Rank‐order analysis for robust multiresponse, multiblock comparisons: Evaluation of herbicide interactions

Studies of interactions among bioactive compounds are often difficult to interpret unambiguously. A priori assumptions about the nature of such interactions can seriously distort analysis of the data. By applying a rank order analysis appropriate to the naturally ordinal scale of response to xenobiotic insult, several co‐herbicides were successfully identified from among numerous candidates in an experiment involving multiple blocks, rates and species. Moreover, underlying herbicide interactions were substantiated and identified which were not apparent by more traditional parametric analysis.