Daily Rumination–Affect Associations in Dyads During the COVID-19 Pandemic

Abstract

Objectives

Negative and repetitive self-oriented thinking (rumination) is associated with lower well-being and health. The social context of rumination remains underexplored and mostly centers on marital relationships. To embrace the diversity of older adult relationships, this study includes a range of different relationships (e.g., spouses, siblings, friends, etc.) and examines the role of rumination by close others on individual well-being during the coronavirus disease 2019 pandemic.

Methods

Using daily diary data from 140 Canadian older adults (M = 72.21 years, standard deviation [SD] = 5.39, range: 63–87 years, 47% women, 71% university educated) and a close other of their choice (M = 59.95 years, SD = 16.54, range: 18–83 years, 78% women, 81% university educated), this project builds on past research examining daily life rumination dynamics from a dyadic perspective. For 10 days, both dyad members reported their daily rumination and affect quality in the evening.

Results

Multilevel models replicate past work showing that individual rumination was associated with higher negative affect (within-person: b = 0.27, p < .001, between-person: b = 0.57, p < .001) and lower positive affect (within-person: b = −0.18, p < .001, between-person: b = −0.29, p < .001). Importantly, we additionally observed that partner rumination was associated with higher negative affect (b = 0.03, p = .038) and lower positive affect (b = −0.04, p = .023), highlighting the social context of rumination.

Discussion

Findings illustrate the significance of rumination for the self and others and underline the merit of taking a dyadic perspective on what is typically viewed as an individual-level phenomenon.

Rumination involves repetitive, self-oriented, negative thinking; it is negatively associated with well-being and health (Clancy et al., 2016; Nolen-Hoeksema et al., 1993, 2008; Riley et al., 2019). Importantly, rumination–well-being associations are not static; they vary across time (Slavish et al., 2018) and are shaped by social partners (King & DeLongis, 2014; Puterman et al., 2010). This project builds on past evidence examining daily life rumination dynamics from a dyadic perspective using repeated daily life assessments from a sample of older adults plus a close other during a time characterized by high stress and uncertainty, namely the coronavirus disease 2019 (COVID-19) pandemic (Alzueta et al., 2021).

Even though older adults are often able to maintain relatively high levels of well-being, they also report rumination (Charles et al., 2001; D’Hudson & Saling, 2010; Marini et al., 2021). In addition to experiencing COVID-19 threats to oneself and loved ones, many older adults were faced with chronic stressors, given that over a long period of time during the pandemic, they had lost control over key aspects of their lives (Morstead et al., 2022; Zheng et al., 2021). Additionally, specific to older adults, the risk of severe consequences is particularly high in this age group (Applegate & Ouslander, 2020). Results from research using daily diaries with younger samples examining everyday rumination across different contexts (i.e., work stress; chronic disease) point to the time-varying nature of rumination and to its association with lower well-being (King & DeLongis, 2014; Müller et al., 2019). We therefore hypothesized that when individuals report elevated rumination, they would also report higher negative affect and lower positive affect both at the daily and overall levels.

Lifespan scholars calling for an interactive minds perspective emphasize that cognitions and feelings are shaped by the people we interact with (Staudinger & Baltes, 1996). Indeed, research has demonstrated emotional transmission among family members and crossover between romantic partners (Larson & Almeida, 1999; Westman, 2001). Consistent with these ideas, a study of paramedics and their partners (King & DeLongis, 2014) showed that daily increases in paramedic rumination were associated with self- and partner-reported increases in marital tensions, pointing to the important role of significant others for stress-related outcomes. In recognition of the fact that many older adults may not or no longer have a spouse (Brown & Lin, 2012; Rook & Charles, 2017) and interact with other close ties (e.g., friends or siblings; Antonucci et al., 2014; Antonucci & Akiyama, 1987; Fuller et al., 2020), we extended previous work on married couples by recruiting older adults with a person they considered close, independent of who this close other was. In line with central propositions from the Social Convoy Model (Antonucci et al., 2014), we expected that close others, including but not limited to romantic partners, profoundly shape well-being in old age. We therefore hypothesized that individuals whose study partners reported elevated rumination would experience more negative affect and less positive affect on that day as well as overall.

This study aimed to shed light on individual and dyadic associations of daily rumination with positive and negative affect in older adults during the pandemic. Using up to 10 consecutive daily diaries from both dyad members, we predicted that individuals who report more rumination would also report more negative affect and less positive affect at the within- and between-person levels. Similarly, we predicted that individuals whose partners report more rumination would report higher negative affect and lower positive affect over and above individual-level effects. Studies using daily life approaches have the unique capacity to capture daily life as it unfolds with high ecological validity, avoiding retrospective memory biases, and offering targets for intervention (Bolger et al., 2003; Hoppmann & Riediger, 2009). These studies are well-suited to address within-person phenomena (e.g., fluctuations of daily rumination within a given individual) as well as between-person difference phenomena (e.g., individual differences in overall levels of rumination across time in study). They further allow us to speak to what sets one day apart from another, as well as examinations of interindividual differences. Methodologically, considering within- and between-person effects is necessary to account for the nonindependence of multilevel data (Bolger & Laurenceau, 2013). To address well-established associations of rumination with well-being, we also controlled for age and gender of both dyad members, dyad type, and relationship status (Bolger et al., 2003; Carstensen et al., 2011; Inglehart, 2002). We further adjusted for day of study and start day.

Method

Participants and Procedures

This project uses data from a larger study conducted between June 2020 and June 2021 (see Zambrano Garza et al., 2023 for details). The sample consisted of 140 Canadian older adults (M = 72.21 years, standard deviation [SD] = 5.39, range: 63–87 years, 47% women, 71% university educated) and a close other of their choice (M = 59.95 years, SD = 16.54, range: 18–83 years, 78% women, 81% university educated). The sample included 82 spouses, 19 friends, seven siblings, 11 adult child–parent, four adult grandchild–grandparent, 16 other family member, one other. For information on eligibility, exclusion criteria, and missing data, please see Supplementary Material. Both study partners attended an instructional Zoom meeting. They then simultaneously completed two online questionnaires (morning and evening) for 10 consecutive days. This study uses data from the evening questionnaires only. Each participant received a $50 Amazon gift card as a token of appreciation. The study was approved by the Behavioural Research Ethics Board of the University of British Columbia, Vancouver, Canada (H20-01645); participants provided informed consent. Participants completed an average of 92% of their evening questionnaires (SD = 1.62, range: 2–10).

Measures

Rumination

Every evening, participants answered two rumination items (“Since this morning, how often did you experience a train of thought that was difficult to get out of your head?” and “Since this morning, how often were you preoccupied with thoughts about the future?”; Slavish et al., 2018; M = 17.61, SD = 20.41, 0 = “not at all” to 100 = “very much”; see Author Note 1).

Positive affect

Every evening, participants reported how they had felt that day using 10 positively valenced items (happy, calm, alert, relaxed, content, enthusiastic, relieved, excited, satisfied, appreciated; M = 61.59, SD = 19.59, R_c = 0.92, 0 = “not at all” to 100 = “very much”; Lay et al., 2019; Tsai et al., 2006).

Negative affect

Every evening, participants reported how they had felt that day using 10 negatively valenced items (sad, overwhelmed, irritated, lonely, frustrated, worried, tired, nervous, disappointed, stressed; M = 15.52, SD = 15.29, R_c = 0.90, 0 = “not at all” to 100 = “very much”; Lay et al., 2019; Tsai et al., 2006).

Covariates

Gender (0 = man, 1 = woman) and age of both dyad members, day of study, and start date were included to account for time-varying COVID-19 restrictions. We also controlled for dyad type (0 = couple, 58%) and relationship status (0 = in a relationship, 1 = single, 85% in a relationship; see Author Note 2).

Statistical Analyses

Due to the hierarchically nested data structure, data were modeled at three levels (day, person, dyad). We examined the origins of variability for positive affect (31% at the day level, 53% at the person level, 16% at the dyad level) and negative affect (33% at the day level, 46% at the person level, 21% at the dyad level). If both members of the dyad were older adults, one was randomly selected as the focal person. Continuous variables were grand-mean centered. Dichotomous variables were dummy coded. Participation date was centered to the study start date (range: 0–366). Rumination was separated into within-person effects (person-centered) and between-person effects (grand-mean centered person means). To examine partner effects, we included partner rumination variables as well as age and gender. We used R to run 1,000 Monte Carlo power simulations with standardized effect sizes of 0.1 of main effects; power was 100% with 280 participants and 2,240 observations. Multilevel models used restricted maximum likelihood estimation including random intercepts and slopes for individual and partner rumination at the day level. We used R package lme4 (see Supplementary Material for code).

Results

Supplementary Tables 1 and 2 provide descriptive information and intercorrelations. Supplementary Figure 1 shows percents of observations by dyad type. Older age was correlated with higher positive affect (r = 0.19, p < .01) and lower negative affect (r = −0.24, p < .01) and rumination (r = −0.16, p < .01). Women reported less positive affect (r = −0.10, p < .01) and more negative affect (r = 0.13, p < .01) than did men. More rumination was correlated with having a woman as a close tie (r = 0.11, p < .01), higher negative affect (r = 0.12, p < .01), and participating with a romantic partner (r = 0.14, p < .01).

As expected, both individual and partner rumination were significantly associated with positive and negative affect in daily life (Table 1, Model A for negative affect and Model B for positive affect). Individual rumination was associated with higher negative affect (Table 1, Model A) at the within-person (b = 0.27, p < .001) and between-person level (b = 0.57, p < .001). Similarly, individual rumination was associated with lower positive affect (Table 1, Model B) at both levels (within-person; b = −0.18, p < .001 and between-person; b = −0.29, p < .001). Partner effects were significant at the within-person level only (negative affect: b = 0.03, p = .038, positive affect: b = −0.04, p = .025). Dyad type (negative affect: b = 5.42, p = .001, positive affect: b = −6.56, p = .051), day of study (negative affect: b = −0.19, p = .001, positive affect: b = −0.38 p < .001), and relationships status (negative affect: b = 4.55, p = .006, positive affect: b = −8.46, p = .009) were also significant predictors. Marginal R² indicates the proportion of total outcome variance explained by predictors via fixed slopes (Rights & Sterba, 2019) for Model A as 0.65 and for Model B as 0.31.

Discussion

The COVID-19 pandemic gave rise to unprecedented stress; this study examined daily associations between rumination and affect quality in older adults plus a close other. Consistent with hypotheses, individuals who reported more rumination reported higher negative affect and lower positive affect. Both daily (within-person) and overall (between-person) rumination were associated with affect quality (see Figure 1); higher rumination was associated with more negative affect and less positive affect. Our results are in line with previous findings (Moberly & Watkins, 2008; Riley et al., 2019), and also uniquely contribute to the literature by placing further emphasis on older adults during times of challenge. Older adults have been shown to experience high levels of rumination (D’Hudson & Saling, 2010); our research advances this knowledge by delving into these associations within the context of daily life and during periods of heightened stress. The integration of daily life contexts allows us to capture the subtleties and variations in rumination that might not be grasped through conventional laboratory designs. In essence, our findings provide a more detailed picture of rumination, highlight its dynamics within the everyday lives of older adults, and underscore the importance of considering these processes within the broader context of aging and resilience.

This study also considered the role of close others in shaping affective experiences in daily life. As such, it highlights how older adults’ affect quality is shaped by their own cognitions and those of close others, providing avenues for future research and interventions (i.e., considering both individuals in a close relationship to decrease the deleterious effects of rumination). Partner effects were significant at the within-person level only, in the expected directions; when a close other ruminated more than their average, individuals reported higher negative affect and lower positive affect. Results dovetail with dyadic findings from younger samples (King & DeLongis, 2014; Müller et al., 2019) and underscore that close others can and do influence each other (Cook & Kenny, 2005). This finding highlights that close others, including but not limited to spouses, shape affect quality in daily life in old age. The significant effects of relationship status and dyad type show that older adults participating with nonromantic partners reported higher negative affect, highlighting the salient role of romantic partners relative to other social ties (Hoppmann & Gerstorf, 2016; Robles et al., 2014). Findings are consistent with prepandemic and pandemic evidence suggesting that individuals who are married tend to report higher well-being than those who are not (Dush & Amato, 2005; Gubler et al., 2021).

We observed a significant day of study effect indicating that individuals reported less positive and negative affect over time. The questionnaires could have created an intervention effect; as participants got used to reflecting on their day, this exercise may have become less emotionally salient over time. We saw no significant effects for age, gender, and start day. This could result from the fact that the sample was predominantly older and female, thereby limiting variability to detect meaningful age and gender differences. The reported associations did not appear to change over the first year of the pandemic, possibly due to the ongoing public health restrictions and continuing challenges.

Limitations and Future Directions

We note limitations of our study sample, design, and measures. First, although we took a step toward including different types of dyads, the age range was wide, gender proportions differed, and we were only able to meaningfully differentiate between heterosexual couples and noncouples. Nonetheless, relationship type and demographic characteristics do not appear to trump what is happening within these relationships (the main effects remained), suggesting that rumination of both members of the dyad play an important role. Future studies should recruit equal group sizes of different dyad constellations to disentangle the possibly unique role of relationship type, age, and gender composition. Second, we are unable to speak to whether rumination increases negative affect (and decreases positive affect) or vice versa; future studies would need to collect more frequent assessments per day and/or more days to comprehensively examine time-lagged associations. Finally, our study does not reveal the source of rumination; future studies could examine specific concerns that participants are ruminating about.

Conclusion

This study builds on past research by exploring the role of rumination for daily affective experiences in older adults and their close others during times of challenge (the COVID-19 pandemic). By taking a dyadic perspective, it highlights the interpersonal contours of what is often seen as an individual-level phenomenon.

Author Notes

1.Item “Since this morning, how often did you think about situations that upset you?” was excluded due to potential overlap with negative affect. Models including this item are in Supplementary Table 3; findings do not differ substantially.

2.Models including employment status and parental status as covariates are reported in Supplementary Table 4; findings did not differ.

Funding

The Canadian Institutes of Health Research (PJT-169093) and a University of British Columbia HIFI award supported this work. E. Zambrano Garza gratefully acknowledges support from UBC’s Four Year Doctoral Fellowship. R. A. Murphy’s time was supported by the Michael Smith Health Research BC (grant #17644). C. A. Hoppmann and M. C. Ashe gratefully acknowledge support from the Canada Research Chairs Program.

Conflict of Interest

None.

Data Availability

Preregistered hypotheses as well as the analytic plan can be found on OSF (https://osf.io/42z6n/?view_only=da666cefacac4500a4f202128606ad2b) and data can be made available upon request.

Acknowledgments

We gratefully acknowledge participants’ contributions to the study.

References