Abstract
Accumulating research shows that gratitude exercises are effective for improving well-being, but the pattern of results also suggests that moderators are likely at play. Researchers have begun investigating moderators of gratitude exercise impact, but it remains unclear which are most important. To help clarify this issue, the present study investigated trait gratitude, baseline depressive symptoms, and inclusion of a rationale as moderators of the efficacy of a two-week gratitude list exercise compared to a daily events list control condition. Results revealed that the gratitude exercise decreased depressive symptoms among participants high in such symptoms at baseline and eliminated the negative effect of low baseline trait gratitude on later happiness, positive emotions, and depressive symptoms. Implications, limitations, and future directions are discussed.
Acknowledgments
The authors are grateful to Drs Jennifer Cheavens and Julian Thayer for their helpful suggestions on this project. We would also like to thank Matthew Free, Joseph Saxon, Lauren Phipps, Mark Wells, Rosalie Volovetz, and Katie Babcock for their assistance with data collection and management.
Supplemental data
Supplemental data for this article can be accessed here: http://dx.doi.org/10.1080/17439760.2014.927905.
Notes
1. This study also included exploratory analyses intended to identify mediators of the effect of the gratitude exercise on well-being. However, because none of the candidate mediators was significant, that portion of the study is described in an online supplement to conserve journal space. We hope that researchers will consider our results when designing studies investigating mediators of gratitude exercises.
2. Further, among participants in the gratitude conditions, number of days missed was not significantly correlated with any DV (p ≥ 0.43) suggesting that, on average, missing days did not affect outcomes.
3. Non-significant G + NR versus G + R contrasts suggest that the two versions of the gratitude exercise did not differ in their effect on the DVs. However, such results do not establish that the two gratitude conditions each differed from the control condition to a similar extent. Thus, follow-up analyses comparing the G + NR and G + R conditions to the control condition separately were conducted. These results are summarized in footnotes and detailed results are reported in the online supplemental materials.
4. As reported in the online supplemental materials, follow-up tests were conducted to examine the results separately for each gratitude condition (i.e. G + NR and G + R). In each case, the pattern of results for ∆CESD was similar to those reported here for the CG condition, although the effect was larger in the G + NR vs. G + R condition.
5. As reported in the online supplemental materials, follow-up tests were conducted to examine the results separately for each gratitude condition. Although in the expected direction, the G + NR vs. control contrast was not significant at low levels of T1GQ-6 whereas the G + R vs. control contrast was. However, as in the CG condition, both the G + R and G + NR conditions resulted in a non-significant correlation between GQ-6 and ∆SHS, whereas T1GQ-6 was significantly positively correlated with ∆SHS among controls.
6. As reported in the online supplemental materials, follow-up tests were conducted to examine the results separately for each gratitude condition. Both the G + NR vs. control and the G + R vs. control contrasts approached significance at low T1GQ-6 and these effects were in the expected direction and similar in magnitude. Also, as in the CG condition, both the G + R and the G + NR conditions resulted in a non-significant correlation between T1GQ-6 and ∆MdailyADES-Pos. whereas T1GQ-6 was significantly positively correlated with ∆MdailyADES-Pos. among controls.
7. As reported in the online supplemental materials, follow-up tests were conducted to examine the results separately for each gratitude condition. The G + R vs. control contrast was not significant regardless of level of T1 CESD, whereas the G + NR vs. control contrast was significant at both high and low levels of T1 CESD (same pattern as reported above for the CG).
8. The effect of condition on change in depressive symptoms was also significant at low levels of baseline symptoms, but the direction of this effect was unexpected. Specifically, when baseline symptoms were low (i.e. T1 CESD ≤ −1.34 SDs), the gratitude condition was associated with an increase in depressive symptoms compared to the control condition. However, given that there was no a priori reason to expect this effect and that there were only three participants in the present sample with T1 CESD scores in this range, we will withhold speculation about possible explanations until this effect is replicated.