Felt understanding as a bridge between social identity and wellbeing among international university students
Abstract
Wellbeing issues among international students in the UK higher education have been recognised as a crisis. To address this, we integrate social identity and felt understanding approaches to wellbeing and mental health, testing whether felt understanding (the belief that others understand oneself) is an important process through which social identity predicts better wellbeing, over and above other, more established mediators (social support, life meaning, and personal control). International university students (including both undergraduates and postgraduates, N = 301) completed an online survey that measured three sets of variables: social identity variables (ingroup identification, multiple identities, multiple identity compatibility); process variables (social support, felt understanding, life meaning, personal control); and wellbeing outcomes (e.g., depression, anxiety, stress). Path analyses confirmed that felt understanding predicted better wellbeing outcomes over and above the other mediators. Additionally, indirect effects from social identity variables to wellbeing via felt understanding were consistently significant, even when adjusting for the other mediators. The results are consistent with the idea that felt understanding is an under-acknowledged resource through which social identities protect wellbeing. The findings contribute to “social cure” research and have implications for promoting wellbeing services from the perspective of group memberships. Please refer to the Supplementary Material section to find this article's Community and Social Impact Statement.
1 INTRODUCTION
The recent, rapid increase in mental health and wellbeing challenges among university students has been recognised as a “crisis” (Guardian News and Media, 2019). This crisis is particularly acute for international university students, who can experience additional stressors such as homesickness due to isolation from family and home culture (Poyrazli & Lopez, 2007), and marginalisation from the wider student body because of miscommunication and cultural differences (Cross, 1995), or “culture shock” (Hsieh, 2006), which can lead to a sense of feeling misunderstood by others, and of social isolation.
To help address some of the social factors that protect or undermine wellbeing, in this study, we tested whether the experience of feeling understood by others (felt understanding; Livingstone, Fernández Rodríguez, & Rothers, 2020; Oishi, Akimoto, Richards, & Suh, 2013; Reis, Lemay, & Finkenauer, 2017) could predict better wellbeing among international university students. Building on research on the so-called “social cure” offered by social identities (Haslam, Jetten, Cruwys, Dingle, & Haslam, 2018), our main aim was to test whether felt understanding can predict wellbeing outcomes over and above other mediators (i.e., social support, life meaning, personal control) suggested in the social identity and health literature. Second, we tested whether the positive effect of social identity on wellbeing operates in part via felt understanding in general social relations.
2 SOCIAL IDENTITY AND WELLBEING: SOCIAL CURE AND CURSE
The protective role of social identities for wellbeing has been examined in terms of three types of social identity variable: social identification, multiple identities, and multiple identity compatibility. Social identification is defined as the extent to which people feel they belong to a certain group (Greenaway et al., 2015; Jetten et al., 2017). Multiple identities in contrast refers to people's perception that they belong to multiple social groups (Haslam et al., 2008; Jetten et al., 2015). Multiple identity compatibility refers in turn to the perceived compatibility between an individual's multiple group memberships (Iyer, Jetten, Tsivrikos, Postmes, & Haslam, 2009). The positive impacts of social identity variables on wellbeing are increasingly recognised and have been dubbed the “social cure” (Cruwys, Haslam, Dingle, Haslam, & Jetten, 2014; Jetten et al., 2017). For example, previous research shows that group memberships are directly associated with lower levels of depression (Sani, Madhok, Norbury, Dugard, & Wakefield, 2015), as well as better health in organisational settings (Steffens, Haslam, Schuh, Jetten, & van Dick, 2017). Additionally, social identity has also been found to protect people from the negative impact of stressors, such as confrontation (Haslam & Reicher, 2006) and illness (Haslam, Jetten, Postmes, & Haslam, 2009).
At the same time, there is also evidence of some potential negative impacts of social identity on wellbeing (a “social curse”). Social identity theory (Tajfel & Turner, 1979) suggests that people have the need for their group identities to be valued and accepted by the society. However, some groups can be negatively stereotyped and socially devalued based on certain characteristic (i.e., disadvantaged groups; Goffman, 1964; Link & Phelan, 2006). Previous research has found that international students' wellbeing could be compromised because of social rejection and discrimination. For example, international students reported distress after perceiving that they were excluded or abused (Rosenthal, Russell, & Thomson, 2006), perceiving local students as having less willingness in making friends with them (Malau-Aduli, 2011).
3 PATHWAYS BETWEEN SOCIAL IDENTITY AND WELLBEING
The positive impacts of social identity on wellbeing have been hypothesised to arise because social identity unlocks psychological resources that mediate the link between social identity and wellbeing. Jetten, Haslam, and Haslam (2012); Jetten, Haslam, Haslam, Dingle, and Jones (2014); and Jetten et al. (2017) outline four such processes. The first process is social support: perceiving others as sharing a social identity with oneself leads people to be more likely to provide social support to, and expect to receive social support from, those fellow ingroup members (Haslam, Reicher, & Levine, 2012). The second process is a sense of meaning and purpose in life, which can provide individuals with a sense of purpose and direction when group members act together towards collective goals (Cruwys, Haslam, Dingle, Jetten et al., 2014; Steger & Frazier, 2005; Steger, Frazier, Oishi, & Kaler, 2006). The third process is personal control. This refers to people's feeling that they have the ability to alter events and achieve desired outcomes (Greenaway et al., 2015). The last process, a sense of belonging, is defined as the extent to which people feel they are a part of a group and close to other ingroup members (Alnabulsi & Drury, 2014; Berkman & Syme, 1979). These four factors have been demonstrated to unlock positive wellbeing effects for different groups, such as participants in mass gatherings, Syrian refugees, and adults diagnosed with depression (Çelebi, Verkuyten, & Bagci, 2017; Cruwys, Haslam, Dingle, Haslam, et al., 2014; Hopkins & Reicher, 2016). We examined each of these potential processes, with the exception of belonging, given its potential redundancy with social identification as a social identity variable defined in this paper.
4 FELT UNDERSTANDING AS ANOTHER POTENTIAL PATHWAY
Beyond the above approaches, we posit that felt understanding could provide another pathway between social identity and wellbeing. Felt understanding is defined as the feeling that one is accurately perceived, understood, appreciated, and cared for (Oishi, Krochik, & Akimoto, 2010). A line of research demonstrates that felt understanding positively influences individuals' wellbeing. For example, felt understanding has been found to relate to happiness (Lun, Kesebir, & Oishi, 2008), positive affect (Oishi, Koo, & Akimoto, 2008), and higher level of life satisfaction (Oishi et al., 2013). However, the positive effect of felt understanding has so far been investigated mainly in close relationships, as a key component which determines the quality of intimate relationships. For example, it facilitates the development of intimacy by providing the basis for acknowledging partners' inner selves (Reis & Patrick, 1996). In other words, revealing selves, a defining feature of intimate relations (Argyle & Henderson, 1984; Reis, 1990), is likely to happen when individuals feel understood by the listeners.
While the wellbeing benefits of felt understanding have tended to focus on close interpersonal relationships (Oishi et al., 2010; Reis et al., 2017) rather than in a broader context where social identities and group memberships also inform whether we feel understood by others (and by whom), research on social identity and wellbeing (Haslam et al., 2018; Jetten et al., 2017) has not considered felt understanding as a source of wellbeing, focusing instead on mediators such as social support (Haslam et al., 2012). Integrating these perspectives, we propose that felt understanding in a general sense can be derived from social identities, and it is therefore a potentially important process through which social identity variables predict better wellbeing. According to self-categorisation theory, social identities arise when people perceive there are meaningful differences between ingroup and outgroup, and ingroup members are believed to share common attributes and experiences (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Shared social identity should provide a basis for felt understanding, insofar as we expect people who share the same identities with us to have the same sets of norms, values, collective experiences, and so forth, and thus that they are better positioned (than if we did not have a sense of shared identity) to understand our own experiences, values, and so forth. This hypothesis is consistent with previous studies in cross-racial relationships in clinical settings. For example, minority group members prefer counsellors who share the same ethnic background (Coleman, Wampold, & Casali, 1995), and they are less likely to drop out from the treatment in this situation (Sue, Fujino, Hu, & Takeuchi, 1991). Combined with the well-established direct association between felt understanding and wellbeing, we therefore propose that felt understanding in a general sense (i.e., that others understand me) would mediate a positive effect of social identity variables (ingroup identification; multiple group memberships; multiple identity compatibility) on wellbeing.
In terms of relations between specific social identity variables and felt understanding, greater identification with a particular group in which we are embedded (e.g., as a student), and a sense of belonging to multiple groups, should predict a stronger sense of feeling understood by others, due to people in the same group sharing similar values, experiences, beliefs, and so forth. Having multiple social identities—that is, many different groups on which to draw—should also predict greater felt understanding, because it extends the range of people who, by sharing a social identity with us, will “get” our experiences, perspectives, aspirations, and so on. In this sense, we expected the relationship between multiple identities and felt understanding would align with “the more the merrier” effect (Iyer et al., 2009), referring to the finding that the more identities an individual has, the better their wellbeing (e.g., self-esteem, resilience; Jetten et al., 2015; Jones & Jetten, 2010).
The perceived compatibility between multiple identities should also predict felt understanding. Previous research found that the more one person finds their new identity fits with the existing identities, the more likely they will accept the new identity and identify with the new group (Iyer et al., 2009). In contrast, if the new group membership conflicts with other memberships, people might reject the new identity and the incompatibility would lead to one's psychological resources being diminished (Brook, Garcia, & Fleming, 2008). In relation to felt understanding, incompatible social identities—and the tension that it implies between the values, experiences, beliefs, etc. that define the respective groups—should predict a lower sense that one is understood by others because it suggests that members of some ingroups will not be understanding of one's membership of other groups.
5 RESEARCH QUESTION AND HYPOTHESES
Integrating social identity and felt understanding approaches, we propose that (1) felt understanding predicts better wellbeing outcomes along with other more established mediators such as social support; and (2) social identity variables can predict better wellbeing via felt understanding. In the present study, we tested whether feeling understood by others has the potential to bridge social identity and wellbeing among international university students, such that international student identification protects wellbeing (e.g., stress, anxiety, depression) via felt understanding. Specifically, social identity variables should indirectly predict more positive wellbeing outcomes via felt understanding, even when adjusting for more established mediators, including social support, life meaning, and personal control. The overall model we tested is presented in Figure 1.
6 METHOD
6.1 Participants and design
Participants were international students (mainly based in the UK) at a university in the south of England. They participated by completing online surveys advertised via university student newsletters as well as Facebook. Data were collected following three separate waves of recruitment: the first timepoint (T1) was November 2021, the second timepoint (T2) was February 2022, and the third timepoint (T3) was May 2022. None of the timepoint was during a period of Covid lockdown. The survey questions at each timepoint were identical. To maximise the sample size given the time and resources available, the T2 and T3 surveys were opened both to people who had completed previous surveys and people who newly joined the study at that timepoint.
Some participants' data were removed because their responses had more than 50% missing values or they did not pass the attention check (exclusion criteria were pre-registered at https://osf.io/9djrw). At T1, 191 students completed the survey. After exclusion, the sample included 153 participants (female: 115, male: 31, non-binary: 5). The mean age was 23.20 years (SD = 5.29), ranging from 17 to 40 years. At T2, 168 students completed the survey. The exclusion left a final sample of 149 (female: 114, male: 28, non-binary: 3, other: 1). The mean age was 24.10 years (SD = 7.07), ranging from 17 to 69 years. At T3, 160 students completed the survey. The exclusion left a final sample of 127 (female: 101, male: 19, non-binary: 4, other: 1). The mean age was 24.50 years (SD = 8.05), ranging from 17 to 70 years. Regarding compensation, participants were entered into a prize draw after finishing each survey: one of five £10 Amazon vouchers at T1, one of eight £15 vouchers at T2, and 1 of 16 £15 vouchers at T3. Design and materials were pre-registered.
Due to insufficient numbers of people completing two or more timepoints (26 people completed T1 and T2, 15 people completed T1 and T3, 31 people completed T2 and T3; only 20 people finished all three surveys), we decided to use a combined, entirely cross-sectional dataset consisting of 301 people's responses from their first time completing the questionnaire (i.e., data from people's survey at their second or third time points were excluded). This final analysed sample included 226 women, 60 men, 8 who identified as non-binary, and 1 who identified as other. The mean age was 23.30 years (SD = 6.12), ranging from 17 to 69 years.
In terms of statistical power for regression analysis, based on sensitivity analyses in G*power 3.1, a sample of 301 would provide 80% power to detect an effect as small as r = .160 (f = .162) for an individual predictor in a model with up to 12 predictors (α = .05; two tailed). The effect falls below Cohen's “medium” effect-size cut-off of r = .3.
6.2 Procedure
The online survey was built using Qualtrics. Participants first read information about the study and gave their consent. Before the main survey, participants needed to generate a unique code using non-identifying personal information (i.e., last digit of phone number; day of the month on which they were born; last letter of their postcode). This ensured that participants' responses at different timepoints could be linked. Then, participants proceeded to the main survey, which included four main sections: social identity measures, process measures, wellbeing measures, and demographics. At the end of the survey, participants were asked whether they would like to receive the follow-up surveys. If they selected “yes,” they were be directed to a separate web page to leave their email address (unconnectable to any responses left in the main survey). After completing the survey, participants were debriefed and thanked for their participation.
6.3 Measures
All of the following scales took the average of item scores, with the exception of depression, anxiety, loneliness, and imposter syndrome, for each of which item scores are summed. Details of measures can be found at the project OSF site: https://osf.io/nmpvz/?view_only=4fb08d2712d145cb8b6f7a6396a21879
6.3.1 Social identity measures
For all of the social identity measures, participants responded on a seven-point scale ranging from strongly disagree (−3) to strongly agree (3).
Social identification
Ingroup identification was assessed for each group using a four-item scale adapted from Doosje, Ellemers, and Spears (1995). The items were “I identify with [members of group],” “I see myself as a [member of group],” “I am pleased to be a [member of group],” and “I feel strong ties with [members of group].” Participants responded to three1 identification scales: identification with other university students (α = .83 at T1, α = .85 at T2, α = .89 at T3), identification with other international students (α = .85 at T1, α = .87 at T2, α = .89 at T3), and identification with home country (α = .89 at T1, α = .90 at T2, α = .88 at T3).1
Multiple identities
The sense of belonging to multiple social groups was assessed using a three-item scale (α = .90 at T1, α = .91 at T2, α = .90 at T3) adapted from the Exeter Identity Transition Scales (EXITS; Haslam et al., 2008). The items were “I belong to lots of different groups,” “I join in the activities of lots of different groups,” and “I have strong ties with lots of different groups.”
Multiple identity compatibility
The perceived compatibility between multiple group memberships was assessed by a two-item scale (r = .43 at T1, r = .64 at T2, r = .66 at T3; Haslam et al., 2018). Items were “On the whole, being a member of a particular group is compatible with being a member of other groups” and “There is harmony between the different groups I belong to.”
6.3.2 Process measures
The following measures assessed the process(es) that could mediate the relationship between social identity and wellbeing (Haslam et al., 2018). Participants responded to each item on a seven-point scale ranging from strongly disagree (−3) to strongly agree (3).
Social support
Perceptions of social support were assessed by a seven-item scale (α = .79 at T1, α = .85 at T2, α = .86 at T3; Haslam, O'Brien, Jetten, Vormedal, & Penna, 2005) including items such as “If I needed it, I would get emotional support from a number of people” and “In general, my family listens to my problems.” Items 4 and 5 were reverse scored.
Felt understanding
The sense of feeling understood by others was assessed using a 10-item scale (α = .89 at T1, α = .92 at T2, α = .88 at T3; Livingstone et al., 2020). This was adapted to refer to others in general rather than a specific outgroup, and to oneself rather to an ingroup. Example items include “In general, others really get what I want to do and achieve,” “In general, others have little idea of what is important to me,” and “In general, others understand my values.” Items 2, 4, 5, 6, 7, 8 were reverse scored.
Life meaning
Meaning in life was assessed using a five-item scale (α = .93 at T1, α = .91 at T2, α = .94 at T3) adapted from the Meaning in Life Questionnaire (MLQ) developed by Steger, Frazier, Oishi, and Kaler (2006). MLQ consists of two dimensions: presence of meaning and search for meaning. This study only used the five items of presence of meaning dimension to measure the extent to which participants feel their lives have meaning. Example items included “I understand my life's meaning” and “My life has a clear sense of purpose.” Item 5 was reverse scored.
Perceived personal control
Perceived personal control was assessed using a three-item scale (α = .78 at T1, α = .76 at T2, α = .70 at T3; Greenaway et al., 2013). Items were “I feel in control of my life,” “I am free to live my life how I wish,” and “My experiences in life are due to my own actions.”
6.3.3 Wellbeing
Stress
Chronic stress was assessed using a six-item scale including three dimensions: exhaustion, lack of accomplishment, and callousness (α = .63 at T1, α = .76 at T2, α = .74 at T3; Steffens, Haslam, Kerschreiter, Schuh, & van Dick, 2014). Items included “I feel like I am working too hard” and “I feel like I am failing to achieve my goals.” Participants responded to each item on a seven-point scale ranging from do not agree at all (1) to completely agree (7).
Depression
Depression was assessed using the Personal Health Questionnaire (PHQ-8) Depression scale (Kroenke et al., 2009). This eight-item scale (α = .88 at T1, α = .88 at T2, α = .92 at T3) included statements such as “Little interest or pleasure in doing things” and “Feeling down, depressed or hopeless.” Participants responded to each item based on how often they had felt this way during the past 2 weeks on a four-point scale ranging from not at all (0) to nearly every day (3). The total score ranged from 0 to 24.
Anxiety
Anxiety was assessed using the Generalised Anxiety Disorder Assessment scale (GAD-7; Spitzer, Kroenke, Williams, & Löwe, 2006). This seven-item scale (α = .90 at T1, α = .91 at T2, α = .93 at T3) included statements such as “Feeling nervous, anxious or on edge” and “Not being able to stop or control worrying.” Participants responded to each item based on how often they had felt this way during the past 2 weeks on a four-point scale ranging from not at all (0) to nearly every day (3). The total score ranged from 0 to 21.
Loneliness
Loneliness was assessed on the three-Item Loneliness Scale (3-ILS; Hughes, Waite, Hawkley, & Cacioppo, 2004). Items were “How often do you feel that you lack companionship?” “How often do you feel left out?” and “How often do you feel isolated from others?” (α = .79 at T1, α = .83 at T2, α = .82 at T3). Participants responded to each item on a three-point scale ranging from hardly ever (1) to often (3).
Self-esteem
Self-esteem was assessed using a five-item scale adapted from the Rosenberg Self-Esteem Scale (RSE; Rosenberg, 1979). Items included “On the whole, I am satisfied with myself” and “I am able to do things as well as most people” (α = .78 at T1, α = .80 at T2, α = .84 at T3). Participants responded to each item on a five-point scale ranging from strongly disagree (0) to strongly agree (4).
Imposter syndrome
Imposter syndrome was assessed on a seven-item Imposterism Scale (α = .91 at T1, α = .92 at T2, α = .93 at T3; Leary, Patton, Orlando, & Wagoner Funk, 2000). Items included “Sometimes I am afraid I will be discovered for who I really am” and “In some situations I feel like an imposter.” Participants responded to each item on a five-point scale ranging from not at all (1) to extremely (5).
Life satisfaction
Life satisfaction was assessed using a single item: “All things considered, how satisfied are you with life as a whole right now?” Participants responded to the item on a 10-point scale ranging from completely dissatisfied (1) to completely satisfied (10).
6.3.4 Demographics
Demographic questions included age, gender, ethnicity background, year of study, first language (English or not), sexual orientation, family education background, degree, subject area, current base, and length of stay in university city (in years) during degree.
7 RESULTS
7.1 Demographics
A summary of sample demographics can be found in Table 1.
| Variable | n | % | M | SD |
|---|---|---|---|---|
| Age | 296 | 23.3 | 6.12 | |
| Gender | 295 | |||
| Female | 226 | 76.6 | ||
| Male | 60 | 20.3 | ||
| Non-binary | 8 | 2.7 | ||
| Other | 1 | 0.3 | ||
| Ethnicity | 283 | |||
| White | 102 | 36.0 | ||
| Asian | 139 | 49.1 | ||
| African, mixed, and other | 42 | 14.9 | ||
| Degree | 285 | |||
| Undergraduate | 160 | 56.1 | ||
| Masters | 63 | 22.1 | ||
| Postgraduate research (e.g., PhD) | 62 | 21.8 |
7.2 Correlations
Table 2 presents details of zero-order correlation (Pearson, two-tailed) among social identity variables, process variables, and wellbeing measures.
| M (SD) | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. US | 1.30 (1.22) | |||||||||||||||
| 2. IS | 1.12 (1.41) | .54*** | ||||||||||||||
| 3. HC | 1.71 (1.32) | .31*** | .42*** | |||||||||||||
| 4. MI | 0.66 (1.50) | .37*** | .32*** | .20*** | ||||||||||||
| 5. MIC | 1.05 (1.18) | .41*** | .37*** | .23*** | .51*** | |||||||||||
| 6. SS | 1.35 (1.01) | .39*** | .29*** | .28*** | .28*** | .36*** | ||||||||||
| 7. FU | 0.21 (1.15) | .38*** | .20*** | .15** | .35*** | .35*** | .57*** | |||||||||
| 8. LM | 0.73 (1.51) | .31*** | .24** | .18** | .27*** | .25*** | .30*** | .39*** | ||||||||
| 9. PC | 1.06 (1.26) | .36*** | .25*** | .28*** | .22*** | .28*** | .42*** | .35*** | .41*** | |||||||
| 10. DP | 9.51 (5.95) | −.22*** | −.16** | −.09 | −.13* | −.23** | −.39*** | −.49*** | −.44*** | −.37*** | ||||||
| 11. AX | 8.28 (5.80) | −.13* | −.11 | −.09 | −.07 | −.12* | −.34*** | −.35*** | −.27*** | −.35*** | .75*** | |||||
| 12. S | 3.75 (1.08) | −.27*** | −.17** | −.14* | −.16** | −.22*** | −.42*** | −.54*** | −.35*** | −.39*** | .67*** | .59*** | ||||
| 13. LL | 6.00 (1.79) | −.29*** | −.20*** | −.12* | −.13* | −.19*** | −.48*** | −.53*** | −.31*** | −.31*** | .53*** | .46*** | .48*** | |||
| 14. SE | 2.11 (0.89) | .28*** | .25*** | .18** | .25*** | .28*** | .37*** | .43*** | .52*** | .47*** | −.55*** | −.50*** | −.54*** | −.44*** | ||
| 15. IP | 17.00 (7.60) | −.19*** | −.15** | −.12* | −.15** | −.19*** | −.31*** | −.37*** | −.34*** | −.28*** | .50*** | .48*** | .47*** | .40*** | −.55*** | |
| 16. LF | 6.27 (1.99) | .35*** | .19*** | .22*** | .29*** | .34*** | .50*** | .53*** | .47*** | .52*** | −.62*** | −.47*** | −.57*** | −.52*** | .61*** | −.38*** |
- Abbreviations: AX, anxiety; DP, depression; FU, felt understanding; HC, home country identification; IP, imposter syndrome; IS, international student identification; LF, life satisfaction; LL, loneliness; LM, life meaning; MI, multiple identities; MIC, multiple identities compatibility; PC, personal control; S, stress; SE, self-esteem; SS, social support; US, university student identification.
- * p < .05;
- ** p < .01;
- *** p < .001.
7.3 Path analysis
We conducted path analysis to test the pre-registered path model (see Figure 1, pre-registered at https://osf.io/9djrw) using the dataset that consisted of all the participants' responses when they completed the survey for the first time (n = 301). In this model, predictors were social identity variables (international student identification, multiple identities, multiple identity compatibility), mediators were process variables (social support, felt understanding, life meaning, personal control), outcomes were wellbeing measures (seven measures, including depression, anxiety, stress, self-esteem, loneliness, life satisfaction, imposter syndrome). The model was run separately for each of the wellbeing measures. Seven models were run in R 4.1.3 (R Core Team, 2022) with maximum likelihood (ML) estimation.
As the model illustrated above was saturated, the model fit was the same for the seven tested models, χ2 (0) = 0.00, CFI/TLI = 1.00/1.00, RMSEA = .00. Rather than assessing overall model fit per se, the aim of this study was instead to test whether felt understanding had a unique role in predicting wellbeing. The model explained 35% of the variance in depression, 21% of the variance in anxiety, 35% of the variance in stress, 39% of the variance in self-esteem, 20% of the variance in imposter syndrome, 35% of the variance in loneliness, and 47% of the variance in life satisfaction.
Table 3 presents the standardised estimates of the paths from process variables to wellbeing outcomes in the model illustrated by Figure 1. Felt understanding significantly predicted all seven wellbeing outcomes. Personal control significantly predicted five out of the seven wellbeing outcomes, with the exceptions being loneliness and imposter syndrome. Life meaning significantly predicted five out of the seven wellbeing outcomes, with the exceptions being anxiety and loneliness. Social support significantly predicted three out of the seven wellbeing outcomes, including anxiety, loneliness, and life satisfaction. Overall, among four process variables, felt understanding was the most consistent predictor of wellbeing outcomes and was the strongest (or jointly strongest) predictor of five of the wellbeing outcomes. In addition, personal control and life meaning also predicted most of the wellbeing outcomes.
| Outcome | Mediators | |||||||
|---|---|---|---|---|---|---|---|---|
| Social support | Felt understanding | Life meaning | Personal control | |||||
| β | 95% CI | β | 95% CI | β | 95% CI | β | 95% CI | |
| Depression | −.10 | −0.22, 0.02 | −.32*** | −0.43, −0.20 | −.26*** | −0.36, −0.14 | −.12* | −0.23, −0.03 |
| Anxiety | −.14* | −0.28, −0.002 | −.21** | −0.33, −0.08 | −.09 | −0.21, 0.04 | −.21** | −0.33, −0.08 |
| Stress | −.10 | −0.23, 0.02 | −.40*** | −0.51, −0.28 | −.11* | −0.22, −0.01 | −.17** | −0.28, −0.06 |
| Self-esteem | .06 | −0.09, 0.19 | .17** | 0.05, 0.29 | .32*** | 0.20, 0.43 | .23*** | 0.11, 0.34 |
| Imposter syndrome | −.08 | −0.23, 0.07 | −.21** | −0.36, −0.06 | −.20** | −0.32, −0.06 | −.08 | −0.21, 0.05 |
| Loneliness | −.23** | −0.35, −0.09 | −.38*** | −0.50, −0.26 | −.08 | −0.21, 0.04 | −.07 | −0.19, 0.05 |
| Life satisfaction | .18** | 0.06, 0.29 | .24*** | 0.13, 0.34 | .20*** | 0.10, 0.30 | .27*** | 0.16, 0.37 |
- Note: Confidence intervals are bias corrected bootstrapped. Bootstrapping sample size = 10,000.
- * p < .05;
- ** p < .01;
- *** p < .001.
7.3.1 Indirect effects
To evaluate the indirect effects, bootstrapping with 10,000 resamples was used to produce bias-corrected bootstrapped standard errors (SEs) and 95% confidence intervals (CIs) for the parameters. Confidence intervals excluding zero are taken as indicating significant indirect effects.
Multiple identities
Indirect paths from multiple identities to wellbeing were significant via felt understanding in the cases of depression (B = −0.26, SE = 0.09, 95% CIs [−0.48, −0.11]), anxiety (B = −0.16, SE = 0.07, 95% CIs [−0.30, −0.04]), stress (B = −0.06, SE = 0.02, 95% CIs [−0.10, −0.03]), imposter syndrome (B = −0.23, SE = 0.11, 95% CIs [−0.50, −0.06]), loneliness (B = −0.10, SE = 0.03, 95% CIs [−0.17, −0.04]), self-esteem (B = 0.02, SE = 0.01, 95% CIs [0.01, 0.05]), and life satisfaction (B = 0.07, SE = 0.03, 95% CIs [0.03, 0.18]). There were also five significant indirect paths from multiple identities to wellbeing via life meaning. Details can be found in Supporting Information.
Multiple identity compatibility
Indirect paths from multiple identity compatibility to wellbeing were significant via felt understanding in the cases of depression (B = −0.34, SE = 0.14, 95% CIs [−0.67, −0.11]), anxiety (B = −0.21, SE = 0.10, 95% CIs [−0.47, −0.06]), stress (B = −0.08, SE = 0.03, 95% CIs [−0.14, −0.03]), imposter syndrome (B = −0.30, SE = 0.15, 95% CIs [−0.68, −0.08]), loneliness (B = −0.13, SE = 0.05, 95% CIs [−0.24, −0.05]), self-esteem (B = 0.03, SE = 0.01, 95% CIs [0.01, 0.06]), and life satisfaction (B = 0.09, SE = 0.04, 95% CIs [0.03, 0.18]). There were also three significant indirect paths from multiple identity compatibility to wellbeing via social support and five significant indirect paths via personal control. Details can be found in Supporting Information.
International student identification
When international student identification was included as the “identification” predictor in the model, analysis showed that significant paths via felt understanding were from multiple identities and multiple identities compatibility, but not from international student identification. When we replaced international student identification with university student identification as the “identification” predictor in the model alongside multiple groups and multiple identity compatibility, paths from university student identification via felt understanding to wellbeing were also significant for all seven wellbeing outcomes (Bs = −0.36 to 0.09, ps = .001–.028). Additionally, there were five significant indirect paths from international student identification to wellbeing via life meaning and personal control respectively, and three via social support. Details can be found in Supporting information.
A model suggested by a reviewer, specifying the seven wellbeing measures as indicators of an overall latent mental health and wellbeing construct, was run as well (please find details in the Supporting Information). The findings were consistent with the results of preregistered analysis presented above, in that (1) felt understanding predicted wellbeing and mental health over and above the other mediators, and (2) indirect effects of each social identity variable via felt understanding were significant.
8 DISCUSSION
Focusing on the context of international students' wellbeing and mental health, the aim of this study was to test whether felt understanding helps to explain the positive association between social identities and wellbeing. Building on research showing that social identity promotes wellbeing by unlocking psychological resources, such as social support, life meaning, and personal control (Cruwys, Haslam, Dingle, Jetten, et al., 2014; Haslam et al., 2018; Jetten et al., 2017), we proposed that felt understanding could be another, under-acknowledged resource in this process. Results of pre-registered path models showed that felt understanding predicted wellbeing outcomes over and above other established process variables (social support, life meaning, personal control) from the social identity literature; moreover, indirect effects via felt understanding from social identity variables (multiple identities, multiple identity compatibility, and student identification) to wellbeing were also significant.
8.1 Felt understanding predicts wellbeing
Coefficients of the paths from mediators to outcomes show that each process variable uniquely predicted at least some wellbeing and mental health outcomes (see Table 3). Importantly, felt understanding significantly predicted all seven wellbeing outcomes, which aligns with previous research that felt understanding can predict a range of wellbeing outcomes, such as greater life satisfaction and fewer physical health problems (Lun et al., 2008). Additionally, felt understanding was the strongest among these process variables in predicting five of the wellbeing outcomes (depression, anxiety, stress, imposter syndrome, loneliness), indicating felt understanding plays a unique and consistent role in predicting wellbeing over and above the processes emphasised in previous social identity work.
Regarding other process variables, personal control predicted five out of seven outcomes. This supports previous studies showing personal control predicts less depression and more life satisfaction (Greenaway et al., 2015; Greenaway, Cruwys, Haslam, & Jetten, 2016). Life meaning predicted five out of seven wellbeing outcomes, reflecting Steger and Frazier's (2005) finding that life meaning positively associated with wellbeing. Social support predicted three wellbeing outcomes, which provides further evidence for the well-established finding that social support protects against stress (Haslam et al., 2005) and prevents burnout in the workplace (Romano, Tang, Hietajärvi, Salmela-Aro, & Fiorilli, 2020). Although social support did not predict all of the wellbeing outcomes, this could simply reflect the fact that the four process variables were also correlated with one another, and may thus have nullified their respective individual associations with each outcome to some extent.
8.2 Social identity predicts wellbeing via felt understanding
In terms of indirect effects, the significant paths via felt understanding to wellbeing are from multiple identities, multiple identity compatibility, and ingroup (student) identification, which is consistent with previous research showing that multiple group memberships and compatibility predict wellbeing especially during life transition (e.g., Haslam et al., 2008; Iyer et al., 2009). However, the indirect effect of international student identification on wellbeing was not significant. This is unexpected as previous research suggests that making friends from one's own culture facilitates international students' wellbeing (Schmitt, Spears, & Branscombe, 2003; Ng et al., 2018). A possible explanation is that international student identification has a two-sided impact on international students' wellbeing. In terms of positive impact, by identifying with other international students who share similar experiences, they can get a sense of connection, feel less lonely, and receive help as well as provide support to others. At the same time, identifying with international students could potentially mean identifying with a minority group which is stigmatised (e.g., disfluent in English; Ruble & Zhang, 2013) and discriminated against (Schmitt et al., 2003; Sherry, Thomas & Chui, 2010). This could result in isolation and rejection from UK students, which are detrimental to international students' wellbeing and impede them developing cross-group relationships, learning about local culture, and adapting to host country environment.
In contrast, university student identification (rather than international student identification) did indirectly predict wellbeing and mental health outcomes via felt understanding. Thus, a more general sense of identification as a student could provide a less double-edged sense of belonging, which in turn predicts better mental health (e.g., Atri, Sharma, & Cottrell, 2007). Apart from this, it could also increase intergroup contact between international students and host national students (Quinton, 2020), which has been found to be directly associated with better wellbeing (Szabó, Papp, & Luu, 2020), and buffer international students' wellbeing from negative impact of prejudice (Imai & Imai, 2019).
The results also support our hypothesis that group identification provides ground for felt understanding, as people would expect ingroup members to share similar experiences and feelings with them, which facilitates a sense of feeling understood. For example, Panayidou and Priest (2021) found that PhD students talking about their difficult experiences during PhD in peer support group could generate felt understanding by peers, which in turn leads to less anxiety and more life satisfaction. This has also been examined in clinical self-help groups where people gain a sense of being understood after speaking to other people who have similar medical condition (e.g., breast cancer; Stang & Mittelmark, 2010).
8.3 Limitations and future directions
There are some limitations in this study. Analyses were conducted based on a cross-sectional sample collected from one university at a certain time, thus the findings cannot provide direct tests of causal relationships. Our conclusions regarding the nature of these relationships are thus cautious, and key propositions here need further research both to replicate effects and to test the causal relationships among social identity, felt understanding and wellbeing via longitudinal designs or experiments. Another issue is that the four mediators (process variables) in our model are correlated with each other, which could lead to the “unique” impact of each one being reduced to some extent. This is also the case with three correlated social identity variables as predictors. To test the reliability of this model, future studies should replicate this model using different samples from other universities or in different contexts. Nevertheless, this study does provide initial evidence that felt understanding can predict wellbeing not only in close relationships but also in a more general sense where social identities are relevant.
Future research could also focus on the potential mediators in turn between felt understanding and wellbeing. Previous research suggests felt understanding fosters wellbeing indirectly by promoting social relationships. It has been well-established that felt understanding helps social bonds form at an interpersonal level (Oishi et al., 2010; Reis et al., 2017), and the quality of social relationships has been found to consistently predict higher levels of life satisfaction (Diener, Suh, Lucas, & Smith, 1999) and reduced distress symptoms, such as depression and anxiety (Fleming, Baum, Gisriel, & Gatchel, 1982; Steffens et al., 2019). A further, related possibility would be to examine whether feeling understood by others in turn feeds back in to and bolsters a sense of social identification; that is, social identities provide a sense of feeling understood by others, and feeling understood in practice in turn helps to sustain social identities over time in a cyclical manner.
9 CONCLUSION
This research provides initial support for an integration of felt understanding and social identity approaches to wellbeing. In particular, feeling understood by others consistently predicted wellbeing and mental health outcomes over and above more established mediators of social identity effects, and in turn was predicted by social identity variables. The findings thus offer initial, if tentative support for the proposition that felt understanding is a key process and resource through which social identities protect wellbeing and mental health.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
ETHICS STATEMENT
This study was approved by Psychology Ethics Committee at the University of Exeter (Ethics Application ID: 489033).
Endnote
Open Research
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are openly available in Open Science Framework at https://osf.io/nmpvz/. https://doi.org/10.17605/OSF.IO/NMPVZ.
