Investigating the Characteristics of Measures Used for Tracking and Evaluating the Effectiveness of Anti-Tobacco Communication Campaigns

This study investigates the measurement and psychometric features of survey questions common to anti-tobacco campaign evaluations. The constructs explored in the study were those designed to measure cigarette usage, adolescent and youth susceptibility to smoking (two sub-samples (1) never and former smokers and (2) never smokers only), receptivity to anti-tobacco messages and promotions and receptivity to pro-tobacco messages and promotions. These items were used in the 1999, 2000 and 2001 Legacy Media Tracking Surveys. The four sets of items were examined in terms of internal structure through their dimensional characteristics and reliability and were evaluated across time and among age, gender and ethnicity sub-groups. Dimensionality was assessed through principal axis factoring, using the scree plot as well as the Kaiser-Meyer-Olkin measure of sampling adequacy and Bartlett's test of sphericity. Reliability was determined through Cronbach's alpha measure of internal consistency. Data analyses indicated that two of the constructs, cigarette usage and susceptibility were unidimensional, while the measures of receptivity to anti-tobacco promotions and messages and receptivity to pro-tobacco promotions were multidimensional. These results were similar across all sample sub-groups (i.e. when items were unidimensional, they tended to be unidimensional across time, age, gender and ethnicity). When they were multidimensional, overall they tended to be multidimensional across the sample sub-groups. The reliability analyses were also similar, though the reliability measures for susceptibility to smoking tended to be stronger among older sample respondents. While the dimensionality and internal consistency results were supportive of items used to measure cigarette usage and receptivity to anti-tobacco promotions, additional items need to be added measuring susceptibility to smoking. Similarly, additional items are needed to develop the constructs of receptivity to pro-tobacco messages and promotions. The criterion measures for anti and pro tobacco attitudes were not consistently supported. Cigarette usage and susceptibility to smoking were both negatively correlated with anti-tobacco attitudes. However, the measures were generally weak. Practical implications of the results include a confirmation of the need to test psychometric features of measures as part of the evaluation of health campaigns. Implications for future research include developing additional reliable and valid evaluation measures of susceptibility to smoking and identifying survey items that can be sued to demonstrate the criterion validity of anti-tobacco campaign evaluation measures.