Cultivation theorists hypothesize a monotonic pattern of means across viewing subgroups, that is, that people in groups defined by higher levels of television viewing will be more likely to give the “television world” answer than will people in groups defined by lower levels of viewing. This prediction, along with the methodological practices used in conducting research on the theory, make it clear that there is an assumption of linearity that is accepted by both theorists and researchers. However, there is reason to speculate that the underlying cultivation relationship may not be linear. This study provides empirical evidence that the relationship is not linear. Because of the nonlinearity of the distribution, the selection of cut points is a critical decision that can significantly influence the results of subsequent analyses. Also, the popular use of certain statistical procedures designed to test linear relationships is also challenged.