. The probability of encountering any particular angler was proportional to: 1) his length of stay in the sample area, 2) the number of traverses that day of the sample area, and 3) the length of the sample day. Expressed mathe- matically, the probability of encountering anglers who said, for example, they were going to fish for three hours during a twelve hour fishing day in which three equally spaced traverses were made is: mi ' hi: _ 3 - 3 3 where, mi = number of traverses on sample day i, hij = total fishing hours on sample day i by interviewed angler j, x- = length in hours of sample day i. Therefore, for every three 3-hour anglers encountered on sample day i, one was theoretically missed and should be included in the estimate. The total number of angler days at a sample area on sample day i was estimated by equations (A) and (B): 38 Vi x 0 Ci = 2:_ .___;L___ (A) j=l mi ' hij if for a particular angler j, mi - hij