Is pseudo-replication biasing results from analyses from the island closure experiment which model individual penguin responses directly?

Butterworth, Doug S; Ross-Gillespie, Andrea

Is pseudo-replication biasing results from analyses from the island closure experiment which model individual penguin responses directly?

Report

2019

Abstract

A simple simulation study is used to investigate the impact of possible pseudo-replication arising from the use of individual penguin observations, in contrast to annually aggregated measures, in analyses of the island closure experiment which attempt to estimate the possible effect on penguins of closure of the neighbourhood of these islands to pelagic fishing. Unlike the case for estimators based on annually aggregated inputs, those based on the use of individual observations are found to lead to possibly substantially negatively biased estimates of the standard errors of the parameter that reflects the effect on penguins of these closures. This means that past results concerning the statistical significance and probabilities that island closures impact penguins from analyses based on individual observations need to be reconsidered. Previous analyses using this approach should ideally be repeated based on year-aggregated inputs, and future analyses need to avoid repeating this earlier approach.

Reference:

Collections

Reports

Full item page