Rigor and transparency in statistical analyses can help to ensure valid research

Abstract

Adherence to important assumptions of statistical methods has significant ramifications for development of new knowledge in landscape ecology for two fundamental reasons: these methods will continue to be used widely and will thus affect much of the research on which advances in landscape ecology will be founded; and the degree to which statistical methods are applied appropriately will influence the statistical validity of that research. Rigorous statistical analyses are essential because no discipline can efficiently advance scientifically if one of its primary approaches for generating new knowledge is used incorrectly. Assessing and communicating compliance with statistical assumptions should be standard practice in confirmatory analyses. Better understanding of the robustness of statistical methods to deviations from assumptions can improve investigators’ decisions about which methods to apply. Explanations about the consequences of actual or possible violations of assumptions can clarify the validity of results. Many of the papers in a sample of 215 research articles published in Landscape Ecology during 2004–2013 exhibited substantial lack of clarity about adherence to statistical assumptions. Brief author statements about whether important statistical assumptions were adequately met would improve confidence in results. Ultimately, rigor and transparency in confirmatory statistical analyses will help to ensure the validity of landscape ecology research.