Social capital as mediating factor on COVID-19 induced psychological distress: The case of college students living through an outbreak.

This investigation examines the interplay between individual-level social capital, psychological distress, and the effects of a coronavirus disease 2019 (COVID-19) outbreak on a college campus. Data were collected from students at two colleges in early 2021. Hypotheses were formed based on prior research on social capital and the COVID pandemic. A structural equation model was constructed to allow the modeling of both latent and observed variables. The Kessler-6 measures were used to create a latent psychological distress variable, which was the dependent variable. Predictor variables include whether the student experienced the outbreak, the student's race and gender, and the level of reported individual social capital (having supportive and rewarding relationships). Our findings suggest that the most robust predictor of reduced psychological distress was individual social capital, which also served as an intermediary between the outbreak and distress. Social capital mediates psychological distress in an outbreak.

INTRODUCTION

The coronavirus disease 2019 (COVID‐19) global pandemic has affected nearly every aspect of life, including the mental health of college students. Though media and trade publications frequently discuss the issue, there has been comparatively little empirical work. A confounding issue is that, due to the timing of the initial shutdowns at midterm Spring 2020, the ability to compare student experiences was limited. This study examines the influence of social capital on psychological distress by taking advantage of the disparate experiences of two colleges: one that suffered a severe outbreak and one that did not.

The colleges, one public and one private, are located about an hour apart and have similar student bodies. During the fall of 2020, the first college experienced a sizable COVID outbreak that resulted in all classes being moved online and students living on campus being sent home. Students at the second college experienced a successful semester on campus. In January 2021, students at both colleges returned to their respective campuses for the spring semester.

LITERATURE REVIEW

Social capital and the COVID pandemic

Research on social capital has been flourishing for the last several years (Claridge, 2018). Typically, the meaning of this concept falls under one of two variations (Fulkerson & Thompson, 2008). The first involves social capital as norms of trust and reciprocity that result in collective benefits for entire populations (Coleman, 1988; Coleman, 1990; Putnam, 2000; Putnam & Leonardi, 1993). The second involves relationships and networks that result in different resources that benefit individuals (Fernandez et al., 2000; Granovetter, 1973; Lin, 2001; Portes, 1998). These two different kinds of social capital are not independent of one another; however, since higher levels of individual social capital are associated with increased levels of collective social capital (Brehm & Rahn, 1997; Jicha et al., 2011). Our primary interest will be examining the mediating effect of individual social capital to determine if it can effectively limit the psychological distress that may have resulted to some extent from the COVID pandemic, especially in the fallout of a localized outbreak on a college campus.

Of particular relevance to the present study is the literature examining social capital and the COVID pandemic. Our review has found that nearly all such studies claim that social capital was a boon for populations, enabling them to produce more agile and effective responses to the pandemic through collective adherence to protective behaviors, such as masking, social distancing, and quarantining as needed (Bian et al., 2020; Borgoni & Anderieu, 2020; Lau, 2020; Lindström, 2020; Ohta & Yata, 2021; Wu, 2021). Each of these studies tends to adopt the normative conception of social capital, noted above. Caballero‐Dominguez et al. (2021) provide one of the only studies, of which we are aware, to examine the interplay between COVID‐19, social capital, and psychological distress. Their investigation of the Columbian lockdown found that social capital could reduce COVID‐induced psychological distress while adopting an individual resource conception of social capital.

Son et al. (2020) found that 71% of students at a Texas University did experience an increase in anxiety due to multiple COVID‐induced stressors, and this is similar to the 77% of adults in a contemporary (Gordon et al., 2021). Students coped with increased stress by turning to others around them for support. Studies of social capital on campus find social capital is an individual resource that reduces binge drinking (Weitzman & Kawachi, 2000), while improving grade point averages and relationships with instructors (Schwartz et al., 2018). Research has also found that drinking parties were not a reliable source of social capital (Buettner & Debies‐Carl, 2012), while participating in athletics, being female, and being white could yield more social capital (Clopton, 2012). Non‐white students were found to increase their social capital on campuses that offered Black Cultural Centers (Hypolite, 2020) and religious organizations (Park & Bowman, 2015).

Psychological distress and the Kessler‐6

Utilizing surveys of college students (see Section 6), psychological distress was measured using the Kessler‐6 scale. The Kessler‐6 is one of two such scales (the other being the Kessler‐10) designed to measure anxiety and mental distress in patients (Kessler et al., 2003). It utilizes six questions to generate a 24‐point scale by using a 5‐point Likert scale, with higher scores indicating greater mental distress.

The questions are: In the past 30 days…

how often did you feel nervous?

how often did you feel hopeless?

how often did you feel restless or fidgety?

so sad nothing could cheer you up?

that everything was an effort?

how often did you feel worthless?

The validity of the scale has performed well in both American and international contexts (Kessler et al., 2010; Prochaska et al., 2012).

RESEARCH QUESTIONS AND HYPOTHESES

We believe that psychological distress was already problematic for the students in our study, predating the pandemic, based on findings of other studies of college student mental health. It should be remembered that students at one college experienced an outbreak the previous term (i.e., an unsuccessful semester) and students at the other colleges did not. Our first hypothesis is that we expect that the COVID‐19 outbreak has a positive statistical effect on students that resulted in elevated psychological distress (a substantively negative outcome), as measured by the Kessler‐6 instrument, upon return to campus for the spring.

Next, while we expect the outbreak to directly elevate distress, based on Caballero‐Dominquez et al. (2021), we expect that social capital will have a mediating effect, wherein the psychological distress from the outbreak is reduced through a negative statistical relationship (Hypothesis 2).

Finally, based on prior research, we expect that students who are white and female will have higher levels of individual social capital, as indicated by high scores reporting supportive and rewarding relationships (Hypothesis 3). In turn, this higher level of individual social capital will translate into lower psychological distress for white and female students, compared to their peers (Hypothesis 4).

METHODS

College student survey data on the impacts of COVID‐19 were collected between the fall of 2020 and the spring of 2021 at a small private college and a medium‐sized public college. This was one element of a wider project carried out by Intermountain COVID‐19 Impact Consortium, a consortium of scholars at three colleges and a medical center. The consortium designed and administered the knowledge, attitudes, beliefs, & behaviors survey, a psychometric instrument designed to measure the relationship between belief, behavior, and mental health. The core survey was administered six times during the 2020–2021 academic year, and additional questions sets were added to the core survey at different waves to test specific hypotheses. This study utilizes the fourth wave, administered in January 2021, that included questions related to social capital. The sample size was 1710 student respondents, of which 478 were at the smaller of the two colleges, while 1232 were at the larger college. The total sampling frame was 8894, for a response rate of 19.2%. While this is a low response rate, lower rates of participation were expected given the crises wrought by the pandemic.

Outbreak

As noted previously, the first college experienced a campus outbreak of COVID infections resulting in all classes being moved online, whereas the second of the colleges was spared such an outbreak. These disparate experiences provided a natural pseudo experimental opportunity to assess the effects of an outbreak. An outbreak binary dummy variable was created.

Social capital

For our examination of individual social capital, we use a measure that asks respondents to state their views on whether they feel like they have strong and supportive networks. The specific wording is: “My social relationships are supportive and rewarding.” Response options ranged from 1 to 7, with 7 indicating a high level of agreement with the statement.

Social inequalities

To examine the research questions that we had about social inequalities, we used questions that measured student gender and race. Due to sample size limitations, we decided to recode the original responses into binary dummy variables for students as white/non‐white and male/non‐male.

Kessler‐6 psychological distress instrument

Our survey adopted the six‐item instrument created by Kessler to assess the level of psychological distress being experienced by students. These items include the following question stem: “During the past 30 days how often did you feel…” The six branches include “nervous,” “hopeless,” “restless or fidgety,” “so depressed that nothing could cheer you up,” “that everything is an effort,” and “worthless.” These agreement items were on a 5‐point Likert scale ranging from strong disagreement to strong agreement. Thus, higher scores indicate greater agreement with the item.

ANALYSIS AND RESULTS

Our analysis makes use of IBM SPSS Statistics and AMOS to complete our analysis of the descriptive statics, correlation matrix, and structural equation model (SEM). SEM is a technique that is well‐suited to analyses that combine observed and latent variables in a predictive model while adding the ability to model mediating or indirect effects. Our research hypotheses test the mediating effect of social capital on psychological distress.

The descriptive statics and zero‐order correlation matrix for this analysis variables are shown in Table 1. As the table shows, about three out of four students experienced the outbreak of COVID on their campus. Racially, the sample is roughly 79% white, slightly less diverse than the wider population of the United States. A greater imbalance is found with gender, males make up less than a third of the student sample. In terms of the average scores for the Kessler‐6 items, we find a range of 2.14 (worthless) to 3.18 (nervous). Turning to the zero‐order correlation matrix, we find that the outbreak variable is weakly associated with individual social capital in a negative direction. We interpret this to mean that the strength of personal relationships may have been weakened by the outbreak as individuals had to become more isolated through social distancing. There was no discernible relationship between the outbreak and race, but a weak negative relationship with gender—these patterns reflect differences in the make‐up of the two colleges, with the outbreak college having fewer male students and similar racial composition. Each of the individual Kessler‐6 items has a positive moderately weak relationship to the outbreak. We interpret this to mean that the outbreak exacerbated psychological distress as students returned to campus for the first time since the outbreak.

Table 1

Descriptive statistics and bivariate correlations

	Statistics	V1	V2	V3	V4	V5	V6	V7	V8	V9	V10
V1 outbreak (experienced)	72.05% (experienced)	–
V2 individual social capital	x̄ = 5.67, SD = 1.30	−0.048a	–
V3 race (white)	79.37% (white)	0.010	0.187b	–
V4 gender (male)	28.55% (male)	−0.086c	0.087c	−0.066d	–
V5 nervous	x̄ = 3.178, SD = 1.27	0.141b	−0.180b	−0.067d	−0.291b	–
V6 hopeless	x̄ = 2.60, SD = 1.33	0.136b	−0.318b	−0.012	−0.155b	0.647b	–
V7 restless	x̄ = 3.07, SD = 1.35	0.133b	−0.232b	0.054d	−0.185b	0.697b	0.660c	–
V8 depressed	x̄ = 2.36, SD = 1.30	0.130b	−0.384b	−0.033	−0.135b	0.536b	0.748b	0.613b	–
V9 everything is effort	x̄ = 2.88, SD = 1.1.36	0.134b	−0.345b	−0.035	−0.126b	0.606b	0.698b	0.655b	0.742b	–
V10 worthless	x̄ = 2.14, SD = 1.30	0.115b	−0.417b	−0.042	−0.108b	0.494b	0.719b	0.563b	0.789b	0.679b	–

α = 0.10.

α = 0.001.

α = 0.01.

α = 0.05.

In reviewing the correlations with individual social capital, all variables are significant. Students who were white and male were slightly higher on individual social capital than their peers. Each of the individual Kessler‐6 items has a moderate negative relationship with individual social capital. We interpret this to mean that individual social capital—having rewarding and supportive relationships—is a strong antidote to psychological distress.

In terms of social inequalities, we find that there are few racial disparities reported by students in terms of the psychological distress items. White students are slightly more likely to report feeling nervous and restless. The effects of gender are more prominent. Males are moderately less likely to experience all of the Kessler‐6 psychological distress items.

Finally, in terms of the Kessler‐6 intercorrelations, we find a robust average correlation, or Cronbach's α = 0.656, with individual correlations ranging from 0.494 to 0.789. We interpret this to mean that the items are highly related and indicative of underlying psychological distress. Indeed, this is the basis for our decision to model distress as a latent variable in the SEM analysis.

SEM analysis

The resulting SEM is shown in Figure 1 with standardized path coefficients. Evaluation of the overall model fit is based on the baseline comparison values, which should approach the value of 1. We find these values to range from 0.83 to 0.906. While the χ 2 (592.48) is statistically significant, we find the ratio to degrees of freedom (31) to be an acceptable 19.1. Finally, the root mean square error of approximation of 0.10 is determined as acceptable. We interpret model fit, based on all of these indicators, to be acceptable. We expect that the inclusion of other relevant variables could improve model fit, though we leave that to future analyses.

Figure 1

Structural equation model with standardized coefficients; χ 2 = 592.38***, degrees of freedom = 31, ratio = 19.1; baseline comparisons: normed fit index = 0.906, relative fit index = 0.834, incremental fit index = 0.911, Tucker–Lewis index = 0.841, comparative fit index = 0.910; root mean square error of approximation = 0.103. SC, social capital

The SEM has been specified so that the dependent variable in our model is the latent psychological distress variable (distress), and this is indicated by the established Kessler‐6 items that measure depression, energy levels, nervousness, hopelessness, worthlessness, and restlessness. Our independent exogenous variables include the binary outbreak variable (whether students experienced the COVID‐19 outbreak), race (white/non‐white binary), and gender (male/non‐male binary). Finally, individual social capital is specified to be an intervening variable between the exogenous independent variables and the dependent distress variable. While the initial model was a saturated model, including all possible paths between items, the final model that we present includes only those items that are statistically significant at the α = 0.05 level or above. The link between gender and individual social capital was removed, as was the link between race and distress, as these were not significant in the initial model.

Turning now to the interpretation of the model, we begin by noting that about 22% of the variation in distress is explained by the model, based on the resulting pseudo‐R 2 value. This suggests the need to identify additional variables in future studies that may fully explain distress. Our goal was less ambitious in this study, seeking to examine the effects of the outbreak and subsequent return to campus on distress, while considering the mediating effect of individual social capital. The direct effect of the outbreak on distress is significant, though somewhat weak (0.12). However, the total effect, shown in Table 2, reveals a slightly stronger effect (0.142), since the outbreak had a weak, though significant, negative direct effect on individual social capital (−0.05). The model also shows that males had a direct effect that indicates lower distress than non‐males (−0.18). There was no direct effect of race on distress, but white students were more likely to report having higher levels of individual social capital (0.09), resulting in an indirect effect on distress (−0.035), as found in the total effects reported in Table 2. The most consequential direct effect is that which flows from individual social capital to distress (−0.40). Consistent with the interpretation of the bivariate correlations with the individual Kessler‐6 items, we interpret this direct effect to mean that one of the best ways to reduce psychological distress among students is to increase their individual social capital. Finally, the robust path coefficients flowing from the latent distress variable to the individual Kessler‐6 indicators are further supporting the strength and validity of this underlying construct.

Table 2

Total effects

	White	Outbreak	Male	Individual SC	Distress
Individual SC	0.088	−0.049	0.000	–	–
Distress	−0.035	0.143	−0.179	−0.401	–
Worthless	−0.029	0.119	−0.149	−0.333	0.832
Everything is effort	−0.029	0.120	−0.150	−0.335	0.836
Depressed	−0.031	0.125	−0.157	−0.351	0.876
Restless	−0.026	0.107	−0.135	−0.301	0.751
Hopeless	−0.030	0.123	−0.154	−0.345	0.861
Nervous	−0.025	0.100	−0.125	−0.280	0.698

Abbreviation: SC, social capital.

DISCUSSION AND CONCLUSION

On the basis of the analysis we have presented, we can provide some tentative responses to our earlier stated hypotheses. First, we expected a positive statistical relationship between the COVID‐19 outbreak and psychological distress (a substantively negative outcome), as indicated by the Kessler‐6 instrument, when students returned to campus. We found weak but significant support for this hypothesis. Moreover, we found strong evidence for the coherence of the distress measurement portion of the model, with the six indicators providing high path coefficients to the latent distress variable.

Our second hypothesis was about the mediating effect of individual social capital. We found that, while the outbreak could reduce individual social capital, for those students with higher levels of individual social capital, they could experience less psychological distress. The outbreak had a weak negative effect on individual social capital, we presume, because personal relations were disrupted by the ensuing lockdown and social distancing that took place in the wake of the outbreak. That the relationship is weak actually suggests that students were fairly resilient at maintaining their social ties.

For our third hypothesis, we tested the idea that there would be racial and gender‐based disparities in terms of social capital, with white and female students having the advantage over their peers. Our model found evidence for a racial disparity favoring white students, but we did not find a gender effect.

Our fourth and final hypothesis was about the mediating effects of social capital and social disparities, in which we predicted, owing to higher levels of social capital, that white and female students would have lower psychological distress. Our table of total effects showed that a weak racial disparity exists that favors white students, while the gender disparity showed it was males who fared better in terms of having lower levels of reported psychological distress.

Overall, our findings fall in line with the extant literature on social capital, COVID, and college students with the exception of the unexpected reverse finding on gender disparity.

Limitations and future research

The main limitation of this study is that it is based on cross‐sectional survey data. This means that we could not model individual‐level change over time. At the same time, the timing of the data collection was opportune, given the bulk of data was collected during lockdown conditions. A second limitation stems from the inability to separate outbreak effects from other college‐level effects—a typical outcome of a pseudo‐experimental design. At the same time, real‐world experiments rarely live up to the rigors of true experimental design. Nevertheless, our findings should be interpreted in light of this limitation.

Future research should continue investigating the predictors of psychological distress. While we have explained a quarter of the variation statistically, there is much to learn. We hope future research will build upon this initial investigation with a goal devoted to maximizing the predictive value of the model.

PEER REVIEW

The peer review history for this article is available at https://publons.com/publon/10.1002/jcop.22731