Lovell and Carter: Sampling weights in regression models of recreational fishing-site choice 
249 
Table 2 
Factors that influenced an angler’s choice of fishing site for groupers (Epinephelus 
spp., Hyporthodus spp., or Mycteroperca spp.) or red snapper (Lutjanus campechanus) 
targeted from private boats in the Gulf of Mexico in 2009: average round-trip distance 
(in miles) between an angler’s permanent residence and a chosen fishing site (standard 
errors of the mean are presented in parentheses), average travel cost, harvest rate per 
angler trip (number of fish caught and kept), and average number of fishing sites in 
a zone (coastal county) where an angler fished. Also shown are the number of angler 
interviews in which anglers indicated they had targeted groupers or red snapper on 
the intercepted fishing trip. Data are from the National Marine Fisheries Service’s 
Angler Access Point Intercept Survey in Louisiana, Mississippi Alabama, and west 
Florida in 2009. 
Factor Groupers Red snapper 
Average round-trip distance (mi) 
Average travel cost (in 2009 dollars) 
Average harvest rate 
Average number of fishing sites per zone 
Number of interviews (trips) in 2009 APAIS data 
Estimated total number of trips 
52.61 (5.19) 
$26.30 (2.60) 
0.39(0.03) 
9.86 (0.61) 
725 
499,931 
72.32 (10.84) 
$36.16 (5.42) 
1.38 (0.07) 
6.42 (0.80) 
265 
216,770 
cate that weighting may matter with regard to the 
parameter and WTP estimates. However, on the basis 
of results from the method of convolutions, we cannot 
reject the null hypothesis that the unweighted and 
weighted estimates are equal at the 5% significance 
level. The P-values from testing the unweighted ver- 
sus the weighted WTP estimates for groupers at the 
mean, lower-limit, and upper-limit estimated popula- 
tion weights were 0.66, 0.16, and 0.71, respectively. 
Similarly, the P-values from testing the unweighted 
versus the weighted WTP estimates for red snapper at 
the mean, lower-limit, and upper-limit estimated popu- 
lation weights were 0.91, 0.09, and 0.99, respectively. 
Discussion 
The primary aim of this research was to examine esti- 
mates of angler WTP that are derived from recreational 
demand models based on the APAIS data with informa- 
tion from the new sampling weights. We used weighted 
and unweighted data to estimate different versions of 
a model of recreational site-choice and measures of an- 
gler WTP for changes in harvest rates for groupers and 
red snapper in the Gulf of Mexico. The structure of the 
models was selected so that it would be as consistent 
as possible with the standard site-choice models used 
for policy-making at NMFS. 
Depending on the weights used and species con- 
sidered, the difference in estimates of angler WTP for 
a one unit increase in the harvest rate between the 
weighted versions of the site-choice model and the 
unweighted version of the model ranged from 13% to 
77%. However, by conventional standards, none of the 
weighted estimates were statistically different from 
the corresponding unweighted estimates. Although not 
statically different, the variation could appear large 
when examining estimates across all anglers that tar- 
geted these species. Interpretation of aggregate esti- 
mates should be done with care so as not to misdirect 
policy recommendations. Using APAIS data for the 
southeastern United States in 2003-04 but different 
types of weights, Hindsley et al. (2011) found that 
weighted estimates of angler WTP were around 40% 
lower than unweighted estimates. Yet, they did not for- 
mally test the differences. Kuriyama et al. (2013) found 
that weighted estimates were around 30% lower than 
unweighted estimates of angler WTP for changes in 
harvest rates by shore anglers in California. The data 
used by Kuriyama et al. (2013) were similar to the 
APAIS data, but they used a more complicated model 
structure (mixed logit) and different types of weight- 
ing factors (e.g., interview effort). They did not formally 
test for statistically significant differences between the 
weighted and unweighted estimates of WTP in their 
shore fishing case study. 
A number of studies have estimated angler WTP for 
groupers and red snapper in the southeastern United 
States. Carter and Liese (2012) estimated that anglers 
were willing to pay between $25 and $80 (in 2003 dol- 
lars) to keep an additional grouper, depending on how 
many fish they had already kept on a trip. Their com- 
parable estimates for red snapper ranged from $20 to 
$62. These results are considerably higher than the es- 
timates reported in our study (Table 4). However, the 
analysis by Carter and Liese (2012) was based on stat- 
ed preference data, which tend to generate relatively 
higher WTP estimates than revealed preference data 
such as the data we used in our site-choice demand 
analysis (Johnston et al., 2006). Haab et al. (2012) pre- 
