The Role of Narrative and Other-Referencing in Attenuating Psychological Reactance to Diabetes Self-Care Messages

ABSTRACT

This study charts pathways through message resistance to enhance the persuasiveness of diabetes self-care messages. A 2 (narrative) × 2 (other-referencing) × 2 (message) × 4 (order) experiment with adult diabetics (N = 58) tested whether packaging overt recommendations as a story rather than an informational argument (i.e., narrative structure) and highlighting the impact of health decisions on family and friends rather than the individual (i.e., other-referencing) can effectively attenuate psychological reactance to messages encouraging healthy diet and physical activity. Narrative and other-referencing each led to lower perceived threat to choice, less state anger and counterarguing, less negative cognitive responses, more positive attitudes toward the ad and the behaviors promoted, and greater intended compliance with message recommendations. Findings illustrate two strategies that communicators may employ in order to benefit from clear, direct health messages while avoiding the reactance they may provoke. Moreover, findings inform message design for diabetes self-care education.

Notes

¹ Four separate presentation orders were manipulated randomly between subjects by the MediaLab experimental software presentation program. As explained in the method section, order was controlled statistically in all tests.

² All messages were black-and-white, were similar in length (ranging from 194 to 212 words), were the same shape and orientation, and featured a single 1.5 × 1.5-inch still photographic image to reinforce the manipulation (e.g., an other-referencing message would include a photo with a grandparent/grandchild pair, whereas a self-referencing message would include a photo of an individual). All recommendation content was adopted from existing diabetes patient education materials, and messages incorporated both positive and negative emotional content rather than emphasizing one emotional tone. The Appendix includes two sample stimulus messages.

³ Much of the current research on reactance uses a thought-listing task to measure negative cognitions, where participants write freely in response to a message; positive, negative, and neutral statements are then coded and summed either by the participants themselves (e.g., Quick & Considine, 2008; Quick & Stephenson, 2007) or by trained coders (e.g., Dillard & Shen, 2005; Quick & Stephenson, 2008) to provide a quantification of negative cognitions. This study used a self-report index (similar to Dillard et al., 1996; Miller et al., 2007; and Silvia, 2006) to capture the cognitive dimension of reactance, in an attempt to circumvent potential sources of bias, such as inflated negativity driven by heightened anger from ruminating on the reactance-inducing stimulus (Gerin, Davidson, Christenfeld, Goyal, & Schwartz, 2006; Rusting & Nolen-Hoeksema, 1998) and response bias driven by participants’ awareness that their thoughts are open to interpretation by the researchers (Dougherty & Hunter, 2003; Paulhaus, 1984), as well as logistic difficulties with incorporating thought-listing tasks into a repeated-measures design.

⁴ Spearman’s rho correlations (used instead of Pearson’s r due to some significant skewness and kurtosis among the criterion variables) indicated that there was significant and substantial collinearity among the dependent variables; therefore, univariate rather than multivariate ANOVAs (MANOVAs) were used. Tabachnick and Fidell (2007) note that MANOVA is inappropriate when the majority of DVs are correlated at .65 or higher, as multicollinearity among dependent measures in MANOVA inflates familywise Type I error. Table 3 shows the zero-order correlation matrix for the dependent and mediating variables.

Additional information

Funding

This research was supported by the Dissertation Research Fund, Maxine Wilson Gregory Chair in Journalism Research, Missouri School of Journalism.