Formative patterns of vegetation responses to cattle grazing can be difficult to detect because of innate het­erogeneity of vegetation communities and grazing patterns, especially in semiarid environments. Nonetheless, some of this heterogeneity can be accounted for using appropriate experimental designs and statistical analyses. Our objective is to determine if postulates proposed by Bork and Werner (interpretation of effects when negative consumption values are manipulated) are applicable to a real data set utilizing a paired experimental design in a heterogeneous, semiarid environment.