PUB - Publikationen an der Universität Bielefeld

A recent meta‐analysis concluded that the badge of status hypothesis is the most parsimonious explanation for an observed association between measures of aggression and the degree or area of coloration across animals. We argue that the premise for the conclusion that aggression and coloration are positively associated relies on a biased, largely incomplete and heterogeneous representation of the available evidence. In addition, we identified severe problems with the calculation and interpretation of the correlations and the heterogeneity, as well as with data extraction and the publication bias tests. Our re‐analyses challenge the original conclusions. Once heterogeneity—which, in contrast to the original study, our re‐analysis shows to be high—and publication bias are properly assessed and accounted for, we find low generality of the relationship between aggression and coloration across animals. In addition, our analyses suggest that evidence changes dramatically depending on the source of the effect size, with evidence obtained directly from correlations and mean group comparisons being statistically nonsignificant and largely different from evidence obtained by transforming inferential statistics ( t , F , and χ ² ), which raises strong methodological concerns. Furthermore, by re‐extracting 11 out the 74 studies included in the original meta‐analysis (15%), we found errors and inconsistencies that raise further concerns about the validity of the full dataset. We give directions to guide future meta‐analytic efforts in this and other topics, showcasing common misunderstandings in systematic reviews and meta‐analyses, including effect size calculation and transformation. We conclude that after 50 years since its conception, the jury is still out regarding the validity and generality of the badge of status hypothesis to explain individual differences in coloration traits across animals.