Are Daily Well-Being and Emotional Reactivity to Stressors Modifiable in Midlife?: Evidence from a Randomized Controlled Trial of an Online Social Intelligence Training Program

Abstract

The complex set of challenges that middle-aged adults encounter emphasizes a need for mental health interventions that promote resilience and positive outcomes. The present study evaluated whether an online, self-guided social intelligence training (SIT) program (8 h) improved midlife adults’ daily well-being and emotion regulation in the context of their own naturalistic everyday environment. A randomized controlled trial was conducted with 230 midlife adults allocated into either a SIT program or an attentional control (AC) condition that focused on healthy lifestyle education. Intent-to-treat analyses examined two bursts of 14-day daily surveys that participants completed pre- and post-treatment. Multilevel models evaluated pre-to post-treatment changes in mean positive and negative affect, as well as daily emotional reactivity to stressors and responsiveness to uplifts. Compared to the AC group, those in the SIT program reported improvements (i.e., decreases) in mean negative affect, positive emotional reactivity to daily stressors (i.e., smaller decreases in positive affect on stressor days), and negative emotional responsiveness to uplifts (i.e., lower negative affect on days without uplifts). Our discussion considers potential mechanisms underlying these improvements, highlights downstream effects on midlife functioning, and elaborates on how online delivery of the SIT program increases its potential for positive outcomes across adulthood.

ClinicalTrials.gov Identifier: NCT03824353.

Decades of research confirm that a sense of connection with other people fuels health, well-being, and resilient adaption to stress across the lifespan (Seppala et al., 2013). Individuals with strong social relationships have healthier cardiovascular functioning, more efficient immune responses to pathogens, suffer less disability in response to illness, and live longer (Hawkley & Cacioppo, 2010; Holt-Lunstad et al., 2010). Anxiety and depression are linked to lost or threatened social bonds, and those without social bonds are at greater risk of suicide (Tsai et al., 2015). A key source of resilience (i.e., positive functioning amid adversity) are sustainable social relationships (Infurna & Luthar, 2018; Luthar & Eisenberg, 2017). However, there is a dire need for mental health interventions capable of improving resilience mechanisms (e.g., social skills, emotion regulation) that are accessible, affordable, and implementable in “real-world” settings (Luthar & Eisenberg, 2017; Wilhelm et al., 2020).

Midlife is a phase of the lifespan when adults rely on social skills and emotion regulation to balance several roles, navigate life transitions, take care of younger and older generations, and confront financial stressors (Infurna et al., 2020). Theoretical work proposes that socio-emotional regulatory skills can be modified in midlife and beyond with training programs that promote self-reflection and intentional prosocial activities (Davidson & McEwen, 2012). Previous research showed that an online, self-guided social intelligence training (SIT) program improved midlife adults’ social skills (Castro et al., 2019). The present study builds on prior work by evaluating whether this SIT program generated secondary spillover effects on daily well-being and emotion regulation in the context of daily negative (stressor) and positive (uplift) events.

Subjective well-being has been measured in diverse ways, such as perceived happiness and life satisfaction (Kahneman & Krueger, 2006). Consistent with daily diary literature, we conceptualized well-being as one’s level of negative and positive emotions on days when no negative or positive daily event is reported (Zautra et al., 2005). Based on the extended process model (Gross, 2015), emotion regulation is defined as processes that unfold over time as one exerts control over emotions in a goal-directed manner. With intensive longitudinal data (i.e., daily surveys), we conceptualized emotion regulation as the temporal dynamics of midlife adults’ negative and positive emotional states as a function of negative and positive events in their own naturalistic context (Kuppens & Verduyn, 2015).

Despite considerable advancements in socio-emotional programs for children and adolescents (Mahoney et al., 2018), there is a lack of effective, easily accessible interventions that improve social and emotional skills of middle-aged and older adults. Research on socio-emotional interventions for adults show inconsistent effects, small effect sizes when present, and rely heavily on group or one-on-one delivery (Fakoya et al., 2020; Siette et al., 2017). The present study innovates by delivering a self-guided socio-emotional program through the internet, and leveraging smartphones to collect daily surveys and evaluate improvements in dynamic processes involving how day-to-day negative (stressor) and positive (uplift) events shape fluctuations in emotional well-being (Almeida, 2005; Hamaker & Wichers, 2017).

Daily process research indicates that exposure to daily stressors negatively impacts well-being and health as much as major life events in midlife (Ong & Leger, 2021). In contrast, engagement with daily uplifts generally promotes well-being, healthy behaviors, and protects against life stressors (Leger et al., 2020; Sin & Almeida, 2018). Combining daily process methods and process-based theories of self-regulation (Almeida, 2005; Gross, 2015), the present study evaluated whether the SIT program improved (i.e., reduced) daily emotional reactivity (i.e., stressor-related changes in well-being), a within-person process of emotion regulation associated with mental health disorders, inflammation, onset of chronic illness, and increased mortality risk (Charles et al., 2013; Mroczek et al., 2013; Piazza et al., 2013; Sin et al., 2015). Importantly, the present study also examined improvements (i.e., reductions) in daily emotional responsiveness (i.e., uplift-related changes in well-being), a separate within-person process of emotion regulation previously associated with lower well-being and major depressive disorders (Bylsma et al., 2011; Grosse Rueschkamp et al., 2020; Khazanov et al., 2019).

Daily process methods, involving repeated measurements of individuals over time and across real-world contexts, benefit randomized controlled trials through less retrospective bias and more ecological validity, while enabling the study of intervention-related changes in dynamic within-person processes (Almeida, 2005). For example, previous research had midlife adults complete 30 daily surveys prior to and following in-person group interventions and found reductions in emotional reactivity to daily stress (Davis et al., 2015). Similar daily process methods have been infused into randomized trials of online interventions (Davis & Zautra, 2013). Recently, the online SIT intervention tested in this study was evaluated with daily diary methods in a randomized controlled trial (Castro et al., 2019). Participation in the SIT program led to increased social connection, emotional awareness, and perspective-taking. Midlife adults were also better able to maintain social connection with close others on stressor days. Given that social connection sustains positive affect and protects against daily stress (Leger et al., 2020; Seppala et al, 2013), the enhanced capacity to maintain social connections on stressor days may translate to improved emotional reactivity (i.e., attenuated stressor-related changes in well-being). Additionally, the improved capacity to maintain social connections may help sustain well-being without relying on “mood-boosting” uplifts (i.e., attenuated uplift-related changes in well-being; Grosse Rueschkamp et al., 2020; Leger et al., 2020).

Building on definitions of social intelligence as interpersonal skills (Eisenberg & Fabes, 1992), the SIT program emphasized gaining knowledge of behavioral processes (Kihlstrom & Cantor, 2011) and commitment to humanizing other people (Castro & Zautra, 2016). The self-guided SIT frames social intelligence as a modifiable attribute, and translates key information derived from social neuroscience, psychology, and related fields (e.g., neuroplasticity, brain development, (non)conscious processing, cognitive biases/schemas, communication techniques, (epi)genetics, attachment styles) into jargon-less psychoeducation via a series of narrated audio-visual animations that are accessed through a website (Castro et al., 2019). Consistent with cognitive-behavioral and interpersonal therapy modalities (Lipsitz & Markowitz, 2013; Lorenzo-Luaces et al., 2016), the SIT course works by encouraging individuals to replace their problematic cognitive-behavioral patterns learned early in life with healthier social schemas and habits via self-reflection and deliberate interpersonal behaviors.

The Present Study

This randomized clinical trial (RCT) examined whether an online SIT program improved daily well-being and emotion regulation among community-dwelling midlife adults (n = 230), compared to a healthy lifestyle information program. To provide a more rigorous examination of theoretically “active ingredients” in the SIT course (MacCoon et al., 2012; Moos, 2007), the latter health program was used as an attention control (AC) condition that mimicked nonspecific intervention factors (e.g., delivery format, course structure, expectations for improvement). We examined daily survey data collected 14 days prior to beginning and 14 days following completion of either the SIT or AC. Building on previous research showing improved socio-emotional skills (Castro et al., 2019), the present study examined whether the SIT program also improved daily well-being (i.e., daily mean levels of positive and negative emotions) and emotion regulation indexed as the within-person processes of reactivity (i.e., stressor-related changes in well-being) and responsiveness (i.e., uplift-related changes in well-being). We hypothesized that the SIT program would lead to higher positive affect, lower negative affect, and reduced emotional reactivity to stressors and reduced emotional responsiveness to uplifts.

Method

Participants and Procedure

Sampling and Recruitment

We contacted 557 individuals from the AS U Live Project who indicated they would like to be contacted for future studies (for more information on the larger sample, see Infurna et al., 2015). Individuals were contacted through (1) mailed recruitment letters that provided an update on the study in which they participated in, and (2) research assistants followed-up 1 to 2 weeks after the newsletter was sent with a phone-call or e-mail. At follow-up, the research team introduced themselves to the potential participant, provided study information, and requested participation. Inclusion criteria for recruitment were that the participant (1) was over the age of 39 and (2) had access to the internet. Participants were compensated up to $173 US for participation in the study (i.e., questionnaires, home visit, phone interview). For participants who were interested, the next step was an in-person meeting (30 min in length), where a research assistant explained the study procedures. Prior to beginning either of the two programs, participants completed a pre-test questionnaire, along with 14 days of daily surveys. After completion of pre-treatment daily surveys, participants were randomly assigned (with an online randomizer) to either the SIT program or AC program, which consisted of healthy lifestyle information presented in a similar format. Participants were instructed to access their assigned program via a website using a computer and complete one module per week (for a total of eight weeks to complete the program). Each module lasted approximately 60–90 min. Program videos could not be fast-forwarded, and module completion was verified electronically. Average completion time was 8.18 weeks (SD = 4.67, range: 1.86 to 25.43) for the SIT program, and 7.94 weeks (SD = 4.20, range: 1.43 to 24.29) for the AC program. At post-treatment, participants completed another 14 days of daily surveys. Figure 1 shows the CONSORT flowchart. Table 1 shows demographic information by group.

Fig. 1.

CONSORT flowchart

Table 1.

Sociodemographic comparisons for the SIT and AC Groups

	SIT	AC	χ2/t (p value)
Sex			0.8 (0.37)
Female	62.3%	56.5%
Male	37.7%	43.5%
Ethnic-racial background			0.0 (0.99)
Caucasian	79.5%	74.1%
Black/African American	2.5%	1.9%
American Indian/Alaska Native	.8%	0.0%
Asian	2.5%	1.9%
2 or more ethnic-racial identities	13.1%	13.9%
Marital status			0.9 (0.35)
With spouse/romantic partner	65.6%	71.3%
Without spouse/romantic partner	34.4%	28.7%
Employment status			0.8 (0.37)
Currently employed	63.1%	57.4%
Not currently employed	36.9%	42.6%
Educational attainment			0.2 (0.69)
< High school diploma/GED	1.6%	2.5%
High school diploma/GED	7.4%	4.6%
Trade/vocation/technical school certificate	4.9%	4.6%
Some college    College degree	25.4% 30.3%	26.9% 28.7%
Some graduate school    Graduate/professional degree	8.2% 18.0%	5.6% 25.0%
Mean
Income	70,631.47	93,119.45	2.6 (0.01)
Age	60.7	60.5	0.2 (0.84)
Daily surveys (across pre and post-treatment)	25.9	25.4	0.5 (0.63)
Program completion (in weeks)	8.2	7.9

Chi-square tests conducted for categorical variables. Ethnic-racial background (1 = Caucasian, 0 = non-Caucasian). Educational attainment (1 = college degree/grad school experience, 0 ≤ college degree)

T tests conducted for continuous variables, SIT social intelligence training (N = 122), AC attention control (N = 108)

Social Intelligence Training Program

The SIT program includes 42 brief audio-visual content videos, structured into 7 thematic modules, that are self-guided and accessed via website (socialintelligenceinstitute.org). Each session contains a video lesson that ranges from 5 to 15 min, followed by reflection questions designed to provoke thoughtful attention to current and past experiences relevant to the material presented. Each session ends with instructions that move the participant from awareness to a practice exercise intended to enhance readiness to change and self-efficacy. The sessions build on one another, gradually increasing in depth of awareness and cognitive-behavioral engagement with material to instill the habit of socially intelligent reasoning and thoughtful behavior. Four meta-cognitive principles guided development of the 42-session, 7-module online curriculum, which is described in more comprehensive detail in previous research (Castro et al., 2019).

Attention Control Condition: Healthy Living Information

The Healthy Living program, referred to as the attention control (AC) condition, provided information about different aspects of health, and was designed to mimic the online delivery and course structure of the SIT program (i.e., 42, 5–15-min sessions organized around 7 modules; for detail, see Castro et al., 2019). The AC controls for non-specific factors associated with providing attention to participants that may yield positive outcomes in the absence of specified treatment, and provides a face valid minimal intervention that prevents differential dropout.

Daily Survey Measures

Negative and Positive Affect

Each day, participants completed the Positive And Negative Affect Schedule, which totaled 17 items (PANAS-SF; Watson et al., 1988). The Negative Affect scale consisted of 9 items, such as feeling anxious, irritable, and distressed. The Positive Affect scale consisted of 8 items, such as feeling happy, loved, or hopeful. Respondents indicated how often they had felt this way during the past 24 h on a 5-point scale ranging from 1 (very slightly/not at all) to 5 (extremely).

Stressor and Uplift Events

During completion of the daily survey each night online, participants answered questions pertaining to negative (i.e., stressors) and positive (i.e., uplifts) events that occurred that day (Infurna et al., 2015). The specific wording for daily stressors was, “Think of the most stressful event that occurred today, even if it may not have been too stressful. Which category was this event in?” Categories were spouse/partner, family, friends, work, finances, health, other, and no stressful event. For daily uplifts, the specific wording was, “Think of the most positive event that occurred today, even if it may not have been too positive. Which category was this event in?” Categories were spouse/partner, family, friends, work, finances, health, other, and no positive event. From these items, we created two dichotomous variables, one for negative events and one for positive events, to indicate whether or not participants reported a stressor or uplift during the given day. If participants reported a stressor or uplift in one of the domains listed, then the stressor or uplift dichotomous variable was coded as a 1, with a 0 for days indicative of no stressors or uplifts. On average, participants reported a stressor on 67% of days pre-test and 57% of days post-test. On average, participants reported a positive event on 89% of days pre-test, and 85% days post-test.

Statistical Analysis

Based on methodological recommendations (Gupta, 2011; Steeger et al., 2021), we conducted analyses on an intent-to-treat basis (i.e., all available data), and tested for baseline equivalency and differential attrition across groups. The initial analytic step was to evaluate whether (1) completers (those who completed an intervention and provided at least one post-treatment daily diary) differed from attritors and (2) SIT group differed from AC group in demographics at pre-treatment by conducting a series of χ² analyses for categorical variables (ethnic-racial background and education level were transformed into binary variables) and t tests for continuous variables in IBM SPSS (Version 26).

Daily survey data were analyzed using Multilevel Models (MLM) in SAS (Version 9.4, SAS Institute, Cary, NC) given that MLM enables construction of statistical models that account for clustered observations in intensive longitudinal data, while providing maximum-likelihood estimation using all available data to retain participants with only partial data and increase power to detect effects (Raudenbush & Bryk, 2002). Pre-treatment differences in key outcomes across SIT/AC groups and completers/attritors, as well as treatment-related changes in daily levels of affect and event-related changes in affect were examined with PROC MIXED (Singer, 1998). MLM partitions variance into two components: daily reports (Level 1, n = 5866) clustered within individuals (Level 2, n = 230). Unconditional mean models were computed for each outcome to yield intraclass correlations (ICC) that indexed the proportion of total variance in daily outcomes at Level 2. ICC estimates indicated that 69% and 50% of the total variance in positive affect and negative affect, respectively, were at the between-person level. ICC estimates indicated ample variance at the within-person level to proceed with the MLM analyses.

Our analyses examined whether participation in the SIT or AC changed mean levels of negative and positive affect, and moderated emotional reactivity to daily stressors and uplifts. Models had denominator degrees of freedom set at between-within, with an unstructured covariance at Level 2. Models included a first-order autoregressive covariance structure at Level 1 to account for autocorrelated residuals given empirical precedent (Davis et al., 2015) and simulation studies that show these structures are preferable (i.e., provides less biased random intercept estimates and fixed effect standard error estimates) to unstructured covariances in intensive longitudinal data with 20–40 observations per cluster (Jahng & Wood, 2017). To increase generalizability to the population of persons from which the sample was drawn, random effects were included for intercepts and both daily event slopes. Level-2 predictors were time (0 = Pre, 1 = Post) and group condition (0 = AC, 1 = SIT). Level-1 predictors of stressor (0 = No Stressor, 1 = Stressor) and uplift days (0 = No Uplift, 1 = Uplift) were left un-centered as is commonly done in the study of within-person emotional processes (e.g., Ong & Leger, 2021).

Analyses were done in stages following previous daily-diary evaluation of intervention effects (Davis et el., 2015). First, we determined whether groups were equivalent at baseline by testing group differences at pre-treatment in key outcomes of mean levels of affects, stressor-reactivity, and uplift-responsiveness, with models including group, stressor/uplift, and group × stressor/uplift interactions as predictors. Next, we tested whether mean levels of affects, stressor-reactivity, and uplift-responsiveness changed significantly from pre-to-post within each group separately with models that included time, stressor/uplift, and time × stressor/uplift interactions as predictors. Then, we combined SIT and AC data, and tested whether groups differed in the magnitude of pre-to-post change with a model including group, time, and a group × time interaction as predictors. Results are based on unstandardized estimates. Effect sizes were calculated by dividing the parameter estimate over the square root of the intercept variance (i.e., standard deviation; for discussion; see Grimm et al., 2017).

Results

Sample Characteristics, Group Baseline Equivalency, and Differential Attrition

The sample of 230 was comprised primarily of females (59.6%) who were middle-aged (M_age = 60.6 years, range = 39–71), Caucasian (77.0%), married/partnered (68.3%), and employed (60.4%). Most participants had at least a college degree (60.2%), and the mean household income was $81,425 per year (range = $1700–$500,000). On average, participants completed 14 surveys at pre-treatment (95% did > 10) and 10 surveys at post-treatment (65% did > 10). The number of daily surveys completed across the RCT did not differ by any demographic (all ps > 0.15). Table 1 shows that the SIT and AC groups were comparable at baseline in demographics (except higher income in AC group) and in the number of surveys completed across the RCT. Results from pre-treatment multilevel models indicated that the SIT and AC groups were comparable in key outcomes of mean negative affect, mean positive affect, stressor-reactivity, and uplift-responsiveness (ps > .12). Therefore, random assignment yielded equivalent groups at baseline on demographics, outcomes, and daily surveys. At pre-treatment, individuals, on average, experienced declines in positive affect (b =  −0.18, p < .01) on days when a stressor was reported and, on average, an increase in positive affect on days when an uplift was reported (b = 0.35, p < .01). A similar pattern was observed for negative affect with stressors, on average, associated with an increase in negative affect (b = 0.30, p < .01) and uplifts, on average, associated with a decrease in negative affect (b =  −0.12, p = .05).

Participants in the SIT and AC groups were equally likely to stay in treatment. Completion rates were similar between groups, with 72% of SIT and 67% of AC participants completing all 7 modules (p > .36). Group comparisons indicated that completers and attritors were comparable across all demographics (ps > .10). Results from pre-treatment multilevel models showed that completers and attritors were comparable in key outcomes of mean negative affect, positive affect, reactivity to daily stressors, and responsiveness to uplifts (all ps > .21). Thus, attrition was not related to demographics nor outcomes. Ratings offered voluntarily by participants after program completion suggested both conditions were viewed favorably.

Intervention Effects on Mean Affects, Stressor-Reactivity, and Uplift-Responsiveness

Pre-to-post multilevel models examined whether those in the SIT or AC changed in mean negative affect, mean positive affect, stressor-reactivity, and uplift-responsiveness. SIT and AC groups were examined separately, then in a combined model. Following treatment, those in the SIT group reported marginally smaller declines in positive affect on stressor days (b = 0.08, p < .08). However, the SIT group reported smaller increases in positive affect on uplift days (b =  −0.13, p < .03) and smaller increases in negative affect on stressor days (b =  −0.07, p < .05). The AC group reported larger decreases in negative affect on uplift days (b =  −0.14, p < .03).

A combined model compared the SIT and AC groups on the magnitude of pre-to-post changes in mean negative affect, positive affect, stressor-reactivity, and uplift-responsiveness. Group comparisons indicated that improvements in outcomes were significantly larger in the SIT group compared to the AC group (Table 2). The SIT group showed smaller decreases in positive affect on stressor days relative to the AC group (Fig. 2; d = 0.16). Moreover, relative to the AC group, the SIT group reported stronger declines in mean negative affect (Fig. 3; d = 0.35), as well as lower negative affect on days without uplifts (Fig. 4; d = 0.37). Effect size estimates indicated the SIT program had a small to medium effect on the outcomes in Figs. 2–4.

Table 2.

Multilevel Models Examining Pre-to-Post Changes in Daily Mean Daily Positive Affect, Mean Negative Affect, Emotional Reactivity to Stressors, and Emotional Responsiveness to Uplifts: Comparison of SIT and AC groups

	Positive Affect	Negative Affect
Fixed Effects	Estimate (SE)[CI]	Estimate (SE) [CI]
Intercept	3.30** (0.10) [3.11, 3.49]	1.36** (0.06) [1.23, 1.49]
Time	0.05 (0.08) [−0.11, 0.20]	0.10 (0.06) [−0.02, 0.23]
Stressor	−0.16** (0.03) [−0.23, −0.10]	0.30** (0.03) [0.25, 0.35]
Uplift	0.33** (0.06) [0.21, 0.45]	−0.09 (0.05) [−0.20, 0.02]
Stressor x Time	−0.06 (0.04) [−0.15, 0.03]	−0.03 (0.04) [−0.10, 0.04]
Uplift x Time	0.05 (0.07) [−0.10, 0.19]	−0.14* (0.06) [−0.26, -0.02]
SIT	0.01 (0.13) [−0.25, 0.26]	0.08 (0.09) [−0.09, 0.25]
SIT x Time	0.05 (0.10) [−0.14, 0.25]	−0.16* (0.08) [−0.31, -0.01]
SIT x Stressor	−0.08 (0.05) [−0.18, 0.01]	−0.02 (0.04) [−0.09, 0.05]
SIT x Uplift	0.13 (0.08) [−0.03, 0.29]	−0.07 (0.07) [−0.21, 0.07]
SIT x Stressor x Time	0.13* (0.06) [ 0.01, 0.25]	−0.04 (0.05) [−0.13, 0.06]
SIT x Uplift x Time	−0.18 (0.09) [−0.36, 0.01]	0.17* (0.08) [0.02, 0.32]
Random Effects
Intercept	0.66** (0.08) [0.54, 0.85]	0.21** (0.03) [0.16, 0.29]
Stressor	0.03** (0.01) [0.02, 0.05]	0.02** (0.01) [0.01, 0.04]
Uplift	0.08** (0.02) [0.05, 0.14]	0.08** (0.02) [0.05, 0.13]
Autocorrelation (1)	0.37** (0.01) [0.34, 0.40]	0.30** (0.01) [0.28, 0.33]
Residual	0.26** (0.01) [0.25, 0.27]	0.16** (0.01) [0.16, 0.17]

Models based on 5866 daily observations (3463 at pre, 2403 at post) nested within 230 individuals

SIT Social Intelligence Training (N = 122), AC Attention Control (N = 108), SE Standard Errors, CI 95% Confidence Intervals, Time (0 = Pre, 1 = Post), Stressor (0 = No Stressor, 1 = Stressor), Uplift (0 = No Uplift, 1 = Uplift)

*p < .05;**p < .01

Fig. 2.

Pre-to-post changes in within-person positive emotional reactivity to daily stressors across social intelligence training (SIT) and attention control (AC) interventions. Flatter slopes indicate less reactivity to stressors

Fig. 3.

Pre-to-post changes in daily mean levels of negative affect across social intelligence training (SIT) and attention control (AC) interventions

Fig. 4.

Pre-to-post changes in within-person negative emotional responsiveness to daily uplifts across social intelligence training (SIT) and attention control (AC) interventions. Flatter slopes indicate less responsiveness to uplifts

Discussion

The objective of this RCT was to test the effectiveness of the SIT program on middle-aged adults’ daily well-being and emotion regulation. Several aspects of this investigation were particularly novel: the focus on improving well-being and emotion regulation processes among midlife adults, intervention delivery via the internet, daily online surveys to evaluate treatment effects, and the translational nature of the SIT program content. Following participation, individuals in the SIT group reported improvements (i.e., decreases) in reactivity to daily stressors and responsiveness to daily uplifts. Those in the AC group reported worse (i.e., increased) responsiveness to uplifts. Compared to the AC group, those in the SIT group reported improvements (i.e., decreases) in mean negative affect, positive emotional reactivity to daily stressors (i.e., smaller decreases in positive affect on stressor days), and negative emotional responsiveness to uplifts (i.e., lower negative affect on days without uplifts). Contrary to our expectations, those in the SIT group (relative to the AC group) did not show improvements in mean positive affect, negative reactivity to stressors, nor positive responsiveness to uplifts. Our discussion considers potential mechanisms underlying improvements in emotion regulation, highlights downstream effects on midlife functioning, and elaborates on how online delivery of the SIT program increases its potential for positive outcomes across adulthood.

Despite advancements in socio-emotional programs for youth (Mahoney et al., 2018), there are few accessible and effective interventions that improve the social and emotional skills of midlife and older adults (Fakoya et al., 2020; Siette et al., 2017). Based on resilience research pointing to social connection and emotion regulation as key modifiable factors in midlife and beyond (Infurna et al., 2020), the SIT program tested here was designed to improve well-being via daily interpersonal skills and emotion regulation mechanisms. Our findings show that the SIT program improved stressor-related changes in positive affect, suggesting that midlife adults were less emotionally reactive to daily stressors, which may have downstream protective effects on well-being, health, and mortality (Mroczek et al., 2013; Sin et al., 2015; Zhaoyang et al., 2019). This improved capacity to sustain daily positive emotions following participation in the SIT program is noteworthy given that positive emotions protect against daily stressors and predict well-being and health outcomes over and above negative emotional states (Leger et al., 2020; Ong et al., 2020). Additionally, those in the SIT group reported lower negative affect on days without uplifts, indicating improved emotion regulation via less reliance on daily uplifts to keep negative affect low (Bylsma et al., 2011; Grosse Rueschkamp et al., 2020; Khazanov et al., 2019). It is also noteworthy that the SIT program reduced mean negative affect, given that negative emotions are associated with more chronic health conditions and worse functional limitations 10 years later (Leger et al., 2018). Though small to medium in magnitude, the effects shown here reflect changes in day-to-day processes that may generate long-term benefits as improved emotion regulation compounds over time.

How does social intelligence training work? A prior evaluation of the SIT program showed that midlife adults were better able to maintain social connection with close others on stressor days (Castro et al., 2019). Given that social connection in the face of adversity contributes to positive affect (Seppala et al., 2013), the SIT program appears to have enhanced one’s capacity to maintain social connection and thus preserve positive emotions amid daily stressful experiences. In turn, the preservation of positive emotions may have enabled individuals in the SIT group to maintain low negative emotions without relying on the “mood-boosting” effects of daily uplifts (Grosse Rueschkamp et al., 2020; Leger et al., 2020). More broadly, the SIT program is similar to interpersonal therapy, including core features of psychoeducation, self-reflection, and behavioral exercises designed to facilitate emotional processing and boost interpersonal skills (Lipsitz & Markowitz, 2013). Consistent with cognitive-behavioral therapy (Lorenzo-Luaces et al., 2016), the SIT program also encouraged individuals to raise awareness of their maladaptive schemas and biases, modify automatic thought patterns, and engage in more adaptive behaviors to shape their environment. From a translational science standpoint, the SIT curriculum provides knowledge of important social processes, such as neuroplasticity, brain development, cognitive biases/schemas, communication techniques, (epi)genetics, and attachment styles. Building on previous research showing improved interpersonal skills (Castro et al., 2019), our findings showed that the SIT program also reduces mean negative emotions and buffers against daily stressor-related decreases in positive emotions, which may have implications for the treatment of mental health disorders and management of chronic physical health conditions (Mroczek et al., 2013; Sin et al., 2015; Zhaoyang et al., 2019).

This study had numerous strengths. First, both the SIT program and AC were delivered online, without manualized protocols or the need for labor-intensive and expensive clinical personnel for group and one-on-one treatments. Leveraging advances in technology, the self-guided SIT program can help bypass treatment barriers (e.g., stigma, high costs) and easily scale to address the unmet needs of individuals and communities in a sustainable way (Wilhelm et al., 2020). Second, intervention effects were evaluated on emotion regulation processes (i.e., stressor-reactivity and uplift-responsiveness) via online daily surveys. These methods allowed us to capture “life as it is lived,” providing a snapshot of improvements in day-to-day emotional dynamics following the intervention (Hamaker & Wichers, 2017). Electronic verification of program completion and the timing of survey completion increase confidence in the validity of the observed intervention-related improvements. Third, the validity of intervention effects is further strengthened by several methodological rigors, such as using multilevel models with full information maximum likelihood for intent-to-treat analyses, and reporting on baseline equivalency and no differential attrition across conditions (Gupta, 2011; Steeger et al., 2021). A fourth strength was inclusion of an AC that was comparable to the SIT program in delivery format, amount of content, and course structure. This supports the notion that the beneficial changes observed in the SIT group were not simply a result of expectancy effects, attention, and passage of time. The automated, self-guided delivery of both programs minimizes “clinician” or other nonspecific factors as explanations for the findings, and provides guarded confidence for the theoretically “active ingredients” within the SIT program (MacCoon et al., 2012; Moos, 2007). Fifth, delivery of the SIT program gives future hope/guidance for its ability to reach the greater population, especially those in middle and older age who may be vulnerable to a lack of social engagement and support. For example, the COVID-19 pandemic has laid bare the fragility of numerous aspects of life, such as relationship quality and functionality, well-being and health of oneself and those around us, and economic, social, and psychological consequences of policy decisions (Infurna et al., 2021; VanderWeele, 2020). With its focus on curbing automatic biases and humanizing other people, the SIT program can provide avenues for broad utilization across population segments and the reduction of mental health disparities evident across race, ethnicity, biological sex, psychological gender, and sexual orientation.

This study had several limitations to consider. First, our sample comprised predominantly White, highly educated, middle-aged community-dwelling adults who were fluent in English and had access to the internet. As such, we cannot comment on the applicability of findings to individuals from different income levels, educational or ethnic-racial backgrounds. Second, a different pattern of findings may have emerged with a more constrained “midlife” age range rather than including participants from 39 to upwards of 70 years of age. Third, given that intervention effects on daily functioning were assessed immediately following program completion, it is unknown whether SIT program benefits were sustained over time. Although promising, inferences of the program’s effects should be limited to what was measured in this study: daily well-being and emotion regulation prior to and after program completion. A fourth limitation is that all measurements were self-reported; additional data from confidants could overcome potential common method variance and reveal how individuals may have used principles they learned from the SIT program with long-standing and new relationships.

This RCT showed that an online, self-guided training program (8 hours) in social intelligence improved midlife adults’ daily well-being and emotion regulation. Combined with prior research (Castro et al., 2019), these findings indicate that the SIT program reduced daily negative emotions and increased daily social connection, emotional awareness, and perspective-taking, which are strong correlates of well-being, health, and resilience (Boden & Thompson, 2015; Galinsky et al., 2005; Seppala et al., 2013). Collectively, our findings show that the SIT program enhanced the capacity to maintain social connection and preserve positive emotions when faced with daily stressors (i.e., lower reactivity), and keep negative emotions low without relying on daily uplifts (i.e., lower responsiveness), which are key emotion regulation processes underlying well-being, health, and resilience (Grosse Rueschkamp et al., 2020; Khazanov et al., 2019; Mroczek et al., 2013; Ong & Leger, 2021). Given that both interventions and surveys were completed entirely online, this study addressed a gap in the intervention literature by examining effectiveness in “real-life” settings (Curran et al., 2012). The potential impact of public health interventions that are widely accessible, affordable, and easy to implement is substantial (Wilhelm et al., 2020). We hope these findings encourage future efforts to boost the resilience of individuals and their communities through affordable technology-based programs.

Funding

This research was supported by the National Institutes of Health (R01AG26006) to Alex J. Zautra and John Hall and National Institutes of Health (R01AG048844) to Alex J. Zautra and Frank J. Infurna.

Declarations

Ethics Approval

This study was approved by the Arizona State University Institutional Review Board, and conducted following ethical standards outlined in the Declaration of Helsinki.

Informed Consent

Informed consent was obtained from all study participants.

Conflict of Interest

Castro (current director) and Zautra (ex-director) served in the not-for-profit company that provides the social intelligence training evaluated in this paper. These roles are entirely honorary, and the authors do not receive financial compensation for these roles.