On not learning from one's own mistakes: A Reanalysis of Gerbner et ally's Findings on Cultivation Analysis Part II

This article critically examines the statistical evidence and conceptual arguments presented by Gerbner et al. to support their assertion that television-viewing “cultivates” distorted perceptions of the “real world.” In Part II, I examine three reformulations or “refinements” of the original hypothesis of across-the-board, linear effects shown in Part I to lack empirical support. Reformulation 1 requires that heavy viewers in subgroups “portrayed” on television as most victimized perceive the world as “more mean” or “scarier” than light viewers in the same subgroups. They do not. Reformulations 2 and 3. proposed by the Annenberg group and called “mainstreaming” and “resonance,” are then critically examined in terms of their logical adequacy and claims of empirical support. These are shown to fail on both counts, as neither has been stated in a form offering predictions which can be supported or rejected. More generally. we conclude Gerbner et al.’S formulation(s) and assertions about cultivation effects are so inclusive that any response to survey items can be argued to support one or another version of the hypothesis. This makes the assertion both irrefutable and untestable, which argues for its rejection as a scientific proposition. Three alternative explanations of responses in the NORC data set are outlined, each providing a more statistically adequate and theoretically sensible interpretation than those offered by the Annenberg team.