When estimating commercial fishing costs, selection bias can impact any model derived from non-census sampling methodologies. In the northeastern United States, commercial fishing operating cost models may suffer from selection bias, as they are often estimated using data collected for biological, rather than economic, purposes. We investigate the effects of sampling bias on trip cost model estimations using weighted/unweighted least squares and Heckman sample selection models. Results suggest that (1) the propensity for a trip to carry an observer is not random with respect to costs and that (2) selection bias exists in the majority of cost models investigated. To gauge the magnitude of selection bias, we compare results of the unweighted least squares and Heckman models. The differences between models can lead to erroneous conclusions at the subfleet level and in estimating trip cost maxima. Results suggest that assessing and correcting for selection bias is necessary when using sampled fishing cost data.

中文翻译：

---

使用海上采样数据估算商业捕鱼旅行成本

在估算商业捕捞成本时，选择偏差会影响源自非普查抽样方法的任何模型。在美国东北部，商业捕捞运营成本模型可能会受到选择偏差的影响，因为它们通常是使用为生物而非经济目的收集的数据进行估算的。我们使用加权/未加权最小二乘法和 Heckman 样本选择模型研究抽样偏差对行程成本模型估计的影响。结果表明 (1) 携带观察员旅行的倾向在成本方面不是随机的，并且 (2) 选择偏差存在于大多数调查的成本模型中。为了衡量选择偏差的大小，我们比较了未加权最小二乘法和 Heckman 模型的结果。模型之间的差异可能会导致在子舰队级别和估计行程成本最大值时得出错误的结论。结果表明，在使用抽样捕捞成本数据时，有必要评估和纠正选择偏差。