250 
Fishery Bulletin 112(4) 
Table 3 
The effect of incorporating survey weights into a conditional logit model of angler site- 
choice is shown below for 4 different versions of the model. The unweighted version 
included no adjustments for survey weights. The other 3 versions incorporated weights 
on the basis of the ratio of the proportion of total angler trips by zone to sample propor- 
tions. Different weighted versions of the model, with mean values or the lower or upper 
limit of 95% confidence intervals of the total proportion of estimated trips, were used 
to estimate parameter values with a weighted exogenous sampling maximum likeli- 
hood function (WESML). Estimated parameters (with standard errors of the mean in 
parentheses) are given for the independent variables. Data used in analyses are from 
the National Marine Fisheries Service’s Angler Access Point Intercept Survey for trips 
in the Gulf of Mexico in 2009 that targeted groupers (Epinephelus spp., Hyporthodus 
spp., or Mycteroperca spp.) or red snapper ( Lutjanus campechanus). 
WESML 
Variable 
Unweighted 
Mean 
Lower limit 
Upper limit 
Travel cost 
(2009 dollars) 
-0.060 
(0.002) 
-0.056 
(0.002) 
-0.067 
(0.003) 
-0.054 
(0.002) 
Log of number of 
sites per zone 
0.251 
(0.099) 
0.242 
(0.092) 
0.571 
(0.142) 
0.063 
(0.085) 
Grouper harvest 
rate per angler 
(number of fish) 
caught and kept) 
0.784 
(0.189) 
0.819 
(0.174) 
0.543 
(0.280) 
0.829 
(0.168) 
Red snapper harvest 
rate per angler 
(number of fish 
caught and kept) 
0.471 
(0.099) 
0.603 
(0.091) 
0.303 
(0.126) 
0.748 
(0.088) 
Log likelihood function 
-914.42 
-942.68 
-838.64 
-971.55 
Number of zones in model 
28 
28 
16 
28 
sented the most recent study that is most comparable 
to our study. Their estimates of angler WTP for red 
snapper and groupers calculated with APAIS data from 
2000 in the Gulf of Mexico were also much higher than 
our estimates. On the basis of the most comparable 
model (conditional logit), they estimated angler WTP 
for an additional red snapper at $123 and for an ad- 
ditional grouper at $91 (in 2000 dollars). At least part 
of the differences in WTP estimates for groupers and 
red snapper between the Haab et al. (2012) study and 
our study likely was due to the inclusion of the for- 
hire mode (with higher associated travel costs) and the 
opportunity cost of time in their model. However, it is 
difficult to tell how much those inclusions contributed 
to the differences in the WTP estimates without a more 
detailed comparison of the models. 
This study was based on the private-boat angler’s 
choice across boat launch sites. However, anglers who 
launch from either a fixed boat slip or a boat storage 
facility will not regularly choose to launch from other 
places. Unfortunately, the APAIS data does not identify 
whether a boat was launched from a trailer, a slip, or 
a fixed dock. In the case of the APAIS data used in our 
study, more than 64% of the intercept sites along the 
Gulf of Mexico where anglers were interviewed in 2009 
were only boat ramps with no associated boat slips or 
boat storage areas. The majority of anglers included in 
our study trailer their boats and, therefore, have a choice 
regarding what launch site they use. The potential lim- 
itation of the standard site-choice model with regard 
to choice of boat launch site should be explored in the 
context of future work, but it does not affect our results 
from the use of APAIS weights in our site-choice models. 
Conclusions 
In this study, different versions of a site-choice model 
for analysis of fishing site choices in the Gulf of Mexico 
in 2009 were estimated with recently available sam- 
pling weights from the APAIS survey in order to il- 
lustrate how site-choice models can incorporate a 
complex sample design and reduce potential biases in 
estimation. Model results indicate that the addition 
of sampling weights affected the estimated param- 
eters for historic catch-and-keep rates. The difference 
