Charts showing the spread of the novel coronavirus (COVID-19) have become, unfortunately, all too familiar. Outbreaks first struck the West Coast, then the New York metro area, and soon hot spots spread around the country.
It is a bitter irony that this contagion weakens us exactly where we are strong, capitalizing on group activities, religious and neighborhood gatherings, and large events to spread exponentially. The response to the rapid spread of this disease has been to practice social distancing and curb those kind of activities—socializing, services, sports, schools—that are at the heart of social capital.
One train of thought holds that localities with stronger social life may be at higher risk for exposure to coronavirus. Some have suggested that communities that had a harder time putting on hold social interactions at church, schools, and playgrounds could have allowed the virus to spread faster and without notice. Globally, for example, the Wall Street Journal found that religious gatherings often served as “super spreading” events, where just one attendee without symptoms could infect dozens or hundreds of unsuspecting fellow worshipers.
Others have argued that more social trust may actually mean that communities would be more responsive to requests to social distance and place activities on hold out of a broader desire to watch out for one another. A preliminary paper from York University, using data from the Social Capital Project, suggested that states with higher levels of social capital may have seen quicker policy responses to the spread of COVID-19.
Now that the virus has worked its way across the United States, some characteristics of states that have been hit hardest to date can be preliminarily identified. The confirmed cases are those tallied by the Johns Hopkins University Coronavirus Resource Center. Due to differences in testing regimes across states, the absolute number of caseloads should not be treated as a measurement of true prevalence of the disease. The Social Capital Project’s Social Capital Index, which featured in our prior report “The Geography of Social Capital in America,” is made up of six sub-indices, which can be used independently to examine different dimensions of social life.
Our “social support” sub-index incorporates information on the share of adults who say they always “get the social and emotional support [they] need,” the share of adults who do favors for neighbors at least monthly and say they trust most or all of their neighbors, and the average number of “close” friends that adults report having. (More information on the Index is available here.) A simple linear regression shows the correlation between our social support sub-index and coronavirus caseloads.
Although correlation, as any social scientist would quickly remind you, does not necessarily imply causation, the simple correlation can still provide interesting preliminary information. For example, this graph indicates an association between higher levels of interpersonal social support and fewer reported coronavirus cases. A one-standard-deviation decrease in the social support sub-index is associated with a 24% increase in the number of confirmed COVID-19 cases. That relationship is statistically significant—that is, it is considered unlikely to occur by chance in the data if there were no relationship between the measures in the full population—and it remains so after controlling for state GDP, days since the first confirmed case, and region.
The other social capital sub-indices don’t show similar patterns. This suggests that if there is a pattern between social capital and confirmed cases, it could have something to do with the quality of relationships, not just the prevalence of non-profit institutions, intact families, or other indicators of social capital. To cite one indicator that is not part of the index, the relationship between self-reported weekly church attendance and COVID-19 cases is a weakly negative one that is not statistically significant. Instead, the strongest relationship appears to be in the interpersonal dimension of having close friends, interacting with neighbors, and feeling part of a community that provides emotional support.
The peak in coronavirus deaths has not yet hit most U.S. states, according to most epidemiological models, so death data should be treated as even more provisional than the confirmed cases. But even provisionally, the same pattern between higher levels of interpersonal social support and lower death rates presents itself. A one-standard deviation decrease in social support is associated with a 36% increase in deaths, and the relationship remains statistically significant even adding in controls. Breaking it out by region, the correlation between the two variables is negative in all regions, and is statistically significant in the Northeast and the Midwest.
To reiterate, none of this analysis should be taken as proving a causal relationship. There are a multiplicity of factors that contribute to a state’s coronavirus risk (integration into global trade and travel, demographics, weather) and the willingness of the state’s population to engage in social distancing (trust in state and local governments, population density, political valence). These confounding aspects could outweigh the effect that interpersonal trust and sense of neighborly solidarity has on willingness to halt social gatherings and economic activity. The graphs are also only as reliable as the data being reported by states.
While the data needed to construct a social support sub-index at the county level is not available, our county-level social capital index is correlated with the number of COVID-19 cases. Many counties have had no major outbreaks so far, but viewing the relationship by region shows a similar pattern in the Northeast, South, and Midwest. (A similar pattern holds for deaths by county, though there are even more counties that report having had zero deaths so far.)
Social capital is a notoriously fuzzy thing to measure, and there is still much to learn about the spread of COVID-19. These numbers will be worth revisiting once the worst of the crisis has passed, and researchers may yet be able to find a clever strategy for identifying the causal effects of social capital on the spread of COVID-19. A new National Bureau of Economic Research working paper found that online social networks have a predictive role in ascertaining the spread of the disease; more socially-driven links in behaviors and distancing could one day be studied more rigorously as well.
Again, these simple correlations do not prove much. They are, however, what might be expected if a strong sense of community and interpersonal support did correspond with a tendency to put others first and adopt voluntary social distancing. In thinking about social institutions, we often tend to think of formal, brick-and-mortar organizations and associations. But organic relationships and the daily interaction within neighborhoods are just as, if not even more, important to keep in mind during the ongoing analysis of this pandemic by academics and policymakers.
The coronavirus, our societal response, and the eventual rebuilding efforts will change much about our economy, our society, and the way we interact with each other. A greater attentiveness to interpersonal support, like doing favors for neighbors, would be a welcome outgrowth of this time when society has become especially aware of how vital communities are to a life lived to its fullest.
Patrick T. Brown
Senior Policy Advisor