Measurement Error in Angler Surveys

Tuesday, August 19, 2014: 2:30 PM
2104B (Centre des congrès de Québec // Québec City Convention Centre)
Alexander Alexiades , Natural Resources, Cornell University, Ithaca, NY
Patrick J. Sullivan , Department of Natural Resources, Cornell University, Ithaca, NY
Clifford Kraft , Natural Reources, Cornell University, Ithaca, NY
Ben Marcy-Quay , Cornell University, Ithaca, NY

Information on fishing effort, catch, harvest, and survival is critical for formulating management poli­cies in freshwater fisheries and understanding the dynamics of aquatic ecosystems. Fisheries managers often use creel surveys to assess fisheries statistics parameters. The mean-of-ratios estimator is traditionally the accepted method for estimating catch rates from incomplete angler trips, while the ratio-of-means estimator is preferable for estimating catch rates from complete trips. Recent studies have demonstrated persistent bias when comparing the two estimators using catch data from incomplete and complete trips from the same sample of anglers and promoted the use of linear regression models to correct for apparent bias, however,  ordinary least squares linear regression may be inappropriate to correct for this apparent bias because of measurement error in both the response and explanatory variables, which underestimates the slope of the relationship. In this context, model II regression methods provide less biased estimates. Using roving creel interview data (incomplete trips) and a catch card survey (completed trips) conducted on the same sample of anglers, we compared catch rates derived from both estimators to show that linear regression underestimates the slope of the relationship and that model II regression reduces bias and performs better as a corrective model.