UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Subject variability in estimates of impact saliency and impact acceptability for wilderness conditions Rutledge, Ronald Brent 1995

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1996-091570.pdf [ 6.81MB ]
Metadata
JSON: 831-1.0075243.json
JSON-LD: 831-1.0075243-ld.json
RDF/XML (Pretty): 831-1.0075243-rdf.xml
RDF/JSON: 831-1.0075243-rdf.json
Turtle: 831-1.0075243-turtle.txt
N-Triples: 831-1.0075243-rdf-ntriples.txt
Original Record: 831-1.0075243-source.json
Full Text
831-1.0075243-fulltext.txt
Citation
831-1.0075243.ris

Full Text

SUBJECT VARIABILITY IN ESTIMATES OF IMPACT SALIENCY AND IMPACT ACCEPTABILITY FOR WILDERNESS CONDITIONS by RONALD BRENT RUTLEDGE B.B.A. Texas Tech University, 1971 M.Sc. Texas Tech University, 1973 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in FACULTY OF GRADUATE STUDIES (School of Forestry) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA December, 1995 © Ronald Brent Rutledge, 1995 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia Vancouver, Canada DE-6 (2/88) ABSTRACT This study examined sources of subject variation in user estimates of impact saliency and the acceptability of impacts on ecological and social conditions found in wilderness. Data collected in a questionnaire sent to visitors to the Spruce Lake Trails Area in British Columbia, Canada were used to test six research hypotheses. The influence of impacts on visitors' wilderness experiences were compared among three sub-groups of the study sample characterized by: 1) how their trip was organized (commercially outfitted vs. privately outfitted); 2) their length of stay (one day vs. two to four days vs. five or more days); and 3) their place of origin (BC residents vs. US residents), respectively. Variations in impact acceptability levels were also compared for the above three sub-groups. The results of hypotheses testing indicated that, overall, the three variables examined (trip organization, length of stay and place of origin) offered little explanation for subject variation in impact saliency and impact acceptability by respondents to the Spruce Lake Trails Area Visitor Study. By and large, survey respondents agreed on the relative influence of most of the impacts evaluated. Behavioral impacts (e.g., discourteous behavior) were judged by all respondents as most important in determining the quality of their wilderness experience. Respondents tended to ignore the social setting (i.e., where the impact occurred) when evaluating the effect of the impacts. Survey respondents had the most difficulty in quantifying an acceptable level of vegetation loss and bare ground at campsites and on or beside trail. They had the least difficulty in quantifying acceptable levels of litter at all locations. ii TABLE OF CONTENTS Page Abstract ii Table of Contents iii List of Tables vii List of Figures be List of Maps x Acknowledgements xi Chapter 1 Introduction 1 Study Purpose 1 Research Questions 2 Study Area 3 Research Hypotheses 5 Chapter 2 Literature Review 7 Introduction 7 Current Wilderness Management Frameworks 7 Methods and Modeling Approaches for Identifying 8 Salient Impacts Conjoint Analysis and Stated Preference and 10 Choice Models Previous Empirical Studies 11 Methods and Modeling Approaches for Identifying 19 Impact Standards The Social Norm Concept 20 The Return Potential Model 21 Previous Empirical Studies 24 The Human Judgement Model 29 Mathematical Representation of the 31 Human Judgement Model The Factorial Survey Approach 33 The Factorial Object Universe Concept 36 Previous Empirical Studies 37 Hypotheses 39 iii Page Chapter 3 Methodology 42 The Survey Instrument 42 The Survey Sample 42 Sub-Groups of the Survey Sample 43 Representativeness of the Survey Sample 48 Identifying Salient Impacts 51 Factorial Object Universe 52 Factorial Object Sample and Respondent 54 Sub-Samples The Rating Task 57 Data Analysis Procedures for Hypotheses 1A, IB and 1C 57 Detennining Impact Acceptability 60 Data Analysis Procedures for Hypotheses 2A, 2B and 2C 60 Study Limitations 62 Chapter 4 Results 65 Introduction 65 Results for Null Hypothesis 1A 65 Relative Impact Factor Influence 67 Ratings Thresholds 68 Rating Variance 68 Rating Error 68 Systematic Variation 68 Coefficient of Determination 69 Results for Null Hypothesis IB 69 Relative Impact Factor Influence 69 Ratings Thresholds 71 Rating Variance 71 Rating Error 71 Systematic Variation 71 Coefficient of Determination 71 Results for Null Hypothesis 1C 72 Relative Impact Factor Influence 72 Ratings Thresholds 72 Rating Variance 72 Rating Error 74 Systematic Variation 74 Coefficient of Determination 74 Results for Null Hypothesis 2A 74 Results for Null Hypothesis 2B 80 Results for Null Hypothesis 2C 83 Summary of Results for Maximum Acceptable Amounts 91 Data iv Page Chapter 5 Conclusions 92 Introduction 92 Null Hypothesis 1A 94 Null Hypothesis IB 96 Null Hypothesis 1C 98 Null Hypothesis 2A 101 Maximum Acceptable Amounts On or Beside Trails at 101 Spruce Lake Maximum Acceptable Amounts at Camps at Spruce 102 Lake Maximum Acceptable Amounts On or Beside Trails 103 Elsewhere in the Area Maximum Acceptable Amounts at Camps Elsewhere in 104 the Area Summary for Null Hypothesis 2A 105 Null Hypothesis 2B 105 Maximum Acceptable Amounts On or Beside Trails at 105 Spruce Lake Maximum Acceptable Amounts at Camps at Spruce 106 Lake Maximum Acceptable Amounts On or Beside Trails 107 Elsewhere in the Area Maximum Acceptable Amounts at Camps Elsewhere in 107 the Area Summary for Null Hypothesis 2B 108 Null Hypothesis 2C 109 Maximum Acceptable Amounts On or Beside Trails at 109 Spruce Lake Maximum Acceptable Amounts at Camps at Spruce 110 Lake Maximum Acceptable Amounts On or Beside Trails 110 Elsewhere in the Area Maximum Acceptable Amounts at Camps Elsewhere in 111 the Area Summary for Null Hypothesis 2C 111 Implications for Management 112 Summary 116 References 120 Appendix 1 Sample Questionnaire 127 Appendix 2 Trailhead Sign. 139 Appendix 3 Contact Card 141 Appendix 4 First Mailing Introduction Letter 143 Appendix 5 Reminder Letter 145 Appendix 6 Second Mailing Introduction Letter 147 v Page Appendix 7 Third Mailing Introduction Letter 149 vi LIST OF TABLES Table 1: Questionnaire Mailing Schedule and Response Rates Table 2: Frequency Values for Sub-Groups by Categories Table 3 Pearson Correlation Coefficients Among Survey Sample Sub-Groups Table 4: Survey Comparisons of Use / User Characteristics Table 5: Impact Dimensions and Factors Table 6: Pearson Correlation Coefficients Among Impact Dimension Factors Table 7: Regression of Hypothetical Situation Ratings on Impact Factors for Trip Organization Sub-Groups Table 8: Regression of Hypothetical Situation Ratings on Impact Factors for Length of Stay Sub-Groups Table 9: Regression of Hypothetical Situation Ratings on Impact Factors for Place of Origin Sub-Groups Table 10: Average Maximum Acceptable Amounts of Impact On or Beside Trails at Spruce Lake for the Trip Organization Sub-Groups Table 11: Average Maximum Acceptable Amounts of Impact At Camps at Spruce Lake for the Trip Organization Sub-Groups Table 12: Average Maximum Acceptable Amounts of Impact On or Beside Trails Elsewhere in the Area for the Trip Organization Sub-Groups Table 13: Average Maximum Acceptable Amounts of Impact At Camps Elsewhere in the Area for the Trip Organization Sub-Groups Table 14: Average Maximum Acceptable Amounts of Impact On or Beside Trails at Spruce Lake for the Length of Stay Sub-Groups Table 15: Average Maximum Acceptable Amounts of Impact at Camps at Spruce Lake for the Length of Stay Sub-Groups Page 44 47 49 50 53 55 66 70 73 75 77 78 78 81 82 vu Table 16: Average Maximum Acceptable Amounts of Impact On or Beside Trails Elsewhere in the Area for the Length of Stay Sub-Groups Table 17: Average Maximum Acceptable Amounts of Impact at Camps Elsewhere in the Area for the Length of Stay Sub-Groups Table 18: Average Maximum Acceptable Amounts of Impact On or Beside Trails at Spruce Lake for the Place of Origin Sub-Groups Table 19: Average Maximum Acceptable Amounts of Impact At Camps at Spruce Lake for the Place of Origin Sub-Groups Table 20: Average Maximum Acceptable Amounts of Impact On or Beside Trails Elsewhere in the Area for the Place of Origin Sub-Groups Table 21: Average Maximum Acceptable Amounts of Impact At Camps Elsewhere in the Area for the Place of Origin Sub-Groups viii LIST OF FIGURES Page Figure 1: A Return Potential Curve for Encounters in Wilderness Settings 22 Figure 2: Mathematical Expression of the Overall Judgement Process of the Human 32 Judgement Model Figure 3: Comparative Measures of the Extent of Sub-Group Agreement 34 Figure 4: Reproduction of Part IV of the Survey Questionnaire 61 ix LIST OF MAPS Page Map 1: The Spruce Lake Trails Area 4 x ACKNOWLEDGMENTS The author would like to thank committee members Dr. Peter J. Dooling, Dr. Neil Guppy and Dr. Al Chambers for their guidance in this research and in the preparation of this manuscript. In addition, special thanks to Dalton McArthur and Terje Void of the BC Forest Service for their assistance and encouragement and for sharing their expert knowledge of the study area and wilderness management in general. This study could not have been conducted without the permission and assistance of the Lillooet Forest District and the visitors to the Spruce Lake Trails Area. Lastly, the author wishes to express his gratitude to his children, Tom, Jeremy and Rebecca for acceptance of their father's absences and preoccupations while completing this research. xi Chapter 1 INTRODUCTION STUDY PURPOSE It is widely recognized that a key challenge in the management of recreation use in wilderness areas is finding means to effectively protect both ecological and social conditions. Current wilderness management principles of the British Columbia (B.C.) Forest Service place emphasis on protecting the wilderness resource from the negative impacts associated with human use (B.C. Ministry of Forests, 1989a). Potential impacts include those on vegetation, soils, water and wildlife as well as on the social conditions necessary for experiencing solitude and a natural setting free from the evidence of human interference (Hendee et al., 1990). Current wilderness recreation planning and management systems recognize that any recreational use of an area leads to change and seek to keep the character (i.e., naturalness) of the area and the rate of change in wilderness conditions due to human factors within acceptable or tolerable levels. Both the Limits of Acceptable Change (LAC) System for Wilderness Planning (Stankey et al., 1985) and the Visitor Impact Management in Wildland Settings (VIM)(Graefe et al., 1986) frameworks share a common focus toward wilderness recreation management: 1) the identification of management objectives regarding desired ecological and social conditions; 2) the selection of measurable impact indicators which reflect the degree of change in ecological and social conditions; and 3) setting impact indicator standards which define the maximum acceptable level of change in the impact indicators and thus the wilderness conditions they reflect. Ecological conditions typically include considerations of the type, severity, prevalence and apparentness of impacts. Social conditions usually consider the levels and types of encounters occurring in different wilderness settings (Stankey et al., 1985). A host of impact indicators have been identified and suggested for monitoring purposes. However, problems have been encountered in selecting appropriate impact indicators and setting impact standards acceptable to wilderness users, managers and interested stakeholders. The purpose of this 1 study is to examine theoretical concepts and methods of determining the influence of impacts on ecological and social conditions found in wilderness and thus on the experiences of wilderness visitors. More specifically, the study seeks to determine whether estimates of impact saliency, impact acceptability and the extent of agreement for these estimates vary between sub-groups of wilderness recreationists. RESEARCH QUESTIONS One problem encountered in selecting impact indicators stems from the difficulty in determining impact saliency: that is, which impacts most negatively influence wilderness visitors' experiences. Because wilderness managers cannot afford to monitor all possible impact indicators, it has been suggested that a smaller number of the most salient impact indicators be chosen to represent ecological and social conditions. The saliency issue is addressed in this study by the following research question: Research Question (1): "What is the relative importance of specific impacts on ecological and social conditions to wilderness visitors?" Researchers and managers have also encountered difficulties in developing impact indicator standards. One problem is related to determining whether wilderness visitors can articulate maximum acceptable levels for impact indicators. This problem is addressed in this study by the following question: Research Question (2): "What are the maximum amounts for specific impacts wilderness visitors will accept before their experience would be changed?" Wilderness managers also need evaluative information on the amount of agreement wilderness visitors and relevant sub-groups have concerning salient impact indicators and indicator standards (Shelby and Vaske, 1991). This information can help managers justify the imposition of impact standards which can potentially restrict visitors' activities. The following research questions address this need: Research Question (3): "Is the extent of agreement on the relative importance of specific impacts greater for certain sub-groups of wilderness visitors?" 2 Research Question (4): "Is the extent of agreement on the maximum acceptable levels of specific impacts greater for certain sub-groups of wilderness visitors?" STUDY AREA The term "wilderness" as used in this study includes both those areas which are legally designated wilderness under Section 5.1 of the B.C. Forest Act and those areas administratively zoned as wilderness. The B.C. Ministry of Forests' policy framework Managing Wilderness in Provincial Forests (1989a, p.2) defines wilderness as: ...an area of land generally greater than 1,000 hectares that predominantly retains its natural characteristics and on which human impact is transitory, minor and in the long-run substantially unnoticeable. The B.C. Ministry of Forests uses additional criteria to distinguish wilderness from non-wilderness areas. Wilderness settings are normally: greater than eight kilometres from primitive roads; have few, if any, permanent structures; have relatively low-use levels; and provide opportunities for solitude and recreation (B.C. Ministry of Forests, 1989). These criteria and the following considerations were used to select the specific area in which to conduct this study. To fully examine the theoretical and methodological issues suggested by the literature review (Chapter 2) and to have practical applications to land management concerns in B.C., other considerations were included. It was felt that the area used for the empirical study needed to provide a broad range of wilderness recreation activities, user characteristics, ecological and social impacts, and settings. The Spruce Lake Trails Area (map 1) is one of a number of areas proposed by the B.C. Government's Protected Areas Strategy to be studied for Wilderness Area designation. The area is located in the South Chilcotin Mountains approximately 200 kilometres west of Kamloops and 200 kilometres north of Vancouver and is managed by the B.C. Forest Service, Lillooet Forest District. Access to the 90,000 hectare area is from Lillooet via Highway 40 to Goldbridge - a five-hour drive from Vancouver. There are no roads inside the area and air travel is the major mode of access. Other primary means of travel to the area are by 3 Map 1 Spruce Lake Trails Area Source: BC Ministry of Forests 4 horse and foot in the summer use season and by snowmobile and helicopter during winter months. The area is located in a transition zone between coastal and interior climates and includes four biogeoclimatic zones from the dry interior Douglas-fir to alpine tundra. Along with its many alpine areas with scenic lakes, there are numerous grassy meadows and open, forested valley bottoms. Geological formations in the area include sedimentary and volcanic rock with limestone outcroppings and fossil deposits. The area has many wildlife species and is an important California bighorn sheep range. A favourable climate and areas of gentle terrain make the Spruce Lake Trails Area suitable for a wide range of recreational pursuits. The B.C. Forest Service has a network of 164 kilometres of hiking and horse trails that follow major valleys and provide access to most of the area. Current recreation uses include: hiking and camping; guided hunting and trail riding; fishing and hunting; ski touring and cross-country skiing; wildlife viewing; and photography. One licensed guide-outfitter and one packer offer commercial services in the area on a continual basis. Visitors travel into and camp in dispersed locations in the Spruce Lake Trails Area as well as at designated sites at Spruce Lake. Recreation visitors to the Spruce Lake Trails Area come from B.C., other parts of Canada, the U.S. and many other foreign countries. The majority of visitors are male (approximately 70%) and range in age from under 20 years to 65 years and over (B.C. Ministry of Forests, 1989b). The broad range of user characteristics and recreation activities, diverse physical settings and remoteness made the area a suitable place in which to conduct the empirical study. RESEARCH HYPOTHESES To answer the research questions posed in a preceding section a number of research hypotheses have been formulated. Additional rationale for the hypotheses' formulation will be presented in Chapter 2. The following research hypotheses will be tested: Null Hypothesis 1A: The influence of specific impacts on visitor's wilderness experiences will not vary for sub-groups of study area 5 Null Hypothesis IB: Null Hypothesis 1C: Null Hypothesis 2A: Null Hypothesis 2B: Null Hypothesis 2C: visitors characterized by their trip organization (i.e., commercially outfitted versus privately outfitted). The influence of specific impacts on visitor's wilderness experiences will not vary for sub-groups of study area visitors characterized by their length of stay. The influence of specific impacts on visitor's wilderness experiences will not vary for sub-groups of study area visitors characterized by their place of origin (i.e., residents of B.C. versus residents of the U.S.). Average maximum acceptable amounts of impact will not vary for sub-groups of study area visitors characterized by their trip organization (i.e., commercially outfitted versus privately outfitted). Average maximum acceptable amounts of impact will not vary for sub-groups of study area visitors characterized by their length of stay. Average maximum acceptable amounts of impact will not vary for sub-groups of study area visitors characterized by their place of origin (i.e., residents of B.C. versus residents of the U.S.). The results of testing these research hypotheses can constitute one source of information to help wilderness managers provide acceptable ecological and social conditions in the study area. As Shelby and Vaske (1991) pointed out (p. 174): Normative information can be combined with information about legal and administrative mandates, agency policy, historical perspectives, public opinion, and/or interest group politics in developing standards for resource management. 6 Chapter 2 LITERATURE REVIEW INTRODUCTION This chapter will examine the conceptual and methodological approaches researchers have used in the past to investigate impact saliency, determine impact standards and analyze the degree of shared agreement among wilderness visitors. Limitations of the various modeling approaches and unresolved research issues will be described which support the relevance of the research hypotheses presented in Chapter 1. An alternative methodological approach, the Factorial Survey Approach, developed in the social sciences (but as yet not used by wilderness recreation researchers) will be presented. The chapter begins with a discussion of management frameworks currently used (or proposed) by wilderness managers. CURRENT WILDERNESS MANAGEMENT FRAMEWORKS Most past wilderness management frameworks, as developed by recreation researchers, have utilized the carrying capacity model. Stated briefly, the model suggests that impacts on ecological and social conditions can be managed by detennining the use level at which unwanted impacts occur and then limiting use not to exceed this level. Recently, researchers and resource managers have recognized that the carrying capacity model has significant limitations. Empirical evidence suggests that other variables such as the type of use are often better predictors of the amount of impact than the amount of use (Roggenbuck et al., 1993). Washburne (1982) points out that the behaviour of visitors and the timing and distribution of use are also significant factors which can cause adverse impacts of recreation use in wilderness areas. The recognition of the traditional carrying capacity model's limitations has led to the development of impact management concepts such as Limits of Acceptable Change (LAC) (Stankey et al., 1985) and Visitor Impact Management (VIM) (Graefe et al., 1986). In addition, both concepts represent efforts to tie wilderness management planning activities "more directly to management concerns, i.e., establishing the appropriate levels of use by 7 considering the various management scenarios, as opposed to merely maximizing one particular use" (Haider, 1993, p.6). These frameworks call for the: 1) identification of management objectives regarding desired ecological and social conditions; 2) selection of measurable impact indicators which represent the degree of change in ecological and social conditions; and 3) setting of impact indicator standards which define the maximum acceptable level of change in wilderness conditions. Monitoring and other management activities are used to keep the impacts from significantly degrading both the ecological and social conditions and, thereby, negatively affecting wilderness visitors' experiences. However, as pointed out in Chapter 1, managers cannot afford to monitor all potential indicators of impact due to budget and personnel constraints. It is, therefore, necessary to be selective and to identify the most salient impacts to use for measurement and monitoring efforts. "However, relatively little attention has been given to the selection process, at least with respect to social impacts" (Whittaker, 1992, p. 14). In summary, current wilderness management frameworks have sought to overcome limitations in the traditional carrying capacity model. Both the LAC and VIM concepts share a common focus and seek to identify all the variables which significantly impact wilderness conditions. Additionally, Management frameworks such as these are inevitably based on such notions as quality and values, usually articulated in the management objectives. These frameworks do not deal with values directly. Instead they use evaluative standards, which can be expressed in terms of user satisfaction, perceived crowding or contact preference standards (Shelby and Heberlein, 1986 as quoted in Haider, 1993, p. 11). METHODS AND MODELING APPROACHES FOR IDENTIFYING SALIENT IMPACTS. The B.C. Ministry of Forests policy definition of wilderness described in Chapter 1 provides two rather broad concepts which can be used to define the quality of wilderness conditions and to identify potential impacts. These are the concepts of naturalness and solitude. Potentially, any impacts associated with these concepts could be judged salient by wilderness visitors and other interested stakeholders. 8 The concept of naturalness as it is often used by biologists and applied to natural ecosystems implies a natural condition with no (or at most minimal) disturbance by man (Margules and Usher, 1981). Thus, naturalness is related to an ecological environment which for the most part has not been altered by humans. This biological definition of naturalness is much akin to the policy definition of wilderness referred to above. The concept of privacy or solitude was recently examined by Hammitt and Madden (1989). They identified the following four dimensions of the concept as it relates to a wilderness experience: 1) being in a natural environment - being in a tranquil, peaceful and remote environment free of man-made intrusions and noises 2) cognitive freedom - the freedom to limit your attention to whatever you choose, to choose actions and use of time and to control your thoughts 3) intimacy - privacy from most people, yet the opportunity to socialize with family or friends in small group experiences 4) individualism - being released from the rules and constraints of society, daily body maintenance and expectations of others In addition to the above criteria related to the concepts of naturalness and solitude, other "saliency" criteria have been proposed. "Stankey and others (1985) suggested that (impact) indicators should relate to the amount and type of wilderness use, permit measurement in cost-effective ways at acceptable levels of accuracy, and be potentially responsive to managerial intervention" (Roggenbuck et al., 1993, p. 187). Another pertinent issue related to identifying the saliency of impact indicators involves who's opinions should be sought and examined. Most researchers agree that "the views of clientele groups are critically important because wilderness is largely a cultural resource" (ibid, p. 188). This idea is reflected in the importance placed on deterrnining evaluative standards in wilderness management frameworks such as LAC. The sections which follow will examine various approaches taken by recreation researchers and researchers from other disciplines to deal with value identification and determining evaluative standards. 9 Conjoint Analysis and Stated Preference and Choice Models Haider (1993) recently presented an overview of quantitative modeling approaches developed in the behavioural sciences to deal explicitly with individual values or preferences. Stated briefly, conjoint analysis has as its goal the "quantitative measurement of the relative importance of one attraction (value) as opposed to another" (Aakar and Day, 1986, as quoted in Haider, 1993, p.20). Of particular interest here is the use of conjoint analysis in stated preference models to identify important attributes and attributed levels related to preferred recreation settings. Stated preference models have been developed by the behavioural sciences to address the multi-attribute nature of choice or value decision-making. Unlike revealed preference models which infer preference from direct observation of behaviour, stated preference models require individuals to express their preferences directly (Haider, 1993). The stated preference approach has been used extensively by recreation researchers. (One of the first applications directed toward wilderness uses can be found in Hendee et al., 1968.) Louviere and Timmermans (1990) note a number of assumptions made by the stated preference models. First, individuals use some type of process rule in making their preference choices. Second, individuals may not consider all the variables or levels of the variables when stating their preferences or making value judgements related to choices. Third, because other unidentified variables can influence preferences, "individuals can be observed to vary in their choices from observation to observation" (Louviere and Timmermans, 1990, p. 13). Lastly, since the individual's preferences reveal the values they place on the variables of choice alternatives, the functional form of an individual's or group's process rule can be diagnosed and tested using statistical techniques applied to the stated preference data set. The following are the steps/decisions involved in constructing a stated preference model: 1) Identification of salient attributes for decision-making 2) Specification of attribute levels 3) Selection of a method for combining attribute levels into descriptions of choice alternatives 4) Choosing an experimental design to place choice alternatives into sets in which stated preference responses are observed 10 5) Choosing a realistic way to present the choice alternatives 6) Choosing a procedure to measure stated preferences 7) Choosing a method for estimating preference functions 8) If the objective of the experiment is to predict choice behaviour, one must choose an approach to transform predicted values into choices. (See Louviere and Timmermans, 1990, for a more detailed discussion of these steps.) Two fundamentally different approaches have been identified within stated preference research - compositional and decompositional. "In the compositional approach it is assumed that individuals can directly express the importance of each separate attribute and the relative and absolute position of each attribute of each alternative" (Haider, 1993, p.20). For example, wilderness visitors would be given a list of typical impacts on ecological and social conditions (e.g., tree damage or encountering others) and be asked to rank or rate them as to how much they would affect the quality of their wilderness experience. (As the discussion below will point out, this has been the predominate approach in the wilderness recreation research literature.) The researcher would then "compose" or combine the preference scores into a ranking of attribute saliency. Previous Empirical Studies Numerous empirical studies have used stated preference and choice models to identify the importance wilderness visitors attach to specific impacts. Nearly all research studies have used the compositional approach. Much of the recreation research described below has come from studies on sources of satisfaction/dissatisfaction and carrying capacity. Although not a direct measure of importance or saliency, sources of dissatisfaction definitely affect the quality of the recreation experience and thus reflect on what "matters" to wilderness visitors. One of the most documented sources of impact is the presence of litter. It has consistently been perceived as a negative impact on the quality of the wilderness environment. Downing and Clark (1979) showed that 92 percent of managers and 50 percent of visitors identified litter as the most commonly perceived problem. Whittaker (1992) reported that river users in Alaska consistently rated the presence of litter as detracting from their experiences (86 to 91 percent on various rivers). Numerous other studies have cited similar 11 findings (See Lucas, 1979; Knopf; 1982; McCool and Petersen, 1982; and McCool et al., 1990). Differences in opinion as to the degree to which the presence of litter affects the quality of the experience have also been reported. Whittaker (1992) reported differences between floaters and powerboaters on Alaskan rivers. These findings and others (see McCool et al., 1990 and Krumpe et al., 1989) seem to indicate that different activity sub-groups vary in terms of the importance they place on the negative influence of the presence of litter. It is clear, however, that most wilderness visitors consider litter to be a negative impact on the naturalness (i.e., a natural setting with minimal disturbance by man) of a wilderness area. Another "negatively valued attribute" found by Anderson (1980) in a study of Boundary Waters Canoe Area Wilderness was "seeing pulled bark on trees." While this is often viewed as an effect of vandalism, this impact is often associated with firewood gathering (Cole and Dalle-Molle, 1982). Cole (1986) noted the very visible effect of tying recreational stock to trees in campsites. Not only are tree trunks noticeably damaged by the abrasive actions of rope, but when tied to small trees, girdling can kill the tree. The overall effect of tree damage is to depreciate the naturalness of the campsite environment. In the Bear Trap Canyon study (McCool et al., 1990), 23 percent of private floaters and 22.9 percent of hikers considered tree damage a slight problem. In their study of visitor preferences in the Pacific Northwest, Hendee et al. (1968) reported substantial opposition to permanent fireplaces in wilderness. Lucas (1985) reported that loose rock campfire rings were more favoured in 1982 than in 1970. More recently, the presence of campfire rings was reported as "moderately" to "very much" influential on the quality of the wilderness experience by the respondents in a study by Roggenbuck et al. (1993). McCool et al. (1990) found that 27 percent of private floaters and almost 24 percent of hikers felt that the number of campfire rings was a slight problem. The amount of bare ground and vegetation loss in a campsite has also been cited by wilderness visitors as negatively affecting their wilderness experience. Mean response ratings found by Roggenbuck et al. (1993) showed this impact to "moderately" to "very much" 12 influence the quality of the wilderness experience. Sub-group differences in the importance of this impact is illustrated by the findings in the Bear Trap Canyon study (McCool et al., 1990). Forty-two percent of private floaters felt that devegetated campsites were a problem compared to 14.8 percent for outfitted floaters and 38.5 percent for hikers. Other studies have noted the effect of vegetation loss from trampling (e.g., soil compaction) on reduced water infiltration rates of soils (Blom, 1976) and decreases in germination rates of seeds (Harper et al., 1965). Thus, ecological as well as experiential impacts associated with vegetation loss have been documented. Another important source of impact much examined by researchers is encounters with other people. Impacts identified under this category are related to the concept of solitude, namely: cognitive freedom - the freedom to limit your attention to whatever you choose; and intimacy - privacy from most people. Whittaker (1992) found that 78 percent of floaters on low-use Alaskan rivers felt that river encounters detracted from their experience and 88 percent of the same group felt that campsite encounters detracted from their experiences. The study by Roggenbuck et al. (1993) seems to support the above findings that wilderness visitors see encounters as an important impact. In this study of the influence of encounter impacts for four wilderness areas in the U.S., mean ratings fell in the range between "somewhat" to "very much" influenced the quality of their wilderness experience. The highest mean ratings ("moderately" to "very much" influential) were for the impacts of: "number of horse groups camped within sight or sound of my campsite"; "number of biker groups camped within sight or sound of my campsite"; "number of hiker groups that walk past my campsite"; and "number of large groups seen along the trail." Whittaker (1992) reported differences among activity sub-groups' feelings regarding both river encounters and campsite encounters. While 44 percent of powerboaters on high-use rivers reported that river encounters detracted from their experience, 49 percent of floaters on the same rivers reported that river encounters detracted from their experiences. Sub-group variations in the relative impacts of seeing large groups is illustrated by the findings in the Bear Trap Canyon study (McCool et al., 1990). Forty-two percent of hikers reported 13 that seeing large groups detracted or very strongly detracted from their experience. Similar amounts for private and outfitted floaters were 35.1 percent and 28.9 percent, respectively. Researchers have also identified a number of impacts typically labeled "sight and sound intrusions." Human-made structures such as trail shelters are often found in some wilderness areas. Merriam and Ammons (1967) found that while they were favoured (80 percent) in Glacier National Park, most visitors opposed such structures in an adjacent National Forest Wilderness. Womble et al. (1978) reported that "nearly 85 percent of hikers felt shelters were appropriate on the Chilkoot Trail." Other types of structures typically found in wilderness include: tent frames, primitive corrals and other facilities for recreation stock, primarily horses. A number of studies have reported strong opposition to these types of structures (See Hendee et al., 1968; Stankey, 1973; and Lucas, 1985). Whittaker (1992) also reported considerable opposition to a number of human-made structures including public use cabins, concessions and long-term camps. However, like many of the other impacts evaluated in the study, activity sub-groups differed in their opposition. For example, while 16 percent of powerboaters on high-use rivers felt that private cabins detracted from their experience, almost twice as many floaters (31 percent) on the same rivers reported that private cabins detracted from their experience. The above studies clearly indicate that to many visitors the presence of human-made structures in wilderness is an important impact on the naturalness of the area. The impact of management signs in wilderness areas has received mixed opinions from visitors. Roggenbuck et al. (1982) reported that "problems with signing ranked among the top ten as perceived by users" with about 25 percent describing it as a problem. Thirty-two to 64 percent of visitors to five rivers in Alaska (Whittaker, 1992) reported that hazard signs would detract from their experiences. In the same study, 38 to 44 percent felt that interpretive signs would detract from their experiences. Hendee et al. (1968) reported visitor support for directional signs only. More recent data (Lucas, 1985) suggest that limited interpretive signs are acceptable. Whittaker's study (1992) shows that the specific location can affect the relative importance of the impacts of management signs. Sixty-four percent of 14 floaters on low-use rivers felt that hazard signs would detract from their experience while only 37 percent of floaters on high-use rivers felt that hazard signs detracted from their experience. Thus, the location of the impact does seem to make a significant difference among users. One impact which is typically found in B.C. Wilderness Areas and backcountry settings is the sighting of aircraft and helicopters. Very little research has been conducted in B.C. or elsewhere to document the extent of this impact and its influence on visitors' experiences. However, research in the Arctic National Wildlife Refuge in Alaska (an area similar to much of B.C.'s wilderness) indicates that this sight and sound intrusion affects visitors' experiences negatively. Warren (1987) reported that over 50 percent of survey respondents would view "improving light aircraft landing strips" as undesirable while over a third of hunters and over half of non-hunters favoured limiting the number of planes that can land in one management zone in a week. More recently, Whittaker (1992) reported that 33 percent of power boaters and 49 percent of floaters on the same rivers felt that airplane landings would detract from their experiences. Seventy percent of floaters on low-use rivers felt that helicopter landings would detract from their experiences. Another source of impact in wilderness is the behaviour of people encountered. Impacts identified under this category are related to being in a natural environment with minimal disturbance by man (naturalness) and cognitive freedom - the freedom to limit your attention to whatever you choose. Stankey and Schreyer (1987) noted that perceptions of crowding and hence, dissatisfactions are not solely determined by the number of encounters. The behaviour of those people encountered might have just as much impact on the quality of the desired wilderness experience (Stankey, 1973). In a study examining types of norms for recreation impacts Whittaker and Shelby (1988) reported the amount of discourteous behaviour Whitewater floaters would tolerate on a 40 mile segment of the lower Deschutes River in Oregon. Eleven different impacts were evaluated by floaters. Impacts examined included encounters, waiting times, camp sharing, number of fire rings and others. Of the eleven, discourteous behaviour had the lowest median tolerance level. Fifty percent of respondents reported they would tolerate 2.2 or fewer incidents of discourteous behaviour per 15 day before the experience became unpleasant. West (1982) reported that 17 percent of respondents in his study of hikers were bothered by the behaviour of others. The most bothersome behaviours included, in decreasing order: noise, yelling and loud behaviour; littering and polluting lakes; and non-compliance with rules. Impacts originating outside of wilderness have received little attention from researchers. In their study of four U.S. Forest Service Wilderness Areas, Roggenbuck et al. (1993) asked respondents to rate two such impacts on the quality of their experience. The visibility of lights originating from outside the wilderness" received influence ratings of "somewhat" to "moderate". The amount of man-made noise originating from outside the wilderness was rated as "moderately influential". No information on sub-group variations were reported. However, it is clear that visitors to these four areas which are geographically distinct (Georgia to Montana) view these impacts as influencing the quality of their wilderness experience. Relatively few other impacts have been examined by researchers seeking to identify salient impacts on wilderness conditions. With the exception of Whittaker (1992) who looked at the relative influence of powerboat, airboat and rafting/canoeing use, most studies have been limited to the above impacts. In all of the stated preference research studies described above, the compositional approach was used. Wilderness visitors and interest groups were usually asked to evaluate lists of different impacts on wilderness conditions. Survey respondents would rate, for example, the influence of encounters on their wilderness experience using Likert-type scales. Regardless of the specific impacts included on the lists to be evaluated, most of the approaches required the respondents to make evaluative judgements outside of a specific situational context. (Some studies have specified a single experiental setting such as the type of trip expected: wilderness Whitewater, scenic Whitewater or social recreation, see Roggenbuck et al., 1991.) Usually, however, the judgement was made regarding only their visit to the area and not for varying locations or for multiple situations encountered in the area. This practice is contrary to past research findings. Recreation researchers have long 16 recognized that the relative impact of encounters, for example, on the wilderness experience depends on more than just the number of people encountered (Stankey and Schreyer, 1987). Other factors such as where the contact occurred and the behaviour of the people contacted also influence the degree to which the number of encounters affects the quality of the wilderness experience. "Real-world" evaluations or judgements regarding the influence of, for example, encounters on visitors' experiences are made in the context of specific situations. In other words, the overall judgement concerning the importance of a specific impact on wilderness ecological or social conditions is made relative to sub-decisions about other factors such as the specific location. One problem with incorporating these extenuating circumstances or combinations of impacts acting in concert is a limitation of the survey instrument. To ask visitors to evaluate impact saliency while keeping in mind the condition of a number of simultaneous factors would be quite onerous using questionnaires. In most cases, the length of the questionnaire needed to evaluate even a dozen potential impacts in combination with multiple behavioural factors and occurring at multiple locations in the area visited would far exceed the maximum length one could use to get an acceptable survey response rate. However, researchers in the behavioural sciences have developed approaches which can overcome these potential problems. In contrast to the compositional approach used in the preceding stated preference modeling approaches, the decompositional approach measures individual preferences of attribute profiles or hypothetical combinations of various attribute levels. The resulting overall preference scores for each synthetically designed choice alternatives can be decomposed into the weights and evaluations (utilities) of attribute positions (levels), given an a priori assumed combination rule that represents how individuals integrate the weights and utilities to form overall preferences (Louviere and Timmermans, 1990, as quoted in Haider, 1993, p.21). The decompositional approach usually uses factorial designs to construct the hypothetical attribute profiles which are then rated one at a time by survey respondents. 17 Haider et al. (1993b, p. 8) summarized the following advantages of research experiments including stated preference models using the decompositional approach: 1) A large number of attributes can be combined to describe hypothetical attribute profiles. 2) Survey respondents evaluate the profiles holistically instead of single attributes or attribute levels. 3) Attribute levels are uncorrelated, reducing multicolinearity problems often encountered in revealed preference models. 4) Researchers control profile choice sets presented to respondents. 5) Attribute profiles which presently may not exist can be designed and evaluated by respondents. 6) Results can be incorporated into current management frameworks which require value-based decisions. A number of empirical studies have used stated preference and choice models to examine a wide variety of problems in general recreation research. The studies discussed below are not intended to be exhaustive, but rather, to demonstrate the application of the approaches in examining factors commonly perceived to impact the quality of the recreation experience. In a series of studies (Lieber and Fesenmaier, 1984, 1985 and Lieber et al., 1988), five physical features which potentially influenced recreation satisfaction of trail areas in Chicago were examined. In this stated preference decompositional approach , a fractional factorial design was used to produce trail profiles combining three levels of the following attributes: type of trail surface, type of terrain, length of trail, changes between woods and open areas, and proximity to residence. A line mark scale was used by respondents to rate each trail profile. The influence of the various physical features were estimated using utility functions in this example of a traditional conjoint analysis. Louviere and Timmermans used a more sophisticated approach (Hierarchical Information Integration) to model a much more complex stated preference choice design. They "examined the likely effects on recreational choices of variations in 19 salient attributes of parks and forest preserves" in the Eindhonen region of the Netherlands (Louviere and Timmermans, 1990, p.24). Four sets of decision constructs were used, each having five, seven, four and three attributes. The decision constructs were: 1) natural environment and accessibility; 2) facilities and things to do; 3) maintenance; and 4) social use. Respondents 18 then rated profiles that described the four decision constructs in terms of how good each construct would be if a hypothetical park had those attribute levels. A rating scale of 0 (very poor) to 10 (very good) was used. The authors then used multinominal logit analysis to estimate the degree to which each decision construct influenced respondents' choices. Haider et al. (1993b) used a stated preference approach based on a decompositional multiattribute research technique (stated choice experiment) to model decision variables that influence tourist choices. Survey respondents were asked to express preferences for hypothetical profiles combining levels of eight remote tourism settings, including: type of fish present, daily catch limit, fishing expectations, type of accommodation and facilities, access to other lakes, scenery and setting, fly-in cost, and distance from home. The examples of stated preference and choice modeling approaches using decompositional techniques discussed above are just a few of the empirical studies that can be found in the behavioural research literature. The author could not find any evidence of similar approaches in research specifically aimed at determining salient impacts on traditional wilderness conditions. However, similar models and approaches have been used by sociologists. These will be described in a later section. METHODS AND MODELING APPROACHES FOR IDENTIFYING IMPACT STANDARDS As the previous discussion of current wilderness management frameworks pointed out, the need to identify quantifiable impact indicator standards is seen as a critical step in achieving management objectives related to providing quality ecological and social conditions in wilderness settings. In addition to providing wilderness managers and users a means of determining when impacts have reached unacceptable levels, impact indicator standards also serve to guide the development of management strategies when impacts exceed agreed upon standards (Stankey et al., 1985). The development of management frameworks such as VIM and LAC and the importance placed on identifying impact standards grew, in part, from the recognition that 19 decisions regarding the quality of ecological and social experiential conditions found in wilderness were based on value judgements. These value judgements reflect philosophical, emotional, spiritual, experiential, and economic responses of those making the judgements. Obviously, few people will have identical responses, and therefore few will make identical value judgements. Thus the task facing agency managers is one of deterrnining consensus value judgements regarding what constitutes desired wilderness conditions and how those conditions should be maintained... (Hendee et al., 1990, p.218). Thus, the idea of setting impact standards is intimately connected with concepts such as consensus or shared agreement. This has led wilderness managers and researchers to turn to social research to provide information about user's concerns and help in the identification of ecological and social condition impact standards (Lucas and Stankey, 1985, as quoted in Williams et al., 1992). The empirical study of consensus has long been a critical issue in the social sciences. Social researchers have investigated "public opinion, value, norms, attitudes and many other areas of human behaviour" (Rossi and Berk, 1985, p.333). Hence, wilderness researchers have increasingly sought to apply the concepts and measurement approaches developed by sociologists, particularly in the field of societal norms, in their attempt to identify shared agreement for impact standards. The Social Norm Concept Many differences exist among sociologists and social psychologists on the nature and definition of norms. However, most definitions of norms contain these elements: norms are standards of evaluative rules and obligatory actions and are acknowledged and recognized as binding by the social group in question. In other words, a norm implies consensus (Rossi and Berk, 1985). One issue which has been raised by researchers regarding the concept of norms involves the distinction between personal norms and social norms. Personal norms are associated with the individual's feelings of moral obligation and motivate behaviour through the individual's desire to enhance their sense of self-worth. The adherence to social norms, on the other hand, is motivated by the individual's perception that they are shared by social group 20 members and have sanctions or rewards which tend to enforce them (Roggenbuck et al., 1991). Other issues related to the social norm construct and its application in determining impact standards for wilderness conditions concern how norms are measured. As Vaske et al. (1986), Shelby and Heberlein (1986), and Roggenbuck et al. (1991) pointed out, the conceptual framework for measuring norms most widely used by recreation researchers is derived from the work of Jackson. The Return Potential Model In the article titled Structural Characteristics of Norms, Jackson (1965) set out a model for the analysis and measurement of norms and described the three characteristics of norms which could be used to reveal the structure of social norms (See Vaske et al., 1986 for a complete review of Jackson's model). Stated briefly, Jackson's model recognized two dimensions of the norm concept. The first essential element is a behavioural dimension. A norm is ordinarily about some act of behaviour that can be judged as appropriate or inappropriate. "Intrinsic to a norm is the specification of the amount or quality of behaviour expected of the actor by relevant others" (Jackson, 1965, p.302). Second, the idea of evaluation is either implicitly or explicitly present in the norm concept. "The evaluation of an act of behaviour can vary from strong approval to strong disapproval through some middle point of indifference" (Jackson, 1965, p.302). The graph in Figure 1 was adapted from Heberlein (1977) and can be used to illustrate the three characteristics of a norm described by Jackson. In Heberlein's example, norms for encountering other people in a wilderness setting are being examined. The behavioural dimension (number of encounters) is shown on the horizontal axis while the evaluation dimension (favourability) is shown on the vertical axis. "A personal norm can be described by a line connecting the individual's ratings at each point along the horizontal axis, while the social norm can be described by a line connecting the mean ratings for all members of a group." (Vaske et al., 1986, p. 140.) The first characteristic of a norm described by Jackson (1965) is called the "range of tolerable behaviour" and is shown in Figure 1 to be from 0 to 4 encounters whether for an 21 Figure 1 A Return Potential Curve For Encounters In Wilderness Settings FAVOrtA8LE »5 • 4 •3 »2 • 1 NEUTRAL • 1 z o o < < cr > < UNFAVORABLE WILDERNESS EXPERIENCE I I I i I I I 4 5 6 7 8 910 20 30 SO 100 OTHER PEOPLE ENCOUNTERED(NUMBER) Source: adapted from Heberiien (1977) 22 individual or the members of a group. A second characteristic, called "norm intensity" is expressed by the height of the curve, above and below the neutral line. "Defined in this manner, intensity is the distance from the neutral line at each point on the contact norm curve, independent of the direction of the evaluation (i.e., favourable or unfavourable)." (Vaske et al. 1986, p.42.) The third characteristic of a norm is called "norm crystallization" and is defined by Jackson (1965) as the amount of agreement about a norm. It is derived by computing the variance for each point on the horizontal axis (behavioural dimension) and then calculating 100 - D (Cronback and Gleser, 1953, as quoted in Jackson, 1965). The lower the variance, the more crystallized is the social norm and the higher the degree of shared agreement or consensus. When the dispersion is great and crystallization is therefore low, member's ideas of appropriate or inappropriate behaviour do not coincide. As the following section will indicate, the approaches taken by researchers to investigate impact standards (and the degree of shared agreement for the standards) have varied greatly and so have the results. In some cases, impact norms have been identified, but in some instances, only certain types of norms of users seeking specific experiences are evident (Patterson and Harnmitt, 1990). Roggenbuck et al. (1991) and Shelby and Vaske (1991) suggest that the contrasting results of past studies beg a further examination of certain theoretical and methodological issues before the reliability and validity of the concepts of norm theory can be established for wilderness recreation management. Theoretical issues which need to be examined include additional characteristics of norms identified by current sociological theorists like Rossi and Berk (1985) and Schwartz (1977). Specifically, wilderness recreation researchers need to ask questions concerning: 1) the "obligations to behave" which arise either from the individual's personal norms or from the norms of reference groups or others such as society at large; 2) sanctions (either ratifications or restrictions) which exist to assure compliance with norms; and 3) the level of agreement among sub-groups about the relative influence of impacts and acceptable impact levels. Methodological issues include which measures of central tendency and dispersion to use when comparing aggregated group and sub-group evaluations. Also, Roggenbuck et al. 23 (1991) cite the failure of studies to sufficiently address the full range of salient issues. For example, they make the case for the need to examine the behaviour of other people as a relevant impact and one which has not been sufficiently addressed in the research on impact norms. Another question which needs to be asked involves whether the same wilderness visitors may have different impact standards for different locations in a wilderness area. The LAC management framework in particular includes the provision of zoning a wilderness area according to different management objectives with different impact standards as part of the zoning criteria. The research questions described in Chapter 1 examine two of the issues raised above. Related to norm theory, the study seeks to determine the specific circumstances under which there is consensus or shared agreement among sub-groups about the relative influence of impacts and acceptable impact levels. Study methods address the potential for varying impact acceptability at different locations in the study area. Previous Empirical Studies The norm construct has been used by recreation researchers to search for: encounter norms among hunters (Warren, 1987); encounter norms in camp and river settings for floaters on Whitewater rivers (Whittaker and Shelby, 1988; Roggenbuck et al, 1991); ecological impact norms at wilderness campsites (Shelby et al, 1988); norms for the number and size of party encounters of hikers and campers (Patterson and Hammitt, 1990; Roggenbuck et al, 1993); and Whitewater rafters' norms about encounter levels for both private and commercial boaters (Whittaker and Shelby, 1988). Impact norms have been found to vary for wilderness visitors engaged in different activities (e.g., hiking, hunting, camping); employing different methods of travel (e.g., on foot, on horseback, boating); participating in commercial and non-commercial activities; and with different trip and demographic characteristics. The description of empirical studies which follows will illustrate some of the impacts investigated and methods used by researchers to analyze empirical data on impact standards. All of the studies reviewed here used some adaptation of stated preference models and employed the compositional approach discussed earlier. 24 Compared to the number of studies examining impact standards and norms for encounter levels, there have been relatively few studies which have investigated ecological impact standards. One such study used data collected at campsites in the Mt. Jefferson Wilderness in Oregon (Shelby et al., 1988). Two ecological impacts were examined: the area of bare ground (in square feet) and the outside diameter of fire rings (in inches). A total of five study areas within the wilderness were selected with a total of 20 campsites evaluated. Campsites in only one study area were evaluated by individual respondents. Respondents were asked to evaluate the impact (either fire ring diameter or bare ground) using a five-point scale ranging from totally unacceptable to totally acceptable with a neutral point of zero. User standards were developed for each site by computing a group mean. Each of the sites represented a different level of impact. Impact acceptability curves (similar to Jackson's return potential curves) were then plotted for each type of impact. The range of tolerable sizes for fire rings ranged from no fire rings to approximately a 20-inch diameter ring at one site and from no fire rings to the 36-inch fire ring at the other site. In terms of the crystallization of norms or the amount of agreement about standards, over 70 percent agreed that no fire ring was acceptable at the two sites. More than 90 percent agreed to accept the 16 and 18-inch rings while the 22-inch and 38-inch rings were unacceptable to 72 and 85 percent, respectively at one site. At the other site, there was no clear majority agreement about the 38-inch fire ring. Nearly everyone (99 percent) found the 73-inch fire ring unacceptable at the one site. The range of tolerable amounts of bare ground went from 0 to 650 sq. ft. at one site; 0 to 750 sq. ft. at another site; and 0 to 1,450 sq. ft. at the third site. In terms of norm crystallization, over 90 percent agreement was found at one site for 156 to 462 sq. ft. of bare ground. At another site, 78 percent agreed that the 1,050 sq. ft. of bare ground was acceptable while 90 percent agreed that the 2,275 sq. ft. areas were unacceptable. Small fire rings and small areas of bare ground were found to be more acceptable than no fire rings and no bare ground at some sites. Also, acceptable impact levels were different at different locations. Lastly, average ratings were generally well above or below the neutral line, indicating that the norms 25 for the size of fine rings and the amount of bare ground were of moderate to high intensity (Vaske et al., 1992). In their study of the Deschutes River in central Oregon, Whittaker and Shelby (1988) asked survey respondents to assess the acceptability of fire rings at campsites. Respondents were asked to give the number of campsites (out of four) with a fire ring that they would tolerate before their experience became unpleasant. Evaluations were given for three different segments of the river having different resource or experience characteristics with each respondent reporting on the segment they knew best. An "it doesn't matter to me" category was available for respondents who were unable to give a number. Personal norms were aggregated into a norm curve used to characterize social norms. Median responses were used to define the social norm with the range of tolerable impacts defined as being from zero to the median. The standard deviation from the mean was used to describe norm crystallization (i.e., the extent of agreement for the norm). Results indicated that social norms ranged from 0.5 camps with fire rings on segment three to 1.0 camps with fire rings on river segment one to 1.1 camps with fire rings on river segment two. Norm agreement was greatest for segment three (s=1.40). However, there were no statistically significant differences (p=.05) for the three segments. Whittaker (1992) reported on the results of a study examining evaluations of acceptable instream flows for recreation activities on the Dolores River in Colorado (Shelby and Whittaker, 1990). "Experienced users were asked to evaluate the flow levels which provided minimum or optimum opportunities for recreation" (Whittaker, 1992, p.32). The authors developed impact acceptability curves to indicate the satisfaction users had for different flow levels for rafting, kayaking and canoe craft. The range of tolerable flows ranged from about 200 cubic feet per second (cfs) to 2,500 cfs for canoers; about 650 cfs and up for rafters; and about 800 cfs and up for kayakers. As indicated earlier, the majority of empirical studies examining impact acceptability have investigated various types of encounter levels. In their study of visitor use in the Bear Trap Canyon Wilderness, McCool et al. (1990) asked survey respondents to indicate the 26 maximum number of floater and land-based groups they could accept seeing per day before those groups began to detract from their enjoyment. Six response levels were provided for each type of group: 1 - none; 2 - one to two; 3 - three to five; 4 - six to ten; 5 - eleven to twenty; and 6 - more than twenty. Results were reported in the form of frequency distribution tables and tested the null hypothesis that there was no difference in the distribution of responses among three groups: private floaters, outfitted floaters and hikers. There were significant differences in how the three groups answered the question for both land- based and floater groups. Forty percent of outfitted floaters would accept seeing more than five land-based groups compared to nearly two-thirds of hikers and private floaters. In terms of meeting floater groups, 60 percent of hikers would accept seeing more than five other groups compared to about 54 percent for private floaters and 34 percent for outfitted floaters. The researchers did not calculate means, medians or other statistics with which to further analyze the degree of shared agreement for acceptable encounter levels. Using data from a 1987 study of recreation use impacts on the lower Deschutes River in central Oregon, Whittaker and Shelby (1988) analyzed a number of encounter norms including river encounters, jet boat encounters, and camp encounters. Randomly selected boater pass purchasers were sent a questionnaire asking them to report the level of encounters they would tolerate. Users then responded to specific statements such as, "It is O.K. to be in sight of other parties as many as hour (s) out of four when boating" a specific segment of the river (p.264). The researchers included an "it doesnt matter to me" category for respondents to use if they couldn't give a specific number. Results were reported in the form of curves which aggregated personal norms into social norm curves. These norm curves were then compared statistically whereby "the social norm was represented by the median response, the tolerable range was defined as being zero to the median and norm crystallization or group agreement was represented by the standard deviation around the mean" (p.265). For different segments of the river, there was variation between norms for river encounters, jet boat encounters, and camp encounters. The median tolerable level for river encounters ranged from 1.8 (hours in sight out of four) for two segments to 2.2 for the third 27 segment. Medians for jet boat encounters ranged from 0.3 (boats per day) on river segment 1 to a high of 9.7 on segment 3 for jet boaters. Median tolerable levels for camp encounters ranged from 1.4 (nights out of four) on segment 3 to 1.9 on segment 2. In conclusion, the authors felt that, overall, the study results indicated that: social norms do exist and reflect shared personal standards; there are three basic types of social norms (no tolerance, single tolerance and multiple tolerance); when social norms differ in different settings, the variation appears to be associated with setting attributes; and, finally, the highest level of agreement appears to be for no tolerance norms i.e., zero impact. Roggenbuck et al. (1991) present contrasting results to the last study by Whittaker and Shelby (1988). The empirical data were collected from commercial river floaters and private boaters on the New River Gorge National River in West Virginia. Both pre-trip and post-trip questionnaires were used to gather information on encounter norms for three different river experiences: wilderness Whitewater, scenic Whitewater, and social recreation trip. The study was unique in that it attempted to determine norm crystallization (shared agreement) for managerially relevant sub-groups of the sample population. Stated briefly, study results showed that "in no case did two-thirds or more of the respondents have norms about encounters" (p. 140). A sizable percentage (over 20 percent of wilderness and scenic Whitewater trips and 13 to 20 percent of social recreation trips for all three encounter types) of respondents who said that encounters made a difference could not give an "acceptable" number. In addition, the authors found almost no difference among sub-groups of the river population concerning the level of norms for two of the three encounter types examined. A number of conclusions can be drawn from the results of the empirical studies presented above on user norms. Relatively few studies have investigated user norms for ecological impacts. In one such study which had this focus, norms of moderate to high intensity were found. Additionally, study respondents were able to express the acceptability of impacts and rarely chose the neutral category (Vaske et al, 1988). As stated earlier, the majority of empirical studies exarnining impact acceptability investigated encounter norms. Results from the studies presented above suggest a number of 28 general conclusions related to this line of research. First, not all researchers have confirmed the existance of encounter norms. Second, where found, encounter norms seem to vary for differently defined user groups and at different locations in study areas. Lastly, the diversity of encounter norms found have prompted researchers to become "increasingly specific in the type of norms measured" (Vaske et al., 1992, p. 30). THE HUMAN JUDGEMENT MODEL As was pointed out earlier, the importance of identifying values and evaluative judgments is a primary focus in current wilderness management frameworks like LAC. A previous section noted how recreation researchers have used models developed in the behavioural sciences to identify important impacts and impact standards based on recreationists1 evaluations or preferences. There is also a considerable body of research and numerous modeling approaches developed in the social sciences concerned with understanding how individuals and groups make evaluative judgments. This section will examine one such approach identified in the sociological literature - the human judgment model. The model's concepts and accompanying methodological approach has much in common with the stated preference and choice models described earlier. In addition, because social scientists are particularly interested in the social components of judgments, the model pays particular attention to the issue of shared agreement or social consensus. In this respect, the model attempts to analyze and measure social norms using measures of central tendency and dispersion similar to Jackson's return - potential model described earlier. Rossi and Anderson (1982) describe a model which has as its primary focus the uncovering of the principles that lie behind human judgments or evaluations. Basic to the model is the assumption that most human judgments have a structure consisting of several elements. First, when making judgments about objects, actions, persons or ideas, people pay attention to only a relatively small number of characteristics of the objects. Rossi and Anderson use the example of buying a car. For example, among the large number of ways in which cars differ, car buyers for the most part consider only a relatively few characteristics or 29 attributes when deciding on a particular car to purchase. Second, judgments about many domains or broad categories of characteristics are socially structured. In other words, there is more or less agreement among persons on how much weight should be given to relevant characteristics and on how such characteristics should be combined in order to make judgments. Thus, for example, car buyers generally may weight fuel efficiency as more important than driver comfort. (Rossi and Anderson, 1982, p. 17.) Lastly, individual judgments tend toward consistency. That is, each individual will depart in relatively regular ways from the "socially defined consensus on how such judgments should be made" (Rossi and Anderson, 1982, p. 17). For example, while most car buyers may weigh fuel efficiency as the most important characteristic, one individual buyer may weigh this characteristic more or less heavily. However, it should be noted that it is possible that in a set of empirical value judgments no structure may exist or that any structure which does appear may be weak. This could be expected when an individual is "presented with a set of completely unfamiliar objects (for example, food from an exotic cuisine...", p. 17). Rossi and Anderson point out that it is relatively simple to determine judgment principles "for classes of objects which differ only with respect to a single dimension" (p. 18). However, the modeling task becomes more complicated when the objects differ with respect to many attributes. Judgments about social objects typically involve objects that have many traits on which they vary. When many dimensions, each of those with many levels, are used, the techniques and experimental designs...break down (Rossi and Anderson, 1982, p. 18). This problem can be illustrated with an example from the wilderness recreation literature discussed earlier. We might assume, for example, that the quality of the wilderness recreation "experience" was determined by the conditions of campsites, encounters with others, and sight and sound intrusions. If these experience attributes were perfectly correlated, we might say, for example, that judgments about the quality of the experience could be determined solely by the condition of the campsite. However, empirical studies have shown that wilderness campsite conditions vary in terms of factors such as the presence of litter, tree damage, vegetation loss, etc. Therefore, sub-decisions about campsite conditions must be considered before the relative importance of site conditions versus encounters and sight and sound 30 intrusions can be determined. Li other words [using Rossi and Anderson's (1982) terminology], the specific features (factors) of site conditions, encounters and sight and sound intrusions are combined in some way when someone evaluates the quality of one wilderness experience versus another. The above example serves to illustrate the complexity of the modeling task and more specifically, highlights the limitations noted earlier of the compositional approach used in stated preference models. (The specific experimental design used in conjunction with the human judgment model to address these limitations will be discussed in a later section.) Mathematical Representation of the Human Judgment Model Figure 2 shows the mathematical representation of the overall judgment process in the human judgment model. The object being evaluated is characterized by a set of variables (factors) X, X2,...Xj c When values of these variables are specified, there will be a predicted judgment of J}, based on the regression of J[ on the object dimensions. That conditional expression jj is equivalent to (bQ + bj k{\ + ...bfc xy ,^ and in regression theory it is sometimes referred to as the systematic component. The residual, ej, represents the deviation of an individual judgment, j}, from that predicted given the x^ characteristics of the object and the best fitting estimates of the b coefficients (Rossi and Anderson, 1982, p.20). The authors point out, rightly, that ej in practice also includes not only individual differences but also measurement errors and other influences such as omitted variables (p.21). But under the assumption that only individual deviations from the social calculus are involved, we can view the expression for a judgment as being comprised of two parts, one part a function of the characteristics and the socially agreed-upon weights attached to the characteristics and one part representing individual deviations from that consensus...this information offers a way to estimate social and individual components of social judgments... The social components can be defined "to be the best estimates of individual judgments and judgment processes that can be made by pooling the judgments made by individuals in a population ( or more usually by a sample of a population)" (p.21). One can then estimate an 31 Figure 2 Mathematical Expression of the Overall Judgement Process of the Human Judgement Model Ji = b 0 + b! xi! +...bkxjic + ei where Jj = the judgement about an object given by individual i b 0 = the intercept term for individual i bk = regression coefficients or weights given by individual i x ik = independent variables or coded characteristics of the object being rated by individual i ei = random error for individual i Source: Rossi, P.H. and Anderson, AJB. (1982) The Factorial Survey Approach. In: Peter H. Rossi and Stenen L. Noch (Eds.). Measuring Social Judgements. Sage Publications, Beverly Hills, California, p.20. 32 expected rating or evaluative judgment for each object by modeling how the individuals in the population studied used the dimensions to arrive at their judgments. Qualitative as well as quantitative variables can be represented in the equation in Figure 2. This can be done by representing qualitative variables as "binary" variables (dummy variables) in which case the b coefficient is the increment (or decrement) in the estimated judgment that is received by an object that has the characteristic in question" (Rossi and Anderson, 1982, p.21). Although dealing effectively with the overall judgment process, the equation in Figure 2 fails to identify the ways members of the sample population vary in their evaluations and thus doesn't explicitly measure the degree of shared agreement. However, there are a number of comparisons which could be made to examine sub-group variations in much the same way as Jackson (1965) measured social norms. Figure 3 shows five measures identified by Rossi and Anderson (1982) which could be compared after computing separate regressions for different sub-groups using the equation in Figure 2. The Factorial Survey Approach The experimental design most often used in conjunction with the human judgment model is referred to as the factorial survey approach. As applied by sociologists, it shares much in common with the decompositional approach used in stated preference and choice models discussed earlier. It has been evolving over the past forty years and has been used by sociologists to examine how individuals and groups make evaluations. "Factorial surveys are so named because they combine ideas from balanced multivariate experimental designs with sample survey procedures" (Rossi and Anderson, 1982, p. 15). The method is described as follows (Rossi and Nock, 1982, p. 10): In simplest terms, factorial surveys consist of providing individuals with contrived hypothetical situations/objects which are to be evaluated according to some process being studied. The construction of such situations/objects follows factorial experimental protocols which ensure orthogonality of all components of the situations/objects. Individuals then respond to a sample of all possible contrived situations/objects. It is this latter component of the factorial survey which most strongly marks its departure from factorial experiments. 33 Figure 3 Comparative Measures of the Extent of Sub-Group Agreement 1. Judgment thresholds (average levels of judgments rendered) Jq+Jm where q and m are sub-groups (e.g., gender) 2. Judgment Variance 0 2 Jq^0 2 Jm 3. Judgment Error (Stochastic error) 1—R2q-/-1—R2m 4. Judgment Process (weights given to different dimensions or levels) bkq-f°km 5. Systematic Variation (intercepts) boqV'bom Source: Rossi, P.H. and Anderson, A.B. (1982) The Factorial Survey Approach. In: Peter H. Rossi and Steven L. Noch (Eds.). Measuring Social Judgements. Sage Publications, Beverly Hills, California, p.20. 34 Thus, the factorial survey approach shares the advantages noted by Haider (1993) for stated preference models using the decompositional approach, namely: large numbers of attributes can be combined to describe hypothetical profiles or evaluation objects which more fully resemble real-world conditions; holistic situations rather than single attributes are presented to survey respondents; profiles or situations/objects can be constructed orthogonally, thereby reducing multicollinearity problems inherent in most surveys; and enable the researcher to clearly identify the relative importance or saliency of many independent variables (attributes or factors). The following example using impacts on wilderness conditions identified by the literature in a previous section can be used to illustrate the construction of hypothetical situations/objects to be evaluated (rated) by survey respondents using a factorial survey. Consider the following two impact dimensions and one setting dimension, each containing two factors: Dimension A: Site Conditions By calculating the factorial (2 factors, Dimension A times 2 factors, Dimension B times 2 factors, Dimension C) we see that there are 8 possible hypothetical attribute profiles. For instance, one possible combination could contain the first factor from each dimension. A survey respondent could then be asked to evaluate the following situation. "While you are hiking along a trail in the wilderness area, you see some litter near the trail. Another hiker is seen approaching." Please rate (by circling a number) the situation described as to the extent to which it would affect the quality of your wilderness experience. Dimension C: Factors: Dimension B: Factors: Factors: 1. The presence of litter 2. Trees damaged by humans Encounters 1. Individuals seen 2. Groups seen other than my own Settings 1. On the trail 2. In camp 35 Extremely negative No effect Extremely positive effect at all effect •4 -3 -2 -1 0 +1 +2 +3 +4 The Factorial Object Universe Concept To clarify the concepts associated with using the factorial survey approach as described by Rossi and Anderson (1982), consider the following definitions and characteristics (pp. 28-30) using examples from the recreation literature cited previously: 1) Dimension - a quality of wilderness environments or a variable characterizing such environments that can vary in kind such as site conditions, encounters, sight and sound intrusions, etc. 2) Factors (levels in the sociological literature) - the specific values that an attribute dimension may take, for example, "trees damaged by people" is a factor of the dimension "Site Conditions," "individuals seen" is a factor of the dimension "Encounters", "aircraft heard" is a factor of the dimension "Sight and Sound Intrusions" 3) Object - a unit being judged that is described in terms of a single factor for every dimension. An "object" may consist of this statement: "A site contains litter, an individual can be seen and an aircraft can be heard overhead." 4) Judgment - rating, rank, or other valuation given by a respondent to an object 5) Factorial Object Universe - the set of all unique objects formed by all possible combinations of one factor (level) from each of the dimensions 6) Factorial Object Sample - an unbiased sample of the objects in a factorial object universe 7) Respondent Sub-Sample - an unbiased sample of a factorial object universe that is given to a single respondent for judgment Characteristics of a Factorial Object Universe include the following: 1) Dimensions are approximately orthogonal, each factor in a dimension appearing equally frequently with each factor in every other dimension. 36 2) Any factor within a dimension will appear in the factorial object universe as frequently as any other factor within that dimension. 3) A factorial object sample is an unbiased sample of the members of a factorial object universe that preserves the essential characteristics of the object population. 4) A respondent sub-sample is a sample of a factorial object sample drawn in a random manner for the purpose of eliciting judgments from a given respondent. Thus, individual respondents are given relatively small sub-samples of factorial objects that can be put together with other respondent sub-samples to form relatively large samples of factorial object populations. The factorial survey method presumes that a researcher possesses some knowledge of the evaluation process being studied. That is, factorial surveys help one model a decision process, but the resulting model depends on the original specification of the process by the researcher. One cannot discover that X is important to a decision process if X was not included in the original specification of that process. As such, factorial surveys will be most fruitfully applied to problems of evaluation which are understood at least tentatively. (Rossi and Nock, 1982, p. 10.) The usefulness of the factorial survey approach derives in part from its ability to overcome some of the limitations of its "parent" designs. First, because situation/objects can be built orthogonally (as in experimental designs), difficulties associated with multicollinearity in survey designs are lessened considerably. Second, because relatively large numbers of dimensions and factors within dimensions can be used (as in sample survey designs), difficulties associated with oversimplified experimental conditions are overcome. In short, from the experimental tradition, the factorial survey borrows and , adapts the concept of factor orthogonality and from the survey tradition, it borrows the greater richness of detail and complexity that characterizes real-life circumstances... (Rossi and Anderson, 1982, pp. 15-16.) Previous Empirical Studies Empirical studies testing the human judgment model utilizing the factorial survey approach have investigated a number of human evaluation issues. Berk and Rossi (1977) used the approach to evaluate typical and desired treatments (sentencing) of convicted offenders. Survey respondents included prosecutors, state and local officials, judges, police, public 37 defenders and various other corrections officials and employees. The rating task consisted of designating one of nine treatments to a set of hypothetical cases of convicted offenders. Each factorial object rated consisted of combinations of factors from three dimensions: age of the offender (5 factors); previous record (7 factors); and crime seriousness (40 factors). In research on status attribution, Nock (1982) "investigated the relative status value of diverse family characteristics, some achieved and some ascribed" (p. 95). Survey respondents were asked to rate the social standing of a set of 50 descriptions of hypothetical households. A nine-point scale was used to rate short vignettes containing information on the following dimensions and factors: occupation of husband (50 occupations); education of husband (fifth grade through college); occupation of wife (50 occupations); education of wife (fifth grade through college); migration status (10 factors); husband's father's education (fifth grade through college); husband's father's occupation (50 occupations); wife's father's occupation (50 occupations); and ethnicity (10 categories). Garret (1982) reported research aimed at defining what constitutes "child abuse." More specifically, the "intent of the research is to provide a model of how various components of an incident combine to influence a seriousness rating of that event" (p. 180). Survey respondents rated 60 hypothetical vignettes which included descriptions of an act involving a guardian and a child, descriptions of the guardian and the child, the social status of the household as measured by the occupation of the main breadwinner and one of six ethnic categories for the household. Rossi and Anderson (1982) described a study by Richard Berk and Marlyn Brewer undertaken to measure the social definition of sexual harassment. In the study, undergraduate students at the University of California at Santa Barbara responded to a mail-back questionnaire which included 25 factorial objects. The respondents were asked to give judgment ratings on a continuum ranging from 1 ("definitely not harassment") to 9 ("definitely sexual harassment"). The factorial object universe consisted of combinations of the following dimensions and factors: status of male (10 factors ranging from "single graduate student" to "married 65 - year old professor"); status of female (4 factors ranging from "single graduate 38 student" to "married graduate student"); the woman's relationship to the man (6 factors ranging from "had rarely had occasion to talk to" to "after being asked, had declined to go out with"); social setting (10 factors ranging from "when class had ended she approached him" to "they were both at a party"); woman's receptivity (12 factors ranging from "she seemed worried and asked about grades" to "blank text" or no description of receptivity); male's verbal behaviour (12 factors ranging from "he asked her about her other courses" to "he talked about his last lecture"); male's physical acts (10 factors ranging from "he reached out and straightened her hair" to "blank text" or no description of the male's physical acts; and male's threat (5 factors ranging from "he promised he would do everything to help her" to "blank text" or no description of the male's threat). The number of unique factorial objects in the universe (excluding "equal status" descriptions from the two status dimensions) was 21,565,440. The studies described above are not exhaustive of the empirical research which has used the factorial survey approach. They were included to illustrate: 1) the range of human judgment topics examined in the literature; and 2) numbers of dimensions and factors within dimensions that can be used to model the complexity and conditions of "real-life" human choices and judgments (Rossi and Anderson, 1982). Techniques used to analyze data sets employing the factorial survey approach will be described in the following chapter. HYPOTHESES The preceding literature review has identified a number of limitations and issues which need to be addressed in research related to identifying salient impacts and impact acceptability for wilderness conditions. The research questions and research hypotheses described in Chapter 1 seek to address some of the issues raised by wilderness recreation researchers. This section will review the research questions and hypotheses and identify the specific issues/limitations associated with them. The first research question raised in Chapter 1 and its companion Null Hypotheses 1A, IB and 1C are related to impact saliency. As the literature review pointed out (Roggenbuck et al., 1991 and Whittaker, 1992) there is a need to more sufficiently address the full range of 39 salient impacts issues. In testing Null Hypotheses 1A, IB and 1C three such impact issues will be included: the behaviour of other people encountered; the specific setting in which impacts occur; and a number of sight and sound intrusions - namely, the presence of aircraft, structures built by people, and forest service signs. (More detail on the inclusion of these impacts and methods to address other issues/limitations will be presented in Chapter 3.) Other issues/limitations related to identifying the relative importance of impacts concerns modeling approaches. As pointed out earlier, few empirical studies have exploited the advantages offered by approaches like the stated preference decompositional method (Haider, 1993 and Louviere and Timmermans, 1990) and the factorial survey approach (Rossi and Anderson, 1982 and Rossi and Nock, 1982). This study will use hypothetical situations or attribute profiles to identify the relative importance of impacts and capitalize on advantages of these alternative methods to: 1) miriimize problems of multicolinearity inherent in many surveys; 2) model the complexity associated with multiple attributes/levels; and 3) provide situational contexts which more fully represent real-life evaluations of the significance of impacts on wilderness conditions. Research question three is related to the issue of determining the extent of shared agreement associated with Null Hypotheses 1A, IB and 1C. Analytical techniques will be used to examine issues related to the measurement of shared agreement as raised by Roggenbuck et al. (1991) and Vaske et al. (1986). More specifically, sources of subject variation will be examined for three relevant sub-groups of the study area's visitor population. The need to investigate variables which might explain variation between different user groups for social impact judgments was considered a "major issue for future research" by Williams et al. (1992, p.175). In this study the extent of agreement for salient impacts and impact standards will be examined for sub-groups characterized by: 1) trip organization i.e., commercially outfitted versus privately outfitted; 2) length of stay; and 3) place of origin of the visitor i.e., residents of B.C. versus residents of the U.S.. (Specific rationale for these sub-groups distinctions is discussed in Chapter 3.) 40 The second research question raised in Chapter 1 and addressed by Null Hypotheses 2A, 2B and 2C concerns the maximum amount of specific impacts wilderness visitors will accept before their wilderness experience would be changed. A number of issues were raised and limitations of previous research were noted in the review of literature. First is the issue of whether wilderness visitors have different impact acceptability levels for different locations in a wilderness. The research by Roggenbuck et al. (1991); Whittaker and Shelby (1988); Shelby et al. (1988); and others seems to indicate that there is a high degree of variability of impact acceptability levels within specific wilderness areas. Null Hypotheses 2A, 2B and 2C address this issue and thereby seek to overcome limitations in many past empirical studies. Another limitation noted by Roggenbuck et al., (1991) in previous studies seeking wilderness visitor opinions on acceptable levels of impact is of a methodological nature. They make the case for including an "I can't give a number" response option when asking survey respondents to indicate maximum amounts (numbers) of impact they would accept. As will be pointed out in the following chapter, such a response category was offered respondents in the Spruce Lake Trails Area study to reduce the possibility of respondents "guessing" or merely stating some number to please the researcher. Finally, the question of the extent of shared agreement raised in Research Question (4) will be investigated. Subject variability for average maximum acceptable amounts of impact will be examined for the three sub-groups discussed above using sample statistics described in Chapter 3. 41 Chapter 3 METHODOLOGY The hypotheses in this study were tested with data from a 1992 survey of visitors to the Spruce Lake Trails Area, British Columbia during the primary use season (June through September). THE SURVEY INSTRUMENT A pilot test of a sample questionnaire was administered to a number of undergraduate and graduate students in the Faculties of Sociology and Forestry at the University of British Columbia, Vancouver, B.C.. No significant changes were made to the questionnaire as a result of the pilot test. Each questionnaire consisted of five parts. (See Appendix 1 for a sample questionnaire.) Parts I, JJ, UJ and V were used to test Null Hypotheses 1A, IB and 1C. Null Hypotheses 2A, 2B and 2C were tested using responses from Parts U, rv and V. Parts II and V contained questions about trip characteristics and demographic information, respectively. THE SURVEY SAMPLE Names and addresses of potential survey respondents were obtained in three ways. A sign on information boards at major trailheads (see Appendix 2) explained the research study and asked visitors to fill out a brief contact card (Appendix 3) which included their name and address and a few short questions about their trip. Visitors were told that they might be sent a questionnaire later. Other names and addresses were obtained at Spruce Lake itself. The resident custodian approached visitors and explained the purpose of the study. They were asked to fill out the contact card and told to expect a questionnaire to be mailed to them later. Lastly, the guide-outfitter who operates in the study area provided a list of his clients who volunteered to participate in the study. A modification of the Total Design Method described by Dillman (1978) was used to administer the mail-back questionnaire. The initial mailing consisted of an introductory letter (Appendix 4), the questionnaire and a stamped self-addressed return envelope. The first 42 mailing was followed one week later with a reminder letter (Appendix 5) and two follow-up mailings. Each follow-up included an introductory letter (Appendixes 6 and 7) and another copy of the original questionnaire. Contact cards and guide-outfitter client lists were collected once in June, once in July and twice in August and September. The mailing sequence was started about a week after the cards and lists were collected. Table 1 shows that there were six initial mailings of questionnaires. Rerriinder letters were sent one week after the initial mailing with the second and third follow-ups coming three and six weeks, respectively, after the initial mailing. In total, 301 questionnaires were delivered. Of the total, 251 questionnaires were returned. One questionnaire from the first mailing in August was returned blank. As table 1 indicates, a total of 250 usable questionnaires were returned which resulted in an overall response rate of 83.1 percent (250 / 301). Sub-Groups of the Survey Sample Null Hypotheses 1A through 1C seek to determine the relative influence of specific impacts on wilderness conditions for sub-groups of the sample population. The need to examine sources of subject variation in impact saliency and impact acceptability has been identified by numerous researchers, most recently by Williams et al., 1992. The three sub-groups selected as most relevant for the empirical study were determined after consultation with B.C. Forest Service's Lillooet Forest District recreation staff and a review of wilderness recreation literature. While the B.C. Forest Service's primary mandate is to provide opportunities for public recreation, provisions are made for recreation users employing commercial operators for access into the area. Differences in attitude and preferences between commercially-outfitted and privately-outfitted users of publicly owned wilderness lands has been documented in past empirical studies. Numerous studies reviewed in Chapter 2 indicated that significant differences seem to exist between the two groups regarding the relative importance of impacts on ecological and social conditions in wilderness (Shelby and Heberlein, 1986; Krumpe et al., 1989; McCool et al., 1990; and Roggenbuck et al., 1991). In addition to the 43 TABLE 1 Questionnaire Mailing Schedule and Response Rates Initial Deliverable Returned Cumulative Usable Mailings Number Number Percent Number Percent June 40 34 85.0 34 85.0 July 102 78 76.5 112 78.9 August (1st) 68 57 83.8 168 1 80.0 August (2nd) 29 29 100.0 198 82.8 September (1st) 54 46 85.2 244 85.6 September (2nd) 8 7 87.5 250 83.1 TOTAL 301 251 83.4 250 83.1 One questionnaire returned but not filled out 44 above research evidence indicating the relevance of examining the extent of agreement between these two groups, this potential for conflicting values has been recognized specifically for the area under study. In the Integrated Resource Management Plan for Spruce Lake (B.C. Ministry of Forests, 1981) a major potential problem was highlighted (p.9), namely, "potential conflicts between commercial and public recreation". Based on the above rationale, Null Hypothesis 1A compared the relative influence of specific impacts on wilderness conditions for commercially and privately-outfitted survey respondents. Null Hypothesis IB compares the relative influence of impacts for sub-groups characterized by their length of stay. None of the previous research on impact saliency has investigated this variable for its potential to explain differences in normative consensus regarding important impacts. Another reason for including this sub-group of the sample population in the research hypothesis is its relevance for the study area. As mentioned in Chapter 1, float planes are a major means of access into Spruce Lake. In fact, air flights increased from 1990 to 1991 by 26 percent (B.C. Ministry of Forests, 1992). The length of stay for aircraft parties averaged 1-2 days in 1991 with some visitors staying only a few hours. For those visitors accessing the area by horse, the average length of stay was 6-7 days. As was the case with commercially-outfitted versus privately-outfitted visitors, a potential for conflicting values exists for visitors seeking shorter versus longer wilderness experiences. Managers of the Spruce Lake Trails Area are interested to know if these sub-groups view differently the importance of ecological and social impacts. Null Hypothesis 1C compares the relative influence of specific impacts on wilderness conditions by visitors characterized by their place of origin. In 1989 (the last year user characteristics were estimated by place of origin), visitors to the study area from B.C. made up over 83 percent of total visitors. Non B.C. residents of Canada accounted for nearly four percent while foreigners accounted for approximately 11.5 percent of total visitors. However, Spruce Lake managers have reason to believe that use by non-residents of B.C. and non-residents of Canada is increasing. This is thought to be due to the increasing number of foreigners visiting the adjacent resort area of Whistler, B.C. and seeking "wilderness type" 45 excursions in the area. Managers are interested to know if these new visitors from other cultures have different attitudes toward impacts and impact standards. A case for investigating this source of sub-group variation in values was made by Rollins (1985, p.8): ...it could be argued that residents of British Columbia will hold different expectations than will visitors from other parts of Canada, or visitors from other countries. These differences may be due to differences in cultural norms, differences in experiences (e.g. with west coast climate and geography), or differences in available information... With the above considerations in mind, Null Hypothesis 1C examines variations in impact saliency for two sub-groups characterized by their place of origin: B.C. residents and U.S. residents. Respondent numbers for other places of origin (non B.C. residents of Canada and non-U. S. foreign visitors) were judged too small to test the hypothesis. Information provided by survey respondents on trip characteristics (Part II of the questionnaire) and demographics (Part V of the questionnaire) was used to develop categories , for the sub-groups used in testing the hypotheses in this study. Table 2 shows frequency and percentage values for each category of the three sub-groups. Two categories were used to characterize the 250 survey respondents by their trip organization. As table 2 shows, nearly 60 percent of the respondents were privately-outfitted while about 40 percent were outfitted commercially. Included in the latter are those respondents who were guided by the commercial guide who operates in the study area and those respondents who were flown to Spruce Lake by commercial float plane pilots. Three respondents (1.2 percent) failed to indicate how their trip was organized and their responses were not included in hypotheses testing for this sub-group. The second sub-group characterized respondents in terms of their length of stay. The first category consisted of day users of the study area and represented 16.8 percent of survey respondents. Category two ("two to four days") represented just over 40 percent. This category includes those overnight visitors on weekend trips (two to four day weekends). The last category includes those visitors staying for extended visits in the study area. Respondents 46 TABLE 2 Frequency Values for Sub-Groups by Categories Sub-Group Cumulative Cumulative Categories Frequency Percent Frequency Percent Trip Organization: 1- OommerciaUy-outfitted 99 39.6 99 39.6 2- Privately-outfitted 148 59.2 247 98.8 Missing 2 _3 1.2 250 100.0 TOTAL 250 100.0 Length of stay: 1- Oneday 42 16.8 42 16.8 2- two to four days 101 40.4 143 47.2 3- More than four days 105 42.0 248 99.2 Missing 2 __2 .S 250 100.0 TOTAL 250 100.0 Place of Origin: 1- ResidentofB.C. 197 78.8 197 78.8 2- ResidentofU.S. 44 17.6 241 96.4 Other3 8 3.2 249 99.6 Missing2 __1 0.4 250 100.0 TOTAL 250 100.0 1 Based on 250 usable questionnaires returned. 2 Values shown represent questionnaires with missing responses and were not included in data analysis for sub-groups. 3 Non-B.C. residents of Canada not included in data analysis for sub-groups. 47 staying more than four days made up forty-two percent of the total. Two respondents (.8 percent) did not answer the survey questions on nights spent in the study area. Two categories were used to characterize respondents by their place of origin. Over three-fourths (78.8 percent) of respondents returning usable questionnaires were residents of B.C. while 17.6 percent were U.S. residents. Eight respondents (3.2 percent) were non-B.C. residents of Canada but were not included in the data analysis because of the small size of the sample. One respondent failed to indicate where he permanently resided. To examine whether the above three sources of subject variation were indeed distinct, Pearson correlation coefficients were computed to measure the strength of the relations between the sub-groups. Table 3 shows the correlation coefficients and p-values. The p-value of 0.0001 associated with the sub-groups of trip organization and place of origin gives strong evidence that their true population correlation is not zero. The r coefficient (-0.3493) indicates that U.S. residents tend to be commercially outfitted while B.C. residents tend to be privately outfitted. Overall, the results of the correlation analysis indicate that while there is some continuity in sub-group membership, the three sources of subject variation were, for the most part, distinct. Representativeness of the Survey Sample As discussed earlier, names and addresses of the sample population were obtained from visitors who agreed to participate in the survey. Because reliable access patterns were not available for the area, it was felt that a classical random sample design was not the best way to collect the questionnaire mailing list. Thus, there is the question of the degree to which the sample population is representative of the total visitor population of the study area. To address this question, use and user characteristics of the Spruce Lake Visitor Study respondents are compared in table 4 to estimates of similar characteristics taken from a 1989 B.C. Forest Service recreation survey (B.C. Ministry of Forests, 1989b) of the study area. In the Forest Service study, visitors to the area were surveyed during the period from June 4 to September 29, 1989, the primary use season. Both personal interviews and a mail-back questionnaire were used to gather information on recreation use, user demographics, 48 TABLE3 Pearson Correlation Coefficients Among Survey Sample Sub-Groups Sub-Groups Length of Stay1 . Trip Organization2 Place of Origin3 Length of Stay -0.13491 0.04857 (p=0.0337) (p=0.4464) Trip Organization -0.34933 (p=0.0001) Place of Origin 1 Length of Stay category values: 1= 2 Trip Organization category values: 3 Place of Origin category values: 1= =one day; 2=two to four days; 3=five and more days 1= Commercially outfitted; 2=Privately outfitted =BC residents; 2=US residents 49 TABLE 4 Survey Comparisons of Use / User Characteristics 1989 B.C. Spruce Lake Forest Service Survey Visitor Study Characteristics n=106 n=250 Average party size ' 5.82 4.67 Average length of stay 4.08 3.55 Place of origin (%): Resident of B.C. 84.6 79.4 Non-B.C. resident of Canada 3.8 2.8 Non-resident of Canada 11.5 17.8 Travel method (%): Hike 21.9 38.1 Horse 25.7 24.6 Aircraft/helicopter 48.6 32.4 Bike 3.8 4.9 Gender (%): Male 70.3 69.2 Female 29.7 30.8 Age(%): Under 20 years 0.0 3.7 20-24 years 1.0 2.4 25-34 years 19.4 24.8 3544 years 32.0 33.3 45-54 years 31.1 23.2 55-64 years 10.7 10.2 65 years and over 5.8 2.4 Trip organization (%): Commercially-outfitted 45.3 40.0 Privately-outfitted 54.7 60.0 50 recreation expenditures and values and to identify management issues. Documentation of this survey does not specifically address the issue of the representativeness of the survey sample. While it is possible that the 1989 study was flawed in terms of "representativeness," it is impossible to know for sure. As the results in table 4 indicate, most statistics for the two samples are similar. The most notable differences occur in the comparisons of average party size and the percent of hikers and aircraft users. On the other hand, the statistics are very similar for the three sub-groups (i.e. characterized by trip organization, length of stay and place of origin) used to investigate subject variability in questionnaire responses for the research hypotheses. Thus for the purposes of this study, the sample population is assumed to adequately represent the visitor population of the Spruce Lake Trails Area. IDENTIFYING SALIENT IMPACTS To determine the relative influence of specific impacts on wilderness conditions and the extent of agreement among respondent sub-groups, the human judgement model and the factorial survey approach (Rossi and Anderson, 1982) were used. This approach was selected to take advantage of elements described by both Haider et al. (1993b) for the decompositional approach used in stated preference and choice models and Rossi and Anderson (1982) for the factorial survey approach. More specifically, these approaches: can model large numbers of independent variables characteristic of complex human judgement processes; and minimize the multicolinearity problems inherent in other survey designs by maximizing the orthogonality between the independent variables. The model as used here describes the relationship between an individual's rating (judgement) and the object (situation) being rated (judged) as follows: Ri = f(Sj) + e where Rj = rating given to situation i Sj = characteristics of information of situation i e = random error (adapted from Nock, 1982) 51 In Parts I and m of the questionnaire, respondents were asked to rate hypothetical situations (factorial objects) he / she might encounter while on a wilderness trip. The characteristics included in the situations consisted of combinations of one factor from each of the impact dimensions listed in table 5. It is necessary to point out that because hypothetical situations of impact were evaluated by respondents, the perception of damage and not actual observed damage or impact is under investigation. Additionally, not all potential sources of impact were included in the dimensions and factors. The dimensions and factors include those identified as salient in the literature review. In general, factors in Dimensions A and C are related to ecological impacts or the criteria for naturalness of conditions as described in Chapter 2. The factors in Dimensions B and D pertain to impacts on social conditions or the criteria for solitude (i.e., cognitive freedom and intimacy). Other factors were selected to explore relationships rarely tested in previous studies of this kind. These were: factor 13-Forest Service signs seen; factor 16-discourteous behavior by others, and factor 17-noisy or loud behavior by others. In addition, factors 19-21 were included to examine the importance the specific social setting has in the more general situational context of impacts to wilderness visitors' experiences. The last factor in each impact dimension (factors 6, 15, 18, and 22) was included primarily because of analytical considerations which will be discussed in Chapter 4. Factorial Object Universe By calculating the factorial of the impact factors in table 4, we see that there were a total of 1440 possible combinations (six factors in Dimension A times four factors in Dimension B times five factors in Dimension C times three factors in Dimension D times four factors in Dimension E). However, not all possible combinations of factors were allowed. Following the treatment by Garret (1982), combinations which described impossible situations were excluded. That is, a hypothetical situation could not be used which described the behavior of others if no encounters took place. An example of an impossible combination is as follows: "While you are in the area, you notice some trees damaged by humans. You encounter no other visitors. Some of them are loud and noisy." This situation (combination of factors) is 52 TABLE 5 Impact Dimensions and Factors Impact Dimension Factors Dimension A: Site Conditions 1. Litter seen 2. Trees damaged by humans 3. Vegetation loss and bare ground 4. Campfire rings made by others 5. Positive site condition (corresponding to setting from Dimension E) 6. No description of site condition Dimension B: Encounters 7. Individuals seen 8. Groups seen other than my own 9. Large groups seen other than my own 10. No encounters Dimension C: Sight and Sound Intrusions 11. Aircraft sighted or heard 12. Structures sighted built by others 13. Forest Service signs seen 14. No structures seen 15. No description of sight or sound intrusions Dimension D: Behaviour of Others Dimension E: Social Setting 16. Discourteous behaviour by others 17. Noisy or loud behaviour by others 18. No description of behaviour 19. On the trail 20. In camp 21. By the lake 22. In the area 53 impossible because you cannot describe the behavior ("loud and noisy") of someone not encountered. Therefore, those situations containing factor 10 from Dimension B: Encounters could not be combined with situations containing factors 16 and 17 from Dimension D: Behaviour of others. Thus, there were 240 "impossible" combinations which were excluded (six factors in Dimension A times one factor in Dimension B times five factors in Dimension C times two factors in Dimension D times four factors in Dimension E). As Rossi and Anderson (1982, p.66) point out, "the introduction of prohibited combinations compromised the orthogonality of the factorial object universe and derivatively the resulting factorial object sample". If complete orthogonality is maintained, intercorrelations between the independent variables (factorial object characteristics or S[ in our model) should be zero (Garret, 1982). Table 6 shows Pearson correlation coefficients for the impact dimension factors (factorial object characteristics or S\). As the table indicates, most intercorrelations among factors from different impact dimensions are minimal. (Correlations among intra-dimensional factors are a product of the research design.) The intercorrelations among factors in the prohibited combinations ranges from -.22 for factor 10 (no encounters) and factor 16 (discourteous behaviour), -.21 for factor 10 and factor 17 (loud or noisy behaviour), and .40 for factor 10 and factor 18 (no description of behaviour). While these results indicate that some degree of colinearity was introduced by the exclusion of some combinations of impact factors, it is felt that the orthogonality of the factorial object universe was not severely compromised. After excluding the "impossible" combinations, the working factorial object universe contained 1200 hypothetical situations. Using a word processor, one to four sentences were written which described each of the 1200 situations (factorial objects). Factorial Object Sample and Respondent Sub-Samples Because of the size of the factorial object universe (1200 possible situations), it was impractical, if not impossible, to ask the 301 potential survey respondents to rate each unique situation or factorial object. In fact, as Rossi and Anderson (1982) point out, this type of rating task is not required (p.32): 54 TABLE 6 Pearson Correlation Coefficients Among Impact Dimension Factors Impact Factors 1 2 3 4 5 6 7 8 9 10 11 1. Litter seen -.20 -.21 -.20 -.20 -.20 .02 .00 -.01 -.03 -.03 2. Trees damaged -.20 -.19 -.20 -.20 .03 .01 .01 .02 .01 3. Vegetation loss -.20 -.21 -.21 .01 -.04 .01 .02 -.02 4. Campfire rings -.20 -.20 .00 .00 .00 .01 .01 5. Positive condition -.20 .00 .02 .00 -.02 -.01 6. No site description -.01 .01 .00 .00 .02 7. Individuals seen -.41 -.43 -.21 -.02 8. Groups seen -.44 -.22 .00 9. Large groups seen -.23 .01 10. No encounters .00 11. Aircraft 12. Structures 13. Signs 14. No structures 15. No intrusions 16. Discourteous behavior 17. Noisy or loud behavior 18. No behavior description 19. On the trail 20. In camp 21. By the lake 22. In the area Dimension A: Site Conditions. Factors 1-6 Dimension B: Encounters, Factors 7-10 Dimension C: Sight and Sound Intrusions, Factors 11-15 Dimension D: Behavior of Others, Factors 16-18 Dimension E: Social Setting, Factors 19-22 55 TABLE 6 CONTINUED Impact Factors 12 13 14 15 16 17 18 19 20 21 22 1. Litter seen -.01 -.01 -.01 .03 .01 .00 -.01 -.01 .00 .01 .00 2. Trees damaged .01 -.01 .00 -.01 -.01 .01 .00 .01 -.01 -.01 .01 3. Vegetation loss .03 .00 -.03 .02 .00 -.02 .01 .01 .00 -.01 -.01 4. Campfire rings -.04 .03 .00 .00 -.01 .00 .00 -.03 .01 -.01 .03 5. Positive condition .00 .00 .00 .00 .02 .00 -.01 .02 -.02 .02 -.02 6. No site description .00 -.02 .03 -.04 -.01 .01 .00 .01 .02 -.01 -.02 7. Individuals seen .00 -.01 .01 .00 .05 .06 -.10 .03 -.01 -.02 .00 8. Groups seen .01 -.01 .01 -.01 .03 .08 -.10 -.04 -.01 .04 .01 9. Large groups seen .00 .01 -.02 .00 .07 .01 -.08 -.01 .02 .00 -.01 10. No encounters -.01 .00 .00 .00 -.22 -.21 .40 .01 .00 -.01 .00 11. Aircraft -.25 -.24 -.24 -.25 .01 .00 .00 .03 -.02 -.01 -.01 12. Structures -.25 -.25 -.26 .01 .01 -.03 .00 .01 -.02 .01 13. Signs -.24 -.25 -.01 .01 -.01 -.02 .02 .01 -.01 14. No structures -.26 -.02 .01 .01 -.02 -.01 .02 .01 15. No intrusions .00 -.03 .03 .01 -.01 -.01 .00 16. Discourteous behavior -.43 -.54 .00 -.02 -.02 .03 17. Noisy or loud behavior -.53 -.01 .02 .00 -.02 18. No behavior description .01 -.01 .01 -.02 19. On the trail -.32 .33 -.34 20. In camp .33 -.33 21. By the lake -.34 22. In the area Dimension A: Site Conditions. Factors 1-6 Dimension B: Encounters, Factors 7-10 Dimension C: Sight and Sound Intrusions, Factors 11-15 Dimension D: Behavior of Others, Factors 16-18 Dimension E: Social Setting, Factors 19-22 56 Fractional replication methods are often used in experimental designs and have much to recommend them, especially for relatively small factorial object universes. For very large factorial object universes, random sampling of the objects can be used, with each respondent-judge being given an independently drawn random sample of factorial objects, respondent sub-samples... In this study, a computer program was used to produce random samples containing 20 randomly selected situations from the 1200 possibilities. Each of the 301 potential respondents were thus given a different set of 20 situations to rate. Because of the nature of the situations being rated (in general, negative descriptions of wilderness conditions), it was felt that a negative response bias could develop in individual response sets. It was decided, therefore, to give each respondent two relatively "positive" situations to rate. These additional situations were added to the randomly selected respondent sub-samples in the questionnaires and were the same for all respondents. Part I of the survey questionnaire contained 10 of the randomly selected situations and one of the fixed "positive" situations (question two). Part Ul contained the other 10 randomly selected situations and the other fixed "positive" situation (question 21). By separating the questions on hypothetical impact situations (Parts I and DX) with questions on trip characteristics (Part U), it was felt that respondent burden would be reduced. The two fixed "positive" situations were not included in the data analysis. The Rating Task Respondents were asked to rate each hypothetical situation as to the extent to which encountering the situation would affect the quality of their wilderness experience. A nine-point Likert scale was used to measure each rating. The scale ranged from -4 ("extremely negative effect") to +4 ("extremely positive effect") with integer steps in between. The mid-point of the scale was zero ("no effect at all"). Data Analysis Procedures for Null Hypotheses 1A, IB and 1C Data from the survey of visitors to the Spruce Lake Trails Area were processed and analyzed using the SAS micro computer statistical package Version 6.04. The analytical techniques and SAS statistical procedures used to test Null Hypotheses 1 A, IB and 1C are described below. 57 As discussed earlier, each of the survey respondents was given 20 randomly selected hypothetical situations to rate. In the analysis, respondent ratings were changed from the original scale (-4 to +4) to a continuum ranging from 1 to 9. For purposes of the study, the unit of measurement for the ratings was assumed to be at the interval level. The three null hypotheses dealing with the relative influence of impacts on wilderness conditions were analyzed using the SAS procedure for general linear models (GLM). GLM is an analysis of variance model which uses dummy (binary) variables to specify model parameters (intercept and regression coefficients) for independent variables. Ratings (dependent variables) given by survey respondents to the hypothetical situations were regressed on the impact factors (independent variables) using the following linear additive model: Rj = b 0 + b! X ! + b 2 x 2 ...+bi cxi c where Rj = rating given to situation } b 0 = intercept term bk = weights given for x^ xfc = impact factors present in situation [ The independent variables were treated as dummy (binary) variables. If a certain factor was included in the situation being rated, its value was one. If it was not included, its value was zero. As described above, the analytic model used in this study was a linear additive equation. Rossi and Anderson (1981) discussed a number of issues related to the appropriateness of using this type of model with data obtained from a factorial survey (pp.60-61): In the event that the model seems incorrect in form (that is, nonlinear and / or non-additive), it is ordinarily possible to rewrite the expression by adding higher-order terms or interactions or by taking logarithms of the variables or by using other methods. Findings related to the issue of non-additive components of the model used here will be discussed in the next chapter. For now, however, the following rationale is presented to justify use of the model tested. 58 The objective of this study is not to develop the best model with which to predict future responses to impact situations. This modeling approach is being used to: estimate the relative importance of factors within each impact dimension in order to develop a sub-set of impact factors which best represents that impact dimension; and to examine the extent of sub-group agreement surrounding the relative influence of the impact factors. Therefore, our interest here (as will be pointed out shortly) is to use the model parameters and other statistics generated by the analysis for purposes other than "to obtain the best fit of the linear model to data." Researchers in previous studies using the factorial approach have used different functional forms (e.g. logistic curves). Rossi and Anderson (1982) examined such an expression and compared the results to a single linear expression using the same data set. Their findings indicated that "even though the logistic expression more clearly fits the data" (a somewhat higher R 2), the signs and size rank order of the regression coefficients were identical for both equations (p.58). In addition, there is the problem of the interpretation of the higher order terms. This is especially difficult for the binary variables used here for the qualitative impact factors. To test Null Hypotheses 1A, IB and 1C, the GLM analysis was run separately for each of the three sub-groups. After regressing the ratings given by respondents to the hypothetical situations on the impact factors, the weights (b-coefficients) given to different factors were tested to see if they were statistically different from zero. For those b-coefficients that were significantly different from zero, a two-sample t-test was used to compare the coefficients between sub-groups. To examine the relative influence of the factors within each impact dimension, each statistically significant b-coefficient was ranked according to its absolute value. The ranks of impact factors within each dimension were then compared between the sub-groups to determine if there were any differences in the relative influence. To further examine the relative influence of the impact factors, sub-group variations and the extent of agreement, a number of statistics described by Rossi and Anderson (1982) were examined: 1) Average ratings were compared for each sub-group to examine rating thresholds. 59 2) The variability of ratings (as measured by the error variance) were compared for each sub-group. 3) The amount of stochastic error (1-R2) was compared for each sub-group to examine rating error. 4) Intercepts were compared for each sub-group to examine the systematic variation in ratings. 5) Coefficients of determination (R )^ values were compared for each sub-group to examine the amount of variation in sub-groups ratings that is explained by the impact factors included in the model. DETERMINING IMPACT ACCEPTABILITY To determine the maximum amount of specific impacts survey respondents would accept, an open-ended response format was used. In Part TV of the survey questionnaire respondents were asked to give one of three possible responses to a list of 10 impacts for four locations in the Spruce Lake Trails Area (see Figure 4). Ideally, the impacts included in this part of a visitor study would be derived from information on impacts judged most salient by the area's visitors and managers. However, since this a priori information was not available for the study area typical impacts found in the literature review were used. Data Analysis Procedures for Null Hypotheses 2 A 2B and 2C As discussed in Chapter 2, researchers have used several measures of central tendency in their attempts to measure and compare the extent of agreement for maximum acceptable amounts of impact. I agree with Shelby and Vaske (1991) that had the specific purpose of the research hypotheses (2A, 2B, and 2C) been to determine a management impact standard, the median would have been the most appropriate measure to examine. However, as pointed out in Chapter 1, Null Hypotheses 2A, 2B, and 2C were formulated to identify sources of sub-group variations. The purpose here, then, is one of deterrnining the extent of group agreement (norm crystallization in social norm theory terminology) rather than to suggest specific impact standards to be applied to the study area. For this reason, the average maximum acceptable amount of impact and the variance were used to facilitate making 60 FIGURE 4 Reproduction of Part IV of the Survey Questionnaire PART IV Conditions typically vary between different wilderness areas and at different locations with the same wilderness area. For the following items, please indicate the maximum amount you would accept in the Spruce Lakes Trails Area before your wilderness experience would be changed. Use "0" for none. If you cant give a number for a particular item or location, leave it blank. AT ITEMS SPRUCE LAKE On or Beside At Trails Camps ELSEWHERE IN THE AREA On or beside At Trails Camps 30. Pieces of litter seen on any one day 31. On any site, the number of campfire rings made by others 32. On any site, the percent of trees damaged by humans 33. On any site, the square metres of vegetation loss or bare ground (1 sq. metre = approx. 9 sq. ft.) 34. Number of individuals seen other than in your own group on any one day 35. Number of other groups seen on any one day 36. Number of large (more than 6 people) groups seen on any one day 37. Number of aircraft/helicopters sighted or heard (excluding high altitude jets) on any one day 38. On any site, the number of human-made structures seen 39. On any site, the number of forest service signs seen 61 statistical comparisons and hypotheses testing. The author acknowledges the limitations associated with the use of these statistics (e.g., potential distortion due to outliers) but agrees with the statement of Shelby and Vaske (1991, p.181): As with most statistical questions, the answer to the question of which measure is most useful for describing norms is "it depends"; partly on the data one has available, and partly on what one is trying to show. The SAS procedure TTEST (a two-sample t-test) was used to test the Null Hypotheses 2A, 2B, and 2C. The p=.05 level was used. As discussed earlier, one additional statistic was used to analyze variations in sub-group responses on the maximum amount of impact they would accept for the impact items at the four locations. The variances (s2) associated with the average values of sub-groups were tested for statistically significant differences using one-way analysis of variance (F-test) to compare the extent of agreement or normative consensus between and within each of the categories for the three sub-groups. STUDY LIMITATIONS Certain possible limitations associated with the use of the Factorial Survey Approach to solicit user opinions on impact salience need to be acknowledged. While the validity of the "human judgement model" has been well documented in the sociological literature (see Rossi and Nock, 1982), the author acknowledges that the reliability of these methods in examining judgements about wilderness conditions has yet to be verified in other studies. The complexity of the relationship between the conditions of wilderness settings and their effects on the quality of user's wilderness experiences were discussed in Chapter 2. One cannot assume, therefore, that the wilderness condition evaluations examined in this study are analogous to the judgements investigated by sociologists like Rossi and Anderson (1982). Indeed, wilderness researchers have acknowledged that such complex judegements may be difficult for wilderness visitors to perform (see Roggenbuck et al., 1991; Shelby and Vaske, 1991; and Roggenbuck et al., 1993). This is, therefore, one of the purposes of this study, to examine whether visitors to our study area can rate hypothetical situations which model these complex relationships. 62 Other possible limitations associated with this study are related to the ability of respondents to recall specific information and the validity of respondent's verbal reports about the "...cognitive process underlying our choices, evaluations, judgements and behavior" (Nisbett and Wilson. 1977). The author acknowledges the debate in the psychological literature over these issues as discussed by psychologists, sociologists and behavioral scientists such as Ericsson and Simon (1980) and Broadburn et al. (1987). Researchers have documented a number of factors which have the potential to affect the accuracy of survey responses such as: interpreting survey questions; placing survey questions "in the context of their general knowledge and their knowledge of the survey's subject matter;" and determining whether their answers are socially desireable (ibid, p. 157). The design of our survey questions took these and other factors into account in an attempt to mitigate any abverse effects. Nonetheless, the possibility of respondents to employ such tactics, for example, as resorting to inferences that use partial information from memory to construct numerical answers (like our maximum acceptable amount questions) is inherent to most contemporary survey designs (ibid). The exact effect these types of factors have on any one survey are, of course, problematic and difficult to quantify. Another limitation of this study is related to the size of the factorial object universe developed to test Null Hypotheses 1A, IB, and 1C. Conceptually, the factorial object approach as described by Rossi and Anderson (1982) could include a significantly greater number of impact dimensions and factors. Indeed, the study of sexual harassment described in Chapter 2 had a factorial object universe which totaled over 21 million hypothetical situations (compared to this study's 1,200). Respondent sub-samples in that study were generated and questionnaires printed using a computer program devised specifically for this purpose. Such a program was not available for our use and each of the 301 survey questionnaires mailed had to be constructed individually on a word processor by the author. This limitation was therefore of a practical nature rather than as a result of limitations of the modeling approach. Another limitation concerns the list of impacts in Part IV of the questionnaire used to test Null Hypotheses 2A, 2B and 2C. As discussed earlier, the ideal situation would be to 63 only include those impacts judged most salient by visitors, interested stakeholders and managers of the specific study area. Visitors would then be asked to provide maximum acceptable amounts for these salient impacts. Since no a priori information of this kind was available for the Spruce Lakes Trail Area, a general list of impacts identified in a review of the literature was used. Hence, it is possible that: 1) other salient impacts were not evaluated for impact acceptability; 2) respondents did not indicate a "maximum amount" because the impact was not salient to them and thus, they could not give a number or had no preference; and 3) respondents simply "guessed" and provided a number to please the researcher. A final limitation of this study, raised above, relates to sampling. No list of Spruce Lake visitors exists and so random sampling from a fixed sample frame was not possible. Comparing distributions of key variables derived from the sample used here with available data on those distributions for the Spruce Lake population showed no major discreptances. Nonetheless, the generalizability of my sample results must be approached cautiously. 64 Chapter 4 RESULTS INTRODUCTION The purpose of this study was to examine sources of subject variation in user estimates of impact saliency and the acceptability of impacts on ecological and social conditions found in wilderness and thus on the experiences of wilderness visitors. More specifically, the study sought to determine which impacts most negatively affect the experiences of recreation visitors to the Spruce Lake Trails Area. In addition, the concept of impact acceptability was investigated. Study area visitors were asked to indicate their maximum acceptable levels of various impacts. Lastly, the concept of shared agreement or normative consensus was explored by exarruriing the degree of sub-group variation in visitors' responses regarding salient impacts and acceptable levels of impact. This chapter describes the results of testing the six research hypotheses presented in Chapter 1 using data collected in the mail-back questionnaire sent to 301 study area visitors. RESULTS FOR NULL HYPOTHESIS 1A A total of 5,000 hypothetical situations were presented to the 250 respondents who returned completed questionnaires. Ratings were obtained for 4,947 hypothetical situations. (Some of the respondents failed to rate all the situations in their questionnaires.) Results of regressing the ratings (dependent variables) on the impact factors (independent variables) for the commercially outfitted and privately outfitted sub-groups are shown in table 7. As discussed in Chapter 3, only main effects (i.e. additive elements) of the SAS GLM procedure are presented. While certain interaction effects (non-additive components) were tested, none of the regression coefficients for the interactions proved to be statistically significant, R 2 values were only minimally increased (<.03) and thus parameter estimates for the interactions are not presented. A number of statistics are presented in table 7 which will be used to examine the relative influence of the impact factors and the extent of agreement among the two sub-group categories. 65 TABLE 7 Regression of Hypothetical Situation Ratings on Impact Factors for Trip Organization Sub-Groups Commercially Outfitted Privately Outfitted b SE Rank b SE Rank Impact Dimensions and Factors A: Site Conditions a Litter seen -1.13* .11 1 -1.18* .09 1 Damaged trees -1.06* .11 2 -0.96* .09 2 Vegetation loss -0.50* .11 4 -0.59* .08 4 Campfire rings -0.35* .11 5 -0.19* .09 5 Positive site condition 0.56* .11 3 0.66* .09 3 B: Encounters b Individuals seen Groups seen Large groups seen C: Sight and Sound Intrusions c Aircraft/helicopters Structures Forest Service signs No structures -0.73* .12 2 -0.71* .12 3 -0.96* .12 1 -0.34* .10 2 -0.38* .10 1 -0.02 .10 0.25* .10 3 -0.48* .10 3 -0.56* .10 2 -0.86* .10 1 -0.54* .08 1 -0.51* •08 2 -0.01 .08 0.25* .08 3 D: Behavior d Discourteous behavior -1.46* .08 1 -1.32* .06 1 Noisy or loud behavior -1.08* .08 2 -1.07* .06 2 E: Social Setting e On the trail In camp By the lake -0.17 .09 -0.16 .09 -0.05 .09 0.02 .07 -0.08 .07 -0.06 .07 Intercept 5.46 R 2 .35 S 2 1.95 N 1957.00 Average raring 3.39 a Reference impact is "no site condition described'' b Reference impact is "no encounters" c Reference unpad is "no sight and sound condition described" d Reference impact is "no behavior described" e Reference impact is "in the area" * Significantly different from zero at the p=.05 level Note: Results oft-test comparing b-coefficients between sub-groups produced no significant differences for any impact factors .15 5.17 .36 1.80 2931.00 3.30 .12 66 The extent to which ratings were influenced by the presence of an impact factor in the hypothetical situation being rated is expressed by the regression coefficient in column b. For example, the coefficient in table 7 for the impact litter for the commercially outfitted sub-group can be interpreted as follows: When the site condition described in the hypothetical situation is the "presence of litter", the rating on average is 1.13 points lower (negative sign) than a reference situation in which no factor for Dimensions A through D were included and in which a general social setting ("in the area") was specified. The impact factors 6, 10, 15, 18, and 22 from table 5 are the "reference factors" and are listed at the bottom of tables 7, 8, and 9. The empirical findings in table 7 indicates that of the 17 impact factors included in the model, 13 had regression coefficients that were statistically significant at the .05 level. In addition, the signs of the coefficients indicate that respondents from both sub-group categories perceived nearly all of the impacts as negatively affecting influence ratings and thus the quality of their wilderness experience. The exceptions for the commercial group are for the impact factors "positive site condition" and "no structures seen" which both have positive signs. The reason for the positive signs for these factors are intuitively obvious. No significant differences were found between the b-coefficients of the two sub-groups for any impact factors. Relative Impact Factor Influence The numbers in the column "Rank" indicate that factor's rank in terms of the size (in absolute values) of the coefficient relative to the coefficients of other factors within the impact dimension. Ranks were presented only for factor coefficients which were statistically significant. Thus, we see that both commercially outfitted and privately outfitted groups agreed on which impacts on Site Conditions mattered most to them, although there were differences in the weights they attached to each impact factor. There were some slight disagreements on the relative influence for the impact factors within Dimensions B and C but complete agreement for the two factors in Dimension D. Overall, the findings in table 7 67 suggest that commercially outfitted and privately outfitted respondents were using similar criteria in evaluating the hypothetical situations. Ratings Thresholds One measure which can be used to describe variations between the two sub-groups is the average rating given to all the situations rated. The average rating given by commercially outfitted respondents was 3.39 while the average rating given by privately outfitted respondents was 3.30. Thus the commercially outfitted group tended to rate all examples of impact situations as less negatively affecting the quality of their wilderness experience than did the privately outfitted group. However, the difference is quite small (.09). Rating Variance Another measure which can be used to describe sub-group variations in rating tendencies is the error variance. As table 7 shows = 1.95 for commercially outfitted respondents compared to = 1.80 for the privately outfitted group. Like the average ratings for the two groups, the difference between rating variances is small (.15). Rating Error The extent of stochastic or random error (1-R )^ is a third measure which can describe sub-group variations. From table 7 we see that, again, like the previous statistics, the difference between the two groups is small; the commercially outfitted respondents' error of .65 (1-.35) compared to the privately outfitted respondents' error of .64 (1-.36). A difference of only .01. Systematic Variation A comparison of intercepts indicates the systematic variation of influence ratings regardless of which impact factors were included in the hypothetical situations being rated. The intercept values in table 7 show that, typically, the commercial group rates each impact situation .29 points (5.46-5.17) higher than the private group. In other words, the commercially outfitted respondents viewed any hypothetical impact situation as less negatively affecting the quality of their experience than did privately outfitted respondents. 68 Coefficient of Determination One last measure which can be used to examine sub-group variations is the coefficient of determination. From table 7 we see that the difference in R.2 values for the two sub-groups was quite small (.01). The 17 impact factors included in the model accounted for about 35 and 36 percent of the variation in the hypothetical situation ratings for both commercially outfitted and privately outfitted respondents, respectively. RESULTS FOR NULL HYPOTHESIS IB As was the case for the last hypothesis, only main effects of the SAS GLM procedure are presented for Null Hypothesis IB. None of the regression coefficients for a number of interaction effects tested were statistically significant. Table 8 presents the results of regressing the hypothetical situation ratings on the impact factors for the sub-group characterized by the length of stay. Only two impact factors (large groups seen and no structures seen) had b-coefficients that were significantly different between two of the three sub-group categories. Relative Impact Factor Influence The findings indicate that of the 17 factors included in the model 13 had regression coefficients that were statistically significant at the .05 level for respondents staying one day; 12 factors had significant coefficients for respondents staying two to four days; and 12 factors had significant coefficients for respondents staying more than four days. The signs of the coefficients were similar for all three categories of the sub-groups with two exceptions. All three sub-group categories agreed on the intra-dimension rankings of factors for 5 of the 12 impact factors that were statistically significant. Length of stay categories one and two agreed on rankings for 8 factors; categories one and three agree on rankings for 6 factors; and categories two and three agree on rankings for 7 factors having statistically significant coefficients.. Overall, these results indicate weak support for the research hypothesis that the relative influence of impacts would vary for study area visitors characterized by their length of stay. 69 TABLE 8 Regression of Hypothetical Situation Ratings on Impact Factors for Length of Stay Sub-Groups One Day Two to Four Days More than Four Days b SE Rank b SE Rank b SE Rank Impact Dimensions and Factors A: Site Conditions * Litter seen -1.38* .18 1 -1.11* .10 1 -1.13* .10 Damaged trees -1.14* .18 2 -1.08* .10 2 -0.90* .11 Vegetation loss -0.58* .18 3 -0.60* .10 3 -0.52* .10 Campfire rings -0.48* .18 5 -0.26* .10 5 -0.20 .11 Positive site condition 0.52* .18 4 0.56* .10 4 0.72* .10 B: Encounters b Individuals seen -0.48* .20 2 -0.72* .12 3 -0.51* .12 Groups seen -0.41* .20 3 -0.81* .12 2 -0.53* .12 Large groups seen -0.78* .20 1 1 -1.12*1 .11 1 I -0.73*1 .12 C: Sight and Sound Intrusions c Aircraft/helicopters -0.40* .16 2 -0.59* .09 1 -0.35* .10 Structures -0.54* .16 1 -0.59* .09 2 -0.29* .10 Forest Service signs 0.00 .16 -0.13 .09 0.11 .10 No structures 0.15 .16 0.07 .09 | 0.48*| .10 D: Behavior d Discourteous behavior -1.33* .13 1 -1.25* .13 1 -1.52* .08 Noisy or loud behavior -1.02* .13 2 -0.99* .13 2 -1.18* .08 E: Social Setting e On the trail -0.10 .14 -0.02 .08 -0.06 .09 In camp -0.39* .14 1 -0.04 .08 -0.10 .09 By the lake -0.27 .14 0.09 .08 -0.11 .08 Intercept 5.54 .24 5.37 .14 5.12 .14 R2 .32 .36 .37 S2 2.08 1.76 1.88 N 840.00 2013.00 2054.00 Average rating 3.43 3.26 3.37 a Reference impact is "no site condition described" " Reference impact is "no encounters" c Reference impact is "no sight and sound condition described" d Reference impact is "no behavior described" e Reference impact is "in the area" * Significantly different from zero at the p=.05 level Note: b-coefficients in boxes were significantly different between sub-groups at the p=05 level 70 Ratings Thresholds The average rating given by the three length of stay categories ranged from a high of 3.43 for respondents staying one day to a low of 3.26 for respondents staying two to four days. The difference, however, is very small (.17). Rating Variance Similar to the above finding the difference in error variances is greatest between categories one and two. Thus, variability in ratings judgements between the three categories exists but is small (.32). Rating Error The amount of random error (1-R )^ associated with each sub-group's model varies from .68 (1-.32) for category one to .64 (1-.36) for category two to .63 (1-37) for category three. The maximum difference is only .05 (.68-.63) between day users and those respondents staying more than four days. Systematic Variation Intercept values were similar for the three sub-group categories. The greatest difference (.42) can be seen between categories one and three (5.54 versus 5.12). Thus, respondents staying one day, typically rated each hypothetical impact situation .42 points higher than respondents staying more than four days, regardless of which impact factors were included in the situation being rated. In other words, day users viewed any hypothetical impact situation as less negatively affecting the quality of their experience than did respondents staying longer in the area. Again, however, the difference is relatively small, less than one-half point. Coefficient of Determination The last measure used to examine variation between the three length of stay categories is the R2 values. Like the past measures of variation discussed, there is little difference between the three categories. The 17 impact factors in the model accounted for from 32 to 37 percent of the variation in the hypothetical situation ratings for respondents staying from one day to more than four days. 71 RESULTS FOR NULL HYPOTHESIS 1C Tests for interaction effects for Null Hypothesis 1C using the SAS GLM procedure produced no statistically significant regression coefficients and, thus, only results from main effects are presented for this hypothesis. Table 9 shows the results of regressing the ratings given to hypothetical situations on impact factors for the respondent sub-group characterized by place of origin. Only two impact factors (groups seen and large groups seen) had b-coefficients that were significantly different between the two sub-groups. Relative Impact Factor Influence The findings indicate that 13 of the 17 impact factors included in the model had statistically significant regression coefficients for both B.C. and U.S. residents. In addition, these factors were the same for both groups. For the most part, the signs of the coefficients indicate that respondents from both groups perceived nearly all of the factors as negatively affecting influence ratings. Both groups had positive signs for the coefficients for the impact factors "positive site condition" and "no structures seen" Respondents from the two sub-group categories gave similar rankings to eight of the 13 impact factors which had statistically significant coefficients. They agreed on ranks for all factors within Dimensions B and D, and disagreed on some factor ranks in Dimensions A and C. These results indicate weak support for the research hypothesis that the relative influence of impacts would vary for B.C. versus U.S. residents. Ratings Thresholds Average ratings given by the two place of origin sub-groups were nearly identical. B.C. residents' average rating was only .01 higher than their U.S. counterparts. Thus there was almost no variation in the two groups' ratings thresholds. Rating Variance Variation in rating tendencies as measured by the error variance indicates that B.C. residents have somewhat greater agreement than do U.S. residents. 72 TABLE 9 Regression of Hypothetical Situation Ratings on Impact Factors for Place of Origin Sub-Groups B.C. Residents U.S. Residents b SE Rank b SE Rank Impact Dimensions and Factors A: Site Conditions a Litter seen -1.13* .07 1 -1.34* .17 1 Damaged trees -0.99* .08 2 -1.17* .17 2 Vegetation loss -0.55* .07 4 -0.66* .16 3 Campfire rings -0.26* .08 5 -0.18* .17 5 Positive site condition 0.61* .07 3 0.58* .17 4 B: Encounters b Individuals seen -0.55* .08 3 -0.82* .19 3 Groups seen -0.57 .08 2 -1.02* .19 2 Large groups seen -0.83 .08 1 -1.26* .19 1 C: Sight and Sound Intrusions c Aircraft/helicopters -0.47* .07 1 -0.31* .16 3 Structures -0.43* .07 2 -0.54* .15 1 Forest Service signs 0.004 .07 -0.10 .15 No structures 0.23* .07 3 0.34* .15 2 D: Behavior d Discourteous behavior -1.34* .06 1 -1.53* .13 1 Noisy or loud behavior -1.05* .06 2 -1.12* .13 2 E: Social Setting e On the trail -0.01 .06 -0.22 .14 In camp -0.09 .06 -0.17 .14 By the lake -0.04 .06 -0.13 .14 Intercept 5.21 .10 5.75 .22 R2 .34 .38 S2 1.79 2.12 •N 3893.00 878.00 Average rating 3.34 3.33 a Reference impact is "no site condition described" b Reference impact is "no encounters" c Reference impact is "no sight and sound condition described" d Reference impact is "no behavior described" e Reference impact is "in the area" * Significantly different from zero at the p=.05 level Note: b-coefficients in boxes were significantly different between sub-groups at the p=.05 level 73 Rating Error The findings in table 9 indicate that like the previous statistics, the difference in the amount of random error is quite small for the two groups. B.C. residents' random error is .66 (1-.34) compared to the comparable statistic for U.S. residents of .62 (1-.38); a difference of .04. Systematic Variation A comparison of intercepts indicates that typically, U.S. residents rated each hypothetical situation .54 points (5.75-5.21) higher than B.C. residents, regardless of which impact factors were being rated. Thus, the U.S. group viewed any hypothetical situation as less negatively affecting the quality of their experience than did B.C. residents. Coefficient of Determination The 17 impact factors in the model accounted for 34 percent of the variation in the hypothetical situation ratings given by B.C. residents. This compares to 38 percent for U.S. residents. The difference in R^ values is, therefore, quite small (.04). RESULTS FOR NULL HYPOTHESIS 2A Null Hypothesis 2A compares maximum acceptable amounts of impact at each of four locations in the study area for survey respondents characterized by how their trips were organized. Table 10 shows the findings for the 10 impact items on or beside trails at Spruce Lake for each sub-group. The table indicates that only one significant difference was found for average values between the commercially outfitted respondents and privately outfitted respondents for any impacts. It is interesting to note that a greater percentage of each category of respondents indicated a maximum amount for the litter impact than for the other nine impacts. In contrast, the vegetation loss impact had the lowest percentage of respondents indicating an amount. In addition to analyzing the amount of variation in average values for the impact items, variances were tested to compare differences in the extent of agreement among the two sub-group categories. Although there was only one statistically significant difference between average values, the variances associated with many of the 74 TABLE 10 Average Maximum Acceptable Amounts of Impact On or Beside Trails at Spruce Lake for Trip Organization Sub-Groups Impact Item Commercially Outfitted (N=98) Privately Outfitted (N=148) 1. Litter % responding 84.8 76.4 Average 2.00* 1.12* s 2 9.61* 5.52* 2. Campfire rings % responding 76.8 68.9 Average 1.01 1.19 s 2 1.88* 9.92* 3. Damaged trees % responding 77.8 71.0 Average 1.56 1.52 s 2 8.47 7.84 4. Vegetation loss % responding 54.6 52.7 Average 1.80 1.18 s2 12.32* 3.84* 5. Individuals % responding 75.8 73.0 Average 7.85 7.28 s 2 40.20* 58.98* 6. Groups % responding 76.8 70.3 Average 2.61 2.51 s 2 6.45 4.28 7. Large groups % responding 77.8 73.6 Average 1.40 1.45 s 2 2.16 2.66 8. Aircraft % responding 76.8 72.3 Average 2.84 2.74 s 2 5.81* 28.73* 9. Structures % responding 71.7 66.2 Average 1.79 1.57 s 2 4.33* 7.02* 10. Signs % responding 65.7 61.5 Average 3.40 2.25 s 2 75.34* 4.16* * denotes sample statistics which were significantly different (p=.05 level) between sub-groups 75 average values did vary. .The most noticeable differences were for impact items 1, 2, 4, 5, 8 and 10. Table 11 presents the results of comparing average values for the 10 impact items at campsites at Spruce Lake between the commercially versus privately outfitted survey respondents. No significant differences were found for nine of the 10 impact items. A significant difference was found for the average maximum acceptable amounts of litter seen. A greater percentage of the commercially outfitted respondents provided a maximum amount for each of the impact items than did their privately outfitted counterparts. The impact of litter had the highest percentage and vegetation loss had the lowest percentage for both categories for this sub-group. There were significant differences for the extent of agreement as measured by the variances for impact items 1,2,4, and 7-10. Another location associated with null hypothesis 2A is on or beside trails elsewhere in the Spruce Lake Trails Area. Results of comparing the impact items at this location for the commercially outfitted and privately outfitted respondents are shown in table 12. No significant differences were found between average values provided by the two categories of this sub-group. Consistent with past findings, the litter impact had a greater percentage of respondents from both categories indicating a maximum amount. The vegetation loss impact again had the lowest percentage of respondents indicating a maximum amount. Significant differences between variances were found for impact items 2, 6, 8 and 9. The last location associated with research null hypothesis 2A is at campsites elsewhere in the area. Results of the statistical tests are shown in table 13. The average maximum amounts were not significantly different between the two sub-group categories for any impact items. The vegetation loss impact again received the lowest percentage of respondents in each category indicating a maximum amount. On the other hand, the impact item receiving the highest percentage of compliance was different for the two categories. The litter impact received the highest percentage for the commercially outfitted respondents while the tree damage impact had the highest percentage for the privately outfitted respondents. There were significant differences in variances associated with average values of impact items 1-4 and 8-9. 76 TABLE 11 Average Maximum Acceptable Amounts of Impact at Camps at Spruce Lake for Trip Organization Sub-Groups Impact Item Commercially Outfitted (N=98) Privately Outfitted (N=148) 1. Litter % responding 84.8 75.0 Average 2.10* 1.28* s2 8.07* 3.92* 2. Campfire rings % responding 80.0 73.0 Average 2.38 2.65 s2 4.16* 11.49* 3. Damaged trees % responding 80.0 72.3 Average 3.18 2.81 s2 29.16 29.16 4. Vegetation loss % responding 64.6 56.1 Average 6.39 5.23 s2 132.94* 60.68* 5. Individuals % responding 75.8 73.0 Average 7.67 6.88 s2 76.39 56.70 6. Groups % responding 76.8 71.0 Average 2.29 2.40 s2 6.25 4.80 7. Large groups % responding 77.8 74.3 Average 1.46 1.24 s2 3.39* 2.16* 8. Aircraft % responding 76.8 73.0 Average 3.04 2.70 s2 7.78* 26.52* 9. Structures % responding 77.8 73.0 Average 3.47 2.29 s2 24.9* 5.11* 10. Signs % responding 68.7 64.9 Average 2.28 1.96 s2 4.12* 2.79* * denotes sample statistics which were significantly different (p=.05 level) between sub-groups 77 TABLE 12 Average Maximum Acceptable Amounts of Impact On or Beside Trails Elsewhere in the Area for Trip Organization Sub-Groups Impact Item Commercially Outfitted (N=98) Privately Outfitted (N=148) 1. Litter % responding 75.8 82.4 Average 1.55 1.10 s2 6.3C 6.30 2. Campfire rings % responding 64.6; 75.7 Average 0.92 1.25 s2 1.72* 15.29* 3. Damaged trees % responding 67.7 78.4 Average 1.45 1.64 s2 9.18 12.96 4. Vegetation loss % responding 49.5 58.8 Average 1.92 1.43 s2 13.47 10.37 5. Individuals % responding 65.7 77.0 Average 7.15 6.41 s2 36.97 42.77 6. Groups % responding 67.7 75.0 Average 2.09 2.49 s2 4.12* 6.66* 7. Large groups % responding 68.7 80.4 Average 1.32 1.25 s2 2.22 1.85 8. Aircraft % responding 67.7 79.0 Average 2.67 2.36 s2 6.15* 26.32* 9. Structures % responding 66.7 71.0 Average 1.35 1.53 s2 2.13* 16.65* 10. Signs % responding 63.6 66.9 Average 2.24 2.15 s2 6.20 4.71 * denotes sample statistics which were significantly different (p=.05 level) between sub-groups 78 TABLE 13 Average Maximum Acceptable Amounts of Impact at Camps Elsewhere in the Area for Trip Organization Sub-Groups Impact Item Commercially Outfitted (N=98) Privately Outfitted (N=148) 1. Litter % responding 73.7 79.7 Average 1.63 1.16 s2 5.86* 2.99* 2. Campfire rings % responding 67.7 77.7 Average 2.13 2.31 s2 3.92* 6.50* 3. Damaged trees % responding 68.7 80.4 Average 2.59 2.36 s2 22.47* 14.36* 4. Vegetation loss % responding 57.6 62.8 Average 6.09 4.77 s2 109.20* 35.40* 5. Individuals % responding 65.7 76.4 Average 5.54 5.31 s2 32.83 23.14 6. Groups % responding 66.7 75.7 Average 1.79 2.03 s2 5.52 4.04 7. Large groups % responding 68.7 79.0 Average 1.04 1.14 s2 1.74 2.43 8. Aircraft % responding 65.7 77.7 Average 2.62 2.17 s2 7.19* 24.30* 9. Structures % responding 69.7 73.6 Average 3.25 1.54 s2 115.78* 2.72* 10. Signs % responding 67.7 68.9 Average 1.92 1.81 s2 3.76 3.10 * denotes sample statistics were significant different (p=.05 level) between sub-groups 79 Overall, the preceding results indicate strong support for the hypothesis that maximum acceptable amounts will not vary for commercially versus privately outfitted survey respondents. However, one impact item (the number of pieces of litter seen per site) had significantly different average values between commercially and privately outfitted respondents at two locations in the study area. RESULTS FOR NULL HYPOTHESIS 2B Null Hypothesis 2B compares maximum acceptable amounts of impact at each of the four locations in the study area for three categories of a sub-group of survey respondents characterized by their length of stay. As discussed in Chapter 3, the first category includes those respondents who were day users (i.e., they did not spend the night in the study area). Category two includes respondents who stayed two to four days (i.e., one to three nights). The last category of the length of stay sub-group was composed of respondents who stayed in the study area for five days or more (i.e., more than four nights). The results of data analysis for maximum acceptable amounts of impact on or beside trails at Spruce Lake are shown in table 14. Only one significant difference was found for any maximum amounts between the three sub-group categories. This was for the number of aircraft/helicopters sighted or heard per day between day users and visitors staying five or more days. The litter impact had the highest percentage of respondents giving a maximum amount for respondents staying one day and respondents staying five or more days. The vegetation loss impact had the lowest percentage for each of the three length of stay categories. Variances associated with the average maximum acceptable amounts of impact were significantly different between some sub-groups for all the impact items except the litter and damaged trees impacts. Another location associated with Null Hypothesis 2B is at campsites at Spruce Lake. As table 15 indicates no significant pairwise differences were found between average maximum acceptable amounts for any of the 10 impact items. Figures for the percent of respondents providing a maximum acceptable amount for the impact item are similar to those at the previous location, on or beside trails at Spruce Lake. Significant differences in variances 80 TABLE 14 Average Maximum Acceptable Amounts of Impact On or Beside Trails at Spruce Lake for Length of Stay Sub-Groups Impact Item One Day (N=42) Two to Four Days (N=101) Five Days or More (N=105) 1. Litter % responding 81.0 79.2 80.0 Average 1.74 1.22 1.58 s2 6.55 7.24 7.45 2. Campfire rings % responding 71.4 75.2 68.6 Average 1.27 1.29 0.83 s2 2.34A 1 2 . 7 4 A B 1.59B 3. Damaged trees % responding 71.4 80.2 67.6 Average 2.20A 1.41 1.32A s2 11.56 7.20 6.25 4. Vegetation loss % responding 47.6 62.4 47.6 Average 1.40 0.94 2.04 s2 5.43^ 1.93AC 1 4 - 4 4 B C 5. Individuals % responding 78.6 76.2 69.5 Average 6.36 6.73 8.71 s2 25.70A 30.25B 83.17^ 6. Groups % responding 69.0 74.3 72.4 Average 2.45 2.45 2.67 s2 2.89A 4.54 6.71A 7. Large groups % responding 69.0 79.2 73.3 Average 1.41 1.25 1.57 s2 1.46A 2.31 2.92A 8. Aircraft % responding 76.2 76.2 70.5 Average 3.69A 2.77 2.35A s2 8.06A 36.97^ 5.24B 9. Structures % responding 66.7 71.3 67.6 Average 2.07 1.85 1.27 s2 4.88 9.18A 2.72A 10. Signs % responding 57.1 65.4 64.8 Average 2.62 3.55 2.02 s2 5.71^ 74.30AC 3.283° Sample statistics with the same letters were statistically different (p=.05 level) between sub-groups 81 TABLE 15 Average Maximum Acceptable Amounts of Impact at Camps at Spruce Lake for Length of Stay Sub-Groups Impact Item OneDay(N=42) Two to Four Days (N=101) Five Days or More (N= 105) 1. Litter % responding 83.3 78.2 78.1 Average 1.86 1.40 1.70 s2 4.88 4.75A 7.08A 2. Campfire rings % responding 76.2 75.2 76.2 Average 2.38 2.78 2.34 s2 4.12^ 5.06B 13.18^ 3. Damaged trees % responding 71.4 80.2 71.4 Average 3.33 2.76 2.93 s2 55.80A 27.25^ 20.61B 4. Vegetation loss % responding 54.8 63.4 59.0 Average 5.65 6.47 5.32 s2 124.32^  118.81AC 58.83c 5. Individuals % responding 78.6 75.2 70.5 Average 7.00 6.72 7.66 s2 84.09A 34.34^ 88.55B 6. Groups % responding 71.4 73.3 72.4 Average 2.97 2.26 2.20 s2 8.01A 4.97 4.75A 7. Large groups % responding 69.0 79.2 74.3 Average 1.69 1.06 1.42 s2 4.80A 1.25^ 3.28B 8. Aircraft % responding 71.4 77.2 73.3 Average 3.33 3.05 2.38 s2 10.89^ 33.99AC 6.00c 9. Structures % responding 71.4 77.2 75.2 Average 4.00 2.65 2.38 s2 50.41^ 6.97A 5.71B 10. Signs % responding 61.9 69.3 66.7 Average 2.31 2.21 1.94 s :2 2.96 3.69 3.17 Sample statistics with the same letters were significantly different (p=.05 level) between sub-groups 82 associated with the average maximum acceptable amounts were found for impact items 1-9. The third location at which maximum acceptable amounts of impact are examined for the length of stay sub-group is on or beside trails elsewhere in the Spruce Lake Trails Area. Only one significant difference was found between average amounts as indicated in table 16. The impact item for which this difference was found was the aircraft/helicopter impact. As was the case with trail locations at Spruce Lake, day use respondents would accept seeing or hearing more aircraft or helicopters per day on or beside trails elsewhere in the area than would respondents staying five or more days. Category three respondents had, overall, less difficulty providing maximum acceptable amounts than did either of the other two categories. The highest percentage (84.8) was by category three respondents for the litter impact while the lowest percentage (52.4) was by category one respondents for the vegetation loss impact. Significant differences were found between variances for some sub-groups for all the impact items. The final location associated with Null Hypothesis 2B was at campsites elsewhere in the area. The results in table 17 indicate that significant differences were found between average maximum acceptable amounts for impact items 6 (groups seen), 8 (aircraft seen or heard) and 9 (structures seen). Percentages of respondents providing maximum amounts are similar to results for the impact items at the previous locations. Categories one and three had the highest percentages for the litter impact while the highest percentage for category two respondents was given for the tree, damage impact. As before, the vegetation loss impact received the lowest response percentages by all three categories. Significant differences in variances were indicated between some sub-groups for all the impact items. RESULTS FOR NULL HYPOTHESIS 2C The last research hypothesis compares maximum acceptable amounts of impact at the four study area locations for survey respondents characterized by their place of origin. Maximum acceptable amounts for the 10 impact items on or beside trails at Spruce Lake are 83 TABLE 16 Average Maximum Acceptable Amounts of Impact On or Beside Trails Elsewhere in the Area for Length of Stay Sub-Groups Impact Item One Day (N=42) Two to Four Days (N=101) Five Days or More (N=105) 1. Litter % responding 78.6 76.2 84.8 Average 1.91 1.14 1.15 s2 8.76A 7.78B 4 0 0 A B 2. Campfire rings % responding 71.4 69.3 73.3 i Average 1.27 1.47 0.77 s2 2.56^ 23.33AC 1.46^ 3. Damaged trees % responding 69.0 78.2 72.4 Average 1.59 1.94 1.18 s2 5.66A 1 7 8 9 A B 7.08B 4. Vegetation loss % responding 52.4 58.4 53.3 Average 2.23 0.98 1.98 s2 27.88^ 3.42AC D . I O 8 0 5. Individuals % responding 73.8 68.3 76.2 Average 6.06 6.94 6.60 s2 25.30A 5944AB 30.58B 6. Groups % responding 69.0 67.3 78.1 Average 2.26 2.62 1.99 s2 3.65A 9.67^ 2.99B 7. Large groups % responding 69.0 72.3 82.9 Average 1.34 1.26 1.25 s2 1.02^ 2.50A 1.88B 8. Aircraft % responding 71.4 70.3 80.0 Average 3.53A 2.66 1.92A s2 9.18^ 40.70AC 3.57^ 9. Structures % responding 64.3 67.3 74.3 Average 1.52 1.82 1.09 s2 2 7 2 A B 24.70AC 1.66*° 10. Signs % responding 57.1 65.4 70.5 Average 2.79 2.42 1.85 s2 8.00A 7.51B 2.28^ Sample statistics with the same letters were statistically different (p=.05 level) between sub-groups 84 TABLE 17 Average Maximum Acceptable Amounts of Impact at Camps Elsewhere in the Area for Length of Stay Sub-Groups Impact Item OneDay(N=42) Two to Four Days (N=101) Five Days or More (N=105) 1. Litter % responding 78.6 73.3 81.9 Average 1.88 1.12 1.33 s2 5.06A 2.76^ 4.84B 2. Campfire rings % responding 71.4 70.3 i 79.0 Average 2.43 2.44 2.02 s2 3.84A 4.58B 6.92^ 3. Damaged trees % responding 69.0 76.2 78.1 Average 3.10 2.35 2.29 s2 20.52A 23.04B 11.09^ 4. Vegetation loss % responding 57.1 59.4 64.8 Average 5.71 5.15 5.53 s2 114.06^ 45.97A 68.23B 5. Individuals % responding 73.8 67.3 76.2 Average 5.90 5.06 5.40 s2 36.97A 20.61A 27.40 6. Groups % responding 69.0 66.3 78.1 Average 3.03^ 1.84A 1.63B s2 8.76^ 4.66AC 2.59^ 7. Large groups % responding 69.0 72.3 79.0 Average 1.45 1.10 0.98 s2 2.04 2.86A 1.61A 8. Aircraft % responding 69.0 68.3 80.0 Average 3.10A 2.65 1.79A s2 38.56AC 3.53^ 9. Structures % responding 66.7 68.3 79.0 Average 2.54A 2.91 1.46A s2 4.71^ 116.64AC 2.043° 10. Signs % responding 61.9 64.4 76.2 Average 2.46 1.94 1.65 s2 5.15A 4.28B 1 9 3 A B Sample statistics with the same letters were statistically different (p=.05 level) between sub-groups 85 presented in table 18 for B.C. residents and U.S. residents. A significant difference was found between the maximum acceptable amount of vegetation loss for the two categories. B.C. residents would accept 1.06 square metres while U.S. residents would accept 3.00 square metres at any one site on or beside trails at Spruce Lake. No other significant differences were found for impacts at this location. Consistent with previous results, the litter impact received the highest percentage of respondents providing maximum amounts, while the vegetation loss impact received the lowest percentages. Significant differences in variances were found for impact items 2-5, 8 and 10. Table 19 presents results for Null Hypothesis 2C at campsites at Spruce Lake. Statistically significant differences were found between average maximum acceptable amounts for impact items 2, 3, 5 and 6-9. Maximum amount response percentages are again consistent with previous results. Variances were significantly different between the two sub-groups for six of the 10 impact items. The third location associated with Null Hypothesis 2C is on or beside trails elsewhere in the Spruce Lake Trails Area (see table 20). Similar to results found for trails at Spruce Lake, only one significant difference was found for average maximum acceptable amounts between B.C. residents and U.S. residents. B.C. resident respondents would accept a maximum of 1.27 square metres of vegetation loss per site while U.S. residents would accept 3.05 sq. metres. Maximum amount response percentages were again the highest for the litter impact and lowest for the vegetation loss impact at this location for both B.C. and U.S. resident respondents. Variances were significantly different for seven of the 10 impact items. Results for Null Hypothesis 2C at campsites elsewhere in the area are presented in table 21. Only one significant differences was found for average maximum acceptable amounts between B.C. and U.S. resident respondents. B.C. residents would accept seeing 1.18 large groups per day compared to .70 large groups for U.S. respondents. Once more, maximum amount response percentages parallel those found previously for this hypothesis with the litter impact receiving the highest and the vegetation loss impact receiving the lowest 86 TABLE 18 Average Maximum Acceptable Amounts of Impact On or Beside Trails at Spruce Lake for Place of Origin Sub-Groups Impact Item B.C. Residents (N=197) U.S. Residents (N=44) 1. Litter % responding 79.7 81.8 Average 1.34 2.03 s2 6.81 8.35 2. Campfire rings % responding 72.1 70.4 Average 1.13 1.29 s2 7.56* 2.40* 3. Damaged trees % responding 74.1 70.4 Average 1.37 2.45 s2 7.02* 12.89* 4. Vegetation loss % responding 54.8 47.7 Average 1.06* 3.00* s2 3.24* 23.72* 5. Individuals % responding 73.1 72.7 Average 7.38 6.81 s2 54.46* 31.58* 6. Groups % responding 73.1 47.7 Average 2.50 2.59 s2 4.58 4.75 7. Large groups % responding 76.1 65.9 Average 1.42 1.31 s2 2.43 1.93 8. Aircraft % responding 74.1 70.5 Average 2.78 2.97 s2 22.37* 7.02* 9. Structures % responding 69.0 65.9 Average 1.65 1.76 s2 6.25 4.71 10. Signs % responding 64.5 61.4 Average 2.89 2.00 s2 40.83* 3.39* * denotes sample statistics which were significantly different (p=.05 level) between sub-groups 87 TABLE 19 Average Maximum Acceptable Amounts of Impact at Camps at Spruce Lake for Place of Origin Sub-Groups Impact Item B.C. Residents (N=197) U.S. Residents (N=44) 1. Litter % responding 78.2 84.1 Average 1.47 2.11 s2 5.11 6.92 2. Campfire rings % responding 74.6 79.6 Average 2.63 2.17 s2 9.06* 4.45* 3. Damaged trees % responding 75.1 75.0 Average 2.95 3.09 s2 31.47* 17.39* 4. Vegetation loss % responding 60.4 59.1 Average 5.83 6.65 s2 94.67 103.43 5. Individuals % responding 73.1 75.0 Average 7.15 6.73 s2 53.88* 114.70* 6. Groups % responding 74.1 65.9 Average 2.34 2.24 s2 4.84 4.84 7. Large groups % responding 76.6 68.2 Average 1.30 1.37 s2 2.22* 4.45* 8. Aircraft % responding 75.1 70.5 Average 2.94 2.48 s2 21.62* 7.34* 9. Structures % responding 76.1 68.2 Average 2.71 3.07 s2 6.60* 51.26* 10. Signs % responding 68.0 61.4 Average 2.16 1.78 s2 3.38 2.96 * denotes sample statistics which were significantly different ( p=.05 level) between sub-groups 88 TABLE 20 Average Maximum Acceptable Amounts of Impact On or Beside Trails Elsewhere in the Area for Place of Origin Sub-Groups Impact Item B.C. Residents (N=197) U.S. Residents (N=44) 1. Litter % responding 81.7 72.7 Average 1.19 1.38 s2 6.25 4.71 2. Campfire rings % responding 72.6 63.6 Average; 1.20 1.04 s2 12.32* 2.04* 3. Damaged trees % responding 75.6 68.2 Average 1.53 2.03 s2 11.76 11.76 4. Vegetation loss 47.7 % responding 56.8 Average 1.27* 3.05* s2 8.47* 23.72* 5. Individuals % responding 72.6 68.2 Average 6.55 6.33 s2 42.77* 24.50* 6. Groups % responding 73.6 63.6 Average 2.37 2.21 s2 6.00* 2.76* 7. Large groups 65.9 % responding 78.2 Average 1.32 1.00 s2 2.16* .71* 8. Aircraft % responding 75.6 68.2 Average 2.44 2.80 s2 21.90* 7.13* 9. Structures % responding 69.5 68.2 Average 1.47 1.33 s2 13.10* 2.37* 10. Signs % responding 67.0 63.6 Average 2.21 2.11 s2 5.76 3.50 * denotes sample statistics which were significantly different (p=.05 level) between sub-groups 89 TABLE 21 Average Maximum Acceptable Amounts of Impact at Camps Elsewhere in the Area for Place of Origin Sub-Groups Impact Item B.C. Residents (N=197) U.S. Residents (N=44) 1. Litter % responding 78.2 75.0 Average 1.26 1.46 s2 3.42* 5.20* 2. Campfire rings % responding 74.1 70.5 Average 2.29 2.06 s2 5.62 3.92 3. Damaged trees % responding 77.2 70.5 Average 2.41 2.52 s2 17.56 13.32 4. Vegetation loss % responding 61.9 59.1 Average 5.20 6.96 s2 60.22* 100.20* 5. Individuals % responding 72.1 68.2 Average 5.49 4.27 s2 27.04 18.23 6. Groups % responding 74.1 59.1 Average 1.99 1.58 s2 4.58 2.82 7. Large groups % responding 76.6 61.4 Average 1.18* 0.70* s2 2.37* 0.76* 8. Aircraft % responding 74.1 68.2 Average 2.37 2.30 s2 21.07* 5.62* 9. Structures % responding 73.1 65.9 Average 2.30 1.55 s2 57.30* 2.46* 10. Signs % responding 69.0 70.5 Average 1.90 1.74 s2 3.42 2.92 * denotes sample statistics which were significantly different (p=.05 level) between sub-groups 90 for both place of origin categories. Significant differences in variances were found for impact items 1,4 and 7-9. SUMMARY OF RESULTS FOR MAXIMUM ACCEPTABLE AMOUNTS DATA Collectively, the preceding results indicate that the three variables chosen to investigate sources of subject variability in maximum acceptable amounts of impact offer little explanatory power. Very few statistically significant differences were found for the 10 impact items for the various combinations of sub-groups for the four locations in the study area. Oh the other hand, the extent of agreement for the average maximum acceptable amounts of impact varied considerably for many impact items across the sub-group comparisons. 91 Chapter 5 CONCLUSIONS INTRODUCTION Current wilderness recreation management frameworks seek to identify, set limits for and monitor those impacts which have the potential to negatively affect natural settings and the experiences of wilderness visitors. However, as was pointed out in Chapter 2, budget and labor force constraints force wilderness managers to be selective and direct management efforts toward the most salient impacts and deterrnining acceptable levels for these impacts. The issue of impact saliency was addressed in this study by seeking to determine the relative importance of specific impact factors within a number of impact dimensions identified in the wilderness research literature. The issue of impact acceptability was addressed by seeking to determine if wilderness visitors could articulate maximum acceptable levels for a number of specific impacts on ecological and social conditions. Another issue related to the above which is of interest to wilderness managers concerns the degree to which relevant sub-groups of wilderness visitors agree on impact saliency and maximum acceptable levels of impact. As much of the literature points out, this type of evaluative information can help managers justify and gain support for the imposition of impact acceptability standards which can potentially restrict visitors' activities. The issue of sub-group agreement or subject variability was addressed in this study by examining a number of variables which might explain or account for differences in wilderness visitors' attitudes about impact saliency and acceptability levels. Six research hypotheses were formulated to examine the issues discussed above. The hypotheses were tested using data obtained in a survey of a sample of visitors to the Spruce Lake Trails Area in southwestern B.C. during the summer use season in 1992. A mail-back questionnaire was formulated to address certain conceptual issues and overcome some of the methodological limitations identified in a review of past empirical studies. 92 Null Hypotheses 1A, IB, and 1C were formulated to examine the impact saliency issue. The specific impacts evaluated by survey respondents included a number of impacts researchers indicated had not been sufficiently addressed in the past. These included impacts related to behaviour, aircraft, structures and management signs. To test the three hypotheses, an alternative modeling approach (the Factorial Survey Approach) was used to provide impact evaluators (survey respondents) with a situational context which more fully represents real-life judgements on the influence of impacts on their wilderness experience. Specific management concerns for the Spruce Lake Trails Area were used to select three variables which might explain variation in impact saliency evaluations between different user groups. The three variables included: trip organization (Null Hypothesis 1 A); length of stay (Null Hypothesis IB); and place of origin (Null Hypothesis 1C). Null Hypotheses 2A, 2B, and 2C were formulated to examine the issue of impact acceptability. As discussed in Chapter 2, a number of past research studies indicate a high degree of variability between acceptable levels of impact for different locations within one wilderness area. This conceptual issue was addressed in this study by asking respondents to indicate maximum acceptable amounts of impact at four locations within the study area. Another issue identified as needing further attention in empirical studies on impact acceptability is of a methodological nature. It concerns the appropriateness of response categories provided to survey respondents. Many past studies have failed to offer respondents an "I cant give a number" response option when asking them to indicate the maximum amount of impact they would accept. Such a response option was provided in this study to reduce the possibility of respondents guessing or stating some number just to please the researcher. The number of respondents who used this response is also clearly indicated in the presentation of empirical findings - a practice not always followed by past researchers. Thus, the basis for conclusions on the extent of agreement for acceptability levels is more fully documented here than in many past studies on the topic. The issue of subject variability is also addressed with Null Hypotheses 2A, 2B, and 2C using the same three explanatory variables used for the previous hypotheses. 93 The remaining sections of this chapter will interpret the results of testing the research hypotheses presented in the last chapter and discuss implications of the findings for managers of the Spruce Lake Trails Area. NULL HYPOTHESIS 1A The first research hypothesis was formulated to test whether variations in impact saliency could be explained in terms of how study area visitors' trips were organized. Impact influence ratings for hypothetical situations were compared for commercially outfitted and privately outfitted wilderness visitors. A number of past empirical studies have indicated that significant differences seem to exist between these two groups regarding the relative importance of impacts on ecological and social conditions found in wilderness. However, since no significant differences were found for the b-coefficients between the two sub-groups, we fail to reject Null Hypotheses 1A The two groups disagreed on the intra-dimensional ranks of only four of the 17 impact factors included in the model. There was complete agreement on the relative influence of each of the five impacts on site conditions. Seeing litter at the site is clearly perceived as most negatively affecting the quality of the wilderness experience regardless of whether the visitor was commercially outfitted or not. This finding is consistent with past empirical findings. The second most highly ranked impact on site conditions by both groups was tree damage. The remaining impacts (with the exception of positive site conditions), while perceived as negatively affecting respondent's experiences, were not viewed as nearly important as litter and damaged trees. Two different intra-dimensional rankings were found for impact factors in the encounters dimension. While both groups agreed that seeing large groups most negatively affected their experience, they disagreed on the relative influence of seeing other groups and individuals. Commercially outfitted visitors ranked seeing individuals more negatively than seeing other groups. This is perhaps intuitively obvious since these visitors are themselves members of a group and therefore view this type of trip organization as more appropriate or even preferable from a safety standpoint. Privately outfitted visitors on the other hand ranked seeing other groups as a relatively more negative impact than seeing 94 individuals. It is probable that more privately outfitted visitors were traveling alone and thus responded more negatively to seeing groups than seeing individuals like themselves. It should be pointed out, however, that while the ranks differed for the "seeing groups" and "seeing individuals" impacts, the differences were slight as the size of the regression coefficients indicate in table 7. The sight and sound dimension also contains two impact factors for which the rankings of the two trip organization sub-groups differed. Privately outfitted visitors perceived seeing or hearing aircraft/helicopters as a relatively more negative influence than seeing structures. Commercially outfitted visitors on the other hand ranked seeing structures as the most negative impact of the two impacts which were both statistically significant and received negative ratings. As was the case for the different rankings in the encounters dimension, the differences between the rankings for the two groups for the aircraft and structure impacts are slight but are clearly distinguishable from the ratings given the other impact factors in the dimension. In terms of the behaviour dimension, the two groups agreed on the relative influence of the two impact factors. Discourteous behaviour was clearly rated as a relatively more negative impact than loud or noisy behaviour by others. It is difficult, however, to interpret with certainty what specific elements of the two impacts were the basis for the difference in relative importance. One might speculate that respondents viewed "discourteous behaviour" as a more direct confrontation and "loud and noisy behaviour" as more indirect or less confrontational. When "discourteous behaviour" was included in the hypothetical situation being rated by survey respondents, the rating was lowered substantially, even more so than "seeing litter" at the site. Thus the recommendation by past researchers that future research on studies of this kind should investigate the importance of behaviour impacts is supported by our findings. Because none of the regression coefficients were statistically significant for both groups in the social setting dimension, ranks were not calculated. Thus, the relative influence of the different factors cannot be compared for the two groups. 95 In addition to the above, a number of additional statistics were discussed in Chapter 4 to present the results for Null Hypothesis 1A. A comparison of rating thresholds, rating variances, rating errors, systematic variations and coefficients of determination for the two trip organization sub-groups indicated that there was a high degree of agreement on the issue of impact saliency between the two groups. This is not to say that there was complete consensus on the relative influence of factors within each of the impact dimensions. However, the results indicate that the variable "trip organization" seems to have little power in explaining subject variation in the relative influence of impacts for visitors to the Spruce Lake Trails Area. NULL HYPOTHESIS IB This research hypothesis sought to explain variations in the relative influence of impacts by the length of stay variable. This variable was included for three reasons. First, none of the previous research on impact saliency has considered its potential to explain differences in group agreement regarding the importance of impacts. Second, past studies have noted significant differences in the motivation to visit wilderness settings between persons exhibiting different trip characteristics such as day versus overnight users (Schreyer et al., 1976). Lastly, managers of the Spruce Lake Trails Area have recognized differences in participation patterns for area visitors and are interested to know if these sub-groups view the importance of impacts differently. Hypothetical situation ratings were regressed on the impact factors for each of the three categories of the length of stay sub-group. The results presented in Chapter 4 indicate that we should reject the null hypotheses for the large groups seen and no structures seen impact factors for visitors staying two to four days and the visitors staying five or more days. We fail to reject the null hypotheses for the remaining impact factors between the three sub-group categories. All three categories agreed on the relative influence of the litter, damaged trees and campfire ring impacts which were rated one, two, and five, respectively within the site conditions dimension. Visitors staying one day and visitors staying two to four days also agreed on rankings for the remainder of the impact factors. Visitors staying more than four 96 days ranked seeing vegetation loss and bare ground lower than visitors staying shorter periods. However, the absolute size of the regression coefficient (-.52) for this group was only slightly different from the coefficients for day users (-.58) and visitors staying two to four days (-.60). Overall, one can conclude from these results that the relative influence of impacts on site conditions are virtually the same regardless of the duration of visitor's trip in the study area. The same conclusion can be made for the impact factors within the encounters dimension. All three categories view seeing large groups as most negatively affecting the quality of their wilderness experience. The two categories staying longer periods ranked "seeing groups" relatively more negative than "seeing individuals". On the other hand, day visitors felt the opposite. A plausible interpretation of this finding is, however, problematic since the size of the regression coefficients for the two impact factors are so similar within each category. The impact dimension containing the most discrepancies in factor rankings by the length of stay categories was the sight and sound intrusions dimension. The most interesting finding related to this dimension concerns the ratings given by day users. The most negatively rated impact by this category was seeing structures. However, a close second was seeing aircraft/helicopters. This finding is surprising in light of the fact that the primary means of access by this group is by float plane. Intuitively, one would think that since these visitors use this travel method, they would not at the same time perceive it as a negative impact. It seems, however, that although day visitors condone their own use of aircraft in this wilderness area, they consider the impact in general (by others?) to negatively affect the quality of their experience. All three length of stay categories agreed on the relative influence of the two impact factors within the behaviour dimension. This finding parallels the results found for the trip organization sub-groups. Similar conclusions apply here as well. 97 In terms of the social setting dimension, only one sub-group category ("one day") had a statistically significant coefficient (the "in camp" factor). Therefore, comparisons of ranks between sub-groups for the factors in this dimension cannot be made. Li summary, the ranks of the b - coefficients for the impact factors were similar for the three length of stay categories. This conclusion is also supported by the results of the comparisons made of rating thresholds, rating variances, rating errors, systematic variations, and the coefficients of determination. Overall, there is a high degree of agreement concerning the relative influence of impacts among study area visitors characterized by their length of stay. We can conclude, therefore, that subject variation in impact saliency cannot be explained by the length of stay variable for data obtained from visitors to the Spruce Lake Trails Area. NULL HYPOTHESIS 1C This research hypothesis was formulated to test whether variations in impact saliency could be explained in terms of the place of origin of study area visitors. Two categories were used to characterize survey respondents by their place of origin. The first group included respondents who were residents of B.C.. The second group were U.S. residents. As was pointed out in Chapter 3, it has been argued that the potential exists for residents of different cultures to have different attitudes and values which could affect the relative importance they ascribe to various impacts on their wilderness recreation experiences. Non B.C. resident visitation seems to be increasing in the Spruce Lake Trails Area and Forest Service managers are, therefore, interested to know how this segment of the visitor population views impacts on wilderness conditions. Based on the results presented in the previous chapter, we reject the null hypotheses for the two impact factors, groups seen and large groups seen, since there were significant differences between the b-coefficients for the two sub-groups. We fail to reject the null hypotheses for the other impact factors with b-coefficients that were significantly different from zero. B.C. residents and U.S. residents agreed on the intra-dimensional rankings of 8 of the 13 impact factors having statistically significant coefficients. Of the five impact 98 dimensions, the two groups agreed on the ranks of all the impact factors in two of the dimensions. Both groups agreed that the litter impact had the most negative influence of the five factors within the site condition dimension. This was followed by the damaged trees factor for both groups. U.S. residents gave a higher relative importance to the vegetation loss impact than did B.C. residents. However, the difference in the absolute size of the regression coefficient for this impact was only slight (.11) for the two groups. Overall, from these findings we can conclude that after litter, both groups view tree damage as the most negative impact on ecological conditions in the site. The presence of campfire rings was rated as the least influential impact by B.C. and U.S. residents alike. In addition, the regression coefficients for this impact was relatively small for both groups indicating that seeing campfire rings affected the quality of their experience very little. B.C. residents and U.S. residents agreed on the ranks for all the factors in the encounters dimension. Seeing large groups was clearly viewed as the most negative impact followed by seeing other groups and lastly seeing individuals. We can conclude that members of both cultural groups hold similar opinions as regard to what constitutes degrees of solitude, at least insofar as this concept is related to seeing other humans in a wilderness setting. On the other hand, there appears to be a slight difference of opinion between the two groups regarding the relative influence of various sight and sound intrusions in wilderness. B.C. residents ranked seeing or hearing aircraft/helicopters as the most negative influence on the quality of their experience compared to the other impact factors in Dimension C. U.S. residents, however, ranked seeing structures as the most negative influence with the aircraft impact ranked third in terms of its influence on their experience for the factors in this dimension. The situation here is analogous to the one observed for the commercially versus privately outfitted sub-groups discussed earlier. Again, one can conclude that since a greater percentage of foreign visitors probably access the area by aircraft than do B.C. residents, the former group view this impact as relatively less important than does the latter group. The implications of this finding for management of the area will be discussed in a later section. 99 Once again, the findings related to the relative influence of the two factors in the behavioural dimension parallel those found for the previous two hypotheses. Discourteous behaviour was ranked by both B.C. and U.S. residents as a relatively more negative impact than land or noisy behaviour. Similar conclusions apply here as well. It appears that both cultural groups apply the same criteria in evaluating potential impacts as they relate to the concepts of solitude and privacy. The findings for the relative influence of factors within the social setting impact dimension are identical to those found for Null Hypothesis 1A which compared ratings for the commercially outfitted and privately outfitted visitors. None of the regression coefficients for any of the impact factors were statistically different from zero for either of the two resident groups. Thus, the respondents tended to ignore these factors when evaluating hypothetical situations which contained them. Of the other statistics reported in Chapter 4 to compare variations in the extent of agreement of impact saliency for the two place of origin categories, only two were more than slightly different. The variation in rating tendencies as measured by the error variance indicated that B.C. residents have somewhat greater agreement than do their U.S. counterparts. In addition, a comparison of intercepts showed that, typically, B.C. residents rated each hypothetical situation as more negatively affecting the quality of their experience than did U.S. residents, regardless of which impact factors were being rated. This would lead us to conclude that the B.C. resident respondents as a whole tended to be somewhat more critical toward any impacts and as a group are more solidified in this position than were the U.S. resident respondents. However, this interpretation, in as much as it relates to group agreement, is not supported by the results of comparisons of rating thresholds, rating errors and coefficients of determination between the two place of origin categories. All of the above considerations for Null Hypothesis 1C lead us to conclude that the place of origin variable provides little explanatory power as regards the variation in the relative influence of impacts for visitors to the Spruce Lake Trails Area. While there were some differences of opinion regarding the ranking of factors within some of the impact 100 dimensions, overall the differences were slight and appear to be overshadowed by numerous similarities. This is not to say that managers might not want to consider such differences in views, however minor, when monitoring impacts and setting impact standards. This and other related issues will be discussed more fully later in this chapter. NULL HYPOTHESIS 2A This research hypothesis was formulated to test whether variations in maximum acceptable levels of impact at different locations in the study area could be explained in terms of how visitors' trips were organized. Average maximum acceptable amounts were compared for commercially outfitted and privately outfitted visitors at each of four locations in the Spruce Lake Trails Area. The discussion which follows will interpret the results for the hypothesis at each of four locations in the study area. Maximum Acceptable Amounts On or Beside Trails at Spruce Lake The first location for which respondents were asked to provide maximum acceptable amounts of impact was at Spruce Lake proper. As discussed in Chapter 1, Spruce Lake is the main destination point of most visitors to the Spruce Lake Trails Area. There are a number of main access trails leading to the lake as well as shorter hiking trails connecting campsites and circling the lake. The results presented in Chapter 4 indicated that we should fail to reject the null hypothesis that average maximum acceptable amounts of impact would not vary for the two sub-group categories. There was only one impact item (litter seen) which had a statistically significant difference in the average maximum amounts between the commercially and privately outfitted respondents. As for the question of whether respondents can give quantitative figures for acceptable levels of impact, one would have to conclude yes for this location. For all impacts listed, over 50 percent of respondents from both groups provided a number. Providing a square metre figure for vegetation was clearly the most problematic for both groups. Overall, a greater percentage of commercially outfitted respondents provided a "maximum amount" figure for each impact item than did their privately outfitted counterparts at this location. 101 To compare the extent of agreement among the two sub-group categories we tested the variances associated with the average maximum acceptable amounts. As shown in Chapter 4, significant differences in variances were found for seven of the 10 impact items. For the campfire ring impact, the variance in maximum acceptable amounts among commercially outfitted respondents was less than 20% than that for privately outfitted respondents (1.88 versus 9.92). This indicates a much higher degree of agreement among commercially outfitted respondents about the maximum acceptable number of campfire rings seen on any one day on or beside trails at Spruce Lake. For the vegetation loss impact the opposite was the case. The variance for the average maximum acceptable amount of vegetation loss at any site on or beside trails at Spruce Lake was over three times greater for commercially outfitted respondents (s2 = 12.32) as for privately outfitted respondents (s2 = 3.84). In terms of seeing or hearing aircraft or helicopters, the commercially outfitted respondents seem to have more intra-group agreement than do privately outfitted respondents (s2 = 5.84 versus s2 = 28.73). The greatest difference in intra-group agreement occurs for impact item 10. The variance for the average maximum acceptable amount of Forest Service signs seen at any site on or beside trails at Spruce Lake was over 18 times greater for commercially outfitted respondents (s2 = 75.34) than for privately outfitted respondents (s2 = 4.16). Thus, intra-group agreement seems to be much higher for the latter group for this impact at this location. Maximum Acceptable Amounts at Camps at Spruce Lake There are a number of public and commercial campsites at Spruce Lake. Thus, the second location for which respondents were asked to provide maximum amounts of impact was for these sites around the lake. A statistically significant difference between average maximum amounts for the two sub-group categories was found for only one of the 10 impact items (litter seen). Thus, overall, we fail to reject the null hypothesis for this location. The average maximum acceptable amount of litter seen per site was higher for commercially outfitted visitors than for privately outfitted visitors. Thus, it would appear that the 102 commercial group has a somewhat greater tolerance for this campsite condition in the wilderness setting. However, the difference in the amount of litter acceptable to the two groups is quite small (2.10-1.28=.82 pieces per site). Results regarding the percent of respondents able to give a specific amount of acceptable impact parallels the findings at the previous locations. Thus, the same conclusions apply as well. Overall, privately outfitted respondents had greater group agreement for the average acceptable amounts of impact than did the commercially outfitted respondents. The later group had greater agreement than did the former for only two impact items (the number of campfire rings per site and the number of aircraft/helicopters seen or heard per day). Maximum Acceptable Amounts On or Beside Trails Elsewhere in the Area While the main destination point of most visitors to the study area is Spruce Lake, a significant number of visitors do travel along trails and camp in many dispersed locations in the Spruce Lake Trails Area. The third location for which respondents were asked to provide maximum acceptable amounts of impact was on or beside trails other than at Spruce Lake. Results indicate that we should fail to reject the null hypothesis that maximum acceptable amounts of impact would not vary for commercially versus privately outfitted visitors at this location. No statistically significant differences in average maximum amounts between the two groups were found. In terms of the quantification question for maximum amounts, results at this location are in complete contrast to the two previous locations. A greater percentage of privately outfitted respondents gave maximum amounts than did commercially outfitted respondents for each impact on or beside trails in dispersed locations. Again the highest compliance percentage was for the litter impact and the lowest was for the vegetation loss impact. For all the other impact items, at least 60 percent of respondents in both sub-groups provided numbers for each impact item. Intra-group agreement among respondents of the two trip organization sub-groups varied considerably for a number of the impact items on or beside trails elsewhere in the area. For impact item 8 (the number of aircraft/helicopters seen or heard per day), intra-group 103 agreement was much higher for commercially outfitted respondents compared to privately outfitted respondents. The variance for the former category was less than one-fourth that for the latter category (6.15 versus 26.32). Commercially outfitted respondents also had much higher agreement for impact items 2, 6 and 9 (campfire rings, groups seen and human-made structures seen) as the variances indicated. Maximum Acceptable Amounts at Camps Elsewhere in the Area The last location associated with Null Hypothesis 2A is at campsites other than at Spruce Lake. As the results in the previous chapter indicated, we fail to reject the null hypotheses for these locations. No statistically significant differences were found among the 10 impact comparisons for the two groups. . The findings related to the quantification of maximum amounts for this location parallel those found for the previous location with one exception. While privately outfitted respondents have higher compliance percentages than did the commercial group, the impact receiving the highest total percentage of quantifiable responses was the percent of trees damaged by others rather than the litter impact. Intra-group agreement varied considerably for a number of the impact items at campsites elsewhere in the area. For impact item 8, intra-group agreement was much higher for commercially outfitted versus privately outfitted respondents. The variance for the former category was less than one-third that for the latter category (7.19 versus 24.30). On the other hand, privately outfitted respondents had much higher intra-group agreement for the amount of vegetation loss and the number of structures seen per site. Especially noticeable is the difference in variances associated with average values for the number of human-made structures seen. The s2-value for the commercially outfitted category is 42.56 times greater than the s2-value for the privately outfitted category indicating a significantly higher level of intra-group agreement for the average maximum acceptable amount of the number of human-made structures seen per site for the latter group. 104 Summary for Null Hypothesis 2A Statistically significant differences in average maximum acceptable amounts between commercially outfitted and privately outfitted respondents were found for only one of the impact items at more than one location. This was for the number of pieces of litter seen per site on or beside trails at Spruce Lake. However, as pointed out earlier, the differences in absolute terms for these values were very slight. On the whole, differences in the extent of intra-group agreement for the average maximum acceptable amounts of impact were highly significant. Values of s2 were consistently lower for commercially outfitted respondents' average maximum acceptable amounts of the number of campfire rings per site and the number of aircraft/helicopters sighted or heard per day at all four locations in the study area. Privately outfitted respondents had greater agreement for the pieces of litter per site and the square metres of vegetation loss at these locations. We can conclude, therefore, that the trip organization variable does indeed help explain variations in group agreement for acceptable levels for these impact items. The variable does not, on the other hand, offer much explanation for variations in either average maximum acceptable amounts or the extent of group agreement for the amounts of other impact items. NULL HYPOTHESIS 2B This hypothesis compared maximum acceptable amounts of impact at the four locations in the study area for survey respondents characterized by their length of stay. The first category included visitors who did not spend the night in the area (i.e., day users). Category two included visitors who stayed two to four days. The third category was made up of visitors who had an extended trip of five days or more. Maximum Acceptable Amounts On or Beside Trails at Spruce Lake The results of comparing acceptability levels on or beside trails at Spruce Lake indicated that, overall, we should fail to reject the null hypothesis. The average maximum acceptable amount of impact was statistically significantly different for only one impact item and then between only one of the three possible pairwise comparisons. The average maximum acceptable number of aircraft/helicopter seen or heard per day was higher for day users than 105 for respondents staying five or more days. This finding is predictable since day users make up the majority of those visitors who access the lake by aircraft and thus would be expected to view this impact as more appropriate than visitors not using aircraft. Overall, most respondents from the three length of stay categories seemed to be able to quantify their maximum acceptable levels of impact. In nearly all cases two-thirds or more of the respondents provided a maximum amount with one noticeable exception. Providing the maximum acceptable square metres of vegetation loss per site was difficult for each length of stay category. Quantifying the maximum acceptable amount of litter per site was relatively easy for all respondents with the percent responding ranging from 79.2 (category 2) to 81.0 (category 1). The extent of agreement with average values varied noticeably among the three sub-group categories for a number of impact items. For the aircraft/helicopter impact (which had one significant pairwise difference) category two respondents had a variance over four times higher than s2-values for category one and and over seven times higher than category three. Visitors staying two to four days, have less agreement as a group as to what a maximum acceptable amount should be for this activity compared to day users and visitors staying for extended periods. Overall, respondents staying two to four days had less group agreement about their average maximum acceptable amounts of impact than did the other two length of stay categories at this location. Maximum Acceptable Amounts at Camps at Spruce Lake We fail to reject Null Hypothesis 2B which compared average maximum acceptable amounts of impact at this location. As the results in Chapter 4 indicated, no statistically significant differences in average values were observed for any of the pairwise combinations. Overall, respondents in all three categories seemed to be able to quantify acceptability levels to a greater extent at this location than at the previous location. This is especially apparent for the vegetation loss impact. Over fifty percent of respondents in each category provided a maximum square metre figure. However, quantification of an acceptable level for this impact remains to be problematic. 106 Differences in intra-group agreement for acceptable levels of impact were noticeable for nine of the 10 impact items. No significant difference was found between the variances for the average number of signs seen per site. Overall, respondents staying more than four days had the highest levels of group agreement. One possible explanation here is that the apparentness of these impacts may grow the longer the visitor stays at the campsites and, thus, opinions about acceptable levels of impact solidifies as well. Regardless of the reasons, these findings on the seemingly lack of across and within group agreement for acceptable levels of impact for these items at this major destination in the study area suggests the potential for major user conflicts. Maximum Acceptable Amounts On or Beside Trails Elsewhere in the Area The results of comparing average maximum acceptable amounts of impact for trails elsewhere in the area were the same as for trails at Spruce Lake. Both sets of results indicate that, overall, we should fail to reject Null Hypothesis 2B. Statistically significant differences in average values were found for only one pairwise comparison for the aircraft impact item. Again, day use respondents would accept seeing or hearing more aircraft per day than would respondents staying five or more days. A similar conclusion holds here as well. Results regarding the "quantification of maximum acceptable amounts" issue are also similar to the findings for the other trail location. Noticeable differences in intra-group agreement levels were evident for all of the ten impact items. Visitors staying two to four days had lowest amount of group agreement for following impacts: campfire rings, damaged trees, individuals seen, groups seen, large groups seen, aircraft/helicopters, and human-made structures. Levels of group agreement were quite similar between day users and visitors staying more than four days for most of the impact items. Day users had the lowest amount of group agreement for the litter, vegetation loss and Forest Service signs impact items. Maximum Acceptable Amounts at Camps Elsewhere in the Area Results of comparing maximum acceptable amounts of impact for the sub-groups at this location indicate mixed support for Null Hypothesis 2B. Statistically significant 107 differences were found in average maximum acceptable amounts for three impact items. Pairwise differences were observed between the average value given by day users and the values for the other two categories for the number of groups seen per day. Day users would also accept seeing or hearing more aircraft/helicopters per day than the other two categories. Respondents staying two to four days would accept seeing more human-made structures than the day users and respondents staying five or more days. Again over fifty percent of the respondents from all three length of stay categories were able to give a "maximum acceptable amount" for all of the 10 impact items. Consistent with past findings, providing a square metre figure for the vegetation loss impact was the most problematic. The extent of intra-group agreement for average maximum acceptable amounts varied for most of the impact items. Respondents within the day use category had much less group agreement for their average maximum acceptable amount of vegetation loss than did the other two categories. The greatest difference in group agreement occurred for the human-made structure impact (which also had the one pairwise difference). Visitors staying two to four days had an s2-value nearly 25 times greater than the same value for day users and nearly 57 times greater than the s2-value for visitors staying more than four days. Summary for Null Hypothesis 2B Taken together, the results for testing Null Hypothesis 2B are similar to those found for Null Hypothesis 2A. Statistically significant differences in average maximum acceptable amounts between the three sub-groups were found for only one impact item at more than one of the four locations. This impact item was the number of aircraft/helicopters sighted or heard per day. Day users' average maximum acceptable amounts of aircraft seen or heard by trails at Spruce Lake, on or beside trails elsewhere in the area and at camps elsewhere in the area were statistically different from the average values at the same locations for visitors staying five or more days. Day users' average maximum acceptable number of human-made structures seen at camps elsewhere in the area were statistically different from the same average values at the same location for visitors staying five or more days. However, as was 108 the case for Null Hypothesis 2 A, differences in absolute terms for the above values were, for the most part, slight. Differences in the extent of intra-group agreement for the average maximum acceptable amounts of impact varied depending on the impact and the location. However, the following general conclusions can be made. Visitors staying two to four days had much less intra-group agreement than the other two categories about maximum acceptable numbers of aircraft/helicopters seen or heard per day at all four locations in the study area. Day use visitors had much less intra-group agreement than the other two categories about maximum acceptable amounts of vegetation loss at campsites anywhere in the study area and by trails in dispersed locations. While the length of stay variable does explain some differences in the extent of intra-group agreement for some impacts at some locations, we must conclude that, overall, the variable does not offer much explanation for variations in maximum acceptable amounts. NULL HYPOTHESIS 2C The last research hypothesis was formulated to test whether variations in average maximum acceptable amounts of impact at the four locations in the study area could be explained in terms of visitors' place of origin. Average maximum acceptable amounts for the 10 impact items were compared for B.C. residents and U.S. residents, the two main cultural groups which visit the Spruce Lake Trails Area. Maximum Acceptable Amounts On or Beside Trails at Spruce Lake As noted in Chapter 4, the results of the data analysis indicated that, overall, we should fail to reject Null Hypothesis 2C at this location. Only one statistically significant difference in average maximum acceptable amounts was found between the two groups. U.S. resident respondents would accept a significantly greater amount of vegetation loss at any one site on or beside trails at Spruce Lake than would their B.C. counterparts. It is difficult to interpret why the U.S. visitors would accept almost three times as much (3.00 square metres versus 1.06 square metres per site) disturbance to vegetation at this location. One possible reason could be that U.S. wilderness recreationists have grown accustomed to higher levels of 109 impact associated with traditional high levels of use in U.S. wilderness settings. However, regardless of the reason, this finding clearly indicates a significantly higher tolerance level for this type of impact for many U.S. visitors to the study area. It should be noted that only 47.7 percent of U.S. respondents provided a maximum acceptable amount for the impact which again highlights the problematic nature of quantifying acceptability levels for this impact. Another finding related to differences in opinions between the two groups relative to the vegetative impact at this location concerns the extent of intra-group agreement. As the results showed, the s2-value for U.S. residents' average amount of vegetative loss was over seven times greater than the similar figure for B.C. residents. This leads us to conclude that, as a group, the B.C. respondents had a much higher level of agreement about their average maximum acceptable amount of vegetative impact than did the U.S. respondents. The same conclusion can be made for the damaged trees impact as well. U.S. residents, on the other hand, had higher levels of agreement for acceptable number of campfire rings, individuals seen, aircraft/helicopter sightings and Forest Service signs seen. Maximum Acceptable Amounts at Camps at Spruce Lake We fail to reject Null Hypothesis 2C for this location as no statistically significant differences were found in average maximum acceptable amounts for any of the 10 impact items. The percentage of respondents providing maximum amounts for each impact item was also similar for the two groups. There were significant differences in the extent of intra-group agreement for average values between B.C. and U.S. residents for seven of the 10 impact items. B.C. residents had greater intra-group agreement for their average maximum acceptable amounts of vegetation loss, individuals seen, large groups seen, aircraft/helicopters seen or heard and human-made structures seen at campsites at Spruce Lake. U.S. residents had greater agreement for the number of campfire rings and damaged trees at the Spruce Lake campsites. Maximum Acceptable Amounts On or Beside Trails Elsewhere in the Area The findings for Null Hypothesis 2C for trails elsewhere in the study area were nearly identical to the findings for the hypothesis for trails at Spruce Lake. Overall, we fail to reject 110 the null hypothesis. The only statistically significant difference in average values found between the two groups was for the vegetative loss impact item. The U.S. average maximum acceptable amount was again almost three times greater than the average for B.C. residents. We can, therefore, conclude that for this single impact item U.S. visitors will tolerate a significantly greater amount of vegetation loss at campsites regardless of where the campsites are located in the study area. However, U.S. respondents had considerably less intra-group agreement for the acceptability levels for this impact as well. On the other hand, the U.S. group did have consistently lower levels of intra-group agreement for their average maximum acceptable amounts for most of the other impact items. Another interesting finding here is that a greater percentage of B.C. respondents quantified the maximum amount response for each impact item than did their U.S. counterparts. Maximum Acceptable Amounts at Camps Elsewhere in the Area Only one statistically significant difference in average maximum acceptable amounts was found for impact items at this location. This was for the number of large groups seen per day. B.C. respondents would accept seeing one and one-half more large groups per day than U.S. respondents. The difference in absolute values, however, is quite small (1.18 vs. .70). The B.C. group consistently had a higher percentage of respondents able to give maximum acceptable amounts for the impact items. The most interesting finding in the results between the two camp locations for Null Hypothesis 2C concerns the levels of intra-group agreement for impact item 9. Whereas for camps at Spruce Lake, the B.C. group had a significantly greater level of agreement for the average maximum acceptable number of man-made structures per site, the exact opposite was the case for camps elsewhere in the area. The U.S. group's level of agreement was almost 23 times greater than the B.C. group's level at the dispersed campsites. Summary for Null Hypothesis 2C Statistically significant differences in average maximum acceptable amounts of impact were found for only one impact item at two locations in the study area. This was for the square metres of vegetation loss on or beside trails at Spruce Lake and elsewhere in the area. I l l Differences in the extent of agreement for average maximum acceptable amounts did vary somewhat depending on the impact and the location, but only two consistent differences were apparent. B.C. resident respondents consistently had higher levels of agreement for average maximum acceptable amounts of vegetation loss, regardless of the study area location. On the other hand, U.S. resident respondents consistently had higher levels of agreement for the average maximum acceptable number of aircraft/helicopters seen or heard per day, regardless of the study area location. We can conclude, therefore, that the place of origin variable does help explain variations in group agreement for average acceptable levels for these two impact items. However, the place of origin variable does not offer much explanation for variations in either average maximum acceptable amounts or the extent of group agreement for the amounts for the other impact items. IMPLICATIONS FOR MANAGEMENT Current wilderness management frameworks for areas like the Spruce Lake Trails Area in B.C. call for: 1) the identification of management objectives regarding desired ecological and social conditions; 2) the selection of measurable impact indicators which represent the degree of change in ecological and social conditions; and 3) setting impact indicator standards which define the maximum acceptable level of change in the impact indicators and thus the wilderness conditions they reflect. In addition, wilderness managers need evaluative information on the amount of agreement wilderness visitors and interested stakeholders have concerning which impacts matter most to them and what levels of impact are acceptable to them. Many of the findings of this study can help B.C. Forest Service managers in the development of items 2) and 3) above and in deterrnining variations in the level of agreement for these items between various sub-groups of visitors to the Spruce Lake Trails Area. Study findings related to impact saliency can be directly applied and/or combined with additional information to develop a list of impact indicators to use in monitoring conditions in the area. As we saw, there was very little variation in sub-group judgements of the saliency of the impacts offered for evaluation. In the event that managers could not afford to monitor all 112 the impacts we examined plus additional impacts identified elsewhere, the results presented here give some indication as to which impacts could be used to form a relevant sub-set of impact indicators. To this end, the following suggestions can be made. In terms of the site condition impact dimension, two impacts consistently ranked highest: seeing litter and seeing trees damaged by people. This finding, incidentally, is identical to that found by Roggenbuck et al. (1993) in a 1990 study of four U.S. Forest Service wilderness areas. In some respects, the presence of litter along trails, but especially in camps can be thought as an impact on both ecological and social conditions. Seeing litter impacts visitors perceptions of the naturalness of a setting. In addition, many types of litter can affect elements of the ecosystem as well (e.g. the effect of organic waste on water bodies). The damaged tree impact has connotations similar to the above. Thus, if managers were forced to make hard choices on monitoring site conditions, monitoring these two salient impacts would coincide with visitors' implicit preferences. Campsite monitoring systems like Cole's campsite impact index technique (Cole, 1983), offer wilderness managers a fast cost-effective way to monitor a number of impact parameters including all of those included in our site condition impact dimension. Quantitative measurements of the impact indicators are weighted and combined into a single impact index figure representing the overall campsite condition. Here the findings on impact saliency in this study could be used to develop the weights which denote the relative importance of that particular impact in the overall campsite condition index. The findings related to the saliency of impact factors within the encounters dimension can also be applied to monitoring social conditions in the Spruce Lake area. If a reduced number of impact indicators were preferred for monitoring purposes, the number of large groups would be the recommendation. It was consistently ranked as the most influential negative impact on the quality of visitors' experiences. In a practical sense, it would probably be just as easy and cost-efficient to monitor all three of the encounter impacts. Our findings indicate that in many cases there was little difference in the absolute size of regression coefficients between the three encounter impact factors. 113 Recommendations for a monitoring sub-set for the sight and sound intrusions impact dimension is more problematic than for the other dimensions. Two factors, however, seemed to be consistently rated as relatively more influential: seeing or hearing aircraft/helicopters and seeing structures made by others. Among all the impact factors included in the evaluations, variations in the rankings of these two seemed to be explained to a greater degree by the sub-group characterizations. In addition, there was often very little absolute difference in the size of the regression coefficients for these two impact factors. One very real practical consideration related to these and other impacts concerns the degree to which management actions can limit the impact. In the case of the structure impact, managers have a great deal of control at least as regards the permanence of any structures built in the wilderness area. They can virtually be dismantled at will, constrained only by dollar and manpower budgets. Helicopter landings can also be restricted by limiting the number of permits. Float plane landings (a major access method at Spruce Lake) are more difficult to regulate, especially by B.C. Forest Service managers. Float plane access is specifically controlled by federal Canadian government regulations. Thus, managers of the Spruce Lake Trails Area must rely almost solely on voluntary compliance from aviators in their attempt to limit the impact of this factor considered salient by many area visitors. It was noted in the literature review that the impact of seeing management signs in wilderness areas has received mixed opinions from visitors (Roggenbuck et al., 1982; Whittaker, 1992; Hendee et al. 1968; and Lucas, 1985). The results of this study indicate that respondents tended to ignore this impact when evaluating factors in the sight and sound intrusion dimension. None of the regression coefficients were statistically significant for the Forest Service signs factor for any of the sub-groups investigated. Although not reported in the results presented in Chapter 4, survey respondents had the opportunity to give comments on the last page of the survey questionnaire. A number of visitors specifically asked for more directional signs in the area. It would seem, therefore, that to some visitors, signs enhance their experience 114 Of the two impact factors in the behaviour impact dimension, discourteous behaviour was ranked as most influential by all the sub-groups in this study. Although it was ranked relatively less influential, loud or noisy behaviour also consistently lowered influence ratings one full point. Thus these findings substantiate similar findings by Roggenbuck et al. (1993), Whittaker and Shelby (1988) and West (1982). We, therefore, agree with the suggestion by Roggenbuck et al. (1993, p. 196) that "managers might better protect the wilderness and provide wilderness experiences for more people by shaping behaviour than by limiting use". Spruce Lake Trails Area managers should continue and perhaps intensify efforts to educate visitors on wilderness ethics, especially practices to decrease conflicts arising from discourteous and loud or noisy behaviour. Our findings clearly indicate that visitors to the area would support these types of wilderness stewardship activities. A number of the results and conclusions reached in this study can also be used by study area managers in the development of impact indicator standards to define acceptable levels of change in wilderness conditions. More specifically, the results of testing Null Hypotheses 2A, 2B, and 2C and other findings discussed in Chapter 4 could be combined with other information recognized by wilderness managers and researchers as pertinent and necessary for addressing such value-laden judgements. Roggenbuck et al. (1993, p. 195) identified a spectrum of information useful for such an undertaking. The range in current conditions, planning group or focus group discussions, interest group opinions, visitor preferences, and research to identify thresholds of rapid change or decline in the indicator can all help the manager establish specific and defensible standards. When eliciting opinions regarding impact indicator acceptability, the ideal situation is to have a priori information on the saliency of prospective impact indicators. While this information was not available when this study was begun, the results on impact saliency provided by the sample of visitors surveyed here could certainly be used to gain additional information on impact acceptability. Our results might prove useful as a starting point for a planning group or in soliciting further information from interest groups or resource specialists in a formal LAC-type wilderness planning process. In addition, the results from the 115 examination of sub-group agreement surrounding maximum acceptable amounts of impact provides evidence that subject variation could be a problem for some impacts at certain locations. Managers should, therefore, use caution in transforming these values into impact indicator standards. SUMMARY This study has investigated sources of subject variation in user estimates of impact saliency and impact acceptability in a wilderness setting. Empirical data from a survey of visitors to the Spruce Lake Trails Area in southeastern B.C. was used to carry out the study. Three variables were selected which might explain variation in impact saliency evaluations and user-stated maximum acceptable amounts of impact. The explanatory variables examined included trip organization, length of stay and visitors' place of origin. Li addition to deterrriining which impacts most negatively affect visitors' experiences and the maximum amount of impact visitors' would accept, the issue of sub-group agreement was also examined. Six research hypotheses were formulated and tested with data from a mail-back questionnaire sent to visitors of the study area in 1992. Ecological and social impacts evaluated in the study and the explanatory variables examined were selected based on consultation with study area managers and a review of the wilderness research literature. Null Hypotheses 1A, IB and 1C were tested using the Factorial Survey Approach. Survey respondents evaluated a number of hypothetical situations describing combinations of impacts they might encounter while on a wilderness trip. Each hypothetical situation consisted of one factor from each of five impact dimensions: site conditions, encounters, sight and sound intrusions, the behaviour of others and the social setting. This approach was used to provide the impact evaluators (survey respondents) with a situational context which more fully represents real-life judgment processes. Null Hypotheses 1A stated that the influence of impacts would not vary for sub-groups of visitors characterized by how their trip was organized. No significant differences were found between the weights (b-coefficients) given to the impact factors by the commercially outfitted and privately outfitted respondents. Of the 17 impact factors included 116 in the model, the two groups disagreed on the intra-dimensional ranks of only four of the 13 factors which had statistically significant regression coefficients. Null Hypothesis IB compared impact influence ratings for three length of stay sub-groups: visitors staying one day; visitors staying two to four days; and visitors staying more than four days. Significant differences betweeb b-coefficients were found for only two impact factors(large groups seen and no structures seen) for two of the three sub-group categories. The three sub-groups agreed on the intra-dimensional ranks of five of the 12 impact factors having statistically significant regression coefficients. Null Hypothesis 1C stated that the influence of impacts would not vary for visitors characterized by their place of origin. Significant differences between b-coefficients for the B.C. resident and U.S. resident sub-groups were found for only two of the impact factors- groups seen and large groups seen. B.C. residents and U.S. residents had similar intra-dimensional ranks for eight of the 13 impact factors having statistically significant regression coefficients. Sub-group camparisons of rating thresholds, rating variances, systematic variations and the coefficients of determination, in general, supported the above conclusions for Null Hypotheses 1A, IB and 1C. Overall, study results indicated that the three variables examined explained relatively little variation in the relative influence of impacts on study area visitors' wilderness experiences. That is not to say that there was complete agreement among visitors on the relative importance of the impacts evaluated. However, the results if combined with additional pertinent information and specific management objectives can facilitate the development of a list of impact indicators managers can use to monitor conditions in the area. Null Hypotheses 2A, 2B and 2C examined average maximum acceptable amounts for 10 selected impacts at four locations in the study area. Null Hypothesis 2A stated that these amounts would not vary for visitors characterized by their trip organization. While study results indicated support, overall, for the hypothesis, there were some statistically significant differences in average maximum acceptable amounts for a few impact items at some of the different locations. The extent of group agreement was consistently higher for commercially outfitted than for privately outfitted respondents for average maximum acceptable amounts of 117 the number of aircraft/helicopters sighted or heard at all locations. Privately outfitted respondents had consistently higher levels of agreement for impacts on site conditions. Thus, the trip organization variable does help explain variations for this impact. Null Hypothesis 2B compared average maximum acceptable amounts of impact at four locations for the length of stay sub-groups. Differences in intra-group agreement for the average maximum acceptable amounts of impact did vary for the three categories depending on the impact and the location. Thus some variations in group agreement levels were explained by the length of stay variable. Null Hypothesis 2C stated that the average maximum acceptable amount of impact would not vary for visitors characterized by their place of origin. Study results indicated, overall, support for this hypothesis. There were differences in the extent of intra-group agreement for average amounts depending on the impact and the location. B.C. residents consistently had higher levels of agreement for maximum amounts of vegetation loss per site and U.S. residents had higher levels of agreement for the maximum acceptable number of aircraft/helicopters seen or heard per day, regardless of the study area location. The place of origin variable did, therefore, offer some explanation for variations in group agreement for certain impact items. In conclusion, the three variables examined in this study seem to offer little overall explanation for subject variation in user estimates of impact saliency and impact acceptability by visitors to the Spruce Lake Trails Area. By and large, survey respondents agreed on the relative influence of most of the impacts evaluated. Behavioural impacts and sight and sound intrusions were found to be important factors affecting the quality of the wilderness recreation experience of visitors. This finding supports the inclusion of these impacts in user evaluations of this kind as suggested by the research literature. The study findings also suggest that when examining user estimates of impact acceptability, researchers should look at multiple locations. There is evidence that visitors had different acceptability levels for some impact items at different locations in the study area. In addition, visitors had consistently higher levels of agreement for impact acceptability at dispersed campsites and lower levels of agreement for acceptable impact levels at primary destination point campsites. This finding is 118 consistent with past studies of this kind. Lastly, the study found that on average the majority of survey respondents could quantify impact acceptability levels for the impacts evaluated. However, the consistently lower quantification rates for the vegetation loss and bare ground impact clearly suggests that alternate evaluation methods should be investigated for this impact in future research. 119 REFERENCES Aaker, D . A and Day, G.S. 1986. Marketing Research. 2nd Edition. New York: Wiley. (Cited in Haider, 1993.) Anderson, D.H. 1980. Displacement of Visitors Within the Boundary Waters Canoe Area Wilderness. Fort Collins, Colorado: Colorado State University. Ph.D. Dissertation. Berk, R A and Rossi, P.H.1977. Prison Reform and State Elites. Cambridge, MA: Ballinger. Berk, R A and Rossi, P.H. 1982. Prison Reform and State Elites - A Retrospective. In: Rossi, P.H. and Nock, S.L. (eds.) Measuring Social Judgements: The Factorial Survey Approach. Beverly Hills, California: Sage Publications. 145-175. Blom, C.W.P.M. 1976. Effects of Trampling on the Occurrence of Some Plantago Species in Coastal Sand Dunes. In: Soil Compaction, Soil Moisture and Seedling Emergence. Oceological Plantarium. 11:225-241. (Cited in Cole, 1987.) Bradburn, N.M.; Rips, L.L.: and Shevell, S.K. 1987. Answering Autobiographical Questions: The Impact of Memory and Inference on Surveys. Science 236 (April 10, 1987): 157-161. British Columbia. Ministry of Forests. 1981. Integrated Resource Management Plan for Spruce Lake. Victoria, British Columbia. British Columbia. Ministry of Forests. 1989a. Managing Wilderness in Provincial Forests: A Policy Framework. Victoria, British Columbia. British Columbia. Ministry of Forests. 1989b. Spruce Lake Recreation Survey Report. Victoria, British Columbia. British Columbia. Ministry of Forests. 1992. Spruce Lake Recreation Survey Report. Victoria, British Columbia. Cole, D.N. 1983. Monitoring the Condition of Wilderness Campsite. U.S.D.A. Forest Service, Res.Pap. INT-302, Intermountain Forest and Range Experiment Station. Ogden, UT. Cole, D.N. 1986. Research on Soil and Vegetation in Wilderness: A State-of-Knowledge Review. In: Lucas, R.C. (compiler). Proceedings-National Wilderness Research Conference: Current Research; 1985. July 23-26; Fort Collins. Colorado. U.S.D.A Forest Service Gen. Tech. Rep. rNT-212, Intermountain Research Station. Ogden, Utah. 135-177. Cole, D.N. and Dalle-Molle, J. 1982. Managing Campfire Impacts in Backcountry. U.S.D.A Forest Service Gen. Tech. Rep. INT-13 5, Intermountain Forest and Range Experiment Station. Ogden, Utah. (Cited in Cole, 1987.) 120 Cronback, L. and Gleser, G.C. 1953. Assessing Similarity Between Profiles. Psychology Bulletin. 50:456-473. Dillman, D. 1978. Mail and Telephone Surveys: The Total Design Method. Toronto: John Wiley and Sons. Downing, K. and Clark, R.N. 1979. Users' and Managers' Perceptions on Dispersed Recreation Impacts: A Focus on Roaded Forest Lands. In: Ittner, R; Potter, D.R.: Agee, J.K. (eds.) Conference Proceedings: 1978 October 27-29: Seattle. Washington. U.S.D.A. Forest Service, No. R-6-001-1979. Pacific Northwest Region. Seattle, Washington. 18-23. Ericsson, A.K. and Simon, H.A. 1980. Verbal Reports as Data. Psychological Review 87(3): 215-251. Garret, K. 1982. Child Abuse - Problems of Definitions. In: Rossi, P.H. and Nock, S.L. (eds.) Measuring Social Judgements: The Factorial Survey Approach. Beverly Hills, California: Sage Publishing. 177-203. Graefe, A.R.; Kuss, F.R.; Loomis, L. 1986. Visitor Impact Management in Wildland Settings. In: Lucas, R.C. (compiler) Proceedings-National Wilderness Research Conference: Current Research: 1985 July 23-26: Fort Collins. Colorado. U.S.D.A. Forest Service, Gen. Tech. Rep. INT-212. Ogden, Utah. 432-439. Haider, W. 1993. Modeling Approaches-Tourism and Recreation. Paper presented to Proceedings Forest Planning-The Leading Edge. Ontario Forestry Research Committee Symposium. Haider, W.; Anderson, D.A.; and Louviere, J.J. 1993a. Improving the Interpretation of a Discrete Choice Experiment with a Maximum Difference Conjoint Model. Paper presented at the Third Canadian Conference on Environmental and Natural Resource Economics, Ottawa, Ontario, October 1-3, 1993. Haider, W.; Anderson, D.A.; and Louviere, J.J. 1993b. The Choice Behaviour of Remote Tourists in North Algoma, Ontario- An Experimental Research Approach. Centre for Northern Forest Ecosystem Research, Ontario Ministry of Natural Resources. Lakehead University. Thunder Bay, Ontario. Hammitt, W.E. and Madden, M.A. 1989. Cognitive Dimensions of Wilderness Privacy: A Field Test and Further Explanations. Leisure Sciences. 11:293-301. Hammitt, W.E. and Patterson, M.E. 1991. Coping Behaviour to Avoid Visitor Encounters: Its Relationship to Wildland Privacy. Journal of Leisure Research. 23 (3): 225-237. Harper, J.L.; Williams, J.T.; and Sagar, G.R. 1965. The Behaviour of Seeds in Soil. In: The Heterogeneity of Soil Surfaces and Its Role in Determiriing the Establishment of Plants from Seed. Journal of Ecology. 53:273-286. (Cited in Cole, 1987.) 121 Heberlein, T.A. 1977. Density, Crowding and Satisfaction: Sociological Studies for Determining Carrying Capacities. In: Proceedings-River Recreation Management and Research Symposium. U.S.D.A. Forest Service, Report NC-28. Mineapolis, Minnesota. 67-76. (Cited in Vaske et al., 1986.) Hendee, J.C.; Catton, W.R.; Marlow, L.D.; and Brockman, C F . 1968. Wilderness Users in the Pacific Northwest: Their characteristics, Values and Management Preferences. U.S.D.A. Forest Service, Res. Pap. PNW-61, Pacific Northwest Forest and Range Experiment Station. Portland, Oregon. Hendee, J.C.; Stankey, GH. ; and Lucas, R.C. 1990. Wilderness Management. Golden, Colorado: 2nd Edition, North American Press. Jackson, J. 1965. Structural Characteristics of Norms. In Steiner, I.D. and Fishbien, M.F. (eds.). Current Studies on Social Psychology. New York: Holt, Rinehart and Winston, Lie. 301-309. Jackson, W. 1988. Research Methods. Rules for Survey Design and Analysis. Scarborough, Ontario: Prentice-Hall Canada, Inc. Knopf, R.C. 1982. Management Problems in River Recreation - What River Floaters Are Telling Us. Naturalist. 33 (2): 12-17. (Cited in Stankey and Schreyer, 1987.) Krumpe, E.E.; Allen, J.A. and McCoy, L. 1989. Hells Canyon Visitor Profile and Recreation Use Study. Department of Wildland Recreation Management, College of Forestry, Wildlife and Range Sciences. Moscow, Idaho: University of Idaho. Lieber, S.R. and Fesenmaier, D.R. 1984. Modeling Recreation Choice: A Case Study of Management alternatives in Chicago. Regional Studies. 18:31-43. (Cited in Louviere and Timmermans, 1990.) Lieber, S.R. and Fesenmaier, D.R. 1985. Physical and Social Conditions Affecting Recreation Site Preferences. Environment and Planning A. 17:1613-27. (Cited in Louviere and Timmermans, 1990.) Lieber, S.R., Fesenmaier, D.R.; and Bristow, R.S. 1988. Social and Environmental Characteristics Affecting Alternatives for Outdoor Recreation Participation. In: Giolledge, R.G. and Timmermans, H, J.P. (eds.) Behavioural Modeling Approaches in Geography and Planning. London: Croom Helm. (Cited in Louviere and Timmermans, 1990.) Littell, R.C. and Schlotzhauer, S.D. 1987. SAS System for Elementary Statistical Analysis. Cary, North Carolina: SAS Institute, Inc. Louviere, J.J. and Timmermans, H.J.P. 1989. Hierarchical Information Integration Applied to Recreation Destination Choice. Environment and Planning A. Forthcoming. (Cited in Louviere and Timmermans, 1990.) 122 Louviere, J. and Tinimermans, H. 1990. Stated Preference and Choice Models Applied to Recreation Research: A Review. Leisure Sciences. 12:9-32. Lucas, RX. 1979. Perceptions of Non-Motorized Recreational Impacts: A review of Research. In: Ittner, R_; Potter, D.R; Agee, J.K.; Anschell, S. (eds.) Recreational Impacts on Wildlands - Conference Proceedings; 1978 October 27-29; Seattle, Washington. U.S.D.A Forest Service, No. R-6-001-1979. Pacific Northwest Region, Portland, Oregon. 24-31. (Cited in Stankey and Schreyer, 1987.) Lucas, R.C. 1985. Visitor Characteristics, Attitudes, and Use Patterns in the Bob Marshall Wilderness Complex. 1970-1982. U.S.D.A Forest Service. Res. Pap. INT-345, Intermountain Research Station, Ogden, Utah. Lucas, R.C. and Stankey, G.H. 1985. The Role of Research in Applying the Limits of Acceptable Change System. In: Watson, A E . (ed.) Proceedings Southeastern Recreation Research Conference. Stalesboro, Georgia: Georgia Southern College. (Cited in Williams et al. 1992.) Margules, C. and Usher, M.B. 1981. Criteria Used in Assessing Wildlife Conservation Potential: A review. Biological Conservation. 21:79-109. Martin, S.R.; McCool, S.F.; and Lucas, R C . 1989. Wilderness Campsite Impacts: Do Managers and Visitors See them the Same? Environmental Management. 13 (5):623-629. McCool, S.F. and Peterson, M. 1982. An Application of the Two Factor Theory of Satisfaction to Recreational Settings. U.S.D.A Forest Service, Tech. Rep., Intermountain Forest and Range Experiment Station. Missoula, Montana. (Cited in Starkey and Schreyer, 1987.) McCool, S.F.; Martin, S.R.; and Yuan, M. 1990. The 1989 Bear Trap Canyon Visitor Study. Institute for Tourism and Recreation Research, School of Forestry, Research Report 13. Missoula, Montana: University of Montana. Merriam, L.C. and Ammons, R.B. 1967. The Wilderness User in Three Montana Areas. St. Paul, Minnesota: University of Minnesota School of Forestry. (Cited in Stankey and Schreyer, 1987.) Nisbett, R.E. and Wilson, T.D. 1977. Telling More Than We Can Know: Verbal Reports on Mental Processes. Psychological Review 84(3): 231-259. Nock, S.L. 1982. Family Social Status. In: Rossi, P.H. and Nock, S.L. (eds.) Measuring Social Judgements: The Factorial Survey Approach. Beverly Hills, California: Sage Publications. 95-118. Patterson, M.E. and Hammitt, W.E. 1990. Backcountry Encounter Norms, Actual Reported Encounters, and Their Relationship to Wilderness Solitude. Journal of Leisure Research. 22:259-275. 123 Roggenbuck J.W.; Watson, A.E.; and Stankey, G.H. 1982. Wilderness Management in the Southern Appalachians. Southern Journal of Applied Forestry. 6(3): 147-152. (Cited in Stankey and Schreyer, 1987.) Roggenbuck J.W.; Watson, A.E.; and Williams, D.R. 1993. Defining Acceptable Conditions in Wilderness. Environmental Management. 17 (2): 187-197. Roggenbuck, J.W., Williams, DR.; Bange, S.P.; and Dean, D.J. 1991. River Float Trip Encounter Norms: Questioning the use of the Social Norms Concept. Journal of Leisure Research. 23:133-153. Rollins, R. 1985. Measuring Recreation Satisfaction Within a National Park Setting: The West Coast Trail Area of Pacific Rim National Park. Seattle, Washington: University of Washington. Ph.D. Dissertation. Rossi, P.H. and Anderson, A.B. 1982. The Factorial Survey Approach. In: Rossi, P.H. and Nock, S.L. (eds.) Measuring Social Judgements: The Factorial Survey Approach. Beverly Hills, California: Sage Publications. 15-67. Rossi, P.H. and Berk, R.A. 1985. Varieties of Normative Consensus. American Sociological Review. 50:333-347. Rossi, P.H. and Nock, S.L. (eds.) 1982. Measuring Social Judgements: The Factorial Survey Approach. Beverly Hills, California: Sage Publications. Schreyer, R.; Roggenbuck, J.W.; McCool, S.F.; Royer, L.E.; and Miller, J., 1976. Dinosaur National Monument Whitewater river recreation study. Institute for the Study of Outdoor Recreation and Tourism, Department of Forestry and Outdoor Recreation, Logan, Utah: Utah State University. Schwartz, S.H. 1977. Normative Influences on Altruism. In: Berkowitz, L. (ed.) Advances in Experimental Social Psychology. 10:221-279. Shelby, B.; Vaske, J.J.; and Harris, R. 1988. User Standards for Ecological Impacts at Wilderness Campsites. Journal of Leisure Research. 20(3):245-256. Shelby, B. and Heberlein, T.A. 1986. Carrying Capacity in Recreation Settings. Corvallis, Oregon: Oregon State University Press. Shelby, B. and Vaske, J.J. 1991. Using Normative Data to Develop Evaluative Standards for Resource Management: A Comment on Three Recent Papers. Journal of Leisure Research. 23:173-187. Shelby, B. and Whittaker, D. 1990. Recreation Values and Instream Flow Needs on the Delores River. Paper presented at the Third Conference on Social Science and natural resources. College Station: Texas. (Cited by Whittaker, 1990.) 124 Shew, R.L.; Saunders, P.R.; and Ford, J.D. 1986. Wilderness Managers' Perceptions of Recreational Horse Use in the Northwestern United States. In: Lucas, RX. (compiler) Proceedings-National Wilderness Research Conference: Current Research: 1985 July 23-26: Fort Collins. Colorado. U.S.D.A. Forest Service, Gen. Tech. Rep. INT-212, Intermountain Research Station, Ogden, Utah. 320-325. Stankey, G.H. 1973. Visitor Perceptions of Wilderness Recreation Carrying Capacity. U.S.D.A Forest Service, Res. Pap. INT-142, Intermountain Research Station. Ogden, Utah. Stankey, G.H. 1980. A Comparison of Carrying Capacity Among Visitors in Two Wildernesses. U.S.D.A Gorest Service, Gen. Tech. Rep. INT-242, Intermountain Research Station, Ogden, Utah. (Cited in Vaske et al., 1986.) Stankey, G.H. 1986. Dispersed Recreation Use and Users in Kosciusko national Park, Australia: A Profile and Comparison with the United States. In: Lucas, R.C. (compiler). Proceedings-National Wilderness Research Conference: Current Research; 1985 July 23-26: Fort Collins. Colorado. U.S.D.A Forest Service, Gen. Tech. Rep. INT-212, Intermountain Research Station. Ogden, Utah. 287-296. Stankey, G.H. and Schreyer, R 1987. Attitudes Toward Wilderness and Factors Affecting Visitor Behaviour: A State-of-Knowledge Review. In: Lucas, R.C. (compiler) Proceedings-National Wilderness Research Conference: Issues, State-of-Knowledge, Future Directions; 1985 July 23-26; Fort Collins, Colorado. U.S.D.A Forest Service Gen. Tech. Rep. INT-220, Intermountain Research Station. Ogden, Utah. 246-293. Stankey, G.H.; Cole, D.N.; Lucas, R.C.; Petersen, M.E.; and Frissel, S.S. 1985. The Limits of Acceptable Change (LAC) System for Wilderness Planning. U.S.D.A Forest Station, Gen. Tech. Rep. INT-176, Intermountain Forest and Range Experiment Station. Ogden, Utah. Stokes, G.L. 1991. New Wildland Recreation Strategies: The Flathead Experience. Western Wildlands. Winter, 1991. 23-27. Vaske, J.J.; Donnelly, M.P.; and Shelby, B. 1992. Establishing Management Standards: Selected Examples of the Normative Approach. In: Defining Wilderness Quality: The Role of Standards in Wilderness Management-A Workshop Proceedings. U.S.D.A Forest Service Gen. Tech. Rep. PNW-GTR-305, Pacific Northwest Research Station. Portland, Oregon. 23-37. Vaske, J.J.; Shelby, B.; Graefe, A R ; and Heberlein, T.A. 1986. Backcountry Encounter Norms: Theory, Method and Empirical Evidence. Journal of Leisure Research. 18:137-153. Warren, G. 1987. Activities, Attitudes and Management Preferences of Recreationists on the Arctic National Wildlife Range, Alaska. In: Lucas, R C . (compiler) Proceedings-National Wilderness Research Conference: Issues. State-of-Knowledge. Future 125 Directions: 1985 July 23-26: Fort Collins. Colorado. U.S.D.A, Forest Service Gen. Tech. Res. INT-220, Intermountain Research Station. Ogden, Utah. 278-286. Washburne, R.F. 1982. Wilderness Recreation Carrying Capacity: Are Numbers Necessary? Journal of Forestry. 80(ll):726-728. (Cited in Stokes, 1991.) West, P.C. 1982: Effects of User Behaviour on the Perception of Crowding in Backcountry Forest Recreation. Forest Science. 28 (1):95-105. (Cited in Stankey and Schreyer, 1987.) Whittaker, D. 1992. Selecting Indicators: Which Impacts Matter More? In: Defining Wilderness Quality: The Role of Standards in Wilderness Management - A Workshop Proceeding. U.S.D.A. Forest Service Gen. Tech. Rep. PNW-GTR-305. Pacific Northwest Research Station. Portland, Oregon. 13-22. Whittaker, D. and Shelby, B. 1988. Types of Norms for Recreation Impacts: Extending the Social Norms Concept. Journal of Leisure Research. 20(4) :261-273. Williams, D.R.; Roggenbuck, J.W.; Patterson, M.E.; and Watson, A.E. 1992. The Variability of User-Based Social Impact Standards for Wilderness Management. Forest Science. 38(4):738-756. Womble, P.; Wolf, W.; and Field, D. 1978. Hikers on the Chilkoot Trail: A Descriptive Report. College of Forest Resources, Cooperative Park Studies, Sociology Studies Program. Seattle, Washington: University of Washington, (Cited in Stankey and Schreyer, 1987.) 126 APPENDIX 1 Sample Questionnaire 127 Spruce Lake Trails Area Visitor Study T H E U N I V E RS I T Y O F B R I T I S H C O L U M B I A Research Coordinator Ron Rutledge Faculty ot Forestry University of British Columbia 270-2357 Main Mall Vancouver, B.C. Canada V6T 1Z4 (604) 822-5092 128 PART I We would like to begin by asking you to rate a series of short statements describing situations you might encounter while visiting wilderness areas like the Spruce Lake Trails Area. After reading EACH statement, please rate the situation described as to the extent to which it would AFFECT the quality of YOUR wilderness experience. As an example, consider the following statement: "As you walk along a forest trail, you meet another visitor. You can see wildlife nearby." Please rate (by circling a number) how this would affect your wilderness experience. Extremely negative no effect Extremely positive effect s~\ a t effect -4 -3 -2 A l / 0 +1 +2 +3 +4 If you feel that encountering this situation would have a slight negative affect on your wilderness experience, you might circle -1 as shown above. If you feel that encountering this situation would have a slight positive effect on you wilderness experience, you might circle +1. Please remember that there is no right or wrong answer. What is important is you rating the effect upon your wilderness experience. Now, please rate (by circling a number) how each of the following situations would affect your wilderness experience. As you rest by a lake, a large group of visitors comes up. Some of them are discourteous to you. There are some campfire rings made by others by the lake." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 129 2 . "While you are in the area, you see no other visitors. Trail and campsite conditions are good, and there is little evidence of serious site impacts." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 3 . "while you are in the area, you notice some trees damaged by humans. Visitors from another group pass by you. Some of them are loud and noisy." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 4 . "You see a visitor camped near your campsite. Some vegetation loss and bare ground is present in the site." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 5 . At your forest campsite, you can see forest service signs. There are some campfire rings made by others around the site. You encounter a visitor who is discourteous to you." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 6 . A large group of visitors passes by you when you are in the area. There are some trees damaged by humans in the area. You can see no structures built by people." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 130 7 . As you are resting at your campsite, a large group of campers passes by you. Some of them are discourteous to you. There are structures built by people near the campsite, and some trees damaged by humans are nearby." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 8. There is a loud group of visitors camped near you. You can see some campfire rings made by others around the campsite, and structures built by people are visible." Extremely negative no effect Extremely positive effect at all effect -2 -1 0 + 1 + 2 + 3 + 4 9. "While you are in the area, you can hear a visitor who is loud and noisy. There are some campfire rings made by others in the area. You see no structures built by people." Extremely negative no effect Extremely positive effect at all effect I - 4 - 3 - 2 - 1 0 +1 +2 +3 +4 1 0 . While you are at a lake, you can see a float plane approaching the lake. The lake and shore area are in good condition. You encounter no other visitors." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 11. "A large group of visitors passes by your campsite. There are some trees damaged by humans at the site." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 131 PART II Now we would like to ask some questions that relate to your most recent visit to the Spruce Lake Trails Area. 1 2 . What was the date of your most recent visit to the Spruce Lake Trails Area? 1 3 . What travel method did you use to get into the Spruce Lake Trails Area? (circle one number) 1 Hike 2 Horse 3 Aircraft/Helicopter 4 Mountain Bike 1 4 . What was your primary reason for going to the Spruce Lake Trails Area? (circle one number) 1 Fishing 5 Picnicking 2 Hunting 6 Camping 3 Nature study 7 Other (please specify) 4 Hiking 1 5 . Did you use a commercial guide, outfitter, or pilot? 1 Yes 2 No 1 6 . How many nights did you stay in the Spruce Lake Trails Area? Nights spent at Spruce Lake Nights spent elsewhere in the Spruce Lake Trails Area 1 7 . How many people were in your group (including yourself)? If more than yourself, 1 8 . What type of group were you with? (circle one number) 1 Family 4 Club or organized group 2 Friends 5 Other (please specify) 3 Family and friends 132 P A R T III Again, we would like to ask you to rate another series of short statements describing situations you might encounter while visiting wilderness areas like the Spruce Lake Trails Area.1 Remember, after reading EACH statement, please rate the situation described as to the extent to which it would AtthCT the quality of YOUR wilderness experience. Now, please rate (by circling a number) how each of the following situations would affect your wilderness experience. 19. While at your campsite, you can see a float plane approaching the area." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 44 20. "A group of visitors passes by you at a forest lake. The group is loud and noisy. You can see an aircraft preparing to land on the lake." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 + 1 +2 +3 +4 21. "As you have traveled and camped in the area, you have not seen or heard any other visitors. You haven't observed any damage to vegetation or trees, and havenl observed any structures made by people." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 133 2 2 . While you sit at a lake, a visitor walks by. The person is loud and noisy." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 2 3 . As you travel through the forest, you notice some campfire rings made by others along the trail. A group of visitors passes by you. You can see no structures built by people." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 + 1 + 2 + 3 + 4 \ 2 4 . "You can see structures built by people near your campsite. A group of other campers are nearby. Some of them are discourteous to you. You can see litter in the campsite." Extremely negative no effect Extremely positive effect at all effect A -3 2^ 1^ 0 +1 +2 +3 +4" I 2 5 . While you are in the area, a large group of visitors comes by you. You see a float plane circling overhead. You have noticed some trees damaged by humans in the area." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 134 2 6 . There is a loud group of visitors camped near you. You can see some vegetation loss and bare ground around the campsite, and structures built by people are visible." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 2 7 . "At a lake in the area, you encounter a visitor. The person is loud and noisy. The lake and shore area is in good condition. You see an aircraft circling the lake." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 +4 2 8 . "As you sit in you camp, a large group of campers come up. Some of the other campers are loud and noisy. You can see forest service signs posted near the site. There is some litter on the ground." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 +1 +2 +3 44 2 9 . "As you travel on a forest trail, you see some vegetation loss and bare ground by the trail side. A visitor passes by on the same trail." Extremely negative no effect Extremely positive effect at all effect -4 -3 -2 -1 0 4-1 4-2 4-3 4-4 135 PART IV Conditions typically vary between different wilderness areas and at different locations within the same wilderness area. For the following items, please indicate the maximum amount you would accept in the Spruce Lakes Trails Area before your wilderness experience would be changed. Use "0" for none. If you can't give a number for a particular item or location, leave it blank. AT SPRUCE LAKE ITEMS On or Beside At Trafe Camps E L S E W H E R E IN THE AREA Onorbeside At Trails Carnps 30 . Pieces of litter seen on any one day 31 . O n any site, the number of camp-fire rings made by others 32 . O n any site, the percent of trees damaged by humans 33 . O n any site, the square metres of vegetation loss or bare ground (1 sq. metre = approx. 9 sq. ft.) 34 . Number of individuals seen other than in your own group on any one day 3 5 . Number of other groups seen on any one day 36 . Number of large (more than 6 people) groups seen on any one da> 3 7 . Number of aircraft/helicopters sighted or heard (excluding high altitude jets) on any one day 38 . On any site, the number of human-made structures seen 39. On any site, the number of forest service signs seen 136 PART V Finally, we would like to ask you a few questions about yourself to help us interpret our results, to further our understanding of all wilderness visitors, and to better protect B.C.'s wilderness recreation areas. Remember, your responses will be kept strictly confidential. 40. Are you? 1 male 2 female 41. In what year were you born? 19 42. Please circle your highest level of formal education. 1 Grade school (1-8) 4 Completed college/university 2 High school (9-13) 5 Post graduate 3 Some college/university 43. With which ethnic or cultural group do you identify? 44. Where do you permanently reside? Province/state Country 45. Is your present residence? (circle one) 1 On a farm 4 In a town 1,000-9,999 2 On rural non-farm acreage 5 In a city 10,000-99,999 3 In a town less than 1,000 6 In a metropolis 100,000 + 137 Thank you very much for your time and effort in completing this questionnaire. Your participation is greatly appreciated and will contribute to the special wilderness research study being conducted in the Spruce Lake Trails Area. In addition, your involvement in wilderness recreation research like this can have an effect on the way B.C.'s unique wilderness are are cared for and protected. If you have any comments about these issues or specific questions in the questionnaire, please write them in the space below. ONE FINAL REQUEST It would be very helpful to this Wilderness Research Study (UBC # 95784) if we could contact you again in about a year's time. Are you agreeable to this? (circle a number) 1 Yes ( please note your address so we can keep you on the mailing list) 2 No Please return this questionnaire in the stamped envelope provided 138 APPENDIX 2 Trailhead Sign 139 Spruce Lake Trails Area Visitor Study Trailhead Information Sign S P E C I A L W I L D E R N E S S R E S E A R C H S T U D Y A L L V I S I T O R S P L E A S E R E G I S T E R W H E N E N T E R I N G T h e U n i v e r s i t y o f B r i t i s h C o l u m b i a i n c o o p e r a t i o n w i t h t h e B . C . F o r e s t S e r v i c e i s c o n d u c t i n g a r e s e a r c h s t u d y i n t h e S p r u c e L a k e T r a i l s A r e a . T o h e l p p r o t e c t a n d m a n a g e t h e a r e a w e n e e d t o k n o w m o r e a b o u t y o u t h e b a c k c o u n t r y v i s i t o r a n d w h a t y o u t h i n k . P l e a s e w r i t e t h e n a m e s a n d a d d r e s s e s o f e a c h p e r s o n o v e r 1 8 i n y o u r p a r t y o n a c a r d f r o m t h e s p e c i a l l y m a r k e d b o x a n d d r o p i t t h r o u g h t h e s l o t . S o m e o f y o u w i l l b e p i c k e d a s s a m p l e v i s i t o r s a n d m a i l e d a q u e s t i o n n a i r e . A l l r e s p o n s e s w i l l b e k e p t s t r i c t l y c o n f i d e n t i a l . T H A N K S F O R Y O U R H E L P 140 APPENDIX 3 Contact Card 141 Spruce Lake Trails Area Visitor Study Visitor Contact Card S P E C I A L W I L D E R N E S S R E S E A R C H S T U D Y V I S I T O R C O N T A C T C A R D Date Name Street Address City Province (State) Country Postal Code (ZIP) Travel Method: Hike Horse Aircraft Mountain Bike_ Primary Activity: Fish Nature Study Hunt Other Did you use a commercial guide or a commercial outfitter? Yes No Planned length of stay: days Number in your group: THANKS FOR YOUR HELP 142 . APPENDIX 4 First Mailing Introduction Letter 143 APPENDIX 5 Reminder Letter 145 APPENDIX 6 Second Mailing Introduction Letter 147 APPENDIX 7 Third Mailing Introduction Letter 149 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0075243/manifest

Comment

Related Items