Abstract
For this article, we tested whether and to what extent young people’s rates of informed voting are influenced by laws and policies that regulate the electoral system and by civic education policies. Education policies and state voting laws vary widely and are in rapid flux; their impact is important to understand. Immediately after the 2012 election, a sample of 4,483 youth was surveyed that included at least 75 respondents in each of the 50 states and national oversamples of African Americans and Latinos. Their experiences with civic education and support from their families predicted their informed political participation as young adults, but variations in the existing state policies did not matter. This may suggest that the kinds of policies that states have enacted—such as allowing early voting or requiring one course on government in high school—are not helpful but policies that promote extracurricular participation and discussion of current issues in schools could be much more effective.
In this article, we asked two central questions related to the relationship between policy and youth civic engagement. Specifically, we chose state laws related to registration and voting and state civic education policies (e.g., standards and required assessments) as independent variables and examined how aspects of each independent variable related to informed voting, electoral participation, and political knowledge.
A primary purpose of education is to produce capable, informed, and engaged citizens. An important specific objective is informed voting, understood as choosing to vote while understanding the candidates’ positions and fundamental facts about the political system. Unfortunately, the turnout rates and the civic knowledge of young Americans are unacceptably low, and there are deep disparities in both turnout and civic knowledge by class, educational attainment, and race/ethnicity.
When the most recent National Assessment of Educational Progress civics assessment was released, the New York Times described it under the headline “Failing Grades on Civics Exam Called a ‘Crisis’” (Dillon 2011). Only one quarter of high school seniors had reached the “proficient” level.
The voter turnout of young adults (18–29) varies by election, from 1% in the 2012 Nevada caucuses to 50% in the 2012 national election, but is generally low compared to rates in other nations and compared to the voting rates of older adults.
Both turnout and civics test scores show large gaps by class and race. Individuals with more educational attainment vote at higher rates. In fact, according to Nie, Junn, and Stehlik-Barry (1996), this is “the best documented finding in American political behavior research” (p. 31). Sondheimer and Green (2010) say that this relationship has been found in “literally thousands of cross-sectional surveys” (p. 174) since 1924. To some extent, educational attainment may be a proxy for social status or personal motivation and ability, but some careful studies find a positive impact from obtaining more education (Dee, 2004; Sondheimer & Green, 2010). Students with higher socioeconomic background and White students perform better on the National Assessment of Educational Progress civics test than students with lower socioeconomic background and Black and Hispanic students, respectively, at all grade levels (Kawashima-Ginsberg, 2013). Turnout varies by race and ethnicity, but White and African American youth are generally more likely to vote than youth from Asian and Hispanic backgrounds. In fact, African American youth turned out at the highest level of all racial groups in 2008, and young African Americans have voted at higher rates than other racial/ethnic groups in many elections since the 1980s (e.g., Lopez & Kirby, 2005).
Virtually all American students take at least one civics or American government class by the 12th grade. A fair amount is known about the effects of specific educational experiences in these courses and in other aspects of school. But little is known about the impact of policies that require such courses or otherwise promote or constrain civic learning. The impact of laws and regulations is an important topic in a time of rapid change in educational policies—marked by No Child Left Behind, the Common Core state standards, charter schools, and vouchers.
Civic education courses and programs in high school (including extracurricular programs such as mock trial and modules like Kids Voting USA that are embedded in courses) have significant, positive effects on voting after graduation (Bachner, 2011; Campbell, 2008; Gimpel, Lay, & Schuknecht, 2003; Kiousis & McDevitt, 2008). These courses may boost voting by enhancing students’ knowledge, interest in politics and issues, habits of discussing politics, and sense of membership and obligation. Extracurricular participation in high school, with the exception of sports, is also positively related to voting (Thomas & McFarland, 2010). But civic learning opportunities vary widely across school districts and schools, and different student populations receive different levels and kinds of instruction. Prior educational success, family socioeconomic status (SES), and school SES have each been shown to affect the quality of civics education a student is likely to receive.
Overall school atmosphere can also affect civic engagement. Research using a variety of methods but generally incorporating controls for socioeconomic background finds that students who attend schools with positive school climate can develop a positive sense of belonging, connection to peers, and trust in institutions and eventually in the broader society and its democratic system (Berman, 1997; Flanagan, Bowes, Jonsson, Csapo, & Sheblanova, 1998). Connection to school and a perception of the school as a fair and supportive system are related to a perception that society is just, a stronger sense of civic commitment (Kahne & Sporte, 2008; Torney-Purta, 2002), an intent to intervene when peers have dangerous plans (Syvertsen, Flanagan, & Stout, 2009), and voting (Campbell, 2005). Dill (2009) finds that private schools have a positive effect on voting (net of demographic factors), which, if the controls in his study are adequate to account for background factors, may suggest that the “climate and context” of private schools is generally more favorable for civic development (p. 1282). That claim is consistent with the premise that school climate matters.
State laws regarding civic education vary. All states have standards for social studies, a broad category that includes civics/government along with other disciplines such as history and geography. Thirty-nine states require at least one course in American government or civics. In the 2012–2013 school year, 21 states required a state-designed social studies test. This number is similar to that in 2006 but is a dramatic reduction compared to that in 2001, when 34 states conducted regular assessments on social studies subjects (Buckles, Schug, & Watts 2001). Just 9 states require students to pass a social studies test to graduate from high school. Eight states have statewide, standardized tests specifically in civics/American government. Social studies assessments have shifted from a combination of multiple-choice and performance tasks to almost exclusively multiple-choice exams since 2000 (Godsay, Henderson, Levine, & Littenberg-Tobias 2012). Martin West (2007) found that state-level history and science course requirements were inversely correlated with the amount of classroom time that students spent on history. Yet a study that examined whether state laws affected knowledge of the First Amendment found no impact, even though the First Amendment is generally included on state standards and specifications for standardized tests (Lopez, Levine, Dautrich, & Yalof, 2009).
Likewise, state laws regarding voting vary widely: The United States is unusual in allowing states to set widely different policies governing elections, from Oregon’s exclusive use of mailed ballots to states that require individuals to appear during limited hours on Election Day (having registered more than 30 days earlier) and present photo ID. Some recent reforms in election laws are likely to improve voter turnout. Two studies have found that Election Day registration (EDR) is an effective mechanism for bringing young people to the polls because it allows people to participate even if they only become interested in an election during the most intense final phase of the campaign (Berinsky, 2005; Burden, Canon, Mayer, & Moynihan, 2012). Several other reforms modestly improve turnout. For example, when schools encourage students to register or allow 16- and 17-year-olds to “preregister,” available evidence suggests an increase in turnout (McDonald & Thornburg, 2010); mailing information about polling locations and sample ballots to households seems to raise turnout significantly among less advantaged young people (Wolfinger, Highton, & Mullin, 2004).
Preliminary research findings on measures that place restrictions on voting are mixed. Stringent residency requirements may dampen the participation of some college students, but voter ID requirements do not appear to have a sizable effect on overall turnout, even though many college students and urban youth lack acceptable identification (Alvarez, Nagler, & Wilson, 2002; Ansolabehere, 2009; Hershey, 2009; Mycoff, Wagner, & Wilson, 2009; Pastor, Santos, Prevost, & Stoilov, 2010; Vercellotti & Anderson, 2006). However, the available research has not looked closely at youth, who may be especially affected. Many of the most controversial requirements are new and have not been studied. That is true, for example, of the particularly stringent photo ID laws adopted since 2010 in several states.
Hypotheses
Based on the review of available literature, we aim to answer three research questions that address the gaps in the current knowledge about the relationship between state policies and youth civic engagement. Although civic engagement is a broad category, we focus here on three outcomes, each of which is a construct composed of several concrete measures: electoral engagement (registering, voting, and following the news), informed voting (voting while knowing information about the candidates’ positions on issues), and political knowledge (possessing more general and perennial information about the political system).
These outcomes may be influenced by many factors at the individual, family, school, community, state, and national level. For the purpose of this study, we focus on a few potential influences that are subject to influence by state governments. Other influences are treated as control variables. We hypothesize the following:
High-quality civic education boosts electoral engagement, informed voting, and political knowledge.
States can promote high-quality civic education by requiring and testing civics at the high school level. Thus, the presence of those requirements and tests boosts the three outcomes of interest.
There is no strong empirical or theoretical basis to assume that state voter laws would affect civic engagement outcomes other than voting itself. Thus, we predict that state voter laws do not affect informed voting, political knowledge, or electoral engagement.
Documented evidence indicates that people are more likely to vote when voting is easier. Thus, turnout is higher in states with provisions that facilitate voting (such as early voting or same-day registration) and lower in states with rules that may make voting more difficult (such as photo ID requirements). Thus, we predict that at least some aspects of state voter laws are related to a higher propensity to vote.
Methods
Participants
Table 1 shows the distribution of participants before weighting. Note that the participants were quite balanced ideologically, with 28.1% identifying as “conservatives,” 27.4% as “liberals,” and 33.2% as “moderates” (11.3% chose “none of those”). However, consistent with other survey research on youth, they were more likely to identify as Democrats (38.2%) than Republicans (21.3%). Of the sample, 28% considered themselves independents, 9.8% said they had no political views, and 2.6% said they identified with another party.
Demographics of the Sample (Unweighted).
Design and Procedure
The Center for Information & Research on Civic Learning & Engagement (CIRCLE) at Tufts University worked with Universal Survey, Inc., to recruit 4,483 individuals to participate in a 17-min random-digit-dialing phone interview. The survey firm was responsible for implementing a screener, obtaining verbal informed consent, and providing a de-identified data set. The sample was designed so that two thirds of the respondents come from cell phone numbers and the rest from landline numbers. The interviews began the day after the 2012 presidential election and went on for 6 weeks, until all the data were collected. The sampling was designed so that at least 75 participants came from each of the 50 states and Washington, DC (n range = 75–131 per state). Participants of Black and Hispanic backgrounds were slightly oversampled to obtain a large sample of each group (i.e., more than 800 individuals in each group, coming from states with moderate to high portions of Black and Hispanic residents).
The respondents were asked to participate in an interviewing lasting about 18 min, made up of approximately 75 questions. 1
After a total sample of 4,483 respondents aged 18 to 24 (all U.S. citizens) had been surveyed, we computed a nonresponse and poststratification weight so that the weighted frequencies match the Census Current Population Survey March 2012 Supplement population data for citizens by gender and race/ethnicity. Because of the oversamples of African American and Hispanic respondents, they are weighted slightly lower than respondents who identified as White or as members of other racial groups. The weight value ranges from 0.79 to 1.19, suggesting an overall very representative sample.
Measures
The survey asked questions in several domains of civic and political engagement, civic knowledge, and educational context. Only the measures that are relevant to the study’s hypotheses are presented below. 2
Dependent variables
Turnout
This is a dichotomous variable, reflecting whether the respondent voted or did not vote in the 2012 election. As discussed below under Limitations and Future Directions, it is self-reported.
Informed voting
Informed voting is a hierarchical index that is based on six criteria: (a) registering, (b) voting, (c) answering at least one (out of two) campaign knowledge questions correctly, (d) answering four or more general political knowledge questions correctly (out of six), (e) voting consistently with their personal opinion on a campaign issue of their choice, and (f) following the news “fairly” or “very closely” during the election season. Based on this rating, a respondent’s score could range from 0 to 6. Conceptually, respondents who score high on this index are those who not only voted in the 2012 election but also had a sufficient level of knowledge both about the general U.S. political system and about at least one specific campaign issue and voted for the candidate whose policy stance reasonably represented their viewpoints. The informed voting index was not designed specifically as a scale, meant to measure a psychological construct. It was rather a single indicator of a desirable trait of citizens in a democracy. Nevertheless, we assessed its internal consistency using the Kuder-Richardson coefficient (as all items in the index were dichotomized). The internal coefficient for the informed voting index was .68, which was quite close to the minimum cutoff of .70. However, principal components analysis showed that the index actually included two factors, one factor on the voting behaviors and the other on political and campaign knowledge and news consumption. With these two factors, the index accounted for 58.4% of the variance.
Electoral engagement
Electoral engagement was an additive index that was a sum score of (a) registration, (b) voting, and (c) keeping up with the news about the election at least somewhat frequently. The score ranged from 0 to 3.
Political knowledge
We asked respondents six factual questions, all of which have been used before in national polls and some of which are recommended as good measures of general knowledge (Delli Carpini & Keeter 1997). For example, we asked which political party is more conservative, who can vote in U.S. elections, and whether the U.S. spent more on Social Security or foreign aid. Participants’ estimated political knowledge scores could range from 0 to 6.
Independent variables
Young people were conceptualized as nested in schools and communities that are within states. To learn about their school and community environments, we asked them about their experiences and their perceptions of the local climate for political engagement. We then added state-level data from public sources: federal surveys and scans of state laws and policies.
The high school curriculum and extracurricular involvement
We asked participants, most of whom are out of high school, to retrospectively report their experiences with high school civics or American government courses. The Civic Mission of Schools report (Gibson & Levine, 2003) recommended certain practices for civics instruction, based on a literature review. Those included discussion of current events, controversial discussions in classes, conducting research on social or political issues, projects on community issues or in the community, and keeping up with news media. Thus, we asked respondents to recall whether they had experienced these practices. Based on the distribution of the response, we categorized respondents into the low-quality, medium-quality, and high-quality groups. The high-quality group recalled experiencing four to five (out of five) practices.
To obtain a richer understanding of the content of their courses and how they had received them, we asked them to name the most important theme that they recalled from their courses, whether they considered their courses relevant to present-day civic life, and whether the teacher had specifically addressed voting and/or encouraged them to vote. Based on the general view in the field that testing influences what is taught, we asked respondents if they remembered taking a standardized test in civics.
Since service-learning (the combination of community service with academic study) is a prevalent practice in social studies, we asked whether youth had conducted service, whether it had been required, and whether it involved any discussion of the root causes of the issues that were addressed through service projects.
We also asked about group participation and whether any of the groups to which a respondent belonged dealt with social-political issues. Also included was one composite measure indicating the number and breadth of groups in which the respondent had participated.
To assess the perceived school climate, we asked a series of questions about respondents’ perception of student voice, openness, and climate for disagreements. These factors were captured by one composite measure, “High school democratic climate.”
Some research suggests that it may be more difficult to encourage civic participation when a community, such as a student body, is diverse (Alesina & La Ferrara, 2000; Campbell, 2007; Mutz, 2002). In diverse contexts, it can feel challenging to raise controversial issues, and individuals may feel less of the motivation that comes from solidarity and peer norms. For the purpose of this article, we identified a group of respondents (about one third) who had attended schools that are considered “racially pluralistic”: where multiple racial groups were represented, with no clear majority group. We did so because past research had indicated that individual racial background (i.e., being non-White) does not necessarily predict lower turnout. For instance, African Americans had the highest youth turnout among all racial groups in the 2008 presidential election (Kirby & Kawashima-Ginsberg, 2009). Instead, we hypothesized that absence of a majority group in the school could hinder deliberative discussions and therefore civic participation (Kawashima-Ginsberg & Levine, 2014). Thus, in this article, “racially diverse schools” do not include schools where a majority of students are from one ethnic minority group.
Also, since the overall level of social advantage in the school could affect civic engagement, we asked individuals to estimate the proportion of their peers who had gone to college and to identify their high school type (public, private, charter, religious, etc.).
State laws and policies
State-level laws and policies related to voting and civic education were the main predictors in Hypotheses 2 and 3. Each state has a unique set of relevant laws and policies that are within the power of the legislature to alter and that may affect informed voting. But states also differ in other ways, and the relationship between particular policies and other distinctive characteristics of states does not appear to be random. For example, policies that are intended to make voting easier are much more prevalent in the upper Midwest and Pacific Northwest than in the South. Thus, we included state-level controls in the model.
We used the Brennan Center for Justice’s classifications to identify voting-related laws. The laws that we included were requirements to present photo identification or proof of citizenship to vote, opportunities to vote early (before Election Day), and a requirement to register well before Election Day. We also included “pushback” actions in the model—attempts to defeat photo ID laws in the courts—because those efforts might encourage voting. Finally, we incorporated online registration, no-excuse absentee voting, extended poll hours, the comprehensiveness of the information available on the state’s election website, 3 and how well the state’s electoral procedures are run overall. 4
We reviewed state laws and standards related to civic education, using several coders to categorize each law or standard and to resolve any differences (Godsay et al., 2012). We considered which courses are required, whether civics is tested, which types of items are used in social studies tests, and which key concepts are included in state civics standards.
As state-level control variables, we introduced measures of the state’s political climate (the competitiveness of the 2012 race and historical turnout rates for people 30 and older) and its civic climate (social cohesion, density of nonprofits, and civic engagement measures in 2006 such as volunteering, working with others, and going to meetings). 5
Covariates
We used factors that have been known to relate to civic participation, such as educational attainment, age, gender, race, SES, and strength of political identification. We also entered in family support for civic participation and racial makeup of high school as covariates in our analyses because our preliminary findings showed that these two factors also related to civic engagement. We explore the relationship among electoral engagement, racial makeup of the school, and family encouragement in another article (Kawashima-Ginsberg & Levine, 2014) and treat these as covariates in this article.
Educational progress
In this analysis, we used a measure of educational progress, rather than educational attainment, because our sample is young and, in fact, more than two thirds of the respondents who were currently out of school stated that they planned to go back to school for additional education. Therefore, we calculated educational progress as standardized units away from the age-specific mean of educational attainment. The higher score means that the respondents have completed more education compared to the participants who were of the same age. We also asked about current school enrollment and about plans to go back to school.
SES
We asked three questions that were used as indicators of socioeconomic background because individual income is not a good indicator of socioeconomic background among young people, many of whom are in school and have no income. We asked instead about maternal educational attainment, number of books in the home, and whether the family had subscribed to a newspaper on a daily basis. Based on these three indicators, we calculated a standardized SES score. The mean SES score was −.02 (SD = .76, minimum = −.2.20, maximum = 1.52), with a normal distribution (see Table 1).
Other demographic characteristics included in the survey and analysis were gender, age (recoded as a dichotomous category of “voters who were eligible in 2008” and “newly eligible voters”), employment status, involvement with a religious congregation, and ideology.
Strength of party identification
We asked the participants to say which political party they identified with and how strongly they identified with that party. We then stripped the ideology and created a code that referred to the strength of identification with a specific political party, such as “strong partisan,” “weak partisan,” and “no identification.” A higher score means stronger partisan identification.
Family engagement and encouragement
Parents and other family members may influence young people’s beliefs and behaviors by exemplifying engagement or by directly encouraging their children to participate. Thus, we asked respondents to report their parents’ engagement in the community, whether their parents had encouraged them to vote, and whether their parents had encouraged them to express their own opinions about political issues.
Statistical Analysis
We used hierarchical linear modeling and multiple regression models. As noted above, we collected statewide data based on an assumption that each survey respondent is “nested” within the state of his or her residence and that respondents living in the same state are influenced by a unique combination of civic education and voter laws. Therefore, our analytical approach was to first conduct hierarchical linear models with each outcome of interest to assess whether there was enough systematic variation attributable to state-level factors. To do so, we examined the variance components and calculated the intraclass correlation coefficient (ICC) from the null model to estimate the portion of outcome variance that was attributable to state factors. If the variance component was significant and the ICC was sufficiently large (>5%), we proceeded with the model using hierarchical linear modeling (Version 7; Scientific Software International, 2012), software that allows accurate analysis of multilevel data. If the model did not meet the above criteria, multilevel modeling was not warranted due to lack of systematic variation by state. In such cases, we proceeded with individual-level multiple regression. In our regression models, we tested the effects of individual and community-level influences, employing a multiple-stage-entry method. We entered individual demographic indicators and family background, including SES and parent encouragement of discussion, expression of voice, and voting (Stage 1), then high school civics education quality and major themes, democratic climate, learning about voting, type of school, exposure to service learning, and whether respondents belonged to any group during high school (Stage 2). In the third and fourth stages of the model, we entered whether students had attended a racially diverse high school and whether they had experienced positive peer norms for civic engagement.
Results
Hypothesis 1: Relationship Between Civic Education Experience and Civic Outcomes
We first employed multilevel models to test whether there was significant state-by-state variation in outcomes. After controlling for individual covariates, we found no significant variation in outcomes by state, beyond the effect of individual experiences and background factors. Thus, we proceeded with linear regression models thereafter.
Electoral engagement
High school civic education experiences and family’s encouragement of civic participation were modestly related to electoral engagement once we accounted for the effect of educational progress, race, age, and strength of partisan identification. African Americans and older youth were also more likely to be electorally engaged, after controlling for all other factors. Participants who held a stronger political identity were also more likely to be electorally engaged. Family and school factors also predicted this outcome: The quality of high school civics class was a marginal predictor of electoral engagement. Parents’ encouragement of civic participation, defined by parents’ encouragement of youth to vote, youth’s engagement in current issue discussion with parents, and encouragement of youth to express opinions at home, was also positively related to electoral engagement. On the other hand, young people who had attended racially diverse high schools were less engaged (see Table 2).
Multiple Regression Analysis: Predicting Electoral Engagement.
Note. Dependent variable is electoral engagement. Unstandardized coefficients are shown for each predictor, with standard errors in parentheses.
p < .05. **p < .01. ***p < .001 (two-tailed tests).
Informed voting
Consistent with our hypothesis, high-quality civic education predicted more informed voting. Furthermore, more educated and older participants were also more likely to be informed voters, as were African Americans. Strong party identification was related to more informed voting. Conservative young voters tended to be less informed voters, compared to liberal and moderate voters. Parents’ encouragement of civic participation and expression of opinions predicted more informed voting, whereas diversity in schools was negatively correlated with informed voting. 6 Service projects in the community were marginally beneficial if they also involved discussion of root causes. Conversely, when service was required without discussion of the root causes of the problem, it appeared to have a negative effect (see Table 3).
Multiple Regression Analysis: Predicting Informed Voting.
Note. Dependent variable is informed voting. Unstandardized coefficients are shown for each predictor, with standard errors in parentheses.
p < .05. **p < .01. ***p < .001 (two-tailed tests).
Political knowledge
For political knowledge, we found that the quality of civic education did not matter significantly, after considering educational progress and plans to pursue more education, but learning specifically about voting did predict better political knowledge. Generally, demographic indicators had stronger relationships with political knowledge than they had with informed voting or electoral engagement. Educational progress, school enrollment (and plans to return to education), socioeconomic background, and party identification were positive predictors. Males, on average, had better political knowledge, whereas African Americans demonstrated an overall lower level of political knowledge. Parental encouragement and the extent of extracurricular participation also predicted political knowledge (see Table 4).
Multiple Regression Analysis: Predicting Political Knowledge.
Note. Dependent variable is political knowledge. Unstandardized coefficients are shown for each predictor, with standard errors in parentheses.
p < .05. **p < .01. ***p < .001 (two-tailed tests).
Hypotheses 2 and 3: The Relationship Between State Civic Education Policy and Youth Civic Outcomes and Between State Voter Laws and Youth Civic Outcomes
We explored relationships among state civic education and voter laws as well as three indicators of civic engagement (informed voting, political knowledge, and electoral engagement) and found no support for either of our hypotheses. In 2012, as in other elections, states varied widely in rates of youth voting. There were also small but significant variations among states in political knowledge, informed voting, and campaign knowledge, but we found that individual experiences with families and school and socioeconomic class can explain almost all of the differences between states. A broad set of state civic education and voting laws explained very little of the between-states variations. 7 To specify, electoral engagement, informed voting, and political knowledge had nonsignificant variance components and/or very low ICC. Thus, we reject Hypotheses 2 and 3.
Hypothesis 4: Voting as a Dichotomous Outcome
We tested voting as a dichotomous outcome. Using multilevel modeling, we found that an EDR law and the competiveness of the election in a state boosted turnout whereas having many restrictive measures in place dampened turnout. Thus, when we examined voting alone, we found support for Hypothesis 3, which stated that the state voting laws would be related to electoral engagement. The model in Table 5 shows that the competitiveness of the race and EDR boosted turnout, controlling for the baseline civic engagement level, nonprofit density, margin of victory for Obama (a negative value if Obama lost in the state), and turnout of people 30 and older in 2010. Other measures to facilitate voting, such as no-excuse absentee voting, vote by mail, online registration, and extended voting hours, did not matter. Similarly, the effectiveness of election administration of the state 8 did not have a significant impact on youth turnout.
Results for Bernouli Population Average Multilevel Model. 9
p < .05. **p < .01. ***p < .001 (two-tailed tests).
We also tested various indicators of state-level civics course requirements, testing, civic standards, and test contents. None of these indicators had a significant effect on turnout (see Table 5 for coefficients). As expected, we found that college-experienced youth and African Americans were more likely to vote. The pattern of voting by demographic groups (i.e., for African Americans and college-experience youth) did not differ significantly across states (see Table 6). However, not all outcome variance is explained by the model, as indicated in the significant chi-square value for the intercept, suggesting that other factors that were not identified in this research also explain variations in youth voter turnout at the state level.
Final Estimation of Variance Components.
Discussion
Consistent with the results of previous studies (some of which have been longitudinal or experimental), we find that interactive civic education in high school and extracurricular group participation boost informed political participation, and the specific study of voting in high school enhances political knowledge. But even though schools are regulated by states, state policies do not seem to affect political knowledge or informed voting, and same-day registration is the only voting law that raises turnout. (Having more restrictive state voting laws modestly reduces turnout.) In short, we found that individuals who receive high-quality civic education are more civically engaged but individuals’ experience of civic education and civic engagement are not systematically related to state laws and policies. One exception was that EDR appears to have a positive effect on turnout (but not on more robust measures of engagement or knowledge).
This null result could be considered disappointing since states differ in their civic education policies and their voting laws and there are passionate debates about such matters as to whether to require and test civics and whether to require photo identification to vote. But it is important to know that these choices do not affect youth engagement very much. And the data suggest alternative strategies for states.
After all, the states differ today in whether they require one high school course that typically emphasizes the Constitution and other perennial issues in American politics rather than current events and whether they test this course using multiple-choice exams that measure factual knowledge (Levine 2013). We find that what matters is discussing current events and, specifically, teaching students to vote. Although some states’ standards mention these matters (Godsay et al., 2012), they are never strongly emphasized, and they are never included in state assessments. For example, no state includes current events on its standardized tests. Despite evidence in this article and in previous studies that participation in extracurricular groups enhances civic participation, states do not enact influential policies that support or require schools to offer such groups.
Likewise, although state voting laws differ in ways that might make registration and voting more or less difficult, no state affirmatively registers all its 18-year-old citizens who are enrolled in its schools and colleges or teaches them the voting process.
In short, the variation in state laws appears substantial from a statistical perspective, but when one steps back to consider the range of extant laws and policies, they all seem very modest and unpromising. The null results presented here should thus encourage states to experiment with entirely new policies.
Although we found little support for our main hypotheses about policies, some findings warrant further discussion. African Americans were more likely to be electorally engaged and vote in an informed way, after controlling for all other factors, but less likely to have high levels of political knowledge. This is not only a phenomenon of the 2012 election when Barack Obama was on the national ballot; controlling for other factors, young African Americans have voted at higher rates than other racial/ethnic groups in many elections since the 1980s (e.g., Lopez & Kirby, 2005), and this pattern was repeated in our study. The apparent contrast between informed voting and slightly lower levels of general political knowledge could be attributed to young African Americans’ high commitment to voting in general and targeted outreach by campaigns and grassroots organizations. These factors seem to have countered the lower-than-average SES and educational attainment of African American youth. CIRCLE’s and others’ research shows that African Americans continue to be overrepresented among noncollege youth, and educational progress continues to be the strongest predictor of political knowledge (Godsay, Kawashima-Ginsberg, Kiesa, & Levine, 2012).
There was a relationship between involvement in extracurricular activities and political knowledge. This is consistent with Thomas and McFarland (2010), a longitudinal study, but it remains unknown why and how extracurricular involvement is related to political knowledge. One possible explanation might be that a measure of political knowledge is a proxy for academic achievement. Extracurricular involvement has been shown to be related to better overall socio-emotional and academic benefits (Gerber, 1996; Mahoney & Cairns, 1997; Marsh, 1992). Therefore, young people who are involved in extracurricular activities may be more academically engaged and maintain better social-emotional health than those who do not. In turn, academically engaged and socially and emotionally healthy youth may also have better political knowledge than those who are not.
We also found that both strength of party identification and the party young people identified with affected their civic engagement. First, young people who explicitly identified with either political party were more engaged, perhaps because people who identify with a party are more motivated to acquire information and vote, or perhaps because engaging with the election boosts party identification. Our study is not able to draw a causal link in either direction, but this is a fruitful area for further study because at least some teachers feel the need to avoid partisan or controversial topics in the classroom (Commission on Youth Voting and Civic Knowledge, 2013). If young people can form their own opinions and political identities through balanced discussions of partisan ideas, it may be worthwhile to incorporate the ideas and policies supported by various political parties into class discussions. In our analysis, conservative youth were overall less engaged, after controlling for other factors, including the strength of partisan identification. This could be because conservative youth’s voices were not fully represented by the Republican party’s policy agenda in 2012 or because they simply could not identify with either presidential candidate. An unpublished analysis of the Commission for Youth Voting and Civic Knowledge Youth Survey data by the first author indicates that conservative youth are far less likely to feel that Mitt Romney represents their views, compared to liberal youth’s feeling that President Obama represents their views.
Educational progress, as expected, had a strong relationship to electoral engagement, which is consistent with previous studies about electoral engagement (Nie et al., 1996). In our study, we used a measure of educational progress instead of attainment to distinguish the effect of age from the effect of education. By scaling this indicator on an age-centered mean, we essentially captured the effect of being ahead of same-age peers in education. Being ahead educationally is likely related to both future educational attainment and social class because entering college at 18 and finishing at 22 is no longer a typical model of college completion. Higher-SES youth are overrepresented in the group that finishes college by their early 20s, and a recent study finds that community colleges are becoming an entry point for low-income youth, with low success rates regardless of their academic qualifications (Century Foundation Task Force, 2013). Thus, a lack of significant finding related to SES in our study should not be interpreted as a lack of relationship between social class and civic engagement. On the contrary, social class, which translates to how quickly young people pursue and complete college, has a powerful effect on civic engagement among youth.
Limitations and Future Directions
Though the study had many strengths, including a large, nationally representative data set on civic participation among young Americans, strong and diverse measurements of civic knowledge and participation, and comprehensive statistical modeling, some weaknesses should be noted for future attention. First, this was a cross-sectional survey, meaning that we cannot draw any causal inferences from our analysis.
The method also introduces well-known sources of bias. Self-selection is one. People who held favorable views of political participation even before high school might elect to take engaging courses and programs, in which case their predispositions (not measured here) would be partly responsible for their higher rates of civic engagement. Further, participants reported their experiences in high school retrospectively, via self-reports. Young adults who are now engaged might recall and report positive experiences even if they did not receive such experiences at a higher rate than their peers. The dependent variables were also self-reported. We estimate that the actual turnout by 18- to 24-year-olds in 2012 was 48.3%. 10 In the survey discussed here, 59.9% said they voted, and 52.6% said they had voted and told us whom they had chosen. That degree of overestimate is very typical for a political survey and probably reflects a combination of two factors: Voters are more likely to participate in surveys than nonvoters, and some people claim they voted when they did not. Thus, the results of this study should be treated as modestly skewed toward voters.
All that being said, the correlations reported here are consistent with previous studies that have used a wide range of methods, and our main original contribution is an estimate of the impact of state policies, which should be less affected by self-selection, self-report, and response bias. While it is difficult to imagine a plausible experimental study on state laws and policies, there are at least two ways in which the current study could be improved. One would be to follow the same group of young people over time to further understand how their civic education experience and contact with restrictive and facilitative voter laws early in their adulthood relate to their civic participation and knowledge later in life. Another would be to study the process of new policy implementation as a series of case studies to understand which factors affect full implementation of policies (e.g., a new policy on civics assessment) and how long the process takes, with a goal of facilitating equitable and swift implementation of a policy that is designed to improve youth civic engagement and knowledge.
Footnotes
Acknowledgements
The authors thank the members of the Commission on Youth Voting and Civic Knowledge for feedback.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The survey that informs this article was funded by the Spencer Foundation, and the scan of state civic education policies was funded by the S.D. Bechtel, Jr. Foundation. We received support from the S.D. Bechtel, Jr. Foundation, the W.T. Grant Foundation, the Robert R. McCormick Foundation, the Spencer Foundation, and the Youth Engagement Fund to collect and analyze the data for this study.