Gratitude and Loneliness: Enhancing Health and Well-Being in Older Adults

Abstract

We experimentally investigated gratitude’s impact on loneliness and health in older adults. Participants were assigned to a daily gratitude writing exercise (treatment group) or a control group. Self-reported loneliness and health (i.e., subjective well-being, subjective health, health symptoms) were measured daily over a 3-week period. In support of our hypotheses, within-person variability in gratitude predicted differences in loneliness and health. Furthermore, those in the treatment group showed stronger cumulative effects of gratitude on loneliness and health symptoms when aggregated across the 20-day study. Additionally, a series of conditional, multilevel indirect effect models found that loneliness acted as a mechanism for gratitude’s differential impact on subjective well-being and health symptoms across conditions. Taken together, this study provides initial evidence that a simple gratitude exercise can strengthen associations among daily gratitude and loneliness and, consequently, improve health, for older adults.

Keywords

gratitude loneliness health subjective well-being

Gratitude is a positive, socially oriented emotion that plays an impressive role in building and maintaining social relationships (Algoe, Haidt, & Gable, 2008; Bartlett, Condon, Cruz, Baumann, & DeSteno, 2012) and increasing intrapersonal well-being (Emmons & McCullough, 2003), including improving physical health (Hill, Alleman, & Roberts, 2013) and decreasing loneliness (Ni, Yang, Zihang, & Dong, 2015; O’Connell, O’Shea, & Gallagher, 2016). Further, gratitude can be cultivated. Research has shown that the simple “three good things” exercise increases feelings of gratitude with concomitant boosts in positive outcomes (Emmons & McCullough, 2003).

Loneliness is a distressing emotional experience linked to physiological and psychological decline (Hawkley & Cacioppo, 2010). Defined as a discrepancy in desired and actual social relationships, loneliness is a subjective experience distinct from objective levels of social interaction (Russell, Peplau, & Cutrona, 1980). Older adults are at greater risk of experiencing loneliness than other age groups (Luhmann & Hawkley, 2016). Given the various well-documented health correlates of loneliness (e.g., shorter life span, greater disease susceptibility, onset of dementia; Holwerda et al., 2012; Luo, Hawkley, Waite, & Cacioppo, 2012), it is imperative that research explore interventions for reducing loneliness, particularly among at-risk populations such as older adults.

To date, the literature is equivocal about the success of loneliness interventions, including those for older populations (Findlay, 2003). Borrowing from social–emotional research, the current study tests a simple intervention with a population of older individuals living in subsidized housing: cultivating gratitude for 20 days through the three good things exercise (Emmons & McCullough, 2003). Gratitude has been linked to reductions in loneliness (Caputo, 2015; Ni et al., 2015). Indeed, recent research has found that loneliness may be one of the mechanisms by which gratitude positively impacts health (O’Connell et al., 2016). The study reported here is the first research, to our knowledge, to experimentally test gratitude’s impact on loneliness, and consequently self-reported health, in older adults.

Literature Review

Gratitude is a positive emotional experience tied to the recognition that others (e.g., friends, family, God) have brought benefits into one’s life (McCullough, Kilpatrick, Emmons, & Larson, 2001). Researchers understand gratitude to be both a transient emotional state felt when one has been the beneficiary of another’s kind actions and an individual difference in how frequently, how intensely, and toward how many people individuals feel grateful (McCullough, Emmons, & Tsang, 2002; see Wood, Froh, & Geraghty, 2010). Of import for our work, gratitude has been shown to be malleable with the three good things writing exercise boosting individuals’ experiences of gratitude (Emmons & McCullough, 2003; Seligman, Steen, Park, & Peterson, 2005). Considerable research over the last 15 years has shown that gratitude offers benefits at both the relational and the individual level. Indeed, research has found that cultivating gratitude has been as effective in boosting well-being and decreasing depression as clinical therapeutic interventions (Seligman et al., 2005).

Gratitude functions to help build and maintain our relationships, one of the most important factors in how satisfied we feel with our lives (Diener & Seligman, 2002). When individuals feel grateful toward others, they are motivated to care for them, thus prompting reciprocal, prosocial behaviors that help bind individuals more closely to one another (Bartlett & DeSteno, 2006; Bartlett et al., 2012). In addition to altering behavior, feeling grateful boosts positive views of our relationships, increasing perceptions of the relationship quality (Algoe et al., 2008) connection and satisfaction (Algoe, Gable, & Maisel, 2010). Intrapersonally, gratitude has been shown to facilitate lower levels of loneliness as well as increase positive perceptions of one’s physical health and well-being. Those who self-report feeling more grateful are more positive and have higher life satisfaction (McCullough et al., 2002) and report having more restful sleep, better physical health (Hill et al., 2013; O’Connell et al., 2016), being less stressed, and lower in loneliness than less grateful individuals (Caputo, 2015; Ni et al., 2015; O’Connell et al., 2016). In a recent examination, researchers found that differences in self-reported loneliness explained the link between dispositional gratitude and self-reported health outcomes (O’Connell et al., 2016). Thus, suggesting that the relationship between gratitude and physical health may be, in part, due to differences in feelings of loneliness.

Loneliness and Health

Feeling socially isolated (i.e., lonely) is an aversive experience that motivates individuals to seek social connection and to avoid negative social interactions (Hawkley & Cacioppo, 2010). Like the experience of physical pain, loneliness signals a problem (i.e., lack of social connection) and motivates one to alleviate it (J. T. Cacioppo et al., 2006). Whereas loneliness is usually transient and assists in remediating social loss (J. T. Cacioppo et al., 2006), sustained feelings of loneliness (i.e., chronic loneliness) have been shown to trap individuals in a cycle of negative social cognitions and “paradoxically self-defeating behaviors” that impede positive social interactions (J. T. Cacioppo, Cacioppo, & Boomsma, 2014, p. 6).

Evolutionary models of loneliness hypothesize that intermittent loneliness functions to mend poor social connections and, consequently, heightens attentiveness to social threats and negative social cues (S. Cacioppo et al., 2016). While this heightened vigilance for social threat is argued to assist social navigation in the short run by encouraging individuals to carefully assess social situations and to avoid social rejection, it takes a heavy toll when loneliness becomes chronic. A sustained search for threats in one’s social landscape leads to nonoptimal changes in individual’s affect and cognition. For example, experience sampling found that those high in loneliness derived less joy from their daily positive activities and experienced more stress from their daily hassles than less lonely individuals (J. T. Cacioppo & Hawkley, 2003). Chronically viewing one’s social world as a place of potential threat and its concomitant sequelae are linked to serious physical and psychological decline. Research has found that loneliness is linked to depression (Luo et al., 2012), cardiovascular health risks (Caspi, Harrington, Moffitt, Milne, & Poulton, 2006), poor immune system functioning (Kiecolt-Glaser et al., 1984), dementia onset (Holwerda et al., 2012), and mortality (House, Landis, & Umberson, 1988). Thus, while intermittent loneliness is argued to serve an adaptive purpose overtime, motivating individuals to seek and maintain social connections, the research is unequivocal about the deleterious impact of chronic loneliness on psychological and physical health. Yet less research has explored how daily experiences of loneliness relate to self-reported health functioning.

Loneliness in Older Adults

Given the serious health risks that chronic loneliness presents, it is alarming that rates of loneliness are notably high among older adults living in the United States (e.g., Bekhet & Zauszniewski, 2012). Indeed, this population is particularly at risk of loneliness in comparison to other age groups (Luhmann & Hawkley, 2016). Decreases in physical mobility, retirement from work, and increases in the death of spouses and friends contribute to increased risk of loneliness as individuals move into older age (Findlay, 2003). A recent survey found that more than 19% of U.S. adults aged 65 and older reported feeling lonely for much of the previous week (Theeke, 2009); other research has documented that nearly 40% of older adults experience loneliness (Cohen-Mansfield & Parpura-Gill, 2007). Researchers have called loneliness in older adults a leading issue for industrialized nations (Findlay, 2003).

While it is clearly recognized that loneliness is a serious risk factor for the health of older adults, knowledge about how to reduce this risk is limited (Cattan, White, Bond, & Learmouth, 2005; Findlay, 2003). In part, research is limited because external conditions that contribute to loneliness in the elderly (e.g., death of loved ones, living alone) are difficult to modify (Findlay, 2003). Further, there is little experimental-based research to test the success of the loneliness interventions that are used (Cattan et al., 2005; Findlay, 2003; Masi, Chen, Hawkley, & Cacioppo, 2011). One review summarized the empirical support for loneliness interventions for older adults as “almost non-existent” (Findlay, 2003, p. 655). A recent meta-analytic review of existing loneliness interventions across age groups found that those targeting maladaptive social cognitions were the most effective for reducing loneliness (Masi et al., 2011). That is, compared to attempts to improve social skills, enhance social support, and increase social interactions, studies employing an experimental design found that addressing lonely individuals’ negative social cognitive patterns (e.g., employing psychological reframing) was the most effective type of loneliness intervention (Masi et al., 2011). There is reason, then, to expect that cognitive reframing in the form of cultivating gratitude (i.e., asking individuals to actively look for the ways in which others bring benefits into their lives) may offer an important and relatively simple tool for inhibiting loneliness and its negative outcomes.

Gratitude: Reductions in Loneliness Facilitating Boosts in Health

Gratitude rests on recognizing that others have acted out of care for oneself, thus gratitude may reduce feelings of social isolation (O’Connell et al., 2016). Indeed, gratitude has been linked to reductions in loneliness (Ni et al., 2015; O’Connell et al., 2016), and this reduction has been argued to act as a mechanism for gratitude’s positive impact on physical health and well-being (O’Connell et al., 2016). Further, experimental interventions have successfully boosted individuals’ feelings of gratitude with concomitant boosts in positivity, well-being (Emmons & McCullough, 2003), and perceived availability of close others (O’Connell, O’Shea, & Gallagher, 2017) compared to those in control conditions. Researchers expect that recognizing the good in one’s life brought about by benevolent others is key to these positive outcomes (O’Connell et al., 2017). Yet no research, to our knowledge, has examined gratitude’s link to loneliness, and consequently health, in an older population.

The current study addresses this gap in existing research by examining the impact of a gratitude intervention, the three good things exercise on loneliness, and, consequently, health among older adults. This exercise is designed to focus individuals’ attention on what is going right in their lives and the roles that others play in this positivity. As such, the exercise is designed to modify individuals’ daily social cognitions. The study presented here is the first, to our knowledge, to experimentally cultivate gratitude and observe its downstream consequences for loneliness and, consequently, subjective well-being and self-reported physical health in an older population.

Present Study

We examined differences in gratitude, loneliness, subjective well-being, and health over 20 days for older individuals assigned to a gratitude condition (i.e., “the three good things exercise”) or a control condition. We were interested in both daily fluctuations in gratitude as a predictor of loneliness and health outcomes and condition differences in the cultivation of gratitude’s impact on loneliness and health. Further, we examined the recently specified process in which gratitude is argued to boost health outcomes because it decreases feelings of loneliness (O’Connell et al., 2016). Thus, we tested the following four hypotheses. First, we explored whether participants in the gratitude condition showed greater increases in gratitude over the 20-day study, relative to participants in the control condition. Second, we examined whether daily gratitude predicted less loneliness and greater health (subjective-well-being, self-reported health, health symptoms) over the 20-day study. Third, we explored whether daily loneliness served as a mechanism through which daily gratitude predicted health. Fourth, we tested for conditional indirect effects of loneliness, based on experimental condition; that is, we explored whether the indirect effects of loneliness on gratitude-health associations were stronger for the gratitude treatment versus control group.

Method

Participants

A convenience sample of 42 older adults (80% female, average age 73, SE = 6.43) recruited from two independent living facilities in a midsized city in the Pacific Northwest participated in the current study. Eighty-five percent were Caucasian, 5% were Hispanic/Latino/Spanish, 8% identified as First Nation or Alaska Native, and 5% identified as Other/multiracial. Twenty-three percent of the participants were widowed, 51% separated or divorced, 15% never married, and 5% were married and/or cohabitating with a romantic partner.

Design and Procedure

This was a longitudinal, double-blind, randomly assigned group study. Researchers introduced the study at informational sessions in each of the facilities. The study was explained as an examination of daily events and their impact on health. Interested participants were invited to complete informed consent at that time. No residents were excluded from participating in the study. On the study start date, research assistants, blind to condition, delivered the paper packets to each resident’s apartment. Each resident received one packet, thus being assigned to either the treatment or the control group. Packets were distributed, so that approximately equal numbers of treatment (N = 23) and control assignments (N = 19) were made for each facility. Packets were assembled ahead of time, so that research assistants remained ignorant to condition on the day of delivery (see Consolidated Standards of Reporting trials checklist in Supplementary Material).

Participants then completed a 21-day survey study. The first survey day served as a baseline assessment and included participant demographics and trait measures of key study variables; as such, the daily survey portion of the study consisted of 20 consecutive survey days. Participants were asked to complete each daily survey in the evening, before going to bed. Participants were provided with 21 envelopes (Day 1 = baseline survey, Days 2–21 = daily surveys) and were asked to deliver each survey to a locked drop box in the main office of their independent living facility at the end of the day. Surveys and envelopes were dated, so that participants knew when to complete each survey. Research assistants retrieved surveys from drop boxes each afternoon and tracked participant compliance. Participants who missed three survey days in a row were provided with a reminder to complete their surveys each day. Participants were provided with US$50 to a local grocery store—US$10 at the beginning of the study and another US$40 at the end.

Participants assigned to the treatment group were asked to complete a daily gratitude exercise prior to the completion of their daily survey. The gratitude exercise consisted of listing three good things that happened that day and why they happened (Emmons & McCullough, 2003; Seligman et al., 2005). Participants in the control group simply completed the daily survey measures.

Of the 840 possible survey day observations (38 participants × 20 days), participants completed a total of 761 surveys (90% compliance). On average, participants completed 18.63 surveys (SD = 1.84). Consistent with other daily studies (e.g., Mohr et al., 2005), data from participants who provided at least 7 days of surveys were retained for the analyses reported in the current study, resulting in a final sample of N = 36 (608 survey day observations; 80% total observations).

Daily Measures

Gratitude

We used an adaptation of the Gratitude Questionnaire (GQ-6; McCullough et al., 2002) to measure participants’ daily experiences of gratitude. The GQ-6 consists of 6 items (1 = strongly disagree, 7 = strongly agree) that uniquely capture participants’ experience of gratitude. Statements include: “I am grateful to a wide variety of people” and “When I look at the world I don’t see much to be grateful for” (reverse-scored). We instructed participants to indicate how much each statement applied to them that day.

Loneliness

Participants completed the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988). Taken from the PANAS, daily loneliness was assessed with the single negative mood item in which participants indicated how lonely they felt in the moment (1 = not at all, 5 = extremely). Previous research has successfully validated single-item measures of loneliness against multi-item measures, such as the UCLA Loneliness Scale (Russell et al., 1980). Further, single-item assessments of moods offer several advantages, including increased face validity, brevity, and reduced participant fatigue, which were of particular concern with this sample.

Subjective well-being

We measured participants’ subjective well-being with the single item: “How did you feel about your life as a whole today?” (−3 = terrible, 3 = delighted; Emmons & McCullough, 2003).

Subjective health

We assessed subjective health via a single-item probing participants’ perception of their health each day (Kodzi, Gyimah, Emina, & Ezeh, 2011). Specifically, participants responded to the question: “Today would you say your health is…” (1 = poor, 5 = excellent).

Health symptoms

We used a 13-item checklist to assess participants’ daily physical symptoms (e.g., headaches, faintness/dizziness, shortness of breath). Items were summed to create the health symptoms measures (Emmons & McCullough, 2003).

Activities of daily living

Participants indicated whether they had difficulty (yes/no) performing each of six activities (e.g., bathing or dressing, eating, climbing stairs; Emmons & McCullough, 2003).¹ Activities to which participants indicated having a difficult time were summed to create a composite “difficulty” score each day; higher scores on this composite represented a greater number of daily activities which participants had a difficult time completing on a given day.

General positivity

A daily general positivity score was computed via the positive affect (PA) items on the PANAS (Watson et al., 1988); items were summed to create a composite PA score. PA (general positivity) was considered as a control variable in all key analyses.

Data Analytic Plan

As each participant in the current study provided multiple reports of daily gratitude, loneliness, and health outcomes, data conformed to a multilevel structure. As such, it was important to take into account nesting of observations within participants, across time. Multilevel modeling was appropriate for the current study, as it enables the modeling of within- and between-subject variability (Bolger & Laurenceau, 2013).

To test for group differences for change in gratitude over the 21-day study, we tested a linear growth model, wherein daily gratitude was modeled as a function of time, condition, and the Time × Condition interaction. In addition, to examine associations among daily gratitude, loneliness, and health, we tested a series of two-level random effect models, modeling daily loneliness and health as a function of daily gratitude. Following recommendations by Raudenbush and Bryk (2002), all Level-1 predictors were person centered and Level-2 predictors were grand mean centered. To account for the autocorrelation of observations closer together in time, we specified an autoregressive structure for residual correlations. All analyses were conducted via multilevel modeling (mixed models; SPSS Version 22.0), using restricted maximum likelihood estimation.

Lastly, we tested for simple indirect effects via a 1-1-1 multilevel mediation model, wherein daily gratitude predicted daily health via daily loneliness. This approach allowed for the partitioning of within- and between-subject indirect effects. That is, we explored whether within-person fluctuations in daily gratitude predicted fluctuations in daily loneliness and, subsequently, daily health (within-person indirect effects). Further, we examined whether individuals who experienced greater gratitude in general (aggregated across 20 days) also reported less loneliness and subsequently better health over the 20-day study (between-person indirect effects). Within-person indirect effects capture the day-to-day process through which gratitude relates to health, via loneliness; between-person indirect effects represent the more cumulative effects of gratitude on loneliness and health overtime. As a next step, we examined condition differences (gratitude group vs. control group) in these indirect effects by testing a multilevel moderated mediation model (conditional indirect effects). Multilevel conditional indirect effects were tested via the MLMed macro in SPSS (Rockwood & Hayes, 2017).

Results

Descriptive Statistics

Correlations between relevant demographic variables (i.e., age, gender, marital status) and gratitude, loneliness, and health were examined in preliminary exploratory analyses to determine any necessary covariates; correlations among potential demographic covariates and key study variables were not significant at the .05 level (see Supplementary Material for table).

Linear Change in Gratitude

We specified a linear growth model of gratitude that allowed each participant to have their own initial level and rate of change in gratitude over the 21-day study. We predicted that individuals in the treatment group should show a linear increase in gratitude over the course of the study. Fixed and random effect estimates from linear growth models are presented in Table 1. Experimental condition was coded as 0 for the control group and 1 for the treatment group. Time was recoded, so that 0 represented Day 1 of the daily survey (not including the baseline assessment on Day 1). As such, parameter estimates in Table 1 can be interpreted as follows: (1) the intercept is the level of gratitude at Day 1 for the control group, (2) condition is the difference in gratitude among treatment and control groups at Day 1, (3) time is the change in gratitude in the control group over the 20-day study, and (4) Condition × Time interaction is the difference in gratitude change between the treatment and control groups.

Table 1.

Parameter Estimates for Linear Growth Model of Daily Gratitude as a Function of Intervention Group.

Fixed Effects (Intercept, Slope)	Estimate (SE)	t(df)	p	95% CI
Fixed Effects (Intercept, Slope)	Estimate (SE)	t(df)	p	Lower	Upper
Intercept (level at Day 1)	5.40 (0.30)	17.99 (34)	<.001	4.79	6.01
Time^a	0.00 (0.01)	0.69 (32)	.498	−0.01	0.01
Treatment^b	0.33 (0.42)	0.78 (34)	.441	−0.53	1.20
Treatment × Time	−0.00 (0.01)	−0.53 (35)	.599	−0.02	0.01
Random effects([co]-variances)	Estimate (SE)	z	p	95% CI
Random effects([co]-variances)	Estimate (SE)	z	p	Lower	Upper
Level 2 (between-person)
Intercept	1.58 (0.39)	4.00	<.001	0.97	2.58
Time	0.00 (0.00)	1.06	.289	0.00	0.00
Intercept × Time	0.01 (0.01)	1.66	.096	−0.00	0.02
Level 1 (within-person)
Residual	0.14 (0.01)	13.83	<.001	0.12	0.16
Autocorrelation	0.25 (0.05)	4.86	<.001	0.15	0.35

Note. N = 36. CI = confidence interval.

^aTime is coded 0 = Day 1, 19 = Day 20. ^bTreatment = condition, coded 0 for the control group and 1 for the gratitude exercise group.

As reported in Table 1, there were no group differences in gratitude on Day 1 of the daily survey. Both gratitude and control groups reported levels of gratitude on Day 1 of approximately 5.40–5.73, on a 1–7 scale (control = 5.40; treatment 5.40 + 0.33 = 5.73). Interestingly, over the 20-day study, neither the treatment nor the control group showed a significant increase in gratitude, nor were there group differences in change in gratitude, among treatment and control groups. That is, experimental and control groups did not differ in their rates of change in gratitude over the course of the study. However, as reported in Table 1, there was significant between-person variability in intercepts, as revealed by random effect estimates ( ${\hat{τ}}_{00}$ = 1.57, z = 4.00, p = .000, 95% confidence interval [CI] = [0.968, 2.58]). Otherwise stated, on the first day of the study, there was considerable variability within treatment and control groups in levels of gratitude. Additionally, there was significant within-person variance in gratitude across the 20-day study ( ${\hat{σ}}^{2}$ = .14, p = .000, 95% CI [0.118, 0.157]; that is, there was considerable fluctuation or variability in gratitude within-person, across treatment and control groups over the 20-day study. Lastly, there was also significant autocorrelation in Level-1 residuals ( $\hat{ρ}$ = .25, p = .000, 95% CI [0.149, 0.353]), which supports our assumption that within-person observations closer together in time were more correlated than observations farther apart in time.

Multilevel Gratitude Associations

Although there were no apparent linear trends in gratitude as a function of condition, as reported above, we were interested in examining how within-person fluctuations in daily gratitude relate to daily loneliness and self-reported health (psychological and physical). Thus, we tested a series of multilevel models wherein daily gratitude predicted loneliness, subjective well-being, self-reported health, and health symptoms. We also tested for condition differences in these associations. As recommended by Bolger and Laurenceau (2013), we included a centered version of observation day (i.e., time) and recoded time into weeks (divided by 7) to help with model convergence.

As reported in Table 2, there were significant within-person associations among daily gratitude and subjective well-being, b = 0.38, t(8.71) = 2.98, p = .016, 95% CI [0.089, 0.661]; self-reported health, b = 0.30, t(14.04) = 3.25, p = .006, 95% CI [.103, .504]; loneliness, b = −0.37, t(12.71) = −2.47, p = .029, 95% CI [−0.701, −0.046]; and self-reported health symptoms, b = −0.44, t(8.82) = −2.56, p = .031, 95% CI [−0.838, −0.051]. On days with greater gratitude, individuals reported greater well-being, better subjective health, fewer health symptoms (e.g., poor appetite, shortness of breath, chest pain), and less loneliness.

Table 2.

Parameter Estimate for Multilevel Models of Loneliness, Well-Being, Subjective Health, and Health Symptoms as a Function of Daily Gratitude and Intervention Group.

Predictor	Loneliness		Well-Being		Subjective Health		Health Symptoms
Predictor	B (SE)	95% CI	B (SE)	95% CI	B (SE)	95% CI	B (SE)	95% CI
Intercept	1.81 (.19)***	[1.43, 2.19]	1.29 (.18)***	[0.93, 1.65]	2.72 (.20)***	[2.31, 3.13]	2.60 (.33)***	[1.93, 3.27]
Time^a	0.01 (.5)	[−0.08, 0.10]	−0.03 (.04)	[−0.12, 0.06]	0.05 (.02)*	[0.00, 0.09]	0.07 (.08)	[−0.08, 0.22]
Wgrat	−0.97 (.15)*	[−0.70, −0.05]	0.92 (.18)***	[0.53, 1.31]	0.30 (.09)*	[0.10, 0.50]	−0.44 (.18)*	[−0.84, −0.05]
Tx^b	0.05 (.26)	[−0.49, 0.59]	0.08 (.25)	[−0.43, 0.58]	0.22 (.28)	[−0.36, 0.80]	−0.03 (.46)	[−0.97, 0.92]
Wgrat × Tx	0.23 (.21)	[−0.21, 0.68]	−0.51 (.25)	[−1.04, 0.01]	−0.22 (.13)	[−0.49, 0.05]	0.10 (.24)	[−0.42, 0.62]
Btwgrat	−0.08 (.11)	[−0.33, 0.16]	0.51 (.11)	[0.28, 0.74]	0.16 (.13)	[−0.11, 0.43]	0.17 (.21)	[−0.27, 0.60]
Btwgrat × Tx	−0.92 (.21)***	[−1.36, −0.49]	0.31 (.20)	[−0.09, 0.72]	0.18 (.23)	[−0.30, 0.66]	−1.06 (.39)*	[−1.85, −0.27]

Note. N = 36, 20 days, 608 reported survey day observations; CI = confidence interval; Wgrat = within-person gratitude association; Btwgrat = between-person gratitude association; statistically significant parameters are presented in boldface.

^aTime, per 7 days. ^bTx = condition, coded 0 for the control group and 1 for the gratitude exercise group.

*p < .05. **p < .01. ***p < .001.

There was also a significant moderating effect of condition on between-person gratitude associations, for health symptoms, b = −1.06, t(29.87) = −2.74, p = .010, 95% CI [−1.85, −0.270], and loneliness, b = −0.92, t(32.61) = −4.35, p = .000, 95% CI [−1.358, −0.492]. Specifically, among those in the treatment group, gratitude had stronger cumulative effects on loneliness and health symptoms, when aggregated across the 20-day study (Figure 1).

Figure 1.

Interaction of gratitude treatment condition on Level-2 gratitude and loneliness.

In line with previous research assessing the health benefits of gratitude, we also tested the models above, controlling for general positivity (as assessed via the PANAS). Results reported above were the same, with the exception of the Leve1-1 gratitude–loneliness association, which was reduced to marginal significance, b = −0.27, t(14.25) = −1.83, p = .088, 95% CI [−0.592, 0.046], and the Level-1 gratitude–health symptoms association, b = −0.24, t(11.60) = −1.33, p = .208, 95% CI [−0.626, 0.152], which was no longer significant when general positivity was included in the model. However, Level-2 findings for condition differences in gratitude effects were not impacted by controlling for general positivity.

Conditional Indirect Effects of Daily Loneliness on Gratitude-Health Associations

Previous research and theory proposes that loneliness serves as one psychosocial mechanism through which gratitude impacts health (O’Connell et al., 2016). Further, as reported in Table 2, daily loneliness significantly predicted subjective well-being and self-reported health within-person and health symptoms when aggregated over the 20-day study, thus warranting the testing of loneliness as a mediator for gratitude-health associations. Results of a series of conditional, multilevel indirect effect models revealed indirect effects of loneliness on associations among gratitude and subjective well-being, and gratitude and health symptoms, at the between-person level (see Figure 2). Importantly, these indirect effects were significant for the treatment group but not the control group for well-being (treatment group: ab = .38, p = .007, 95% CI [0.130, 0.674]; control group: ab = .06, p = .430, 95% CI [−0.079, 0.224]) and health symptoms models (treatment group: ab = −.89, p = .005, 95% CI [−1.603, −0.348]; control group: ab = −.13, p = .428, 95% CI [−0.483, 0.171]). Thus, results support moderated mediation for the indirect effect of loneliness based on gratitude condition, at the between-person level. For both outcomes, moderated mediation was qualified by a significant condition by gratitude interaction on loneliness (see Table 3). That is, aggregated over the 20-day study, there was a strong negative association between gratitude and loneliness for individuals in the treatment group, b = −1.05, t(36) = −6.17, p < .001; however, a test of simple slopes revealed that this negative gratitude-loneliness association was not significant for the control group, b = −0.06, t(36) = −0.53, p = .602. Conditional indirect effects for subjective well-being and health symptom models hold when controlling for general positivity.

Figure 2.

Conditional indirect effects of loneliness on gratitude-health symptoms association.

Table 3.

Parameter Estimates for Conditional Indirect Effect Models Based on Condition.

Effect	Model 1: Well-Being (DV)				Model 2: Health Symptoms (DV)				Model 3: Subjective Health (DV)
	M: Loneliness		Y		M: Loneliness		Y		M: Loneliness		Y
	B (SE)	p	B (SE)	p	B (SE)	p	B (SE)	p	B (SE)	p	B (SE)	p
Within
Constant	2.32 (0.71)	.002	−0.38 (0.68)	.580	2.32 (0.71)	.003	−0.05 (1.28)	.972	7.17 (1.06)	<.001	2.45 (0.84)	.007
Grat	−0.43 (0.16)	.012	0.55 (0.01)	<.001	−0.43 (−0.16)	.013	−0.37 (0.13)	.011	−0.19 (0.15)	.212	0.19 (0.05)	<.001
Grat × Tx	0.24 (0.21)	.266	—	—	0.24 (0.21)	.266	—	—	0.24 (0.21)	.266	—	—
Lonely	—	—	−0.25 (0.08)	.006	—	—	−0.02 (0.09)	.823	—	—	−0.05 (0.03)	.175
Between
Grat	−0.09 (0.13)	.475	0.43 (0.10)	<.001	−0.19 (0.13)	.474	0.19 (0.18)	.319	−0.95 (0.18)	<.001	0.14 (0.12)	.252
Tx	4.86 (1.27)	<.001	—	—	4.86 (1.27)	<.001	—	—	4.86 (1.27)	<.001	—	—
Grat × Tx	−0.86 (0.22)	<.001	—	—	−0.86 (0.22)	<001	—	—	−0.86 (−0.22)	<.001	—	—
Lonely	—	—	−0.40 (0.12)	.003	—	—	0.87 (0.22)	<.001	—	—	−0.22 (0.15)	.153
Index of moderated mediation (moderator: condition)
Within	ab = −0.06 95% BC CI: [−0.19, 0.04]				ab = −0.00 95% BC CI: [−0.07, 0.05]				ab = −0.01 95% BC CI: [−0.05, 0.01]
Between	ab = 0.19 95% BC CI: [0.11, 0.64]				ab = −.74 95% BC CI: [−1.35, −0.27]				ab = 0.19 95% BC CI: [−0.07, 0.50]

Note. Grat = gratitude; Tx = condition; Lonely = loneliness; DV = dependent variable.

General Discussion and Future Directions

Gratitude is the positive emotion felt when one recognizes the benevolent others in one’s life (McCullough et al., 2001). In line with the positive, relationship-enhancing nature of gratitude, previous research has found that gratitude is negatively correlated with loneliness (Ni et al., 2015; O’Connell et al., 2016). That is, gratitude rests on recognizing that others have acted out of care for oneself, thus reducing feelings of being alone (O’Connell et al., 2016). As discussed, extensive research has shown that loneliness is linked to deleterious outcomes (e.g., Hawkley & Cacioppo, 2010). As such, one of the mechanisms argued for gratitude’s positive impact on health is its reduction in loneliness (O’Connell et al., 2016). With this in mind, the current, experimental study tested four key hypotheses. First, we examined whether participants who completed a brief daily gratitude writing exercise showed greater increases in gratitude over the 20-day study, relative to participants in a control condition. Second, we examined whether daily gratitude predicted less loneliness and greater health (i.e., subjective-well-being, self-reported health, health symptoms) over the 20-day study. Third, we explored whether daily loneliness served as a mechanism through which daily gratitude predicted health. Fourth, we tested for conditional indirect effects of loneliness, based on experimental condition; that is, we explored whether the indirect effects of loneliness on gratitude-health associations were stronger for the gratitude treatment versus control group.

Results supported our expectation that gratitude cultivation can decrease loneliness and, consequently, boost self-reported health in an older population. In line with previous research we found that on days with greater gratitude (relative to average levels of gratitude), individuals reported greater well-being, better health, fewer health symptoms (e.g., poor appetite, shortness of breath, chest pain), and less loneliness. Further, at the between-person level, associations among gratitude, loneliness, and health-symptoms were stronger for those in the gratitude treatment group. Similarly, moderated mediation revealed significant indirect effects of average gratitude on subjective well-being and health symptoms, via loneliness, for those in the gratitude treatment group.

Although there were no significant condition differences in average changes in gratitude over the 20-day study, there was significant within-person variability in gratitude, and gratitude subsequently predicted loneliness and health in ways that differed across conditions. That is, those in the gratitude treatment group showed a process by which changes in gratitude decreased loneliness which in turn positively impacted health (i.e., subjective well-being and health symptoms).

While our study provides the first evidence, to our knowledge, that a simple gratitude writing exercise can enhance the effects of gratitude on loneliness and, consequently, boost self-reported health in older individuals, there are limitations and questions that remain. One limitation is clearly our sample size. Finding community-dwelling older adults willing to participate in a 3-week study limited our ability to widely recruit. Yet, as indicated in our literature review, this is the first experimental study that we know of to examine gratitude’s impact on older adults’ health. Thus, it makes an initial but important contribution to the literature.

We found important differences in gratitude’s impact on loneliness and health across the 3-week period. However, we did not find group differences in change in gratitude over the 20-day study. This may have been due to subtle shifts in average gratitude over time that did not reach significance in our sample. Due to the nature of the research (e.g., recruitment of community-dwelling older individuals for a 3-week period), our sample was small. Further, some treatment participants did not follow the gratitude writing exercise in its entirety, for example, writing about fewer than three things that went well or failing to write about why the good things happened. We expect that this diluted the impact the gratitude writing exercise could have. We recommend that future research collect a larger sample and design the gratitude writing exercise to facilitate compliance (e.g., fill in the blank for each response needed). Further, it is possible that changes in gratitude over our study were nonlinear, whereas our hypothesis-driven models tested for linear change in gratitude based on treatment group. Future work should consider unique patterns of change in gratitude among older adults, by modeling the effects of time on gratitude (and condition differences in the time-gratitude association) as nonlinear.

Recent research has found a modest relationship between the gratitude measure used in the current study (e.g., GQ-6) and social desirability. Thus, suggesting that gratitude scores may be inflated because participants recognize gratitude as a socially valued construct (Caputo, 2017). This is a concern for research employing such self-report measures. The current study, however, employed an experimental design. As such, both of our conditions were subject to social desirability pressures. Our between-group findings, then, suggest that the cultivation of gratitude had an impact, separate from social desirability, on the downstream consequences of loneliness and health.

Finally, our study found that the gratitude writing exercise produced health benefits within a 3-week period. Future research should assess the longer term benefits of cultivating gratitude, for instance, investigating the possibility of continued benefits at 1-month or 6-month intervals as previous research has found (Seligman et al., 2005).

Conclusion

In the first research, to our knowledge, to explore gratitude’s ability to decrease loneliness, and, consequently, to boost well-being in an older population, we find that a simple gratitude writing exercise may offer some relief to a population at risk of loneliness and its consequent deleterious outcomes. Older, community-dwelling individuals who participated in the gratitude writing exercise (i.e., treatment group) showed stronger indirect effects of loneliness on gratitude-health associations than the control group. This is a simple exercise that could be easily taught and offered to both community-dwelling and facility dwelling older individuals.

Supplemental Material

Supplemental Material, CONSORT_2010_Flow_Diagram_(2) - Gratitude and Loneliness: Enhancing Health and Well-Being in Older Adults

Supplemental Material, CONSORT_2010_Flow_Diagram_(2) for Gratitude and Loneliness: Enhancing Health and Well-Being in Older Adults by Monica Y. Bartlett and Sarah N. Arpin in Research on Aging

Supplemental Material

Supplemental Material, Descriptive_Table_Supplementary - Gratitude and Loneliness: Enhancing Health and Well-Being in Older Adults

Supplemental Material, Descriptive_Table_Supplementary for Gratitude and Loneliness: Enhancing Health and Well-Being in Older Adults by Monica Y. Bartlett and Sarah N. Arpin in Research on Aging

Footnotes

Acknowledgments

We are appreciative of the undergraduate research assistants who have helped make this project happen: Janelle Bieker, Sophie Oswald, Elizabeth Stewart, and Matthew Roberts.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We received internal funding. We received $2,400 from the Gonzaga University College of Arts and Sciences Faculty Development Fund.

ORCID iD

Monica Y. Bartlett

Supplemental Material

Supplemental material for this article is available online.

Note

References

Algoe

S. B.

Gable

S. L.

Maisel

N. C.

(2010). It’s the little things: Everyday gratitude as a booster shot for romantic relationships. Personal Relationships, 17, 217–233. doi:10.1111/j.1475-6811.2010.01273.x

Algoe

S. B.

Haidt

Gable

S. L.

(2008). Beyond reciprocity: Gratitude and relationships in everyday life. Emotion, 8, 425–429. doi:10.1037/1528-3542.8.3.425

Bartlett

M. Y.

Condon

Cruz

Bauman

DeSteno

(2012). Gratitude: Prompting behaviors that build relationships. Cognition & Emotion, 26, 2–13. doi:10.1080/02699931.2011.561297

Bartlett

M. Y.

DeSteno

(2006). Gratitude and prosocial behavior: Helping when it costs you. Psychological Science, 17, 319–325. doi:10.1111/j.1467-9280.2006.01705.x

Bekhet

A. K.

Zauszniewski

J. A.

(2012). Mental health of elders in retirement communities: Is loneliness a key factor? Archives of Psychiatric Nursing, 26, 214–224. doi:10.1016/j.apnu.2011.09.007

Bolger

Laurenceau

J. P.

(2013). Intensive longitudinal methods. New York, NY: Guilford.

Cacioppo

J. T.

Cacioppo

Boomsma

D. I.

(2014). Evolutionary mechanisms for loneliness. Cognition and Emotion, 28, 3–21. doi:10.1080/02699931.2013.837379

Cacioppo

J. T.

Hawkley

L. C.

(2003). Social isolation and health, with an emphasis on underlying mechanisms. Perspectives in Biology and Medicine, 46, S39–S52. doi:10.1353/pbm.2003.0049

Cacioppo

J. T.

Hawkely

L. C.

Ernst

J. M.

Burleson

Berntson

G. G.

Nouriani

Spiegel

(2006). Loneliness within a nomological net: An evolutionary perspective. Journal of Research in Personality, 40, 1054–1085. doi:10.1016/j.jrp.2005.11.007

10.

Cacioppo

Bangee

Balogh

Cardenas-Iniquez

Qualter

Cacioppo

J. T.

(2016). Loneliness and implicit attention to social threat: A high-performance electrical neuroimaging study. Cognitive Neuroscience, 7, 138–159. doi:10.1080/17588928.2015.1070136

11.

Caputo

(2015). The relationship between gratitude and loneliness: The potential benefits of gratitude for promoting social bonds. Europe’s Journal of Psychology, 11, 323–334. doi:10.5964/ejop.v11i2.826

12.

Caputo

(2017). Social desirability bias in self-reported well-being measures: Evidence from an online survey. Universitas Psychologica, 16, 1–13. doi:10.11144/Javeriana.upsy16-2.sdsw

13.

Caspi

Harrington

M. E.

Moffitt

T. E.

Milne

B. J.

Poulton

(2006). Socially isolated children 20 years later: Risk of cardiovascular disease. Archives of Pediatric and Adolescent Medicine, 160, 805–811. doi:10.1001/archpedi.160.8.805

14.

Cattan

White

Bond

Learmouth

(2005). Preventing social isolation and loneliness among older people: A systemic review of health promotion interventions. Ageing and Society, 25, 41–67. doi:10.1017/S0144686X04002594

15.

Cohen-Mansfield

Parpura-Gill

(2007). Loneliness in older persons: A theoretical model and empirical findings. International Psychogeriatrics, 19, 279–294. doi:10.1017/S1041610206004200

16.

Diener

Seligman

M. E. P.

(2002). Very happy people. Psychological Science, 13, 80–83. doi:10.1111/1467-9280.00415

17.

Emmons

R. A.

McCullough

M. E.

(2003). Counting blessings versus burdens: An experimental investigation of gratitude and subjective well-being in daily life. Journal of Personality and Social Psychology, 84, 377–389. doi:10.1037/0022-3514.84.2.377

18.

Findlay

R. A.

(2003). Interventions to reduce social isolation amongst older people: Where is the evidence? Ageing & Society, 23, 647–658. doi:10.1017/S0144686X03001296

19.

Hawkley

L. C.

Cacioppo

J. T.

(2010). Loneliness matters: A theoretical and empirical review of consequences and mechanisms. Annual Behavioral Medicine, 40, 218–227. doi:10.1007/s12160-010-9210-8

20.

Hill

P. L.

Allemand

Roberts

B. W.

(2013). Examining the pathways between gratitude and self-rated physical health across adulthood. Personality and Individual Differences, 54, 92–96. doi:10.1016/j.paid.2012.08.011.

21.

Holwerda

T. J.

Deeg

D. J.

Beekman

A. T.

van Tilburg

T. G.

Stek

M. L.

Jonker

Schoevers

R. A

. (2012). Feelings of loneliness, but not social isolation, predict dementia onset: Results from the Amsterdam Study of the Elderly (AMSTEL). Journal of Neurology, Neurosurgery, and Psychiatry. doi:10.1136/jnnp-2012-302755

22.

House

J. S.

Landis

K. R.

Umberson

(1988). Social relationships and health. Science, 241, 540–545. doi:10.1126/science.3399889

23.

Kiecolt-Glaser

J. K.

Ricker

George

Messick

Speicher

C. E.

Garner

Glaser

(1984). Urinary cortisol levels, cellular immunocompetency, and loneliness in psychiatric inpatients. Psychosomatic Medicine, 46, 15–23.

24.

Kodzi

I. A.

Gyimah

S. O.

Emina

Ezeh

A. C.

(2011). Religious involvement, social engagement, and subjective health status of older residents of informal neighborhoods of Nairobi. Journal of Urban Health, 88, 370–380. doi:10.1007/s11524-010-9482-0

25.

Luhmann

Hawkley

L. C.

(2016). Age differences in loneliness from adolescence to oldest old age. Developmental Psychology, 57, 943–959. doi:10.1037/dev0000117

26.

Luo

Hawkley

L. C.

Waite

L. J.

Cacioppo

J. T.

(2012). Loneliness, health and mortality in old age: A national longitudinal sample. Social Science and Medicine, 74, 907–914. doi:10.1016/j.socscimed.2011.11.028

27.

Masi

C. M.

Chen

Hawkley

L. C.

Cacioppo

J. T.

(2011). A meta-analysis of interventions to reduce loneliness. Personality and Social Psychology Review, 15, 219–266. doi:10.1177/1088868310377394

28.

McCullough

M. E.

Emmons

R. A.

Tsang

(2002). The grateful disposition: A conceptual and empirical topography. Journal of Personality and Social Psychology, 82, 112–127. doi:10.1037//0022-3514.82.1.112

29.

McCullough

M. E.

Kilpatrick

S. D.

Emmons

R. A.

Larson

D. B.

(2001). Is gratitude a moral affect? Psychological Bulletin, 127, 249–266. doi:10.1037//0033-2909.127.2.249

30.

Mohr

C. D.

Armeli

Tennen

Temple

Todd

Clark

Carney

M. A.

(2005). Moving beyond the keg party: A daily process study of college student drinking motivations. Psychology of Addictive Behaviors, 19, 392. doi:10.1037/0893-164X.19.4.392

31.

Yang

Zhang

Dong

(2015). Effect of gratitude on loneliness of Chinese college students: Social support as a mediator. Social Behavior and Personality, 43, 559–566. doi:10.2224/sbp.2015.43.4.559

32.

O’Connell

B. H.

O’Shea

Gallagher

(2016). Mediating effects of loneliness on the gratitude-health link. Personality and Individual Differences, 98, 179–183. doi:10.1016/j.paid.2016.04.042

33.

O’Connell

B. H.

O’Shea

Gallagher

. (2017). Examining psychosocial pathways underlying gratitude interventions: A randomized controlled trial. Journal of Happiness Studies. doi:10.1007/s10902-017-9931-5

34.

Raudenbush

S. W.

Bryk

A. S.

(2002). Hierarchical linear models: Applications and data analysis methods (Vol. 1). Thousand Oaks, CA: Sage.

35.

Rockwood

N. J.

Hayes

A. F.

(2017, May). MLmed: An SPSS macro for multilevel mediation and conditional process analysis. Poster presented at the annual meeting of the Association of Psychological Science (APS), Boston, MA.

36.

Russell

Peplau

L. A.

Cutrona

C. E.

(1980). The revised UCLA Loneliness Scale: Concurrent and discriminant validity evidence. Journal of Personality and Social Psychology, 39, 472–480.

37.

Seligman

M. E. P.

Steen

T. A.

Park

Peterson

(2005). Positive psychology progress: Empirical validation of interventions. American Psychologist, 60, 410–421. doi:10.1037/0003-066X.60.5.410

38.

Theeke

L. A.

(2009). Predictors of loneliness in U.S. adults over age sixty-five. Archives of Psychiatric Nursing, 23, 387–396. doi:10.1016/j.apnu.2008.11.002

39.

Watson

Clark

L. A.

Tellegen

(1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. doi:10.1037/0022-3514.54.6.1063

40.

Wood

A. M.

Froh

J. J.

Geraghty

A. W. A.

(2010). Gratitude and well-being: A review and theoretical integration. Clinical Psychology Review, 30, 890–905. doi:10.1016/j.cpr.2010.03.005

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.43 MB

0.15 MB