The use of sampling weights in regression models of recreational fishing-site choices

The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Abstract—Improved methods for estimating saltwater recreational fishing catch and effort have been developed by the NOAA National Marine Fisheries Service. Sampling weights that account for a complex sample design in surveys of anglers are now available with NMFS catch and effort estimates. Previously, estimates of the economic value to anglers (known as the “willingness to pay”) for additional fish caught that were based on angler surveys did not typically account for the underlying complex sample design. In this study, a recreational-demand model was used for analysis of fishing site choices in the Gulf of Mexico in 2009 among private-boat anglers who target groupers (Epinephelus spp., Hyporthodus spp., or Mycteroperca spp.) or red snapper (Lutjanus campechanus). Different versions of the model were developed with and without accounting for the complex sample design. Results between the unweighted version and weighted versions of the model varied in estimates of catch between sites and the value anglers place on being able to catch and keep additional fish. In 2012, the NOAA National Marine Fisheries Service (NMFS) released a new method for estimation of recreational fishing catch and effort based on data obtained from its Access Point Angler Intercept Survey (APAIS) of saltwater anglers. Previous methods of estimation of catch and fishing effort from this intercept [interview] survey were subject to a number of different potential biases as pointed out by the National Research Council of the National Academies (NRC, 2006). In particular, the earlier estimation methods did not account for the complex sample design of the intercept survey and instead simple random sampling was assumed. The new method of estimating catch and effort uses specially calculated weights and variance adjustments (Breidt et al.1). The APAIS sampling weights incorporate information from a sepa-