Flagship conservation areas 557 The final list consisted of five attributes (Table 1): (1)
presence or absence of threatened bird species, as the protection of threatened birds can be highly valued in conservation projects (Loomis & White, 1996; Lewis et al., 2018) and the notion of threatened provides a strongly conservation-related attribute, (2) presence or absence of charismatic mammals, as these can be highly valued in flag- ship species campaigns (Smith et al., 2012) and occur in many South African conservation areas, (3) presence or absence of legal protection, as legal designations can make conservation areas seem more legitimate to donors (Hayes & Ostrom, 2005), (4) high or low existing conservation funding, as some flagship species campaigns highlight level of neglect as a key marketing approach (Veríssimo et al., 2017), and (5) ownership by government, private entity, charity or community, as this relates to notions of power and accountability (Borrini-Feyerabend & Hill, 2015)and could affect donor perceptions of how their money would be spent, as well as what the social implications might be. We originally named this final attribute ‘governance’ because this term is used by IUCN when describing oversight of protected area management (Borrini-Feyerabend et al., 2013). However, after carrying out a pilot of the choice experiment with a small group of UK residents, we found that ownership type was more clearly understood by respondents. We recruited respondents for the choice experiment via
Prolific (2019b), an online research platform where respon- dents are paid a small amount for participating in online stud- ies. This ensured that the sample size was large enough to generate reliable model estimates (Bliemer & Rose, 2011)and it allowed access to a broad sample of the UK population. Online platforms provide a valid and reliable method of be- havioural data collection (Horton et al., 2011; Peer et al., 2017) and we chose Prolific because it has a large proportion of UK- based respondents (Peer et al., 2017), is specifically designed for academic research, and has ethical standards for respon- dent payment (Palan & Schitter, 2018).We recruited 852 UK- based respondents in May 2019 whowere paidGBP 0.50 each. We used NGENE 1.2 to generate an unlabelled choice ex-
periment with aDz efficient design, assuming amultinomial logit and null (zero) priors (Choicemetrics, 2018). This con- sisted of 36 choice pairs split into four blocks, so each respondent only answered one block of questions (nine choice pairs) and was less likely to show respondent fatigue (Mangham et al., 2009). At the start of the questionnaire, respondents were presented with a table of attribute descrip- tions. They were then allocated to a block of questions ac- cording to their birth month, producing a relatively even spread of respondents across the blocks. We presented the choice pairs in cards that appeared one at a time (Supplementary Fig. 1). The order of the attributes presented on the choice cards was shuffled for every card to prevent location bias (Mangham et al., 2009). At the end of the ques- tionnaire, we collected socio-economic data (Table 1).
To ensure validity, we checked all responses for unrea-
sonably short completion times, using two minutes as the minimum required time. We also checked the data for signs of respondent fatigue (Bradley & Daly, 1994). This would show as straight As or Bs being chosen towards the end of the questionnaire, or it could appear as different attributes determining respondents’ choices at the begin- ning compared with the end of the nine choice questions (Veríssimo et al., 2009). To test for this fatigue, we used χ2 tests to assess the proportion of As and Bs and the propor- tion of each binary attribute level chosen at each choice position in the questionnaire.
Econometric data analysis
Choice experiments have a strong grounding in random utility theory (Hensher et al., 2015) and it is assumed that an individual’s(n) preferences are the sum of a systematic, observable component and a random component:
Uni = Vni(Xnib)+eni 1
where Uni is the perceived utility of alternative i, Vni is the systematic component of utility that is a function of the at- tributes (Xni) and a vector of the parameter coefficients that relate to the appeal of the attributes (β), and eni is the ran- dom error component (Garnett et al., 2018). We initially employed the multinomial logit model
(Louviere & Hensher, 1982) to estimate our data. The model specification for our multinomial logit analysis was:
Ui =b1i(Tbsi)+b2i(Chai)+b3i(Pasi)+b4i(Ecfi) +b5i(Owni)+b6i(Doni)+ei
2 Explanations of the abbreviated attributes are shown in
order in Table 1. When specifying the full utility functions for the multinomial logit model, we dropped one level per attribute to avoid collinearity and used this as a reference level (Table 1). Thus, the parameter estimates calculated show preferences in relation to these reference levels. We also determined the mean willingness to pay, which involves calculating the ratio of each attribute’s parameter coefficient to the negative of the donation coefficient (Ryan et al., 2012). To add greater nuance to our analysis, we constructed a
latent class model to identify groups (i.e. preference het- erogeneity) within the sample according to respondents’ preferences and socio-economic traits (Swait, 2007). Each respondent was assumed to belong probabilistically to one of the identified groups (Garnett et al., 2018). When esti- mating a latent class model, researchers must determine the number of groups, or classes, to be estimated in the model, plus which variables to use to explain classmember- ship. We ran extensive tests using different combinations of socio-economic variables with different numbers of classes and made our class decisions based on an assessment of
Oryx, 2022, 56(4), 555–563 © The Author(s), 2021. Published by Cambridge University Press on behalf of Fauna & Flora International doi:10.1017/S0030605321000259
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93 |
Page 94 |
Page 95 |
Page 96 |
Page 97 |
Page 98 |
Page 99 |
Page 100 |
Page 101 |
Page 102 |
Page 103 |
Page 104 |
Page 105 |
Page 106 |
Page 107 |
Page 108 |
Page 109 |
Page 110 |
Page 111 |
Page 112 |
Page 113 |
Page 114 |
Page 115 |
Page 116 |
Page 117 |
Page 118 |
Page 119 |
Page 120 |
Page 121 |
Page 122 |
Page 123 |
Page 124 |
Page 125 |
Page 126 |
Page 127 |
Page 128 |
Page 129 |
Page 130 |
Page 131 |
Page 132 |
Page 133 |
Page 134 |
Page 135 |
Page 136 |
Page 137 |
Page 138 |
Page 139 |
Page 140 |
Page 141 |
Page 142 |
Page 143 |
Page 144 |
Page 145 |
Page 146 |
Page 147 |
Page 148 |
Page 149 |
Page 150 |
Page 151 |
Page 152 |
Page 153 |
Page 154 |
Page 155 |
Page 156 |
Page 157 |
Page 158 |
Page 159 |
Page 160 |
Page 161 |
Page 162 |
Page 163 |
Page 164
