The enduring influence of death on health: insights from the terror management health model

Abstract

The terror management health model suggests targeting sources of self-esteem or identity, in conjunction with mortality salience, offers a pathway for health behavior promotion. To date, however, experimental evidence has been limited to single time point studies. Two studies assessed whether similar processes impact behavior over time. In Study 1, mortality salience was paired with exercise (i.e., riding a recumbent bike); two weeks later, individuals primed with mortality reported more exercise than those not primed with death, and this increased fitness-contingent self-esteem and exercise intentions. In Study 2, when smokers visualized a prototypical unhealthy (vs. typical) smoker after mortality salience, they reported more attempts to quit smoking (over three weeks) than participants not primed with mortality. This facilitated continued quit attempts and decreased smoker identification three weeks later. Implications are discussed for a longitudinal process model in which mortality salience catalyzes a reinforcing relationship between behavior and esteem/identity, potentially sustaining health behavior change over time.

Keywords:

• Mortality salience
• self-esteem
• identity
• smoking
• exercise
• health behavior

Acknowledgements

We thank Melissa Spina, Jacquelyn Watson, and Burcin Cihan for their help with data collection.

Notes

1. Given the discrepancy in sample size between male and female participants, we controlled for participant gender in all analyses. Inclusion of this covariate did not influence any of the reported results. We also tested for, and did not find, any interactions of mortality salience with participant gender. One main effect of gender was found: At the two-week follow-up, male participants (N = 20) reported greater intentions to exercise in the future (M = 483.25, SD = 458.06), compared to female participants (N = 125, M = 310.88, SD = 207.86; p = .03). No other gender differences were observed.

2. The first wave of data collection occurred during the fall semester; after analyzing this data and noting the unexpectedly high attrition rate (discussed below), the decision was made to continue collecting data through the spring semester. This was done to ensure that the study was not underpowered and to provide more stable estimates, but critically, the results and statistical significance of effects were unchanged as a result of the additional data collected.

3. Lambert et al. (2014) demonstrated that mortality salience increases fear-specific affect relative to a neutral condition. However, the issue of present interest is whether mortality salience produces affective differences compared to another aversive (but not death-related) topic, and more critically, whether affect has an influence on later processes.

4. Full descriptive statistics for both studies are available in the supplementary material.

5. All results are consistent with and without the imputed data included. Full statistics for these results on the raw (non-imputed) data are available in the supplementary material.

6. We also imputed missing values for the baseline fitness-contingent self-esteem measure for the full sample (N = 282) and found consistent results.

7. A description of other measures in Study 2, included as part of a separate study or intended to capture different facets of smoking attitudes and perceptions, is available in the supplementary material.

8. A second subliminal mortality manipulation was administered at Time 2; due to a methodological error, the data is not included in this analysis. There was no evidence that the manipulation influenced any of the reported results (see supplementary material for a detailed discussion).

9. The smoking cessation measure also asked participants to report how recent their last quit/reduction attempt was and the length of these attempt. However, some response choices corresponded to a time prior to taking part in the study (e.g., “More than one month ago”). Because the items assessing number of quit attempts unambiguously reflected the duration of time between laboratory sessions, and most clearly indicated behavior, we focused on this outcome as our primary index of smoking cessation efforts.

10. The results are consistent without the missing data included with one exception: in the full model, the interaction effect of the mortality and prototype primes on Time 2 attempts to quit is no longer statistically significant; this may be due, in part, to reduced power given the smaller sample size. All conditional indirect effects are statistically significant, and thus, the conclusions drawn from the analyses remain the same.