Who is my neighbor? The effect of community racial in-group representation and residential isolation on volunteering

ABSTRACT

Do individuals residing in racially diverse communities volunteer less than individuals living in homogenous communities? While a growing body of literature explores the relationship between diversity and trust, we know less about how racial diversity affects volunteering. Drawing on the Current Population Survey’s (CPS) volunteering supplement combined with county-level Census data, this article explores how racial diversity moderates the relationship between individual race and volunteering behavior. We find that context moderates the effect of individual race/ethnicity on volunteering. However, these moderating effects differ across racial/ethnic groups. Greater in-group exposure is associated with a decline in volunteering propensity among Blacks, but it does not have a statistically significant impact on whites and Latinos. At the lowest levels of residential exposure/segregation, the likelihood of volunteering for Blacks is statistically indistinguishable from that of whites, but the likelihood of volunteering for Blacks increases as residential in-group exposure increases until it is statistically indistinguishable from the likelihood of volunteering for Latinos. Racial in-group representation does not have a statistically significant impact on the likelihood of volunteering for whites and Blacks, but greater in-group representation decreases the likelihood of volunteering for Latinos.

KEYWORDS:

• Volunteering
• racial diversity
• in-group theory
• constrict theory
• civic voluntarism
• participation
• social contact theory

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 We exclude from our sample all volunteering activity that the respondent indicated was required by a school or court and all volunteering for professional or labor organizations.

2 We limit our sample only to those respondents identifying as one race (i.e. White-only, Black-only, Asian-only, or American Indian-only). Eighty percent of the respondents identify as white, 12% are Black, 7.6% are Asian, Hawaiian, or Pacific Islander, and less than 1% of respondents are American Indian. We drop all respondents who identify as more than one race (10,723 cases).

3 Asian, Hawaiian, Pacific Islander and American Indian (Latino and non-Latino) respondents constitute the remaining 7.7% of the sample.

4 It is important to note that our sample includes a small number of individuals who are less than 18 years of age. While concerns may be raised that these “youth” are already capture as part of the family unit, the CPS methodology only includes those under 18 if they are the person who owns or rents the home. We also note that we exclude all voluntary activity indicated by the respondent to be a requirement of a school or court.

5 Types of multi-level modeling include multi-level logit (Andersen and Milligan 2011; Tolsma, Van der Meer, and Gesthuizen 2009; Traunmüller 2011), hierarchical linear modeling (Glanville, Paxton, and Wang 2016; Hooghe 2007; Oliver 2009; Rocha and Espino 2009; Rotolo and Wilson 2011; Traunmüller 2011; Tolsma, Van der Meer, and Gesthuizen 2009), hierarchical non-linear modeling (Rotolo and Wilson 2014), and non-hierarchical linear models (Rotolo 2000).

6 All models presented are estimated using Stata 14.

7 Given the high correlation between these independent variables reported in Table 2, we conducted additional analyses to diagnose problematic multicollinearity. Results of our perturbation analysis (Hendrickx 2004; Belsley 1991) suggest that these coefficient estimates are not substantially affected, thereby alleviating this potential concern.

8 Values of the key independent variables used to predict volunteering behaviour in each scenario are indicated in each figure, with the value of all other variables held constant at their sample means.