|
Abstract:
|
Meta-analyses often use weighting to account for unequal precision of parameter estimates. The method of weighting estimates by the inverse of the estimated sampling variance is often used in meta-analyses of survival estimates from capture-recapture studies (e.g., Jolly-Seber estimates). Weighting by inverse variance may give biased results, because Jolly-Seber survival estimates are correlated with their estimated sampling variances. We propose alternative weighting methods, including the inverse of the squared coefficient of variation. Using Monte Carlo simulation, we investigate the effect of weighting method on the performance of resulting estimators under scenarios involving stationary and non-stationary processes. The choice of most effective method depends on multiple factors, including analysis objective, degree of stationarity of the survival estimates, and sampling effort.
|
