Do Race, Ethnicity, Citizenship and Socio-economic Status Determine Civic-Engagement?

Circle Working Paper # 62

Do Race, Ethnicity, Citizenship and Socio-economic Status Determine Civic-Engagement?

J. Foster-Bey

Introduction
The Corporation for National and Community Service is an independent federal agency whose mission is to promote service, volunteerism and civic engagement. It pursues this mission in two ways: 1) by directly supporting and funding community and national service programs, through its AmeriCorps, VISTA, Senior Corps, Learn and Serve America, and the National Community Civilian Corps (NCCC) funding streams. These programs serve young people, adults, senior citizens, and individuals with disabilities; 2) by encouraging non-profits and community leaders to promote community volunteerism and civic engagement.
The Corporation recognizes that racial and ethnic minorities, immigrants, low-income youth, non-college educated individuals and the disabled may be underrepresented in service and volunteering programs as well as in other civic engagement activities. In response, the Corporation has decided to make increasing access to service and volunteering for individuals from disadvantaged circumstances one of its strategic focus areas. In order to pursue this focus the Corporation is exploring ways to expand its outreach to underrepresented communities and to build partnerships aimed at addressing the challenges involved in recruiting and retaining underrepresented groups in service and volunteering programs and activities. This paper is aimed at providing background information that will hopefully prove helpful in forming these partnerships.

Background
Research literature indicates that volunteering and other forms of civic engagement— such as voting, participating in community meetings, membership in community associations, and writing letters to the editor—can be predicted by differences in an individual’s education, family income, and race. But some argue that these differences are artifacts of how various surveys measure volunteering and civic engagement. For example, surveys on volunteering may have a certain cultural and class bias because the focus is on unpaid service in formal non-profit organizations. However, many racial
and ethnic minorities may engage in helping behavior in more informal settings, such as helping one’s neighbor or working on a community problem outside of a formal nonprofit.
Using the 2005-2007 Current Population Survey’s Annual Volunteer Supplement, this paper will briefly examine whether certain disadvantaged socio-economic groups are under-represented in volunteering and civic engagement as the research suggests. One challenge, however, is that there is no clear consensus on the best way to measure disadvantage. Because of the racial history of the United States and the disproportionate economic challenges that racial and ethnic minorities continue to face, race and ethnicity are often used as a proxy for social and economic disadvantage. However, what often appear to be racial and ethnic disparities are actually reflections of class and income differences. For some ethnic groups economic disadvantage is further confounded by differences in citizenship and immigration status. Rather than select one measure of disadvantage, this paper will examine volunteering and civic engagement by race and ethnicity, educational attainment, family income, and citizenship status. In addition, the paper will assess variations in volunteering and civic engagement by race and ethnicity, and citizenship status while controlling for differences in family income and education.
Finally, the paper will measure civic engagement using the following measures:

• Percentage of survey respondents who volunteer with a formal non-profit. This provides a measure of the formal volunteer rate by group;
• Percentage of survey respondents who do not volunteer but attend community meetings or work on community problems. This will tell us whether disadvantaged groups exhibit a preference for community-oriented work over what is generally accepted as volunteering with a formal organization, such as a non-profit, school, political or community organization, or religious organization;
• Percentage of individuals who both volunteer and either attend community meetings or work to fix a problem in their neighborhoods. This should provide a better understanding about how volunteers from some groups are also highly engaged community actors, and,
• Percentage of all respondents—volunteers and non-volunteers—who attend community meetings or work on community problems. This provides another view of group differences in civic engagement.

The Effects of Race, Ethnicity, and Citizenship on Civic Engagement
Table 1 provides an analysis of all four measures of civic engagement by race, ethnicity, and citizenship status. The most immediate finding is that there are strong differences in civic engagement by race, ethnicity and citizenship status. Whites are much more likely than blacks, Hispanics or Asians to be civically engaged using any of the four measures. Asians and Hispanics consistently appear to have low rates of civic engagement using any of the four measures. They also appear to be much less actively engaged in their communities as compared to either whites or blacks.

Some of the civic engagement and social capital literature suggests that immigrants and limited-English speakers may be less civically engaged than non-immigrants and native English speakers. Given that some racial-ethnic groups are more likely to be either immigrants or limited-English speakers or both, ethnic differences in volunteering and community participation may reflect the unfamiliarity with the language, lack of knowledge about dominant cultural norms, or lack of access to community volunteer and service opportunities. One way to test whether immigrant status and English language limitations influences volunteering and service is to assess whether there is a relationship between citizenship status and observed differences in volunteering and civic engagement. Table 1 indicates that volunteering and community participation can be predicted by differences in citizenship status. On almost every measure of civic engagement, non-native born US citizens rank well below native born US citizens.

Table 1 provides evidence that race, ethnicity and citizenship status predict civic engagement. However, it is possible that these demographic factors may actually serve to confound other more important differences. For example, blacks and Hispanics have lower average incomes and education compared to whites and Asians. It is possible that the observed differences in civic engagement really reflect income and educational variations and not simply race or ethnicity or citizenship.

Civic Engagement by Income and Education

Figure 1 indicates a very strong positive relationship between civic engagement measures and educational attainment. High levels of education predict high levels of civic engagement. The relationship appears strongest for respondents who volunteer with formal organizations.
Figure 2 presents civic engagement by family income. It shows that increases in civic engagement are strongly predicted by increases in family income. The relationship is
strongest for formal volunteering. The relationship between family income and civic engagement, while substantial, does not appear to be as important as the link between education and civic engagement. The findings here suggest that in addition to race, ethnicity and citizenship status, education and family income also predict civic engagement.

Controlling for Multiple Factors
It seems clear that the propensity to volunteer is confounded by several different factors. In order to sort out some of the complexity, the following table looks at race, ethnicity and citizenship status while controlling for education and family income. Table 2 shows that, even after controlling for family income and education separately, race and ethnicity are still important predictors of civic engagement. Whites consistently demonstrate higher levels of civic engagement on every measure except one.
Moreover, even after controlling for education and family income, Hispanics and Asians remain less likely than either whites or blacks to be civically engaged.

Table 3 presents the same analysis as Table 2 but using citizenship status as the predictor. Again the evidence is quite striking. Even after controlling for family income or education, citizenship status is a powerful predictor of civic engagement. Native born US citizens have higher levels of civic engagement on all three measures than either foreign born citizens or non-citizens. The implication is that beyond differences in education and income there is something about citizenship status which affects an individual’s
propensity to be civically engaged.

Finally, in order to determine the relative importance of each of these group characteristics on the propensity to be civically engaged a logistic regression was performed on each of the four civic outcome measures controlling for race, ethnicity, education, family income, and citizenship status. Logistic regression is used to examine the simultaneous impact of multiple factors on a categorical outcome variable—in this case civic engagement. The regression examines whether there are racial, ethnic differences after controlling for age, gender, education, family income, and citizenship status.1 Table 4 presents the results of the logistic regressions for each of the four civic outcome measures:

• Formal volunteering vs. non-volunteers
• Those engaged in community activities only vs. all other respondents
• Those engaged in both volunteering and community activities vs. those who do not do both
• Those engaged in community activities vs. those not engaged in community activities

Logistic regression analyzes how different characteristics increase or decrease the odds of an event occurring holding all other characteristics constant. In this case, the event is whether an individual is likely to participate in one of the four civic engagement outcomes. Each entry in table 4 is an odds ratio. That is, the factor in the table shows how different characteristics influence the odds of a particular outcome occurring. For example, table 4 shows that being female increases the odds of being a formal volunteer by over 50% relative to being male2. On the other hand, being Latino relative to non-Latino reduces the odds of being a formal volunteer by 24%. Table 4 shows that after controlling for age and gender, citizenship, education, income3, and race and ethnicity have independent effects on civic engagement. Even when holding education, family income, citizenship, age and gender constant, blacks, Latinos and
Asians are less likely than the reference group to be civically engaged, while whites are more likely to be civically engaged than the reference group. It is also possible to say that even after controlling for age, gender, race and ethnicity, non-citizens, less educated and low income individuals are less civically engaged. The regression supports the story that some groups are less civically engaged. But the analysis does not tell us why these group differences exist.

Table 2, 3 and 4 show that race, ethnicity, education, income, and citizenship status have independent effects on civic engagement even after controlling for other characteristics:

• Table 2 shows that blacks, Asians and Hispanics have lower rates of civic engagement even after controlling for either income or education differences.
• Similarly, table 3 shows non-citizens and non-native born citizens have lower rates of civic engagement compared to native born citizens even after controlling for either income or education differences.
• Finally, table 4, controls for several factors at the same time. It shows that holding all other factors constant not having a 4-year college degree, having low income, being a member of a minority race or ethnicity, and not being a native born citizen each reduces the odds of civic engagement,

Accessing Service
The above analysis strongly supports the story that there are group differences in civic engagement based on race, ethnicity, citizenship status, education and family income.
However, these group differences may be the result of supply side factors such as different preferences for civic participation among groups, differential access to the resources needed to participate civically, such as time, or lack of information about the opportunities to be civically engaged. Group variations in volunteering and services may also be the results of demand side factors such as fewer available opportunities to serve, or the failure by non-profits to ask certain groups to serve or volunteer. Unfortunately, it is difficult to use the CPS Volunteer Supplement to directly analyze these possible supply and demand factors. However, it may be possible to indirectly assess some of these issues.

Table 5 presents selected volunteer characteristics by race, education and family income. The table shows percentage of individuals who volunteer by selected organization types, the intensity of their volunteer commitment, and whether or not they became volunteers because someone asked them to volunteer.
The type of organizations where individuals volunteer may provide some insights into both preferences and opportunities to serve. Even after controlling for education and family income, blacks appear to be more likely to volunteer with religious organizations than either whites, Asians or Latinos. On the other hand, Latinos appear to be more likely to volunteer in educational and youth service organizations than blacks, whites or Asians. This might suggest that if blacks and Latinos, for example, prefer serving in religious and educational organizations their volunteering may be affected by the availability of volunteer opportunities in these types of institutions in their communities. It might also indicate that the best way to reach these groups is through these organizations.
There appears to be only marginal group differences between blacks and whites in the intensity of volunteer commitment—i.e. volunteering over 100 hours annually and volunteering 12 weeks or more per year. However, Latinos and Asian volunteers seem to generally have a lower level of volunteer intensity than either blacks or whites. This may indicate that non-profits have more difficulty engaging Latinos and Asians in volunteering.
Finally, while there is no data in the CPS Volunteer Supplement on whether all respondents were asked or not asked to volunteer, there is information on whether or not individuals who did volunteer were asked to volunteer. This provides some insight as to whether non-profits are more or less likely to ask some groups to volunteer. While whites appear slightly more likely to have become a volunteer because someone asked them, it appears that there is very little differences between whites, African-Americans and Latinos in whether or not they became volunteers because someone asked them to volunteer. On the other hand, Asians seem to be noticeably less likely than other groups to have become a volunteer because someone asked them to volunteer.

Dropping Out of Civic Engagement
In any given year, the actual observed number of civically engaged individuals is the sum of the number of individuals who continue participating in civic activities from the previous year and the number of new individuals recruited into civic activities in any given year. The number of people who fail to continue participating in civic activities has a big impact on civic participation rates. This “leaky bucket” phenomenon may also disproportionately affect the civic participation rates of certain groups.
Table 6 presents differences in the civic engagement attrition rates by race and ethnicity controlled by education and family income. Attrition is defined as the number of individuals who volunteered, attended a community meeting or worked to fix something in their neighborhood in one year but failed to continue this activity the next year. Attrition may be a measure of burnout or lost opportunities.

Controlling for education and family income non-Hispanic whites are least likely to drop out of civic activities from one year to the next, while Hispanics are most likely to discontinue civic activity4. The lower rates of civic engagement for non-Hispanic blacks and Hispanics compared to non-Hispanic whites may be driven by very high attrition rates. Interestingly, one measure where African Americans and Hispanics have lower attrition rates than non-Hispanic whites is “volunteers who engage in community activity.” It appears that black and Hispanic volunteers with 4-year college degrees are less likely to stop engaging in community activity from one year to the next than similarly situated whites. High income black volunteers are also less likely to stop engaging in local community activity than either similarly situated Hispanics or non-Hispanic whites.

Finally, table 7 presents the attrition rates for the three measures of civic engagement.
For the most part Native born US citizens tend to have lower civic engagement attrition rates than either foreign born citizens or non-citizens. In formal volunteering the differences are generally sizable and consistent—citizenship predicts lower attrition rates and being foreign born increases the likelihood of not being retained as a volunteer. For the other two measures, being native born seems to predict lower attrition rates but the relationships are far from consistent.
Table 6 and 7 analyze the two-year attrition rate for civic engagement—i.e. the percentage of individuals who dropout of civic activity within a two-year period.—by our measures of disadvantage—education, income, race/ethnicity, and citizenship status.

• In general, compared to non-Hispanic whites, attrition rates for blacks and Hispanics are higher across all measures of civic engagement used in this study. These higher attrition rates hold even after controlling for income or education.
• In general, compared to native born citizens, non-citizens and non-native born citizens tend to have higher attrition rates for formal volunteering, However, the findings are mixed for other forms of civic engagement.
• Black and Hispanic volunteers with 4-year college degrees are less likely to stop engaging in community activity from one year to the next compared to similarly situated whites.
• Compared to similarly situated Hispanics and non-Hispanic whites, high income black volunteers are also less likely to stop engaging in local community activity.

Conclusion
This paper provides descriptive data on differences in civic engagement between advantaged and disadvantaged groups. Because there is not always consensus on what measure best describes disadvantaged, this paper used several indicators—race, ethnicity, citizenship status, family income and educational attainment. Major findings from this data indicate that:

• Family income and education predict both likelihood of civic engagement and the rates of attrition.
• Higher levels of income and education predict higher civic participation and attrition rates.
• Whites tend to have higher rates of civic engagement than blacks, Hispanics or Asians, and they also have lower attrition out of civic activities from one year to the next. Similarly, native-born citizens have higher rates of civic engagement and lower attrition than immigrants.
• The findings for race, ethnicity and citizenship status hold even when family income and educational attainment are considered.

The findings suggest that socio-economic status (SES)—defined by education and family income—are important predictors of civic engagement. Lower SES individuals seemed less civically-engaged than individuals from higher SES groups. However, it appears that community and cultural factors may also play a role in predicting civic participation.
Racial and ethnic minorities and immigrants are much less likely to be civically engaged than whites or native-born citizens.
The findings also suggest that the racial/ ethnic and socio-economic differences in volunteering rates would not be appreciably affected by counting community-based activities such as attending meetings or working on neighborhood problems as a less organized, but still important, form of volunteering. It appears that whites, native-born citizens, the well educated and high income groups are not only more likely to volunteer, but they are also more likely to attend a community meeting or work on a neighborhood problem than racial and ethnic minorities, immigrants and low-income and working-class
individuals.
Finally, this analysis cannot determine whether the observed racial, ethnic, citizenship status and SES differences in civic engagement are the results of supply or demand side issues. It is possible that certain groups are simply less committed to civic engagement.
On the other hand, it may be that certain groups—racial and ethnic minorities, immigrants, low SES individuals—face more and greater barriers to civic participation than native-born citizens, whites, or individuals from high SES groups.

Technical Appendix
Logistic Regression is a type of predictive model that can be used when the target variable is a categorical variable with two categories – for example volunteer or not volunteer. The dependent variable in a logistic regression has a value of either 1 or 0. In this instance, 1 means the respondent is a volunteer and 0 means the respondent is not a volunteer.
As with ordinary least squares regression, a logistic regression model is used to determine the effects of one variable while holding other variables constant. For example, what is the effect of educational attainment on the decision to volunteer when controlling for all other relevant characteristics—such as age, gender, race, etc?
Logistic regression produces coefficients for each variable in the model. The coefficients indicate how a one unit change in the independent (or predictor) variable affects the target or dependent variable. Using volunteering as an example: How does a one year increase in age influence the decision to volunteer? However, unlike linear regression, the coefficients in a logistic regression represent the independent variable’s impact on the log-odds. So for interpretation purposes, logistic regression coefficients are converted into odds-ratios. They can then be interpreted as the effects a particular independent variable has on the odds of a particular event occurring while holding all other factors constant.

Numeric example:

Odds = probability of volunteering / (1-probability of volunteering)

Odds ratio = Female odds of volunteering / Male odds of volunteering; or Hispanic odds of volunteering/ non-Hispanic odds of volunteering

Volunteer odds for entire population = .43

Female to male odds ratio = 1.5

Hispanic to non-Hispanic odds ratio = .76

Holding all else constant, the odds of being a volunteer increase for a female compared to a male by .43 *1.5. On the other hand, holding all else constant, for a Hispanic compared to a non-Hispanic the odds of being a volunteer decline by .43*.76

The logistic regression model used in this paper takes the following form:

Civic engagement Indicator (dependent variable) = Age + Gender + Race + Hispanic +

Education + Family Income + Citizenship Status

The following variables were used in the model:

Dependent or target variables are:

• Volunteered in formal organization, 1 if yes 0 if no
• Did not volunteer but attended community meeting or worked on neighborhood problem, 1 if yes 0 if no
• Volunteered in formal organization and attended community meeting and or worked on neighborhood problem, 1 if yes 0 if no
• Attended community meeting or worked on neighborhood problem, 1 if yes 0 if no.

Independent or Predictor variables:

• Age – continuous variable measuring age at time of survey
• Black – respondents who self-identify as non-Hispanic black or African American 1 = yes, 0 if no
• White – respondents who self-identify as non-Hispanic white, 1= yes, 0 = no
• Asian – respondents who self-identify as non-Hispanic Asian, 1= yes, 0 = no
• Hispanic – respondents who self-identify as Hispanic or Latino, 1= yes, 0 = no
• Female – female =1 male = 0
• Education – three point scale 1 = high school or less; 2 = some college; 3 = 4-year college graduate
• Family Income – four point scale: 1 = income in bottom quartile; 2 = income in second quartile; 3 = income in third quartile; 4 = income in top quartile.
• Citizenship Status – three point scale: 1 = non-US citizen; 2 = non-native born US citizen; 3 = native born US citizen