Estimating Trends With a Linear Model: Reply to Sauer et al

• DOI: 10.1093/condor/106.2.440
• Publication Date: 2019-09-06T05:01:25Z
• Authors: Jonathan Bart; Brian Collins; R. I G. Morrison
• Keywords: Observer (physics); Estimating equations

Sauer et al. (2004) advocate the use of trend estimation models that adjust counts for differences among observers. We agree such adjustments are sometimes needed, and we noted (Bart 2003) they may readily be carried out prior to using method described. Including observer covariates, however, is not always necessary substantially reduces precision, as acknowledge. Furthermore, under plausible conditions, including covariables introduces bias rather than removing it. In addition, weighting scheme used in estimating-equations approach introduce bias. Our avoids these sources bias, simpler more flexible estimating- equations (e.g., carrying power sample-size calculations much easier with our approach), has smaller standard errors approach, especially when fluctuate widely. Model-based methods, also have advantages, particularly assessing complex influences on counts. recommend analysts consider both approaches; comparing results obtained different methods informative.