ABSTRACT
In an experiment (N = 206) using skin cancer prevention messages and a 2 (mortality: salient, control) × 2 (freedom-limiting language: freedom-limiting, autonomy-supportive) independent-group design, we tested the terror management health model and integrated its predictions with the theory of psychological reactance. We used a sample of young adults because they are most at risk for excessive tanning. Consistent with the study predictions about proximal defenses, mortality salience significantly increased intentions to wear sunscreen all year around, relative to the control condition. A significant interaction between freedom-limiting language and mortality salience on behavioral intention to purchase high-SPF lotion revealed that, when a freedom-limiting message was paired with mortality salience, intentions to purchase high-SPF lotion were significantly greater as compared to autonomy-supportive language, indicating that mortality salience mitigated the maladaptive effects of reactance. These results add to a growing body of research identifying boundary conditions for reactance effects and, further, point to the utility of directive (albeit freedom-limiting language) in health-prevention messages attempting to communicate deadly health risks.
Disclosure statement
The authors do not have any conflict of interest influencing these results.
Notes
1. The University of Oklahoma IRB (#6087) approved this data collection.
2. Magnitude scale instructions were as follows: “Suppose we would like to know how much knowledge you have about your major. To answer this question, use a number from zero to infinity. Zero means you have no knowledge at all about your major, and higher numbers represent greater levels of knowledge. If you feel you have moderate knowledge about your major, rate your knowledge as 100. If your knowledge is twice as much as a moderate level, rate your knowledge as 200; if your knowledge about your major is half the moderate level, rate your knowledge as 50. You can use any number from zero on up, such as 18, 193, or 347.”
3. Magnitude scales (see Lodge, Citation1981, for a discussion) are ratio-type scales that range from zero to infinity and are unbounded at the top, which helps avoid ceiling effects. These scales have been used successfully across various disciplines (including communication) because, unlike Likert-type scales and other ordinal measurements, magnitude scales are better at producing interval-level data (Lodge, Citation1981).
4. Transformations affect the data in predictable ways: In a transformation equation, Y*= (Y+k)λ, where Y is the original variable, Y* is the transformed variable, and k is a constant, using λ < 1 is likely to result in a more symmetric distribution for positively skewed data, and using λ > 1 is likely to result in a more symmetric distribution for negatively skewed data (Fink, Citation2009). Note that although transformations affect the scale on which a variable is measured, they do not change “the relative difference between people for a given variable” (Field, Citation2013, p. 201).
5. For example, the initial skewness of one of the anger items (annoyed) was 14.35 (SE = 0.17), indicating a significant departure from normality. Transformed item’ skewness became 0.44, showing a substantial improvement. Different transformation formulas were used for different variables to accommodate the specific type of skewness present in a given variable. In this study: threat perception = (winsorized original item + 1)0.5; anger = (winsorized original item + .01)0.01; intentions to purchase high-SPF lotion = (winsorized original item + 1)0.3; intentions to purchase low-SPF lotion = (winsorized original item + 1)0.3; intention to wear sunscreen all year round = (winsorized original item + 1)0.2
6. Using principal axis factoring (PAF) to create indexes does not change study results. All bivariate correlations between the indexes formed using PCA and PAF were > .97.
7. These descriptives apply to all indexes except negative-relevant thoughts (a single-item measure reported in the original units because it did not require transformation; original skewness = 0.08; SEskewness = 0.17) and the intention to wear sunscreen all year round (a single-item measure reported in the transformed units; original skewness = 3.94, transformed skewness = 0.57, SEskewness = 0.17).
8. Given that mortality-salience induction checks have been shown to inadvertently activate death awareness (Hayes & Schimel, Citation2018; Hayes et al., Citation2010), we chose not to use a lexical-decision/word-fragment/puzzle-completion task as a manipulation check for the mortality-salience induction. To assess the efficacy of the mortality-salience induction (see Massey & Miller, Citation2017, for a similar approach), trained coders scored mortality salience and control essays about the emotions aroused by thoughts of own death in terms of the level of existential anxiety. The scale ranged from 0, indicating no anxiety (e.g., “I’m not threatened by my own death … .;” “[Death] will lead … to a better understanding of Christ’s sacrifice … I am not sad at the thought, I am honored to give my life to the one who gave me his.”), to 3, indicating extreme anxiety (e.g., “Thinking about my own death makes me sick to my stomach … ;” The thought of never existing and being completely gone is terrifying to me.”). The results revealed that the essays in the mortality-salience condition showed significantly (p < .001) greater levels of death anxiety, relative to the control condition, which indicates that our mortality-salience manipulation was successful.
9. Reporting one-tailed tests is warranted for these results given that directional predictions were proposed.