Important issues that are still unresolved in the study of animal social groups are how dominance is measured and how individuals are ranked. Based on observations of mule deer (Odocoileus hemionus hemionus), we constructed hierarchies for the same 26 bucks using five potential correlates of dominance and three measures based on the outcome of agonistic interactions. Data for each of six behavior measures were converted to ranks by six different numerical techniques.

Different measures of behavior resulted in different hierarchies. This was especially so when the hierarchy based on sparring rates was compared to hierarchies based on other criteria. Although our results showed that dominance is not a unitary concept, several measures of dominance were highly correlated with one another. Thus, one measure, in some instances, may be a good but imperfect predictor of dominance defined by other criteria.

For data sets based on index scores or rates of performing behavior, the mean difference and standard-error difference ranking methods had distinct advantages. Both produced equally weighted ranks that minimized the effects of sampling errors.