Premium
Use of a Computer Simulation Model to Determine the Behavior of a New Survey Estimator of Recreational Angling
Author(s) -
Jones Cynthia M.,
Robson Douglas,
Otis David,
Gloss Steven
Publication year - 1990
Publication title -
transactions of the american fisheries society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.696
H-Index - 86
eISSN - 1548-8659
pISSN - 0002-8487
DOI - 10.1577/1548-8659(1990)119<0041:uoacsm>2.3.co;2
Subject(s) - estimator , statistics , recreational fishing , normality , minimum variance unbiased estimator , variance (accounting) , bias of an estimator , sampling (signal processing) , trimmed estimator , fishing , survey data collection , recreation , mathematics , consistent estimator , distribution (mathematics) , econometrics , computer science , fishery , ecology , biology , economics , mathematical analysis , accounting , filter (signal processing) , computer vision
We have recently developed a new estimator of recreational fishing effort. Here we used a simulation model to determine and demonstrate the statistical behavior of this new estimator. The estimator is used with access‐intercept surveys and was designed to give accurate, efficient estimates of fishing effort over a geographically large and diverse fishery. Because this is a new estimator, little is known about its behavior. Specifically, the form of the estimatorˈs sampling distribution, the variance components (within‐day versus between‐day), and its t ‐distribution were unknown and could not be determined analytically. Hence, to assist people who will want to know the statistical properties of this estimator and to characterize it more completely, we studied its behavior numerically by use of a simulation model based on real‐world data. Analysis of the simulation results showed the sampling distribution of the estimator to be non‐normal when limited to a single survey route; it was more closely approximated by a gamma distribution. The estimator approached normality when used to estimate effort from multiple‐route (large‐scale) fisheries with greater fishing effort. Within‐day variance (influenced by starting position along the route and direction of travel) was larger than the between‐day (day‐to‐day) variability. Because the sampling distribution of the estimator was non‐normal, the t ‐distributions were generated empirically to determine the direction and degree of misspecification when the usual Student t ‐distribution was used. Use of the Student t resulted in slightly skewed ∝ values with too large a probability of inclusion in the lower tail and too small a probability in the upper tail.