The development and validation of a coding protocol to measure change in tobacco-control newspaper coverage.

The national ASSIST newspaper coding protocol model was used as a template to adapt a system for measuring tobacco-related newspaper coverage in Colorado newspapers. Over a 3-month period, tobacco-related articles were clipped from 180 daily and weekly newspapers. Variables coded included adaptations of the original ASSIST categories. During development and testing, additional variables were added to make the protocol more comprehensive and sensitive to tobacco policy media coverage. Inter-coder reliabilities were calculated for all non-static variables using Cohen's kappa. Disagreements were resolved through group discussions. Two rounds of testing achieved ratings above .70 for all variables. The protocol improves dramatically upon the ASSIST model by providing greater breadth and depth of analysis and more sensitivity to the nuances of newspaper coverage of tobacco-related issues. Given its simplicity, the protocol could also prove valuable for anti-tobacco advocacy groups who wish to track the changes in public and media opinions.