Violent crime and Poisson regression: A measure and a method for cultivation analysis

A content analysis of network prime-time television found violent crime ove r-represented and property crime (burglary) under-represented when compared to crime statistics. These differences were used to predict survey respond ents' estimations of crime rates in their community. A significant relation ship was found between television exposure and beliefs about crime rates wh en using a non-linear analytic model. Heavier television viewers were more likely to give higher estimates of murder rates and lower estimates of burg lary rates than lighter viewers. These estimation patterns are consistent w ith televised portrayals and offer support for the cultivation hypothesis.