Integrating Self-Affirmation with Health Risk Messages: Effects on Message Evaluation and Response

ABSTRACT

This study tested a new method of using self-affirmation theory to increase adaptive responses to health risk messages. Participants’ self-concepts were affirmed via text integrated with risk messages (public service announcements/PSAs) rather than by methods used in most previous studies that have limited applicability for campaigns employing mediated messages. Participants were exposed to health risk PSAs either including or not including affirming text to examine relative effects on message evaluation and adaptive responses. The moderating effect of personal relevance of the behavior on responses was also examined. Participants exposed to PSAs with integrated, affirming text (affirmed participants) reported more positive attitudes toward the PSA, greater self-efficacy, and greater intention to reduce risky behavior than participants exposed to messages without affirming text (non-affirmed participants). Compared to non-affirmed participants, affirmed participants also reported weaker perceptions of risk severity and equivalent perceptions of susceptibility. The personal relevance of the behavior did not moderate any effects of self-affirmation. Implications for health campaigns that employ mass-mediated messages are discussed.

Notes

^1. It should be noted that initial tests of self-affirmation intended to provide an alternate explanation for why cognitive dissonance occurs and how it could be reduced: the proposition was that dissonance is caused by a threat to the self that stems from knowledge of inconsistencies, rather than from a basic need for psychological consistency among important beliefs (Sherman & Cohen, 2006; Steele & Liu, 1983; Stone & Cooper, 2001).

^2. We note that self-esteem and positive affect do not consistently mediate the influence of self-affirmation on relevant outcomes. Sherman and Cohen (2006) note, however, that such inconsistencies may reflect shortcomings of research designs and statistical processes used to test such effects (c.f., Spencer, Zanna, & Fong, 2005). We therefore suggest the potential influence of these proposed mediating variables on efficacy perceptions in future research.

^3. Sample characteristics and mean value ratings are available from the lead author.

^4. Group identity (N = 162): M = 9.64, SD = 1.60, possible range 0-10.

^5. Sample messages are available from the corresponding author.

^6. Moderating effect for flip-flops: 1) attitude toward the PSA: b = .11, 95% CI [–.10, .32], t = 1.04, p = .302; 2) perceived susceptibility: b = –.05, 95% CI [–.21, .17], t = –.47, p = .639; 3) perceived severity: b = –.20, 95% CI [–.42, .02], t = −1.80, p = .074; 4) self-efficacy: b = .01, 95% CI [–.18, .16], t = -.10, p = .922; and 5) behavioral intention: b = –.07, 95% CI [–.28, .13], t = –.70, p = .488.Moderating effect for drinking bottled water: 1) attitude toward the PSA: b = .07, 95% CI [–.12, .25], t = .71, p = .480; 2) perceived susceptibility: b = .11, 95% CI [–.07, .30], t = 1.23, p = .220; 3) perceived severity: b = .13, 95% CI [–.07, .33], t = 1.30, p = .197; 4) self-efficacy: b = .08, 95% CI [–.08, .24], t = .97, p = .333; and 5) behavioral intention: b = –.05, 95% CI [–.23, .13], t = –.58, p = .563.Moderating effect for drinking highly caffeinated beverages: 1) attitude toward the PSA: b = .07, 95% CI [–.13, .26], t = .65, p = .514; 2) perceived susceptibility: b = .02, 95% CI [–.20, .24], t = .17, p = .865; 3) perceived severity: b = .13, 95% CI [–.05, .30], t = 1.42, p = .158; 4) self-efficacy: b = .12, 95% CI [–.07, .31], t = 1.23, p = .222; and 5) behavioral intention: b = .17, 95% CI [–.05, .38], t = 1.54, p = .127.