Mechanisms of Child Abuse Public Service Announcement Effectiveness: Roles of Emotional Response and Perceived Effectiveness

Abstract

This study tests the processes through which child abuse public service announcements (PSAs) are effective. The proposed model builds upon the persuasion mediation model of Dillard and Peck (2000), which integrates emotional response and perceived effectiveness as antecedents of issue attitudes and behavioral intention. The model tested the mediating role of perceived effectiveness in the persuasion process. Multigroup structural equation modeling was performed for three different types of child abuse prevention PSAs shown on YouTube to 486 college students. The model was well fitted across all three child abuse PSAs. Emotional response seems to exert the largest influence on behavioral intention directly and indirectly through perceived effectiveness and issue attitudes. In addition, perceived effectiveness has both a direct and an indirect impact on behavioral intention.

Notes

¹Participants exposed to PSA1 first had lower scores than those exposed to the other two PSAs first for the following two variables: perceived effectiveness [F(2, 479) = 12.85, p < .001, M _PSA1first = 1.58, M _{PSA2 first} = 2.23, M _{PSA3 first} = 2.20] and behavioral intention [F(2, 479) = 9.77, p < .001, M _{PSA1 first} = 3.85, M _{PSA2 first} = 4.54, M _{PSA3 first} = 4.49]. Participants exposed to PSA2 first had lower scores than those exposed to the other two PSAs first for the following three variables: emotional response [F(2, 465) = 8.31, p < .001, M _{PSA1 first} = 5.15, M _{PSA2 first} = 4.52, M _{PSA3 first} = 5.02]; perceived effectiveness [F(2, 479) = 5.12, p < .05, M _{PSA1 first} = 1.96, M _{PSA2 first} = 1.52, M _{PSA3 first} = 1.91]; and behavioral intention [F(2, 481) = 3.24, p < .05, M _{PSA1 first} = 4.17, M _{PSA2 first} = 4.03, M _{PSA3 first} = 4.41]. Finally, participants exposed to PSA3 first had lower scores than those exposed to the other PSAs first for perceived effectiveness [F(2,479) = 3.22, p < .05, M _{PSA1 first} = 1.66, M _{PSA2 first} = 1.50, M _{PSA3 first} = 1.30].

²Due to space limitations, the full covariance of correlation matrix is not provided here but can be obtained upon request.

³The MGSEM is based on the repeated, within-subject measures and therefore has N = 486 for all the three groups: participants viewed and responded to all three PSAs. To cross-validate the proposed model, we also performed the MGSEM using within-subjects measures: only the first PSA that participants in different classes viewed was tested in the model: n = 128 for PSA1, n = 109 for PSA2, and n = 214 for PSA3. The MGSEM indicates an equally good fit, with [χ²(729) = 1345.49, p < .001, RMSEA = .07, NNFI = .94, SRMR = .057, CFI = .95]. Thus, the proposed model seems valid using either within-subject or between-subjects design.

⁴Although not hypothesized, additional mediation tests were conducted for issue attitudes as a mediator for emotional response and for perceived effectiveness, both of which lead to behavioral intention. According to the same formula and procedure as for H4, the z-score product for the mediation path for emotional response-behavioral intention was calculated as follows: PSA1, z score = 8.33, p < .001; PSA2, z score = 7.51, p < .001; PSA3, z score = 5.35, p < .001. Likewise, the z-score product for the mediation path for perceived effectiveness–behavioral intention was calculated as follows: PSA1, z score = 12.5, p < .001; PSA2, z score = 8.94, p < .001; PSA3, z score = 9.63, p < .001.