Predicting Civic Engagement in Urban High School Students

Jefferson County Public Schools (JCPS) is a metropolitan school district that includes more than 150 schools serving approximately 98,000 students. Fifty-five percent of students in the district come from economically disadvantaged homes and qualify for free or reduced-price lunch. Since the 1996-1997 academic year, JCPS has conducted a Comprehensive School Survey (CSS) of students in Grades 4-12, parents, and staff to measure their perceptions about factors associated with school climate. More than 100,000 questionnaires are returned each year.

In 2007-2008, the CSS was revised to take a whole-child approach and to include a more comprehensive set of civic engagement measures. Many less-informative items were deleted from the two-page questionnaires to make room for new items to augment and strengthen constructs and other indicators of whole-child development generally and civic engagement specifically. The new conceptualization frames CSS within the most fundamental educational context: the personalized engagement and nurturing of the whole child. In the commitment to educating the whole child in JCPS, the survey captures data that go beyond the common focus on academics. The new CSS weaves together the threads that connect not only reading, writing, math, science, social studies, practical living, and arts/humanities, but also the important socioemotional, civic, and moral connections that tend to be fragmented in a more accountability-oriented approach.

The choice of civic engagement items for the CSS questionnaires was based on a conceptual framework derived from current understandings of youth civic engagement, as well as other considerations. Civic engagement is generally understood to mean working for the betterment of one’s community. The American Psychological Association, for example, defines civic engagement as “individual and collective actions designed to identify and address issues of public concern” (American Psychological Association, 2009). The National Conference on Citizenship views civic engagement in terms of electoral activity (e.g., voting, working for a political candidate, contributing money to a campaign), service activity (e.g., volunteering, belonging to a community organization), and citizen-centered activity (e.g., attending a public-issue meeting, working with others to solve a problem) (National Conference on Citizenship, 2008).

In constructing a conceptual framework to represent civic engagement, a primary concern was to include indicators of participation in both politics and civil society (Table 1). Political engagement involves activities such as voting, campaigning, and contributing money to political candidates or parties. The term “civil society” has carried a variety of meanings since its origin in the time of Aristotle (Edwards, 2009; O’Connell, 1999). We use the term broadly to cover engagement in nonpolitical, civil institutions such as religious organizations, youth programs, other community-based groups (London School of Economics, 2004), as well as individual performance of the service-oriented work that these institutions generally do. Civil and political systems, however, are not the whole of civic life. Civic engagement often originates with persons trying to solve a public problem, either local or more distant. Sometimes they solve the problem on their own. Other times they collaborate with political or civil entities (e.g., a city council, the Red Cross) or both. We use the term “problem-solving activity” for this type of civic engagement that begins with citizens working to solve a problem and often involves both the civil and political systems. The civil, political, and problem-solving activities described by our framework roughly parallel the service, electoral, and citizen-oriented classifications of civic engagement defined by the National Conference on Citizenship (2008).

A second concern was to include measures of both individual and social action. Political engagement is fundamentally a social activity, which can be cooperative (e.g., cooperating to collect petition signatures), competitive (e.g., competing with others over the allocation of government funds), or both (McIntosh & Youniss, in press). Yet political engagement also involves solitary acts, such as voting or contributing money to a political candidate or cause. The same conceptualization applies to civil society. Individuals work together in parent-teacher associations, church committees, and other community groups. But they also perform individual civil acts such as recycling their newspapers, shoveling snow from the sidewalk of an elderly neighbor, or donating canned goods to a food pantry. In summary, we considered that a well-rounded citizen would be capable of engaging in civil, political, and problem-solving activities, both individually and socially, and our conceptual framework of civic engagement reflects these ideas.

The full conceptual framework calls for six indicators of youth civic engagement (Table 1), and we selected one measure for each of them. For civil activity, we chose environmental conservation and community service to measure individual and social action, respectively (although environmental conservation may lead to social action, and community service is sometimes performed alone). Also, community service in youth has been linked to increased civic participation in adulthood (e.g., Youniss, McLellan, & Yates, 1997).

As measures of political activity, we chose the variables intention to vote and political discussion to measure individual and social action, respectively. The voting item was worded to capture intention—“planning to vote”—because most high school students are under the legal voting age. Longitudinal research suggests that the intention of high school youth to vote in the future is highly correlated with actual voting in adulthood. For example, Campbell (2007) found that 84% of high school seniors who indicated on the Monitoring the Future Survey that would vote in the future reported 10 years later that they actually did vote. We selected political discussion because it is real political activity that youth can participate in, even when they are too young to vote in public elections. In addition, participation in political discussion is strongly linked to other civic-related outcomes. Adolescents who frequently talk about political and other current events with their parents score higher on measures of political knowledge, news monitoring, and other civic outcomes and, when they enter young adulthood, vote, volunteer, and engage in other civic activities more frequently than do youth who seldom discuss politics with their parents (Andolina, Jenkins, Zukin, & Keeter, 2003; McIntosh, Hart, & Youniss, 2007). Similarly, youth whose teachers discuss politics and current events with them, orchestrate youth discussions of controversial topics, and create “open classrooms” where diverse student opinions are heard and respected score higher than other students on measures of political knowledge and other civic outcomes (Andolina et al., 2003; Torney-Purta, Lehmann, Oswald, & Schulz, 2001).

As measures of problem-solving activity, we selected the variables willingness to contact official about an issue of concern and personal efficacy. The former is a measure of an individual’s disposition to take the initiative to solve a public problem. Personal efficacy—“I have the ability to make a difference in my local community”—captures the belief that one has the ability to change some aspect of the environment, be it a social, political, or other component (Berman, 1997). Although the relationship between efficacy and civic action is complex, studies of social and political activists suggest that persons who feel more personally efficacious tend to participate in political, civic, and other prosocial activities to a greater degree than those who feel less efficacious (Colby & Damon, 1992). Similarly, research suggests that adolescents who score high on measures of political efficacy score higher than other youth on several civic measures such as community service and public communication skill (McIntosh, Hart, & Youniss, 2007).

In summary, the conceptual framework for indicators of civic engagement reflects the desire that students who graduate from JCPS high schools will be well-rounded citizens capable of acting individually or collectively in civil and political arenas or to help solve problems of whatever community they settle in as adults.

What factors predict youth civic engagement? In addition to the set of civic outcome measures, the revised CSS questionnaires contain items that allow us to explore factors that might predict civic engagement, including a range of school-related factors (e.g., academic grades, school engagement, sense of belonging), extracurricular activities at school and in the community, and student background characteristics (e.g., gender, ethnicity). We were particularly interested in four factors which, research suggests, may enhance youth civic engagement: positive character, conflict resolution skill, school discussion climate, and participation in nonsport extracurricular activities.

Theoretical research suggests that character education fosters the civic development of adolescents (Berkowitz & Bier, 2005; Berkowitz, Sherblom, Bier, & Battistich, 2005). Empirical research provides some support for this suggestion in finding that character education is associated with democratic values (Berkowitz & Bier, 2004), prosocial behaviors, civic engagement, and citizenship (Character Education and Civic Engagement Technical Assistance Center, 2008). However, measures of positive character are not often found in studies of civic engagement, and we wanted to identify specific civic outcomes most closely linked with positive character and to learn where positive character falls among other factors in terms of overall importance to civic engagement.

Conflict resolution skill is promoted in schools as a means of reducing violence among students and is sometimes incorporated into civics curricula (e.g., Constitutional Rights Foundation Chicago, 2006). The skill requires an ability to take another’s perspective, which is an important role in social development generally (Selman, 2003) and in political engagement specifically (McIntosh & Youniss, in press). We therefore expected to find conflict resolution skill positively associated with political activity measures and perhaps with measures of civil and problem-solving activity as well.

School discussion climate indicates the degree to which students and teachers feel comfortable expressing their opinions on current political and social issues. The more “open” the discussion climate, the more comfortable students feel in these discussions. Research indicates that the openness of the school discussion climate is linked to youth intentions to become politically active in the future. For example, using U.S. data from the Civic Education Study in 28 countries conducted by the International Association for the Evaluation of Educational Achievement, Campbell (2005) found that ninth-graders who perceived their classrooms to have a relatively open environment scored higher on measures of intention to become civically engaged in the future (e.g., will vote in national elections, write letters to newspapers about a political or social issue, volunteer to help the poor or elderly) than did other students. We therefore expected to find school discussion climate linked to a number of civic outcome measures.

Research indicates that participation in organized youth activities is positively associated with numerous areas of youth development, including civic engagement (Mahoney, Larson, Eccles, & Lord, 2005; Youniss et al., 1997). Participation in extracurricular activities such as student government, performing arts groups, and science or math clubs is linked to increased participation during adulthood in voting, contacting public officials, and other political activity, as well as participation in voluntary community organizations (Hanks & Eckland, 1978; Hart, Donnelly, Youniss, & Atkins, 2007; Otto, 1976; Verba, Schlozman, & Brady, 1995). Similar links between participation in 4-H, Scouts, YMCA/YWCA, or other community-based youth programs and participation in adulthood in civil, political, and other community organizations have also been demonstrated (Ladewig & Thomas, 1987). However, some research has found sports extracurricular activities to be negatively associated with civic engagement (Verba et al., 1995). We therefore created separate measures of sport and nonsport extracurricular activities.

While we expected many of our predictor variables to show positive links to civic outcomes, we were especially interested in determining which ones were more important in terms of breadth of relationship (i.e., associated with several civic outcomes), strength in predicting overall civic engagement (i.e., effect size), and consistency (i.e., showing similar effects 2 years in a row).

CSS data used in this study were gathered in 2008 (during the 2007-2008 school year) and in 2009 (during the 2008-2009 school year) from students in the district’s 21 high schools. Questionnaires were completed by 16,390 (60%) students in 2008 and by 21,100 (78%) students in 2009. The increase in response rate over the 2 years resulted largely from a change in the data collection procedure, which went from a paper-and-pencil system to a Web-based, technology-driven system based on the tailored design method (Dillman, 2000).

Table 2 presents characteristics of students in the two survey groups. Students were divided about equally between males and females. Whites constituted the major ethnic group (about 55%) and Blacks accounted forabout a third of the student population. Around 40% of students received free or reduced-price lunch, and about one in six students lived in a family with only one adult. (JCPS census data for October 2008 indicated 50% females, 56% Whites, 35% Blacks, and 47% free or reduced-price lunch among all high school students). Several students in each of the 2 years indicated no adult in their family, a finding which may reflect homelessness among students. Student homelessness has increased in the district in recent years, rising from 3,892 students in 2003 to 7,341 in 2009.

The community service measure is a dichotomous (yes = 1, no = 0) variable asking students if they are currently performing, or have in the past performed, service to people or other work “to make my community a better place.” The measures of intention to vote (“When I am 18, I am planning to vote in a public election”) and willingness to contact official (“I would contact a public official about an issue of concern”) are both dichotomous items (yes = 1, no = 0). Personal efficacy is a 4-point scale measured by a single item: “I have the ability to make a difference in my local community” (4 = strongly agree, 1 = strongly disagree). The environmental conservation variable is a 7-point scale created from two items about recycling and energy conservation (Table 3). Political discussion is a 10-point scale created by adding the responses to three items asking students how often they discuss politics or national issues with parents, peers, or teachers. Response options for all of the items used for these two scales ranged from 4 (strongly agree) to 1 (strongly disagree). Internal reliability values for the scales met the minimum (Cronbach’s alpha = .50 to .60) recommended for use of additive composite scales in statistical analyses (Nunnally, 1967).

We also constructed a global youth civic engagement variable comprising the six individual civic engagement measures. Scores for each of the items were converted to standardized scores (Z scores) in order to equalize the influence of the measures on the global variable. Table 4, which presents the bivariate correlations of the six standardized variables, shows that all correlations are positive and statistically significant at the .001 level. In addition, all the correlations but one are stronger than .10. Scores from the six items were added to create the global variable, which had internal reliability values of .65 in 2008 and .64 in 2009.

We created eight predictor variables that, to greater or lesser extent, can be influenced by the schools: positive character, conflict resolution skill, academic grades, school discussion climate, school engagement, school belonging, school support, and personal safety. We created a positive character scale by adding the scores of three items that evaluate character, each with a 4-point response scale (4 = strongly agree, 1 = strongly disagree; Table 3). Six other variables—conflict resolution skill, school discussion climate, school engagement, school belonging, school support, and personal safety—each item with response categories ranging from 4 (strongly agree) to 1 (strongly disagree), were constructed similarly. The academic grades variable is a 5-point scale derived from a single item: “Over all my courses, I get mostly …,” (“A’s” = 4, “F’s” = 0).

The nonsport extracurricular activities variable is a 3-point scale created by adding the scores (yes = 1, no = 0) on two items probing involvement in nonsport extracurricular activities at school and in the community (Table 3). The sport extracurricular activities variable is a 3-point scale created by adding the scores (yes = 1, no = 0) on two items asking about involvement in extracurricular sports at school and in the community.

We included five variables that describe the background characteristics of students in the study: gender, ethnicity, grade in school, free/ reduced price lunch, and adults in family (Table 2). Free/reduced price lunch is a measure of poverty, and adults in family refers to the number of adults in the family.

We regressed each of the six civic outcome variables and the global youth civic engagement variable on our set of background, school-related, and extracurricular predictors. Logistic regression was used with dichotomous outcome variables (community service, intention to vote, willingness to contact official), and the results are presented as odds ratios (odds ratios < 1.0 represent negative associations). Ordinary least squares (OLS) regression was used with outcome variables having three or more values (environmental conservation, political discussion, personal efficacy). These results are presented as stan-dardized coefficients (betas), which employ units of standard deviation to evaluate the relative contributions of the predictors in the model. Note that betas and odds ratios are presented together in some tables.

For analyses with the global youth civic engagement variable, we used hierarchical OLS regression. The order in which the predictor variables were entered into the analyses was guided by theoretical considerations, as indicated by hierarchical analytic convention (Cohen & Cohen, 1991). In particular, we were interested in knowing the separate contributions of school-related and extracurricular factors to explaining youth civic engagement, beyond the effects of background characteristics. In addition, given the large data sets, we wanted to determine whether the magnitudes of statistically significant results were meaningful or trivial. To that end, we report adjusted R² and model-improvement values for our regressions as measures of effect size. These values are presented as percentages of explained variance.

We also created summaries of the individual-outcome regressions and the global regressions to help with interpreting the findings interms of breadth of associations, strength of relationships, and consistency over time of the predictor variables. Based on these findings and theoretical considerations, we sorted positive factors into three levels of (a) top, (b) intermediate, and (c) basic importance in predicting civic engagement in this urban population of high school students.

Table 5 presents frequencies and mean scale values for the six civic engagement variables and their components. The results for 2008 indicate that 78% of high school students reported they intend vote to in a public election when old enough to do so, 64% had done or were currently performing community service, and 19% said they had contacted a public official about an issue of concern. Mean scores for environmental conservation and personal efficacy were both above scale midpoints of 2.5 and 5.0, respectively, whereas the mean score for political discussion was slightly below the scale midpoint of 7.5.

The data for 2009 show increases on five of the six civic outcome measures. Intention to vote rose 11% and political discussion increased by 9%, perhaps reflecting increased student interest in the 2008 presidential election, which occurred a few months before the 2009 survey. The rise in political discussion with friends was particularly large (13%). Increased measures of environmental conservation, community service, and personal efficacy, however, suggest the importance of other factors as well. The proportion of students doing community service in the present or past was 6% higher in 2009 than the previous year, and the average scores on the environmental conservation and personal efficacy scales rose 4% and 3%, respectively. A change in the wording of the item on contacting public officials, from actual behavior to intended behavior, made comparison of results for the two years not viable. In 2008, 19% of students reported that they had contacted a public official about an issue of concern in the past, whereas in 2009 more than half of students said they would contact a public official about an issue of concern in the future.

Tables 6 and 7 present the results of regressing the six civic engagement variables on the full set of predictor variables using the 2008 and 2009 data sets, respectively. Table 8 summarizes the results for both years.

Environmental conservation was most strongly related to personal efficacy in both 2008 and 2009. Students who said they could make a difference in their community (personal efficacy) tended to recycle, save energy, and perform other acts of environmental conservation more than other students. Environmental conservation was positively related to four of the other five civic outcome variables in both years but showed a negative correlation with intention to vote.

Community service in 2008 was most strongly linked to having contacted public officials and somewhat less strongly associated with nonsport activities, intention to vote, and personal efficacy. This pattern of associations was essentially repeated in 2009 but with slightly less-robust results. Students who performed community service reported more frequently that they planned to vote in the future than did students who performed no service. They also tended to have contacted, or said they would contact, public officials about an issue of concern. Community service was positively associated with all of the five other measures of civic engagement in 2008 and 2009.

Intention to vote was linked—both positively and negatively—to numerous factors in both years. It was most strongly and positively associated with community service in 2008 and with willingness to contact official the year after. In 2009 the second strongest positive association with intention to vote was for African American (compared with White), presumably because the United States had just elected its first Black president. Intention to vote was positively related to four of the five other civic outcome variables and negatively associated with environmental conservation in both years.

Political discussion was most strongly related to school discussion climate in both 2008 and 2009, a finding that was not surprising, given the empirical evidence outlined above. It is noteworthy that this link was stronger during a time of relatively little political discussion among students (2007-2008 school year) than at the time of an exciting presidential election (2008-2009 school year) and relatively more political discussion (Table 4). This finding suggests that in the absence of a stimulating political campaign, school discussion climate may be particularly important for fostering discussion of political and social issues, as it was in the 2007-2008 school year. Political discussion was positively associated with all five other civic engagement measures in both years.

Willingness to contact official was worded differently on the questionnaires for 2008 and 2009. In 2008, when it read “I have contacted …,” the variable was most strongly related to community service. In 2009,when the variable was worded “I would contact …,” it was most strongly related to intention to vote. Interestingly, both variations of contacting a public official were negatively related to personal safety. Students who felt less safe in and around their school said they had contacted or would contact a public official about an issue of concern more frequently than did students who felt safer. This finding suggests that many students may be contacting a public official in regard to fear for their own safety. Willingness to contact official was positively correlated with four of the other five civic outcome variables in both years.

Personal efficacy was most strongly linked to environmental conservation in 2008 and 2009. It was also related, though not quite as strongly, to positive character, conflict resolution skill, and community service in both years. These findings indicate that youth who practice conservation, help people in need, share things with others, and perform community service tend to have greater feelings of personal efficacy than other youth do. Personal efficacy was positively related to four of the five other civic engagement variables in both years.

The findings presented in Tables 6 and 7 show that each of the six civic engagement measures has a unique set of significant predictor variables. But which variables have the strongest links to civic engagement generally? To find out, we regressed the global youth civic engagement variable, comprised of all six individual civic measures, on the full set of predictor variables. For these analyses, we considered the two versions of contacting public officials (contacted vs. would contact) as equivalent, although the frequencies reported in Table 5 suggest some differences. In interpreting the results, we followed the example of Nie, Junn, and Stehlik-Barry (1996) in defining a strong effect with OLS regression as a standardized coefficient (beta) of .20 and above, a moderate effect as between .10 and .19, and a weak effect as below .10.

Hierarchical regression analysis indicates that our full set of predictor factors explain about 37% of the variation in overall civic engagement among high school students (Table 9). Background characteristics explain very little of overall variation, slightly more than 3% (Model 1). School-related factors, however, account for more than 29% of the total variance in civic engagement, suggesting that being well-integrated into the school community may provide an important foundation for integration into the civic community (Model 2). Extracurricular activities explain an additional 5% of variation beyond that attributable to background characteristics and school-related factors (Model 3). These results indicate that school-related and extracurricular factors explain meaningful proportions of variance in youth civic engagement.

As expected, the four variables of special interest were among the strongest and most consistent (i.e., significant in both 2008 and 2009) predictors of overall youth civic engagement in our model (Table 9). The nonsport activities variable was strongly (beta = .20 or greater) associated with civic engagement in 2008 and almost reached the “strong” level in 2009. Positive character, conflict resolution skill, and school discussion climate showed moderate (beta = .10 or greater) links with civic engagement. In addition, school engagement was also moderately linked with overall civic engagement in both years. School belonging, sport activities, school support, academic grades, and grade in school were weakly related to overall youth civic engagement. Among the background variables, only free/reduced-price lunch was negatively associated general civic engagement.

Since nonsport activities was the strongest predictor of global youth civic engagement in our model, we reran the analysis to evaluate the relative contributions of its two component items: nonsport activities at school and nonsport activities in the community. The results indicated that community nonsport activities were about twice as strong as school nonsport activities in predicting global civic engagement (beta = .190 and .148 for community activities in 2008 and 2009, respectively, versus .074 and .088 for school activities). These findings suggest that schools and community organizations are both important for enhancing youth civic engagement. Furthermore, school-community partnerships might be a particularly effective approach to strengthening youth civic engagement in large urban districts.

The Civic Mission of Schools report (Carnegie & CIRCLE, 2003) laid out an argument for the role of schools in promoting youth civic engagement and presented a range of promising ideas and practices to accomplish that. Yet owing to the widely varied circumstances of communities across the nation, it remains for each school or school district to create an approach that fits its particular situation. Among child well-being researchers it is said that “what gets measured gets emphasized, and what gets emphasized gets measured” (Moore & Lippman, 2005, p. 1). In this study we have presented the civic outcomes that JCPS has chosen to emphasize in its youth by monitoring those outcomes through its annual surveys of students. We have also identified and characterized a number of predictor factors in terms of the breadth, strength, and consistency of their associations with these civic outcomes, with the aim of determining which factors might be emphasized as a means of enhancing youth civic engagement. In the sections below we discuss these findings, as well as other considerations that help determine the relative importance of various positive factors in predicting civic engagement.

There are many positive relationships— some quite strong—among the six outcome measures of youth civic engagement (Tables 6 and 7). For the most part, youth who score high on one civic outcome score high on other civic outcomes as well. In both 2008 and 2009, students who performed community service or participated in political discussions scored higher on all the other civic outcomes than did youth who did not engage in these activities or participated less frequently (Table 8). Results also indicated that environmental conservation, intention to vote, willingness to contact official, and personal efficacy consistently predicted four of the five other civic outcomes.

As these data indicate, all six of the civic outcomes in this study are relatively strong predictors of youth civic engagement, but they can be divided into levels of higher and lower importance based on whether they represent actual civic behavior or a disposition toward civic behavior. Actual civic behavior is activity in which youth engage in the present; they do not have to wait until adulthood. Dispositions, on the other hand, are not behaviors but are inclinations to act in the future. Research has found that intention to vote or perform other civic acts is highly correlated with actual behavior in the future (Campbell, 2005, 2007), and dispositions may be the only type of measure that can be used to evaluate certain activities, such as voting, that are legally forbidden or highly unlikely for youth. Nonetheless, we suggest that actual civic behaviors are relatively more important for predicting youth civic engagement than are dispositions or other factors that predict engagement but are not, themselves, civic actions. Among the six civic outcomes in this study, community service, political discussion, and environmental conservation all measure actual civic behavior, while intention to vote, willingness to contact official, and personal efficacy measure dispositions toward civic action.

The data also show that JCPS high school students who performed community service were 50% to 60% more likely to say they intend to vote in the future than were students who did no service, similar to other recent studies which have found that volunteering in adolescence predicts voting in adulthood (Campbell, 2006; Hart et al., 2007). This relationship is worth emphasizing because of concerns often expressed in public forums that young people choose between community service and political engagement and do not combine the two forms of civic engagement (Boyte, 1991; Keeter, Jenkins, Zukin, & Andolina, 2005). Since 1972, when 18-year-olds first had the right vote, the proportion of eligible youth (ages 18-29) voting in presidential elections declined almost continuously (the exception was in 1992) from a high of 55% to a low of about 40% in 1996 and 2000 (Kirby & Kawashima-Ginsberg, 2009). At roughly the same time, volunteering among youth was rising. Among 16-19-year-olds, for example, volunteering more than doubled between 1989 and 2005 (Corporation for National and Community Service, 2006). Moreover, in the decade of the 2000s, youth voting has rebounded to levels previously seen in the 1970s (Kirby & Kawashima-Ginsberg, 2009) while youth volunteering has essentially plateaued (Lopez et al., 2006).

The findings of the present study, in contrast, suggest that voting and volunteering go together. One possible explanation for this connection is that service and voting intention reflect “engaged” students who choose to participate in several civic activities, and that there is essentially no causal connection between service and voting intention. Another possibility is that community service and intention to vote are both manifestations of a community norm for civic engagement (Campbell, 2006). A third possibility is that the community service performed by these students stimulated their desire to take a more active role in the political arena in the future by voting.

The finding that environmental conservation is negatively associated with intention to vote is interesting in that environmental conservation is positively linked with all the other civic outcomes, particularly personal efficacy (beta ≥ .250, Tables 6 and 7). Overall, these findings suggest that environmentally conscious youth are civically engaged but may prefer activities they can participate in directly (e.g., community service, political discussion, contacting officials) rather than indirectly (i.e., voting—for representatives to act on one’s behalf).

Table 9 presents all the predictor variables that were significantly linked to overall youth civic engagement in 2008 or 2009. As expected, the four variables of special interest (positive character, conflict resolution skill, school discussion climate, and nonsport activities) showed stronger positive correlations with overall youth civic engagement than did most of the other predictors.

Participation in nonsport activities was easily the strongest of all the predictor variables. Its beta values were about 18% larger than those of the next most important predictor in 2008 and 2009, and it was the only one achieving a strong effect size (in 2008) by the criteria used in this study (beta = .20 or greater). Moreover, the predictive value of nonsport activities was relatively widespread. In 2008, participation in nonsport activities was consistently linked to four of the six civic outcomes (but not environmental conservation or intention to vote). These findings confirm earlier research, cited above, that have found positive links between nonsport activities and civic engagement.

The effects of conflict resolution skill were slightly more widespread than those for nonsport activities, although its association with overall youth civic engagement was moderate in size (Tables 6-9). Conflict resolution skill, as mentioned previously, is often cited in the literature as a technique for reducing violence in schools. The results of the current study underscore the idea that the skill also has important applications to civic engagement. Berman (1997), for example, has proposed that teaching conflict resolution skill (along with other skills) is one way to enable students to move beyond discussion of seemingly intractable social or political issues to the point of feeling capable enough to help find workable solutions. In other words, conflict resolution skill may enable youth to engage difficult civic issues, whereas without that skill they may avoid getting involved.

Positive character was also a moderate predictor of overall youth civic engagement, but its effect was less widespread than that of conflict resolution skill (Tables 6-9). Positive character was positively associated with four of the six civic outcomes in both 2008 and 2009. These findings add support to contentions that character education efforts can enhance youth civic development (Berkowitz & Bier, 2004, 2005; Berkowitz et al., 2005; Character Education and Civic Engagement Technical Assistance Center, 2008).

School discussion climate also reached moderate effect size (beta = .10 or greater) in predicting overall youth civic engagement (Table 9). The strength of this finding was driven largely by sizable correlations with political discussion (Tables 6 and 7). This result was not surprising, given findings from earlier research (cited in Torney-Purta et al., 2001) and the logic that environments which encourage discussion of political and social issues would contribute to the frequency of such discussions.

Other evidence suggests that school discussion climate predicts a wide range of civic outcomes, including intention to participate in civic activities such as voting, volunteering and writing letters to newspapers about issues of concern (Campbell, 2005), as well as civic knowledge (Torney-Purta et al., 2001). It was surprising, then, that in the current study school discussion climate consistently predicted only three of the six civic outcomes (political discussion, intention to vote, personal efficacy) and was negatively related to environmental conservation and community service. The negative findings may be a statistical anomaly related to the strong correlation of school discussion climate with political discussion. When the latter was removed from our regression models, school discussion climate was no longer significantly linked to community service in either 2008 or 2009, nor was it significantly related to environmental conservation in 2009. Regardless of the reason for these negative findings, the lack of more-widespread positive effects for school discussion climate in predicting civic outcomes merits further study.

Based on the findings in Tables 8 and 10, we can divide the significant positive predictor variables into two groups of higher and lower importance for our models of youth civic engagement. In the higher group we include those having a moderate or strong effect in predicting overall youth civic engagement (beta = .10 or greater) and relatively widespread associations (positively related to four or more of the civic outcomes in both years): nonsport activities, conflict resolution skill, and positive character.

In the group of lower importance we place positive predictors that were statistically significant but did not meet the criteria for inclusion in the higher-importance group. These variables include school engagement and school discussion climate, which had moderate effects but were positively linked to only half of the civic outcomes in both 2008 and 2009. Four other school-related variables (school belonging, sport activities, school support, and academic grades) also showed significant but weaker associations with youth civic engagement (Table 9). Although these measures did not show as strong connections with youth civic engagement as some of the other variables, it is reassuring to know that qualities valued for their links to school success are also related to success in the civic realm.

Data presented in Tables 8-10 also indicate that when school and community factors are taken into account, background characteristics such as gender and family structure show little or no consistent relation to overall youth civic engagement. The exceptions are poverty (measured by eligibility for free or reduced-price lunch) and grade in school. The link between civic engagement and grade in school is what we would expect to see as youth mature and are increasingly exposed to opportunities for civic engagement at school and in the community. Poverty, however, was associated with decreased civic engagement among youth in both 2008 and 2009. The findings suggest that this group of students may be missing out on activities and venues that enhance youth civic engagement.

The current exploratory study was designed to take advantage of data gathered by an annual survey of high school students in a large urban school district with an ethnically and socioeconomically mixed population. A major advantage of the survey is that its questionnaires include a small but comprehensive set of civic engagement measures embedded in a larger set of measures designed to monitor whole-child educational development. While this arrangement allows for the exploration of a wide range of possible predictors of civic engagement, it does not permit evaluation of some factors specifically intended to enhance civic development and engagement. For instance, the questionnaires contain no measures of political knowledge, which has been found in several studies to predict voting, working in political campaigns, participating in a protest, and other civic actions (Delli Carpini & Keeter 1996; Verba et al., 1995). Nor do the questionnaires have measures focusing more narrowly on social studies instruction (Campbell, 2005), mock trials (Niemi & Junn, 1998), student government (Torney-Purta et al., 2001), or other promising school-based approaches to civic development and engagement (Carnegie & CIRCLE, 2003).

Another limit is that the study is correlational. Although the study identified links between some predictors and outcomes, we cannot draw firm conclusions about either the direction of causality between predictors and outcomes or even its source (i.e., a third factor may cause the change observed in both predictor and outcome). Nonetheless, the findings describe the pattern and strength of relationships among an important set of predictors and civic outcomes, providing insights that can guide future research and—to a certain degree—inform educational practice. In addition, the results demonstrate that the annual surveys and the data they generate constitute a powerful tool for studying youth civic engagement and its predictors in this particular large urban school district.

Future research about youth civic engagement should address its longitudinal nature by assessing growth over time for participating students. Future studies should also address the qualitative aspects that were not addressed in this primarily quantitative research.

The major purpose of the current study was to identify the most important predictors of youth civic engagement in a set of probable and possible predictors as a way of trying to identify what works best. Apart from background characteristics, most of the variables in our models were positively associated with youth civic engagement, but some were stronger, more reliable, and more widely associated with civic outcomes than others. Using these findings and other considerations, we divided both variable types—outcomes and positive predictors—into groups of higher and lower importance for predicting youth civic engagement. In Table 11, we sorted all of these dichotomous groupings into three levels of relative importance to our models of youth civic engagement.

Top Level. At the highest level of importance, we selected community service, political discussion, and environmental conservation, for two reasons. First, these behaviors are real, direct forms of civic action. When students perform community service, participate in political discussions, or recycle used items they are directly engaging in actual civic activity. Second, each behavior was positively related to four or more of the five other civic outcomes. Students who participate in community service or political discussions generally score higher on all other civic outcomes in this model than do other youth. Youth who act in environmentally friendly ways score higher, on average, than other youth on four of the five other civic outcomes. In addition, political discussion and community service are especially amenable to intervention, which is important for school districts and schools around the nation. Schools that arrange for these activities not only provide students with opportunities for civic action but also give them experience in politics and civil society, the two major areas of civic engagement (Carnegie & CIRCLE, 2003). Together, these behaviors constitute a basic first step toward well-rounded citizenship.

Intermediate Level. At an intermediate level of importance, we put nonsport activities, conflict resolution skill, and positive character, as well as the civic outcomes that measure dispositions (intention to vote, willingness to contact official, and personal efficacy). All six of these factors were consistently and positively related to four or five of the six civic outcomes. Although nonsport activities, conflict resolution skill, and positive character showed moderate to strong associations with overall youth civic engagement, we did not include them at the top level of importance because they are predictors of civic engagement, rather than actual civic activity.

Basic Level. At a basic level of importance, we included school engagement, school discussion climate, sport activities, school belonging, academic grades, and grade in school. Each of the factors at the basic level was positively associated with overall civic engagement, but the strength of those associations was relatively small (beta < .10) or the variables were consistently linked with only three or fewer of the six civic outcome variables.

In summary, the conceptual framework of this study presents a vision of civic engagement that one school district has set for its youth as part of a broader, whole-child approach to education. Ideally, the young persons who attend and graduate from JCPS high schools would come to act in environmentally friendly ways, volunteer service to the community, discuss the political and social issues of the time, and vote regularly when they come of age, as well as take individual and collective action to solve community problems.

This study highlights the importance of conducting well-designed annual school surveys as a way of monitoring educational policy and interventions associated with civic engagement outcomes. CSS surveys in the coming years should reveal what effects those efforts have had. Perhaps more important, the findings of the study suggest a number of actions that might be considered by school districts and schools—in collaboration with community partners—to help youth develop into the well-rounded citizens represented in the conceptual framework.

We wish to thank James Youniss of The Catholic University of America and Peter Levine and Kei Kawashima-Ginsberg of the Center for Information and Research on Civic Learning and Engagement, for their thoughtful comments on various drafts of the paper. CIRCLE has graciously supported the work of this project.