Social Connection, Loneliness, and Social Burdens as Experienced by Participants in the WELL for Life Project — Steven Crane’s Master’s Thesis
Social Connection, Loneliness, and Social Burdens as Experienced by Participants in the WELL for Life Project
Steven Crane | Master’s Thesis in Community Health and Prevention Research | 2019
Stanford University School of Medicine
Catherine Heaney, MPH, PhD
Associate Professor (Teaching) of Psychology and of Medicine (Stanford Prevention Research Center)
Tamara Sims, PhD
Behavioral Scientist, Fidelity Health Solutions
Kelly Young-Wolff, PhD
Research Scientist, Division of Research, Kaiser Permanente
Background and objectives: Social connectedness is a key component of health and well-being, yet certain subdomains of social connectedness deserve closer attention in certain populations. This report examines three subdomains of social connectedness in different age and gender groups.
Research Design and methods: Participants (n=4043) completed an online survey as part of the Stanford WELL for Life project. The survey included 13 items tapping social connectedness. Factor analysis was used to construct three subscales from these items. Linear regressions were used to evaluate associations between age and scores on these subscales among all participants and stratified by gender. Transcripts from semi-structured interviews (n=102) in which participants described periods of high and low well-being were reviewed in order to further illuminate the regression results.
Results: Three sub-domains of social connectedness were identified: social support and belonging; loneliness; and social burdens. Female gender was significantly linked to greater experiences of social burdens. Qualitative data from interviews extended this finding: women were more likely to discuss experiences of being drained by helping care for others (particularly family) than men, though men talked more about difficulty in meeting others’ expectations. Older age was significantly linked with lower experiences of loneliness and more social connectedness regardless of gender.
Discussion and implications: Studies of social connectedness and well-being may benefit from including measures of social burdens. Interventions to improve the well-being of those with low social connectedness might focus on high loneliness among young people and high social burdens among women.
Social connection is a central element of a well-lived life. In the first century BC, Lucretius wrote, “We are each of us angels with only one wing, and we can only fly embracing one another” (Trombley, 2016). Indeed, the importance of concepts such as friendship, social support, social connection, healthy familial ties, bonding, attachment, and partnership in the sociological, psychological, and public health literatures is widely evident. The ties that bind us together form a lattice that supports our growth, health, and longevity; thus, special attention must be paid when those ties are frayed or strained.
Social connectedness and health
It has been well established that social connectedness is protective against a number of specific conditions including depression (Sarason, Sarason, & Gurung, 2001), heart disease (Sorkin, Rook, & Lu, 2002), and cognitive decline (Seeman, Lusignolo, Albert, & Berkman, 2001) as well as all-cause mortality (Holt-Lunstad, Smith, & Layton, 2010). Indeed, Berkman and Syme’s (1979) influential early work studying nearly 7,000 residents of Alameda County, California over 9 years found that those who were least socially connected tended to die more than 2 years earlier than those who were more socially connected, even when controlling for socioeconomic status, initial health status, and health habits. House et al. (1988) established that the number and strength of one’s social ties are as powerful or more powerful predictors of health and longevity than smoking, blood pressure, blood lipids, obesity, and physical activity. Looking at the subjective feeling of loneliness, Holt-Lunstad and Smith (2012) compare the effect of loneliness on mortality to smoking 15 cigarettes per day, a finding that was anticipated 30 years ago by House et al. (1988) when they wrote, “[the] evidence on social relationships and health increasingly approximate that available at the time of the U.S. Surgeon General’s 1964 report on smoking and health, with similar implications for future research and public policy.”
More recent work has warned of a “loneliness epidemic” (Holt-Lunstad, 2017) in many Western countries that deserves high prioritization from a public health perspective (Holt-Lunstad, 2018). There is general concern about loneliness throughout the lifecourse including among the Millennial Generation (YouGov, 2019) as well as among adults over age 50, for whom there is sufficient concern about loneliness to merit an ongoing consensus study from the National Academies of Sciences, Engineering, and Medicine (2019). These studies continue to point to reliable evidence that social disconnection is among the most powerful drivers of ill health and medical spending, with $6.7 billion in US federal spending alone attributable to social disconnectedness in older adults (Health Resources & Services Administration, 2019).
Loneliness over the life course
In recent polls, about 20% of Americans report “always” or “often” feeling lonely, but this number is not constant across age groups. When comparing three generations, 30% of Millennials (born 1981–1996) reported loneliness and reported having the fewest friends and acquaintances when compared to Generation X (born 1965–1980, of whom, 20% report loneliness) and Baby Boomers (born 1946–1964, of whom, 15% report loneliness) (YouGov, 2019). Similarly, in New Zealand, adolescence is the peak age for experiencing feelings of loneliness (New Zealand Ministry of Social Development, 2016).
One possibility for this elevated loneliness among younger age groups is differences in the use of technology. Millenials and younger generations are “digital natives” to whom internet-connected devices are second nature. They use their digital devices more frequently and across a greater range of daily activities than older generations, and are more dependent on them, feeling anxious when disconnected (Chesley & Johnson, 2014). Time on mobile devices in adolescents has also been linked to decreases in time with parents in a longitudinal US sample of children aged 13–18 (Lee, 2009). However, greater loneliness among younger age groups was also observed before widespread use of internet technologies (Schultz & Moore, 1988), so there may be broader drivers of loneliness among adolescents and college-aged people, such as conflict with and separation from family as they become more independent and isolation that comes from leaving home to enter new social contexts.
Later in life, relationships center around family ties, particularly marriage (Williams & Umberson, 2004) and birthing and raising young children (Hynes & Clarkberg, 2005), activities which can be very socially connecting, but which may also bring new stressors to one’s social life. One might also suspect that as age increases past middle age that loneliness would increase, and indeed this assumption has good face validity given the life transitions that happen in these age groups: retirement, empty nest syndrome, poorer physical health, and the deaths of network members (Yang & Victor, 2011). Certain measures of social connectedness indeed show declines with age: smaller network sizes, less closeness to many network members, and fewer distal group ties (Cornwell, Laumann, & Schumm, 2008). About 21% of men and 34% of women over age 65 live alone, as well (ACL Administration for Community Living, 2019).
Nevertheless, elderly populations in the United States, on average, are not lonelier than younger groups. AARP surveys consistently show that, even controlling for other factors, as adults get older, they become less lonely than those in middle age, as measured by the UCLA Loneliness Scale (Anderson, 2010; Thayer & Anderson, 2018). Specifically, the 45–49 year old demographic reported loneliness 43% of the time compared to 25% in those over 70 years old, as reported in their 2018 work. This could be, in part, due to the fact that older age groups do more socializing with neighbors, participating in religious institutions, and volunteering (Cornwell et al., 2008). This paradox may also be due to the tendency for older people (who generally perceive shorter time horizons) to deepen their existing social networks to focus on meaning and fulfillment while younger people tend to expand their social networks to focus on exploration and learning, as predicted by socioemotional selectivity theory (L L Carstensen, 1992).
This trend toward less loneliness with age is not without its exceptions, however. Some fail to find much of an effect of age on loneliness, such as a report from the UK that describes a nearly flat trend of loneliness with age (Mental Health Foundation, 2010). So there is no consensus on the effect of age on loneliness, particularly when comparing samples across different countries and cultures as was done by Yang and Victor (2011).
Social connectedness and well-being
Social connectedness affects more than our physical health and longevity; it plays a central role in our well-being, broadly considered. Perceived social support availability can predict overall life satisfaction and feelings of well-being as people live through college experiences (Demakis & McAdams, 1994). Social connectedness is also a topic that people associate with overall well-being. A previous study (Heaney et al., 2019) by the WELL for Life team at Stanford University prompted interviewees to discuss times in their adult lives when they were experiencing particularly high or particularly low levels of well-being, and these participants spontaneously mentioned social connectedness as a component of both types of experiences more than any other domain of well-being (Heaney et al., 2019). Specifically, nearly all (97%) of the participants mentioned social connectedness at least once in their discussion of well-being; by comparison, the next most common domains were lifestyle and daily practices (e.g. diet, sleep, and exercise) mentioned by 88% of participants and experience of emotions mentioned by 87%. When the interview data is analyzed a different way, 25% of the 4,540 data elements (i.e. meaningful excerpts of text taken from the interview transcripts) related to social connectedness. Lifestyle and daily practices were mentioned in 17% of data elements and physical health was mentioned in 12%. Figure 1, below, displays the relative frequency with which data elements corresponded to all ten domains of well-being identified in this earlier work.
Figure 1. WELL for Life Flower. This figure displays proportions of data elements corresponding to 10 domains of well-being (Heaney et al., 2019).
The valence of the relationships between social connectedness and well-being, however, is mixed. Almost as frequently as participants mentioned the contribution of social connectedness to their well-being, they also mentioned social difficulties that detracted from their well-being. As subsequent illustrative quotes in this study will show, our social relationships can also be the source of some of our greatest sorrows.
One such aspect of social difficulties is that of social burdens as experienced by individuals. The present work defines social burdens as feelings of 1) being drained by helping care for others, 2) being upset by others, and 3) struggling to meet the expectations of others (see Table R1 for full question text). Understanding these negative aspects of social relationships will help us better understand the nuances of the relationship between social connectedness and overall health and well-being.
There is evidence that these social burdens may differ by gender and by age group. For example, in the context of formal caregiving for a child or sick or elderly family member, women bear the brunt of caregiving burdens. In the case of Alzheimer’s, two-thirds of caregivers are women, wives are more likely to care for husbands than the other way around, and daughters are 28% more likely than sons to care for elderly parents. Furthermore, evidence suggests that caregivers who are women experience slightly higher levels of overall burden and depression than caregivers who are men (Alzheimer’s Association, 2016; Bott, Sheckter, & Milstein, 2017).
The topic of burdens born by professional or lay caregivers (self-identified or identified by researchers) has been extensively studied, but this work represents a relatively narrow scope, neglecting those who don’t see themselves as caregivers in a traditional sense. Indeed, some exposure to these social burdens is a common human experience, e.g. disagreements with supervisors at work, marital conflict, or arguments with one’s parents. These experiences bring stress, and with chronically elevated social stress comes poor health (Kiecolt-Glaser & Newton, 2001; Sapolsky, 2007).
The present study
The present study investigates the question of how different ages and genders experience different facets of social connectedness, specifically social support and belonging, loneliness, and social burdens. With respect to loneliness, this study aims to contribute to our understanding of how loneliness patterns differ across age groups, in part heeding a call for more research into the sources of loneliness in middle-aged adults (Luhmann & Hawkley, 2016). With respect to social burdens, this study describes the development of a scale that briefly measures socially burdensome experiences in the general population (beyond dedicated caregivers), encompassing everyday experiences of social difficulties with others. Finally, this study presents interview data that explores the experience of social burdens in participants’ own words.
Interview participants and qualitative methodology
Participants (n=102) were recruited in 2016 from a broad sampling of businesses, schools, and community organizations in the San Francisco Bay Area, which produced a diverse sample of community residents. Table 1 lists the participant demographics. In semi-structured individual interviews with a median length of 35 minutes (range: 20–90 minutes), participants were asked to describe a period of their adult lives when they were experiencing particularly high well-being, and later in the interview, they were asked to describe a period when they were experiencing particularly low well-being.
Recordings of the interviews were transcribed, and research staff inductively coded the interview transcripts using NVivo qualitative analysis software. Data elements (i.e. excerpts from the transcripts) were assigned a code, and through iterative coding and discussion, the research team identified ten domains of well-being represented by the participants in their interviews. The methodological details of this work have been described in Heaney et al. (2019).
For the present study, two research assistants compiled and re-examined data elements that had been previously coded to pertain to the well-being domain of social connectedness — including both contributors and detractors from participants’ overall well-being. The research assistants extracted data elements that indicated an experience of a social burden, specifically data elements that pertained to the three survey questions that comprise the social burdens subscale (SBS — see Table R1): feeling drained by helping, feeling that others upset you, and feeling difficulty in meeting others’ expectations. They independently analyzed each data element to detect potential mentions of social burden, and, if relevant, coded each element into one or more of the appropriate social burden sub-codes. The author reviewed these codings and any discrepancies were resolved through discussions to ensure consistency in the application of the codes.
After coding decisions were finalized by group consensus, the frequency of the three main categories of social burdens was measured as a percentage of the data elements, with sub-reports by males and females (no interview participants identified their gender as other than male or female). The three social burden categories were also measured as a percentage of interviewees who mentioned that social burden at any point in the data elements identified. A chi square test for independence was run comparing males and females across each social burden type. Finally, the author synthesized impressions gained through reading all the data elements to describe key themes, and identified illustrative quotes for each theme.
Survey participants and quantitative methodology
The United-States-based sample of participants (n = 4,044) aged 18 years or older was recruited using a variety of community engagement strategies, mostly focused on participants in the San Francisco Bay Area. Research staff conducted a community partner program with local schools, businesses, and other community organizations to establish long-term partnerships in which aggregated participant data was shared with the organization. Participant characteristics are shared in Table 2.
Participants remotely accessed the Stanford Well for Life Scale (SWLS) over the internet. The survey is hosted online at wellforlife.stanford.edu (see Appendix A). The thirteen questions from the SWLS that pertain to the domain of social connectedness are listed in Table R1. Participants responded to the prompt, “During the last two weeks, how often did you feel…” by indicating the frequency with which they experienced states corresponding to the following questions using a 1–5 scale: 1 — Never; 2 — Almost Never; 3 — Sometimes; 4 — Often; 5 — Very Often.
The first activity of the present research project was to perform a principal component analysis of the 13 social connectedness items using varimax rotation, and factors with eigenvalues above 1 (3 in total) were kept and used as the basis for a factor analysis. These 3 factors produced 3 subscales, each comprised of items with similar factor loadings. Those sub-scales are presented in Table 3.
The subscales were examined with Cronbach Alpha for Internal Consistency. Furthermore, a t-test was performed to compare subscale scores between men and women. Subscale scores were plotted by age and then by age separately for men and women. Multivariate linear regression was used to examine age, gender, race, ethnicity, work status, and education as predictors of each of the subscales. All plots and statistical analyses were produced in R Studio version 1.1.463 (Rstudio, 2019).
Principal component and factor analysis
Table 3 presents the factor loadings with the full text of each item. Cronbach alpha scores are also reported for each subscale created from the factors, the Social Support and Belonging Subscale (SSBS), Loneliness Subscale (LS), Social Burdens Subscale (SBS). Each item was thus scored on a 1–5 scale, and each subscale was calculated as an average of its component items. Thus, higher scores on the SSBS indicate more social connectedness, and higher scores on the LS and SBS indicate more loneliness and social burdens, respectively.
Subscale Subscale means, medians, standard deviations, and Spearman’s rank correlations are presented in Table 4.
Subscale results by age
Figures 2–4 display subscale scores across the age continuum.
Figure 2: SSBS results by age. The figure displays a smoothed line with standard error indicated in gray.
SSBS scores rise and fall in the 18–55 year age group, and generally increase from age 55 onward. The slight downturn in the oldest age group and the wide confidence interval there may be due to low numbers of participants in that category.
Figure 3: LS results by age. The figure displays a smoothed line with standard error indicated in gray.
LS scores generally decrease in older age groups from their peak in the youngest age groups.
Figure 4: SBS results by age. The figure displays a smoothed line with standard error indicated in gray.
SBS scores fall then rise again in the 18–40 year age group, and generally decrease from age 40 onward.
Subscale results by age and gender
To reveal whether the trends reported above in subscale score differed by gender, the figure below displays the trends for each subscale, with men (red line) and women (teal line) plotted separately.
Figure 5. Plot of SSBS by age, separately for males (red) and females (teal); standard error indicated in gray. In certain age groups, women appear to have greater scores on the SSBS.
Though there is broad overlap in the standard error between the two genders, women’s average SSBS scores exceed those of men, except around age 40.
Figure 6. Plot of LS by age, separately for males (red) and females (teal); standard error indicated in gray. In certain age groups, women appear to have lower scores on the LS.
Though there is broad overlap in standard error, women’s average LS scores are usually lower than or equal to those of men, except around age 50–60 during which they are higher.
Figure 7. Plot of SBS by age, separately for males (red) and females (teal); standard error indicated in gray. In certain age groups, women appear to have greater scores on the SBS.
Women’s average SBS scores generally exceed those of men, especially during ages 35–45, until after age 60 when they are largely overlapping with those of men.
T-test for subscales
T-test results indicate that without controlling for other factors, women, on average, report higher SSBS and SBS scores and lower LS scores. Specifically, SSBS scores are slightly higher in women (X̄ = 3.98) than men (X̄ = 3.93; t = -1.84) at a level approaching significance at the p < 0.05 level (p = 0.067). SBS scores are higher in women (X̄ = 2.63) than men (X̄ = 2.51; t = -4.42; p < 0.001). And LS scores are lower in women (X̄ = 2.36) than men (X̄ = 2.44; t = 2.57; p = 0.01).
Linear regression for all subscales
The following multivariate linear regression models were used to see if differences by gender and age held true after controlling for covariates (race/ethnicity, work status, and education level). Tables 5–7 display results for demographic predictors of each subscale score. They show the parameter estimates and adjusted r-squared values for men only, women only, and both groups combined.
Using linear models that control for the influence of race/ethnicity, education, and work status, the relationships between age and SSBS look more different for men and women than that portrayed in Figure 5, which shows broad overlap between the two groups. Compared to the 18–19 year-old reference group, men aged 30–70 score lower and women aged 60–80 score higher on the SSBS. Otherwise, age does not appear to be a strong predictor of SSBS scores.
Scores on the LS show a stronger negative relationship with age among women than among men, however the direction of the trend is similar and it may show statistical reliability if the number of men included (in this case, 983) were more similar to the number of women sampled (2,497). In women, compared to the 18–19-year-old group, those aged 20–60 show a decrease of roughly constant magnitude, and then there appears to be a further drop in score in the 60–80-year-old group.
SBS scores do not show any strong trends with age until age 60–80 in women, when they then show a marked decline. Regardless of age, men’s scores on the SBS are lower than those of women after controlling for the factors listed above. For all regressions, examination of residual plots suggested sufficient randomness that the assumptions of linear regression were not violated.
T-test for social burden items
Because Figure 7 and Table 7 both indicated a difference in social burden experience by gender, we here present t-test results comparing the three items that comprise the SBS by gender. Scores on “I feel drained by helping,” were higher in women (X̄ = 2.64) than men (X̄ = 2.39; t = -6.34; p < 0.001). Scores on “people upset me” were also higher in women (X̄ = 2.69) than men (X̄ = 2.58; t = -3.80; p < 0.001). Scores on “hard to meet expectations” were equal between genders (X̄ = 2.55 for both). Given these differences, we then interrogated the qualitative data to see how people talked about their experiences of social burdens.
Types of social burdens from qualitative data
In total, 937 data elements (excerpts from interviews) from the full set of interview transcripts related to the domain of social connectedness. Of those, 204 data elements (22%), representing 73 interviewees, were coded as including a social burden. Three common types of social burdens were described: (1) being drained by caring for others, (2) other people upsetting you, and (3) difficulty in meeting others’ expectations. Specifically, 61 data elements (7% of all data elements related to social connectedness) related to being drained by caring for others, 85 data elements (9%) related to others upsetting the interviewee, and 100 data elements (11%) related to difficulty in meeting others’ expectations. The sum of these percentages exceeds 22% because many data elements were coded as describing more than one type of social burden. Because these data were collected from a smaller number of participants (n = 102) and certain age categories had small n, interpretations of the social burden sources presented below are not stratified by age.
The survey questions did not reference specific sources of social burdens, however the qualitative data allows us to look at the types of relationships people discuss when speaking about these various types of social burdens. All data elements that pertained to social burdens were reviewed to identify key themes and patterns. Among sporadic mentions of less frequent topics such as broader fit with community, keeping up with expectations of professors, feeling ostracised from a religious congregation, etc., three major sources of social burdens emerged: family relationships, workplace relationships, and friendship relationships. These sources of social burden are described in more detail below along with illustrative quotes.
Source 1: family relationships
Familial relationships contribute to the vast majority of social burdens recounted by the participants. Participants frequently mentioned stress related to parenting, especially times when they felt frustrated by their children As one father recalled:
I guess I was angry at our kids for not understanding the importance of working a little harder at school to do well. They had success, but they were passing and that was ok, but it’s like, “No, I think you might want to do a little better if you put in a little more work,” that sort of thing. So, being angry and upset, that I guess I would describe that.
— 69 year old male
Stories of break-ups, divorces, and separations were also commonly shared, with many participants describing periods of conflict with or mental illness in one’s partner. Interestingly, even single data elements sometimes described parallel experiences of social lows alongside highs in other domains of well-being. For example, one participant described falling out with siblings and a romantic partner while also being at a peak of physical well-being.
Interestingly enough, it was at the time when my sibling was not talking to me. So even though I was running, and physically was really good, emotionally it was a really hard time. And I also was going through heartbreak, and a breakup and all this other stuff. So I guess the running kind of helped me get through it to a degree. So it was a really difficult emotional time, but a really good physical time.
— 41 year old female
Notably, many participants experienced parenting and partner conflict simultaneously, exacerbating the experience of social burden attributable to family relationships. The following quote encompasses many of these common themes at once: partner conflict, exasperation with circumstances, and time pressure due to children.
I rarely went out, but one time I went to a bridal shower and came back at 11:30 PM. [My husband] was so mad at me… I didn’t have much of a life because it was just the kids and me, and I didn’t get out that much… I don’t know what he thought I was doing. But he was yelling at me. He did everything but hit me. But I know that’s what he wanted to do. I’d just sit there, “I can’t do this anymore,” because he had no respect for me.
— 66 year old female
Source 2: workplace relationships
A second common source of social burden was in the workplace, either from superiors, peers, or in the case of this woman, interpersonal interactions and empathetic burdens that come with being a teacher.
Given the nature of the work that I do, is that my sense of joy is very tied to the experience of my students. And so when they’re going through tough times, I’m going through tough times.
— 41 year old female
Source 3: friendship relationships
Burdens from relationships with friends was the final source of social burden that emerged from the data. Participants described how supporting and empathizing with friends could feel draining. As one participant noted:
Of course, you have to be there as a friend. You have to listen to them about their problems, too. I remember there’s a lot of those that you have to be there as a friend to listen to them. It actually contributes to you too being depressed because somehow you help them gather some ideas. But in a way, it’s not good.
— 49 year old male
Social burden stemming from relationships with friends, peers, or classmates were especially common among younger participants, who are often navigating a transition from adolescence to adulthood, and encountering new identities and responsibilities that come with life in college. One young participant recounts the struggles of keeping up with academic expectations while juggling student organization responsibilities for which she feels underappreciated.
I also felt underappreciated for the work I was putting in to something. So this year was the year of unprecedented all-nighters, but I also felt like, you know, not sleeping for whatever reason, and bringing it back to whatever organization that I was in, and not feeling like I was appreciated for the amount of work I was putting into it. But also, at the same time… feeling that obligation to continue, even though I wasn’t feeling appreciated.
— 22 year old female
In sum, when asked to describe periods of high and low well-being in their lives numerous specific examples of social burdens shone through in their stories.
Social burdens by gender
Table 8 analyzes all data elements according to how frequently they mentioned the three types of social burden; data are also presented separately for men and women, and a chi square test for independence was performed for each category of social burden. Each participant had one or more data element, so the number of data elements (204) is greater than the number of participants who mentioned any social burden (n=73).
Results show that data elements from women were far more likely to mention being drained by helping others, and less likely to mention difficulty in meeting others’ expectations when compared with data elements from men. Data elements related to being upset by other people were comparable in frequency between men and women.
Table 9 presents the same information as Table 8, but counts each participant only once.
These results again show that women recounted the experience of being drained by helping others more often than men (49% of women compared to 32% of men), however, a chi square test for independence did not show the absolute numbers to be significantly different. In contrast to the results counted by participant, when counted by data element, difficulty in meeting expectations of other people was comparable among men and women.
In reviewing the data elements from the participant interviews, men and women both commonly shared stories of how their romantic partnerships created social burden. Similarly, there were no strong differences between men and women with respect to their mentions of their parents among their social burdens. However, women were more likely to mention their sons or daughters specifically (7 data elements) than men (1 data element).
Summary of findings
Unadjusted charts and t-tests display the following patterns: 1) SSBS scores mostly increase with age, and except for a drop around age 40, women score higher than men overall. 2) LS scores generally decrease with age, especially after age 55, though men have higher scores around ages 25–35 years old and women have lower overall scores than men do. 3) SBS scores are highest around age 18 and around age 40, though these peaks are driven by the female portion of the sample, who have higher SBS scores than men overall.
After adjusting for a number of covariates, results show that, compared to the 18–19 year old reference group, 1) SSBS scores are lower in men aged 30–70 (in marked contrast to the unadjusted results) and higher in women aged 60–80; 2) LS scores are lower in all age groups of women after age 20 (and especially so after age 60) and lower in men after age 50 (marginally significant at the p<0.1 level for age 50–59); 3) SBS scores are lower in women aged 60–80, but independently of age, men have lower scores than women. 4) The items that comprise the SBS demonstrate some differences by gender. In unadjusted t-test comparisons, women are more likely to report being drained by helping and being upset by others.
Data elements from interviews (with a smaller, separate participant sample) show that, overall, three sources of social burdens are most frequently discussed: family relationships (especially intimate partnerships), workplace relationships, and friendships. When examining categories of social burdens by gender, women talk more about being drained by helping, and men talk more about difficulty in meeting others’ expectations.
Social support and loneliness
Contrary to common stereotypes, but consistent with prior studies, social support and belonging was experienced to the greatest extent among those in the oldest age groups, and they were also the least lonely compared to any other age group. These results are in keeping with recent AARP surveys that report lower scores on the UCLA Loneliness Scale among those 75+ years old than those 45–49 years old (Anderson, 2010; Thayer & Anderson, 2018). This may be due, in part, to a trend described by Carstensen and Löckenhoff (2003) in which older age groups improve in their ability to find satisfying relationships. They write, “socioemotional regulation improves with age, which is associated with increased investment in emotionally meaningful others”. People may be heartened to know that long years of isolation and loneliness are far from an inevitable part of getting older: they are the exception. Future research could use the same data sets that underlie this article to investigate how different groups of elderly participants experience social connectedness and emotional experience, e.g. how do people over age 65 who continue working differ in their social connectedness compared to those who retire?
Additionally, a recent representative US sample showed the highest levels of loneliness among Millenials (when compared to older generations), 30% of whom always or often report feeling lonely (YouGov, 2019). This trend is echoed in our sample in which the youngest age groups are the most lonely, and this may be due to heavier use of social media than older groups, the relinquishment of which has been linked to decreases in loneliness and depression (Hunt, Marx, Lipson, & Young, 2018). Younger people (even teenagers) may want to find ways to invest in meaningful and supportive in-person relationships, which may entail less interaction on social media. Future research could investigate which practices best support the social connectedness of this age group. One option suggested by the present research is continuing to finish college and graduate school as those who did so in our sample experienced greater social connectedness and less loneliness.
Measuring and conceptualizing social burdens
This study began by identifying three sub-domains of social connectedness from within the larger SWLS, conceptualized as experiences of social support and belonging, loneliness, and social burdens. The SSBS and LS closely track existing sociological and psychological literature on the measurement and conceptualization of social support and loneliness, respectively. The SBS, however, represents a measure of social burdens that 1) extends beyond what is typically measured, and 2) is applied to a population that is not typically recognized as experiencing social burdens from caregiving. The three items that comprise the SBS were created for the SWLS based on data from participant interviews. A literature review of other measures of social burdens found that most measures are aimed specifically at professional or family caregivers (such as the Scale for Positive Aspects of Caregiving Experience (Grover, Nehra, Malhotra, & Kate, 2017), The Burden Assessment Schedule (Thara, Padmavati, Kumar, & Srinivasan, 1998), the Zarit Burden Interview Questionnaire (Novak & Guest, 1989), and the Perceived Family Burden Scale (Levene, Lancee, & Seeman, 1996)). These all revolve around the experience of caregivers as viewed through the lens of being a caregiver, which narrows the scope of this research and neglects those who don’t conceptualize themselves as caring for someone in a traditional way.
Though applicable to caregiver populations, the SBS also measures important aspects of social life that go beyond specific caregiving roles. Being drained by helping others, being upset by others, and finding it hard to meet expectations of others are experiences that we all have from time to time that can have detrimental effects on our overall well-being, and thus the description of these items and their application to a general population constitute a contribution to the overall conceptualization of well-being.
The importance of social burdens notwithstanding, the experience of these burdens is often excluded from the consideration of even the broadest investigations into individual social relationships and overall well-being. For example, a broad review of data from the World Values Survey, the US Benchmark Survey, and similar surveys in Canada makes no mention of social burdens despite considering social connectedness in very wide-ranging terms (Helliwell & Putnam, 2004). Future research conceptualizing social connectedness and well-being would do well to ask about the experience of social burdens with the items presented here plus further measures for other as-yet underappreciated aspects of social difficulties.
Insights from participant interviews complemented the quantitative survey results by providing examples of the types of social burdens experienced, which may hint at the driving causes of social burdens. We learned that social burdens were very frequently mentioned when participants discussed their well-being in the domain of social connectedness (22% of all data elements), and that all three sub-components of the SBS were widely represented among these data elements. These results demonstrate that just as positive and connecting social experiences are important for bolstering well-being, negative or burdensome social experiences are prominent in interviewees’ minds when recounting periods of low well-being.
These social burdens were further explored through a synthesis of themes that emerged from reading the data elements taken from interview transcripts. Though no obvious thematic differences emerged between men and women (except that women were more likely to mention their children), some themes emerged around the source of the social burden: the family environment was most commonly mentioned (originating from partners, children, or parents), followed by friendships and workplace relationships. Within these categories, experiences recounted by participants were diverse, dealing with everything from struggling with sharing one’s sexual identity with one’s parents, financial burdens from supporting one’s siblings, breaking up with high school romantic interests, sour relationships with supervisors and work, to experiencing racism. These examples help give insight into the types of real-world experiences that might be reflected in the survey data.
Gender and social burdens
We also show how the experiences of these burdens differ by gender and by age group. Women have higher levels of social burden overall than men do. This may have to do with pressures from work, family, and caregiving responsibilities for children and elderly relatives. However, once people reach the age at which children are likely to be out of the house, it seems men’s and women’s experiences of social burdens and loneliness tend to converge and decline.
Interview data indicate: 1) that women more frequently reported being drained by caring for others across the board; 2) women also more frequently reported that others upset them, but men talked at greater length about these upsetting events; and 3) a similar pattern holds for the category of difficulty in meeting the expectations of others. The percent of men who mentioned these did not differ much from the percent of women who mentioned them, but when they were mentioned, men tended to talk at greater length and with greater variety about this type of experience. One might assume that these more extensive stories of these types of social burdens indicate greater importance and salience of these events to the interviewees.
Quantitative evaluations of these trends on the SBS items bolstered the qualitative findings: women were more likely to speak of being drained by helping care for others and scored higher on that survey item; men were more likely to speak of difficulty in meeting the expectations of others in their lives, and this may be the source of elevated SBS scores among men aged 30–39. These results reinforce previous literature that emphasize the extra caregiving burdens that women face: women are more involved than men in caring for their children (Craig, 2016) and ill or disabled relatives, including parents (Pavalko & Wolfe, 2016). Notably, however, it is women who show a reduction in social burdens after age 60, with a sharper drop that reaches statistical significance compared to that in men, in whom it does not. It’s possible that with a larger sample of men, this age would also show a reliable decrease in social burdens; in any case, future research might seek to understand better what psychological and lifestyle changes enable this lowering of burdens in older women. For example, compared to their younger, more socially burdened selves, older women who are free of obligations to care for aging parents or young children might feel relatively less burdened, appreciating the freedom that comes with older age more than men do.
The participant sample that provided survey responses is limited in certain ways. The sample was not drawn to be representative of any larger group, so results here would need to be replicated in other communities or in larger, representative samples in order to draw more generalizable conclusions. Further, there may be some uncontrolled confounders in this observational study (such as health status, income, or a greater tendency to be interested in wellness due to self-selection to participate) that could be influencing the results. Nevertheless, within the sample available, we control for education, race, and work status to attempt to demonstrate trends that differ by age and gender.
Because the participants did not provide any objective or highly structured data on their socially burdensome activities (e.g. number of hours per week caring for a family member, number of people that cause a social burden, etc.), we rely entirely on their memory and self-report to represent their experiences. Thus, it is theoretically possible that the differences we observe between men’s and women’s experiences of social burdens may reflect biases in memory or differences in the way men and women are socialized to think about and talk about social issues and their well-being. Nevertheless, because of the relative ease of collecting self-report data, and the close relationship between self-report and an individual’s mental state, self-report is commonly used in the social sciences and settings such as health care (reporting problems to a medical provider). In fact, self-assessed health status is comparable or superior to more intensive objective measures of health at predicting overall mortality (Kuhn, Rahman, & Menken, 2006).
Finally, the entirety of this article reports on cross-sectional data, and as such can only be used to generate hypotheses about the causes of social burdens rather than determine them. Nevertheless, because the qualitative data represent participants’ perceptions of the causes of their social burdens in their own language, these can strengthen our understanding of the survey data. As the WELL for Life project continues and the same participants return to retake the survey on an annual basis, eventually we will also have longitudinal data to better understand the role of growing older on our outcomes of interest in ways that aren’t biased by cohort effects. In any case, these results suggest a number of intriguing possibilities for experimental research that could establish causal mechanisms, as explored below.
The present study of loneliness and social burdens across ages and genders suggests a number of avenues for future research in addition to those mentioned above.
The qualitative aspects of the present study chose to focus on illuminating the differences between men and women with respect to social burdens. Perhaps equally interesting would be to apply the same methodology to the differences between young and old with respect to loneliness, which is nearly an entire point lower in the 75-year-old demographic compared to the 18–20-year-olds (Figure 6). The data elements that specifically mention aspects of loneliness might be particularly valuable to examine in the 18–20 year-old age group to better understand why they routinely show the worst overall scores on our measures of social connectedness. This could inform social debates on the role of social media vs. in-person connection and help craft policies — in universities, for example — to help support young adults as they transition to greater independence from their families.
Future research could also drill down deeper to better understand the experiences of social burdens in specific sub-groups. It may be the case that many people with the same level of responsibility and objective measures of caregiving (such as hours per week on similar caregiving tasks) do not perceive this “burden” in the same way; however, because our measures lacked specific questions about caregiving, parenting, or volunteering activities, we are unable to detect when these activities are happening but not registering as burdens to individuals. We could also examine the existing data to see under what circumstances individuals have high levels of well-being despite having high social burdens. What allows some people to experience social burdens yet not suffer for them? Furthermore, scores on measures of purpose and meaning may be inversely correlated with scores on the SBS (or, perhaps, the “drained by helping care for others” item specifically). Indeed, prior research has found that greater purpose and meaning is linked with the experience of being uplifted by providing care, which suggests that interventions that encourage cultivating a sense of meaning and purpose through one’s caregiving tasks may bolster overall health and well-being (Polenick, Sherman, Birditt, Zarit, & Kales, 2018).
Women, overall, experience more social burdens than men, yet women also show a decline in social burdens with older age, suggesting that it is younger and middle-aged women who show higher levels of burden (also suggested by figure 7). Again, if this finding were replicated with more widely representative samples, then this age group would be ripe for broader interventions aimed at reducing this burden. Most work on social support interventions does not address the sub-category of burdens directly. It also tends to focus on the elderly (Cattan, White, Bond, & Learmouth, 2005; Dickens, Richards, Greaves, & Campbell, 2011) or on more severe clinical conditions: substance abuse, cancer, or other medical and psychiatric conditions, though some does focus on parents (Hogan, Linden, & Najarian, 2002). Generally, these intervention trials lack rigorous experimental designs and have relatively short follow-up periods, so these would be areas of improvement for future studies. Such studies might feature interventions aimed at particular life transitions (e.g. having a first or second child, one’s children entering middle school or high school, or the death of a parent or parent-in-law) and could include support group interventions with other parents (with childcare included to make them feasible to attend) such as those in Telleen et al. (1989), interventions aimed at spouses or other family members to increase their social support, or even voluntary befriending schemes, such as those described by Cox (1993). This demographic seems to be squeezed from three sides — young children, elderly parents, and work — and thus deserves more focused study to better understand their predicament and policies and intervention programs to help relieve it. The recommendations for intervention research mentioned above might equally well apply to the other age group that experiences the least connection and the most loneliness: those 18–20 years old.
Finally, more research into the utility of including measures of social burdens into routine health and well-being assessments may prove fruitful for identifying those at risk for physical or mental health problems before they manifest as clinical conditions, which would save the individual and broader institutions or society the emotional and financial burdens of those conditions. Longitudinal study could use the SBS as a predictor of the likelihood of developing a clinical condition or all-cause mortality, and if it proves to be a strong predictor of these outcomes, it could become a commonplace screening tool used in workplace, governmental, or medical assessments of population health.
Social connections are among the most important parts of our lives. This study dissects the concept of social connectedness into its different components, and ultimately describes a pattern where 1) older age groups have the most social support and belonging, the least loneliness, and the lowest levels of social burdens; 2) women experience more social burdens than men overall; and 3) women are more often drained by duties of helping and caring for others in their lives than men. In this way, this study identifies certain groups that experience detriments to their social connectedness more than others, and suggests paths forward in both research and practice to explore the causes of and solutions to the frayed and strained social connections in these groups.
I want to add a note of heartfelt gratitude to a number of people who helped me along the way in producing this thesis. Cathy Heaney has been there from even before the start of this thesis, providing my introduction to health psychology over ten years ago! I was so happy that when I returned for my master’s program, Cathy agreed to be my advisor. A huge thanks for all your guidance, wisdom, and encouragement in this process. To Tamara Sims and Kelly Young-Wolff, thank you so much for heeding the call to read the thesis for basically a perfect stranger! Thank you for dedicating your valuable professional time to improving this thesis and helping me learn important parts of the academic writing process. To Jennifer Robinson and T.O. Preising, you are the heart and soul of CHPR, and your program organization and palpable care and concern for your students make this whole scene possible. The entire Stanford Wellness Living Laboratory (WELL) did basically all of the original research that this thesis rests upon, and it fills my heart with happiness to know each of you and to know that WELL is chugging along, every day, so that we may better understand human well-being. What a fantastic cause!
Autumn Albers, who has been my academic and life partner for over ten years, is unmatched in her kindness, generosity, and supportiveness to me and my life pursuits. You took our beloved toddler, Corey Albers Crane, many a morning and afternoon so I could dig in and keep the thesis momentum going, you provided thorough proofreading and content suggestions for the thesis, and your own incredible work ethic inspired me to keep pressing forward with my work even when I was tired. 100% of the writing of this thesis happened in my co-op, Red House. I don’t think I could ever find a more harmonious, beautiful home base from which to develop my own social connectedness and life passions. Finally, for anything of value I ever produce in my life, I must offer gratitude and love to my parents, Jeff Crane and Susan Crane. By always believing in me, encouraging me, and loving me, you made me who I am.
Participant registration on the WELL for Life Website
Berkman, L. F., & Syme, S. L. (1979). Social networks, host resistance, and mortality: a nine-year follow-up study of Alameda County residents. American Journal of Epidemiology, 109(2), 186–204. https://doi.org/10.1093/oxfordjournals.aje.a112674
Bott, N. T., Sheckter, C. C., & Milstein, A. S. (2017). Dementia Care, Women’s Health, and Gender Equity: The Value of Well-Timed Caregiver Support. JAMA Neurology, 74(7), 757–758. https://doi.org/10.1001/jamaneurol.2017.0403
Cattan, M., White, M., Bond, J., & Learmouth, A. (2005). Preventing social isolation and loneliness among older people: a systematic review of health promotion interventions. Ageing and Society, 25(1), 41–67. https://doi.org/10.1017/S0144686X04002594
Cornwell, B., Laumann, E. O., & Schumm, L. P. (2008). The social connectedness of older adults: A national profile*. American Sociological Review, 73(2), 185–203. https://doi.org/10.1177/000312240807300201
Dickens, A. P., Richards, S. H., Greaves, C. J., & Campbell, J. L. (2011). Interventions targeting social isolation in older people: a systematic review. BMC Public Health, 11, 647. https://doi.org/10.1186/1471-2458-11-647
Grover, S., Nehra, R., Malhotra, R., & Kate, N. (2017). Positive Aspects of Caregiving Experience among Caregivers of Patients with Dementia. East Asian Archives of Psychiatry : Official Journal of the Hong Kong College of Psychiatrists = Dong Ya Jing Shen Ke Xue Zhi : Xianggang Jing Shen Ke Yi Xue Yuan Qi Kan, 27(2), 71–78.
Health Resources & Services Administration. (2019). The “Loneliness Epidemic” | Official web site of the U.S. Health Resources & Services Administration. Retrieved from https://www.hrsa.gov/enews/past-issues/2019/january-17/loneliness-epidemic
Helliwell, J. F., & Putnam, R. D. (2004). The social context of well-being. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences, 359(1449), 1435–1446. https://doi.org/10.1098/rstb.2004.1522
Holt-Lunstad, J. (2017). The potential public health relevance of social isolation and loneliness: prevalence, epidemiology, and risk factors. Public Policy & Aging Report, 27(4), 127–130. https://doi.org/10.1093/ppar/prx030
Holt-Lunstad, J. (2018). Why social relationships are important for physical health: A systems approach to understanding and modifying risk and protection. Annual Review of Psychology, 69, 437–458. https://doi.org/10.1146/annurev-psych-122216-011902
Hunt, M. G., Marx, R., Lipson, C., & Young, J. (2018). No more FOMO: limiting social media decreases loneliness and depression. Journal of Social and Clinical Psychology, 37(10), 751–768. https://doi.org/10.1521/jscp.2018.37.10.751
Hynes, K., & Clarkberg, M. (2005). Women’s employment patterns during early parenthood: A group-based trajectory analysis. Journal of Marriage and Family, 67(1), 222–239. https://doi.org/10.1111/j.0022-2445.2005.00017.x
Kuhn, R., Rahman, O., & Menken, J. (2006). Survey Measures of Health: How Well Do Self-Reported and Observed Indicators Measure Health and Predict Mortality? — Aging in Sub-Saharan Africa — NCBI Bookshelf.
Lee, S. J. (2009). Online Communication and Adolescent Social Ties: Who benefits more from Internet use? Journal of Computer-Mediated Communication, 14(3), 509–531. https://doi.org/10.1111/j.1083-6101.2009.01451.x
Levene, J. E., Lancee, W. J., & Seeman, M. V. (1996). The perceived family burden scale: measurement and validation. Schizophrenia Research, 22(2), 151–157. https://doi.org/10.1016/S0920-9964(96)00071-0
New Zealand Ministry of Social Development. (2016). Loneliness: The Social Report 2016 — Te pūrongo oranga tangata. Retrieved from http://www.socialreport.msd.govt.nz/social-connectedness/loneliness.html
Polenick, C. A., Sherman, C. W., Birditt, K. S., Zarit, S. H., & Kales, H. C. (2018). Purpose in life among family care partners managing dementia: links to caregiving gains. The Gerontologist. https://doi.org/10.1093/geront/gny063
Sarason, B. R., Sarason, I. G., & Gurung, R. A. R. (2001). Close personal relationships and health outcomes: A key to the role of social support. Personal Relationships: Implications for Clinical and Community Psychology.
Seeman, T. E., Lusignolo, T. M., Albert, M., & Berkman, L. (2001). Social relationships, social support, and patterns of cognitive aging in healthy, high-functioning older adults: MacArthur studies of successful aging. Health Psychology, 20(4), 243–255. https://doi.org/10.1037//0278-6220.127.116.11
Sorkin, D., Rook, K. S., & Lu, J. L. (2002). Loneliness, lack of emotional support, lack of companionship, and the likelihood of having a heart condition in an elderly sample. Annals of Behavioral Medicine, 24(4), 290–298. https://doi.org/10.1207/S15324796ABM2404_05
Telleen, S., Herzog, A., & Kilbane, T. L. (1989). Impact of a family support program on mothers’ social support and parenting stress. American Journal of Orthopsychiatry, 59(3), 410–419. https://doi.org/10.1111/j.1939-0025.1989.tb01676.x
The National Academies of Sciences, Engineering, and Medicine. (2019). The Health and Medical Dimensions of Social Isolation and Loneliness in Older Adults : Health and Medicine Division. Retrieved from http://www.nationalacademies.org/hmd/Activities/PublicHealth/IsolationandLonelinessinOlderAdults.aspx
Williams, K., & Umberson, D. (2004). Marital status, marital transitions, and health: a gendered life course perspective. Journal of Health and Social Behavior, 45(1), 81–98. https://doi.org/10.1177/002214650404500106
© Steven Michael Crane, 2019. Please contact the author for approval for use of this work (it will almost certainly be granted!). email@example.com or firstname.lastname@example.org.