Antismoking Initiatives: Effects of Analysis Versus Production Media Literacy Interventions on Smoking-Related Attitude, Norm, and Behavioral Intention

Abstract

This study developed inoculation-driven antismoking interventions aimed at changing attitudes, norms and intentions about smoking to influence smoking behavior in adolescents. This study explored the efficacy of 2 intervention approaches designed to help adolescents to refrain from smoking initiation. Participants were junior high students (6th, 7th, and 8th grade) from schools in the Northeast. Two kinds of experimental workshops and a control group were designed as stimulus material in a repeated measure nonequivalent group experimental design. The 2 intervention workshops developed included: analysis + analysis (where participants discussed and analyzed cigarette and antismoking ads) and analysis + production (where participants discussed, analyzed, and then created their own antismoking ads). The analysis + production workshop was generally more successful than the analysis + analysis workshop and control group in changing participants' behavioral intention to smoke and attitude toward smoking but not subjective norms over time. Implications and directions for future research are discussed.

ACKNOWLEDGMENTS

The authors would like to thank Itzhak Yanovitzky and Michael T. Stephenson for their contributions to this project.

Notes

¹There were no differences between schools on measures of interest for this study, so they were combined for this study. One class of students (n = 25) participated in a third workshop type (production + production). However, analyses are not reported for this group because of sample size and nonsignificant results for behavioral intention, attitude toward smoking, and subjective norm across the four times.

²All measures were created by the authors and pretested with subsamples of 110 undergraduate students enrolled in communication classes.

³Descriptive statistics run to explore normality of the attitude toward smoking scale showed a skewed distribution for attitude toward smoking scale (skewness = 1.11, kurtosis = .51). This scale was transformed using frequency distribution so that the intuitive interpretation of the attitude toward smoking scale (with a higher score indicating a more positive attitude toward smoking) was maintained while the scale showed a normal distribution. Descriptive statistics run to explore normality of the normative beliefs of parents, friends, and best friend scales showed a normal distribution for normative beliefs of friends scale (skewness = .62, kurtosis = .09), but skewed distributions for normative beliefs of parents scale (skewness = 2.39, kurtosis = 6.80) and normative beliefs of best friend scale (skewness = 1.50, kurtosis = 1.94). Applying transformation to normative beliefs of parents and best friend scales reduced skewness. The behavioral intention scale was normally distributed and not transformed.

⁴Although belief strength was not used in the creation of the measure of smoking beliefs, motivation to comply was used in the creation of the measure of subjective norm. Belief strength does not significantly improve the prediction of attitude when a standardized set of beliefs is used (Cronen & Conville, 1975; Delia, Crockett, Press, & O'Keefe, 1975). As a construct, motivation to comply does significantly improve the prediction of subjective norm when the items focus on the specific behavior of interest (O'Keefe, 2002).

⁵Details of the workshops are available from the authors upon request. All sessions were conducted by the same trained researcher (with a classroom teacher present but not participating) using a pretested script for each condition.

⁶Steps involved in between-subject analyses: Step 1: To show that groups are equal on outcomes at baseline; Step 2: To show that groups are different on outcomes postintervention; Step 3: To show that outcomes should change the same following the Analysis I workshop; Step 4: To show that the Production and Analysis II workshops generated different changes on outcomes. Steps involved in within-subject analyses: Step 1: To show that the control group is equal on outcomes at baseline and postintervention; Step 2: To show the cumulative effects of Analysis I and Analysis II sessions for A + A workshop; Step 3: To show the cumulative effects of Analysis I and Production sessions for A + P workshop; Step 4: To show the difference in change in outcomes between Times 3 and 2 for A + A and A + P workshops.

⁷For sample comparability, three independent variables were crossed in a series of three-way ANOVAs with participants nested in combinations of workshop types (A + A, A + P, and control group), grade (6th, 7th, and 8th), and school (A and B). There were no significant two- or three-way interaction effects for behavioral intention. The main effects for workshop type, F (2, 214) = .58, p = .56, η² = .01 grade, F (2, 214) = 1.50, p = .23, η² = .01; and school, F (1, 214) = .14, p = .71, η² = .00 were also not significant.

⁸Six Kruskal-Wallis tests were run with workshop type (A + A, A + P, and control group), grade (6th, 7th, and 8th), and school (1 and 2) as independent variables and attitude toward smoking and subjective norm as dependent variables. The results of the tests were not significant.

⁹In addition, to examine the difference between Times 2 and 3 attitude toward smoking scores for participants in the A + A and A + P workshops, a difference variable was created (attitude toward smoking Time 2 Time 3). A nonparametric equivalent of the independent sample t test (Mann Whitney) was conducted with the difference score as the dependent variable and workshop type as the independent variable. The results revealed that the difference score between A + A (Mean Rank = 84.28) and A + P workshop (Mean Rank = 86.61) was not significant (z = .42, p = .67; Step 4 not confirmed).