Sample Size Estimation for On-Site Creel Surveys

Conducting sample size analysis is important to ensure that sample sizes are adequate to meet objectives for precision but not so large that valuable resources are wasted. When simple survey designs are used, sample size analysis is straightforward. However, creel surveys often follow complex designs that can make sample size estimation difficult. The objectives of this study were to provide sample size estimators for commonly used creel survey designs and investigate sample size requirements to achieve varying levels of precision. For estimates of angling effort, the average sample size among fisheries required to achieve relative 95% confidence intervals of 40% (i.e., coefficient of variation of approximately 0.2) was 16 d (range = 7–40 d). To estimate mean catch rate with the same level of precision, 43 survey days (range = 8–95 d) were required when the daily catch rate estimator was used, and 11 d (range = 3–30 d) were required when the multi-day estimator was used. Fifty-five days (range = 14–89 d) and 57 d (range = 19–140 d) were required to estimate total catch with relative 95% confidence intervals of 40% when the daily and multi-day estimators were used, respectively. Although absolute precision will vary among fisheries, our results suggest that approximately 30 d of surveying may serve as a reasonable starting point for planning creel surveys when no prior information is available on expected sample variance, as relatively large increases in sample size beyond 30 d were needed to achieve relatively small increases in precision. Received February 14, 2017; accepted June 9, 2017 Published online August 4, 2017

开展样本量分析至关重要，其核心目标在于确保样本量足以满足精度要求，同时避免因样本量过大而浪费宝贵的研究资源。当采用简单调查设计时，样本量分析过程直观简便。然而，渔获统计调查（creel survey）通常采用复杂的设计方案，这使得样本量估算工作颇具挑战。本研究旨在为常用的渔获统计调查设计提供样本量估算方法，并探究实现不同精度水平所需的样本量需求。针对捕捞努力量的估算，当要求相对95%置信区间为40%（即变异系数约为0.2）时，各渔业研究所需的平均样本量为16个调查日（范围：7~40个调查日）。若以相同精度水平估算平均渔获率，采用单日渔获率估计量时，需43个调查日（范围：8~95个调查日）；采用多日渔获率估计量时，则仅需11个调查日（范围：3~30个调查日）。当采用单日与多日估计量、以相对95%置信区间40%估算总渔获量时，分别需要55个调查日（范围：14~89个调查日）与57个调查日（范围：19~140个调查日）。尽管不同渔业的绝对精度会存在差异，本研究结果表明，若缺乏预期样本方差的先验信息，在规划渔获统计调查时，以约30个调查日作为合理的起始样本量是可行的——因为若要实现精度的小幅提升，往往需要将样本量大幅增加至30个调查日以上。收稿日期：2017年2月14日；录用日期：2017年6月9日；在线发表日期：2017年8月4日