The relationship between nonresponse bias and angler specialization

Abstract

This research examined the effect of nonresponse bias on angler surveys, and estimated the size and number of angler specialization groups from mail surveys with incomplete response. Angler surveys can yield inaccurate results because of the failure of some sampled units to return their questionnaire. Consequently, inferences made from the respondents to the population of interest will be subject to nonresponse bias. Using results adjusted for nonresponse, the size and number of angler specialization groups were estimated. Knowledge of the number and size of specialization groups can be used by fishery managers to relate the different types of experiences sought by the angler population. A 10-page mail questionnaire was sent to a stratified random sample of 9,981 Texas fishing license holders. Information was collected on fishing experience, fishing participation, species preferences, attitudes, and orientation to fisheries management efforts. A response rate of 62% was obtained (excluding non-deliverables). Adjustments for nonresponse were made using the method of response propensity stratification. Individual response probabilities were determined using logistic regression with a binary variable indicating response as the dependent variable and the external variables age, sex, race, and purchase date of fishing license. Response probabilities were inverted, sorted, and grouped into cells to obtain mean nonresponse adjustment weights. Statistically significant nonresponse bias was detected when unadjusted survey variables were compared with adjusted variables. The size and number of angler specialization groups were determined using cluster analysis. Four hierarchical methods were used to determine the number of groups, and a nonhierarchical method was used to determine the size of each group, with the number and starting means specified in advance. Seven specialization groups were recovered, ranging from 11% to 19% of the total population, after correcting for nonresponse bias. Groups were verified with external variables; attitudes toward management options, fishing tournament participation and fishing club membership were significantly different between groups...