Reevaluating Europe’s other debt with improved statistical tools

Abstract

Dullinger et al. (Proc Nat Acad Sci USA 110:7342–7347, 2013) showed that current proportions of threatened species in European countries are more correlated to socio-economic pressures calculated a century ago than to current socio-economic pressures. These results could have important impacts on future political decisions, yet the statistical analyses used may be strongly impacted by pseudo-replication and over-dispersion of data. I reanalysed Dullinger et al.’s data using generalised linear mixed models accounting for pseudo-replication and over-dispersion. These new statistical models indicated that socio-economic pressures had a much less clear impact on proportions of threatened species than indicated by Dullinger et al. In many cases, the effects of socio-economic pressures even vanished. Dullinger et al. (Proc Nat Acad Sci USA 110:7342–7347, 2013)’s results are actually much less “considerable” than their paper implies and therefore should be viewed with caution. Ecologists should more often incorporate pseudo-replication and overdispersion in their statistical analyses.