The Attentional Mechanism of Message Sensation Value: Interaction between Message Sensation Value and Argument Quality on Message Effectiveness

A Multilevel Analysis of Antimarijuana Public Service Announcement Effectiveness

“…In their study, Kang et al (2006) also examined a third dimension, marijuana risk, which included age, SS, and measures of personal involvement with marijuana. Assuming that high-risk adolescents would be more involved in the issue and thus motivated to attend to and process the ad, Kang and colleagues hypothesized that the interaction between MSV and AS ''is more likely to occur among high risk than low risk adolescents'' (p. 358).…”

“…Unanticipated results in recent studies have often been interpreted to confirm and/ or refute one theory over another (e.g., supporting ELM over AMIE in Kang et al, 2006). However, we take issue with these interpretations for two reasons.…”

“…However, several recent studies have found interactions where the theoretically predicted effects of a given message variable are conditional upon other variables (Kang, Cappella, & Fishbein, 2006;Lang & Yegiyan, 2008;Langleben et al, 2009;. Specifically, two important message features in many persuasion theories are (1) the strength of arguments within the message, and (2) the degree to which a message elicits physiological arousal.…”

“…Each study deals with variable interactions, but often with contradictory theory, findings, and/or conclusions. Kang et al (2006) examined the interaction between message AS, MSV, and participant drug risk on measures of message effectiveness, finding that the typical high-MSV, high-AS ads used to target at-risk adolescents were the least effective message type in the highrisk adolescent audience. Stephenson and Palmgreen (2001) tested the effects of SS and marijuana personal involvement (PI), as separate risk-related variables, as well as perceived MSV on cognitive-processing amount and valence.…”

“…Very often these ''message features'' are measured by polling audience members as to what they perceive. The problem with this method is that either (1) researchers ask the same sample questions about their message-feature perceptions (using this as an independent variable) as they also ask about outcomes (e.g., Stephenson & Palmgreen, 2001, with PMSV), or (2) researchers ask a different sample to rate the message feature, but then lose individual-level information by aggregating the scores, assigning a single score to represent the feature regardless of variance in the perception of that feature (e.g., Kang et al, 2006, with AS). The first approach suffers from problems of nonindependence of measurements; the second involves loss of important individuallevel variation.…”