Effects of Recall Bias and Nonresponse Bias on Self-Report Estimates of Angling Participation

Abstract

We addressed the problems of recall and nonresponse biases in self-report angling participation surveys. While each source of bias has been recognized as problematic, previous research has not addressed the interaction effects of these biases. Recall bias was assessed by comparing immediate, 3-month, and 6-month recall periods. A diary format was used for the immediate recall period and mail surveys for the 3-month and 6-month recall periods. Nonresponse bias was assessed by conducting telephone interviews with subjects who did not respond to the mail or diary surveys. The dependent variable was total number of days fished over two separate 3-month periods. Analysis showed significant first-order interactions (P ^ 0.05), suggesting that recall bias and nonresponse bias are not independent factors. Respondents are more likely to report higher levels of participation and nonrespondents lower levels of participation as the length of recall period increases. Findings indicate that studies which use long recall periods, or do not control nonresponse bias, overestimate use. Future studies can control recall and nonresponse biases by combining frequent sampling with telephone interviews that request short recall periods.Effective management, planning, and develop-gent validity. Convergent validity addresses the ment of recreational fisheries resources are in-extent to which different methods give the same creasingly dependent upon accurate and reliable estimates of angling participation. When estimates data sources. An important aspect of this data col-of angling participation vary widely among estilection process involves the retrieval, analysis, and mation methods, it calls into question the validity evaluation of information concerning the behav-of the estimates and limits their utility to manior, preferences, catch, and effort of sport anglers, agers. This information can be useful in a variety of ways,The accuracy of estimates can be improved by including the identification of use trends, projec-examining the magnitude of biases that occur in tion of angler demand, evaluation of the impacts existing survey methods and refining methodolof management decisions, allocation of funding, ogies so they explicitly control for method effects, and examination of user conflict.This study examines the effects of method bias in As fisheries planners and managers have be-the Illinois Sport Fishing Survey (ISFS). The ISFS come more involved in the collection and use of is the primary mechanism used to assess angler angler information, they have become increasingly information in Illinois (Illinois Department of aware of measurement problems that can affect Conservation 1988). the validity and reliability of study results. OneThe ISFS uses a mail survey for collecting statearea of concern examined in our study is conver-wide angler information, as do many other agencies (e.g., USFWS 1982;Harris and Bergersen ----1985), because mail surveys are a cost-effective

How to Bait the Hook: Identifying What Motivates Anglers to Participate in a Volunteer Angler Data Program

“…However, it is consistent with the number of participants in many VAD programs reported in the literature (e.g., Younk and Cook 1992;Tarrant et al 1993). Nonparticipants in this case were individuals who had at some point registered for the AAP but had never participated, and it is possible that the barriers to participation differ for those who had never interacted with the program at all.…”

Angler apps as a source of recreational fisheries data: opportunities, challenges and proposed standards

“…First, because angler app users may be study subjects in addition to data collectors, angler app data may be more prone to biases that result from non-random participation (i.e. App data are likely to suffer from avidity bias (Jiorle 2015) because the willingness of anglers to share data increases with avidity (Harris and Bergersen 1985;Tarrant et al 1993;Connelly and Brown 1995), and to underestimate effort if there is little incentive to log trips that do not result in a catch. For example, data from the fishing app iAngler (Table 1) were biased towards urbanized regions and inshore species (Jiorle 2015).…”

“…can be subsampled according to the demographic(s) of interest or reweighted based on statistically rigorous estimates of the general angler population (e.g. Finally, persistent biases can be addressed via correction factors that are obtained through validation or simulation (for conventional examples seeTarrant et al 1993;Connelly and Brown 1995;Connelly et al 2000;Sullivan 2002;Ashford et al 2010). Finally, persistent biases can be addressed via correction factors that are obtained through validation or simulation (for conventional examples seeTarrant et al 1993;Connelly and Brown 1995;Connelly et al 2000;Sullivan 2002;Ashford et al 2010).…”

Sea Bass Angling in Ireland: A Structural Equation Model of Catch and Effort

“…For example, catch rates may be a predictor of recreational demand, because catch may be perceived as a measure of site quality (Parsons and Needelman, 1992;Englin and Lambert, 1995). Furthermore, using self-reported catch as a demand predictor also leads to measurement error due to recall bias (Tarrant et al, 1993;Morey and Waldman, 1998). Structural equation models have been used to circumvent this issue.…”