Promoting Democratic Engagement during College: Looking beyond Service-Learning.

Abstract Preparing students to be engaged participants in our democratic society has long been an important college outcome. Over the past few decades, postsecondary institutions have primarily attempted to improve civic outcomes by integrating service activities into their curricula. While research on the effects of service-learning are plentiful, research on how other educationally beneficial activities influence democratic outcomes is scarce. In this study, we find that service-learning may not be the only means for promoting democratic outcomes because other high impact practices, most prominently learning communities, have greater or equivalent relationship to two dimensions of democratic engagement.

These trends have caused many institutions to reemphasize their responsibility to develop informed students who contribute to our democracy. Over 1,100 institutions are members of the Campus Compact, which seeks to promote higher education as a public good through promoting engagement in service activities (Campus Compact, 2015). Other initiatives such as the American Democracy and Political Engagement Projects have also sought to embed service activities into the curriculum. In turn, an increasing number of institutions now offer courses with a service-learning component designed to increase students' civic engagement. In 2014, 52 and 62% of first-year and senior students, respectively, reported taking at least one class that included service-learning (National Survey of Student Engagement, 2014).
While many researchers have investigated the relationship between service-learning and civic engagement (e.g., Astin & Sax, 1998;Conway, Arnel, & Gerwien, 2009;Cress, Astin, Zimmerman-Oster, & Burkhardt, 2001), scholars have largely overlooked how other educational practices and activities can influence civic engagement. This oversight fails to distinguish between differences in volunteering and the democratic responsibility to participate in polities and civil society (Bok, 2001). Service-learning is often an apolitical activity that can help improve society in a micro sense but cannot always address systemic problems and bring students into formal politics. Thus, scholars such as Bok (2001) and Finley (2011) cautioned against assuming a direct linkage between service-learning and democratic outcomes.
In this study, we investigate the relationship between selected high impact educational practices and two dimensions of civic engagement: democratic awareness and democratic participation. High impact practices (HIPs) are activities that have been found to impact a variety of educational outcomes. These practices and activities introduce students to diversity, provide them with responsive and meaningful feedback, facilitate interactions with faculty and peers, connect them to settings off-campus, and require them to spend a significant amount of time and effort engaged in educationally beneficial activities (Kuh, 2008;Swaner & Brownell, 2008). However, limited research has investigated how participation in these activities, other than service-learning, influence democratic outcomes. To address this limitation, we compared and contrasted the relationships of selected HIPs, using a large, multi-institution sample of seniors, and found that service-learning may not be the best avenue to promote democratic awareness and participation.

Literature Review
Because civic engagement lacks a clear definition in the literature (Adler & Goggin, 2005;Finley, 2011), it is important for us to begin by outlining our meaning of civic engagement. For the purposes of this study, we chose to utilize the term democratic engagement rather than civic engagement because its meaning encompasses participation in both civil society and polity. We adopted Michael X. Delli Carpini's (2006) definition of democratic engagement: "the combination of democratic awareness and democratic participation" (para. 1). He defined democratic awareness as "cognitive, attitudinal, and affective involvement in BOTH civil society and the polity" (para. 1) and democratic participation as "individual and collective actions designed to address public issues through the institutions of BOTH civil society and the polity" (para.1). By using the term democratic rather than civic, we were able to account for the myriad of ways students can seek to improve American society.
Service-learning and volunteering during college are positively correlated with a wide range of desirable student academic and social outcomes. Service activities have been positively correlated with higher retention and graduation rates (Astin & Sax, 1998;Gallini & Moely, 2003). Astin, Sax, and Avalos (1999) found that students who volunteer tend to be more socially responsible and committed to their communities and education. Service-learning, the main institutional response to improve student engagement in public life, has been positively related to students' interpersonal development, sense of social responsibility, and leadership and communication skills (Astin & Sax, 1998;Cress et al., 2001). Additionally, Conway and colleagues' (2009) meta-analysis found that the effect sizes of participation in a service-learning course were 0.28, 0.36, and 0.30 (small to medium) for personal, social, and citizenship outcomes, respectively.
Research assessing volunteering or community service assumes that these activities bring students into civil society and the polity. However, as Bok (2001) highlighted, volunteering to help the poor is admirable, but the act fails to solve the problem of poverty. This point is buttressed by the high level of volunteering but low interest in politics among undergraduates today (Eagan et al., 2015). Additionally, civic and democratic engagement research has been restricted to small samples and case studies, limiting the ability to generalize findings (Finley, 2011).
Connections between pro-social behaviors and different forms of service and civic engagement have advanced an increasingly nuanced and complex picture of civic engagement in college (Kahne, Westheimer, & Rogers, 2000;Moely & Miron, 2005;Morton, 1995). Morton (1995) initially challenged the theory of a charity-to-justice continuum of service-learning, which posits that students move from unquestioning charity to social justice activism through their experiences with service-learning. Instead, Morton asserted that three paradigms for service exist: charity, project models, and social change models driven, at least partially, by different student preferences for pro-social behavior. Moely and Miron (2005) found that students prefer the charity paradigm, or helping service activities, advanced by Morton (1995).
Furthermore, democratic engagement promotion differs by institutional type. It is argued that larger research universities have been more successful at incorporating civic engagement (Lounsbury & Pollack, 2001). Other environmental factors, such as the collective attitude of the student body, influence students' commitment to civic engagement (Astin, 1993;Sax, 2000). In addition, course structure and campus activities play a role in developing civic skills along with the cultivation of civic knowledge and values (Beaumont, 2005). However, developing socially and civically responsible citizens does not appear to be a universal goal among postsecondary faculty . These factors are especially salient when looking at the disciplines comprising the academy because as Zlotkowski (2001) claimed, civic engagement is not part of the disciplines' perceived goals. For example, Linda Sax (2000) noted that a relationship exists between a student's chosen major and their engagement in democratic activities during and after college. She found that history and political science majors were 55% more likely during college to discuss politics than the average college student. She found that major and different dimensions of democratic citizenship during college lasted beyond graduation, noting that majoring in engineering had a negative effect on social activism activities after college. That effect, she claimed, is consistent with Astin's finding that majoring in engineering is associated with an increase in materialism and a decline in concern for the larger society (Astin, 1993).
While much research has examined community service associated with a course (e.g., Conway et al., 2009;Finley, 2011), little attention has been devoted to assessing how other highly effective educational practices influence democratic engagement. A single study found no relationship between civic engagement and learning community participation after controlling for other factors, but the study relied on a relatively small sample (Rowan-Kenyon, Soldner, & Inkelas, 2007). Literature on the effect of global experiences on civic engagement has focused on service-learning courses overseas (Bringle, Hatcher, & Jones, 2010) or on building a global perspective (Braskamp, Braskamp, & Merrill, 2009;Tarrant, Rubin, & Stoner, 2014). We were unable to find any studies that examine the relationship between democratic/civic engagement, undergraduate research with faculty, and non-service-based senior capstone projects. Thus, to fully understand how the college experience influences and promotes democratic/civic engagement, researchers must look beyond service-learning and examine other student experiences.

Theoretical Framework
Social capital theory guided this study. Social capital is essentially an individual's network of sustained, trustworthy, and reciprocal relationships (Portes, 1998). Social capital networks can be formal, such as members of a football team, or informal, such as friendships among individuals living in the same dormitory. Putnam, Leonardi, and Nanetti (1993) contended that social capital has three essential elements: moral obligations and norms, social networks, and social values.
There are two main types of social capital: bonding and bridging (Halpern, 2005;Putnam, 2000). Bonding social capital tends to form in networks where the individuals have a shared trait, such as members of a Black fraternity. In contrast, bridging social capital is more outward looking and connects heterogeneous students such as members of a volunteering group. Social capital is cultivated during the collegiate experience and used to leverage change throughout life. Its value lies in the potential to facilitate information sharing and collective action, which individuals can leverage to implement solutions for collective problems and issues (Halpern, 2005).

Research Questions
Due to the incongruence between the design of service-learning courses and civic engagement identified by Bok (2001), we took an exploratory approach to identify how different college experiences influence college seniors' democratic engagement. We paid particular attention to the collection of activities known as HIPs due to their known ability to positively influence a variety of desirable educational outcomes. Additionally, we were interested in HIPs that cultivate social capital, which students may use throughout college to acquire and share information, and to organize and collaborate with others concerning issues of the public good. Therefore, guided by social capital theory, we investigated the following research questions on democratic engagement among college seniors: 1.
How are student and institutional characteristics associated with democratic awareness? 2.
How are student and institutional characteristics related to democratic participation? 3.
How does participation in selected HIPs influence students' democratic awareness and participation? 4.
How does the estimated effect of participating in selected HIPs compare to the estimated effect for service-learning on democratic awareness and participation?

Data
To answer the research questions, we utilized data from U.S. college seniors who responded to the 2014 National Survey of Student Engagement (NSSE). NSSE is a large, multi-institutional survey administered annually that examines students' participation in educationally beneficial activities in-and outside of the classroom, time-usage patterns, and satisfaction with the institution. Due to our focus on democratic engagement, we limited our sample to students who also responded to the NSSE Civic Engagement module, a set of questions participating institutions may elect to administer to their students. We also excluded part-time and distance-learning students and those enrolled at special-focus institutions from our sample. We excluded these populations due to our focus on HIPs, which are generally designed for more traditional student populations. The response rate for the sample was 28%. Previous research has demonstrated that NSSE data are not prone to substantial self-selection or non-response bias and produce reliable results at response rates similar to 28% rate achieved in this study (Fosnacht, Sarraf, Howe, & Peck, 2017). After accounting for these exclusions, our data sample contained 10,305 students who attended 46 institutions. About two-thirds of the respondents were female. Approximately, three out of four students were White, while Asians and Latinos each comprised 4% of the sample. Black and international students represented 7 and 5% of the sample, respectively. The largest major fields were business, social sciences, and the health professions, although, the respondents were well distributed across the disciplines. Three out of five students attended a public institution. A majority of students were enrolled in institutions that offered master's degrees, while 30 and 12% of the respondents attended doctoral universities and baccalaureate colleges, respectively.
Our outcomes of interest were democratic awareness and participation. We created these outcomes by applying Samejima's (1969) graded response model (GRM) to items from the civic engagement module (see Appendix). GRM is a generalization of the two-parameter item response theory (IRT) model for ordinal outcomes. IRT is a probabilistic framework and set of methods to evaluate the relationship between a latent trait and item(s) measuring the trait. The democratic awareness variable was derived from items asking how often the respondents informed themselves about or discussed "local or campus" and "state, national, or global" issues. The democratic participation construct was created from items inquiring about how often the student "raised awareness about," "asked others to address," and "organized others to work on" "local or campus" and "state, national, or global" issues. We used GRM rather than a factor analysis approach to create and score these variables because these activities require various amounts of effort and skill, and GRM accounts for these variations. We created a score for each student by using the item parameters in Appendix to calculate the most likely position on the latent trait continuum given the students' responses. The α parameter indicates how well an item discriminates between individuals at different levels in the latent continuum and is derived from the slope of the item parameter curve. The β parameters indicate the threshold where on the latent continuum an individual would most likely choose that response option. The marginal reliabilities were 0.85 for both outcomes. We checked the IRT assumptions of unidimensionality and local independence for both outcomes by performing an exploratory factor analysis of the polychoric correlations. We standardized both variables to have a mean of 0 and a standard deviation of 1.
We also used data on a number of student characteristics such as race/ethnicity, sex, adult status (age > 23), major field, transfer status, nationality, parental education, grades, Greek-life membership, and residing on-campus. To control for variations in the institutions attended by the respondents, we used data on the following institutional characteristics: control, Basic 2010 Carnegie Classification (aggregated), Carnegie Community Engagement Classification (dummy variable indicating being a member of either classification), locale, residential character, undergraduate enrollment, Barron's rating, and region.
We also utilized data on student participation in a number of high-impact practices: learning communities, study abroad, research with a faculty member, senior capstones, and service-learning (recoded to "not done" vs. "done"). These activities have been dubbed "high-impact" because they have been frequently shown to improve student learning and development (Kuh, 2008;Swaner & Brownell, 2008). Sixty-eight percent of the sample had a course that incorporated service-learning, 59% completed a senior capstone, 31% did research with a faculty member, 29% participated in a learning community, and 20% studied abroad.

Analyses
Due to the exploratory nature of this study, we began our analyses by examining the descriptive relationship between participation in HIPs and democratic awareness and participation. We tested these relationships by performing two-group t-tests that grouped students who did and did not participate in each of the HIPs. After examining these relationships, we investigated how they changed after controlling for both student and institutional characteristics. This investigation entailed creating two ordinary least squares models. We did not use multilevel modeling to account for the nesting of students within colleges due to the low intraclass correlation coefficients (ICC) for both of our outcome variables (Lee, 2000), which were less than 0.03. However, we did use robust standard errors that were adjusted to account for the clustering of students within institutions. For each outcome, we then estimated a multivariate model to investigate the relationship between the outcome of interest and the student and institutional characteristics identified above, except for the HIPs. We then estimated a second model for each outcome that added indicators of student participation in the HIPs. With this model, we performed post hoc tests that compared the estimated HIP coefficients to the estimate for service-learning. The post hoc tests examined whether or not the estimated HIP coefficient minus the service-learning coefficient was significantly different from zero using a z test. These post hoc tests were performed using the LINCOM command in Stata. Because we standardized both of the outcome variables, the coefficients from the multivariate models represent the estimated effect size for a one-unit change in the independent variable.

Limitations
This study suffers from a number of limitations. First, this was an exploratory study, and the relationships detailed in the following should be viewed as correlational, not causal. Student participation in HIPs is voluntary in most cases, and participating students may be predisposed to engage in democratic activities. Due to the cross-sectional nature of our data, we were unable to assess the direction of many of the associations found in our results, and the results may be subject to self-selection bias. For example, does participation in service-learning activities promote democratic engagement, or does an interest in democratic activities lead students to enroll in a service-learning course? While the existing literature has established a connection between these activities (e.g., Astin & Sax, 1998;Conway et al., 2009;Cress et al., 2001), it is largely quiet on the direction and order of the relationship. Additionally, little research has investigated why students choose to participate in the other HIPs examined in this study, which precluded us from eliminating the possibility that an interest in democratic activities leads students to participate in programs like learning communities and research with faculty. Before altering practice, the findings should be replicated using other data sources and experimental or quasi-experimental approaches, if possible. If these techniques are not possible due to ethical concerns, a longitudinal study could help establish the direction of the relationship between HIPs and democratic engagement and lead to more concrete evidence on their effectiveness. The sample contains students attending institutions that chose to administer NSSE's civic engagement module. Due to the expressed interest in the module, these institutions may not be representative of bachelor's-granting institutions nationally. Additionally, we relied upon students to accurately self-report in which programs and activities they have participated. We were also unaware of the specific programs and activities the respondents reported doing. We were unable to account for program effect diffusion, which occurs when a participant interacts with a non-participant. Students bring to college a variety of predispositions that effect their participation in service and democratic behaviors (Weerts & Cabrera, 2015). Due to these limitations, our results should be viewed as broad average estimates and not be applied to a specific practice or program.

Results
We present the study's results below in two sections, one for each outcome. We begin the sections with the bivariate results for the high impact practice items. Then we present the multivariate results, and we conclude with results that compare the high impact practice estimates to the service-learning estimates.

Democratic Awareness
We began by comparing engagement in democratic awareness activities by participation in selected HIPs using two-group t-tests. The means were significantly different on all five HIPs examined. The magnitude of the mean differences (in SDs) were 0.14 for service-learning, 0.22 for senior capstones, 0.27 for research with faculty, 0.31 for learning communities, and 0.33 for study abroad.
The multivariate results can be found in Table 1. When we held other factors constant, the models show that Asians and Latina/os were less likely to engage in democratic awareness activities than Whites. However, adult and male students were shown to be more likely to engage in democratic awareness activities when we controlled for other variables. Substantial differences were observed by major field because most fields were significantly lower than the social sciences. The exceptions, arts and humanities, communications, media, and public relations and social service professions, were not significantly different. When we held other variables constant, students who had earned mostly A's were shown to be significantly more likely to engage in democratic awareness activities than students with lower grades. Democratic awareness did not vary much by parental education with the exception of students with a parent who had earned a doctoral or professional degree. These students were more likely to take part in democratic awareness activities than students with a parental education level of bachelor's. After we controlled for other characteristics, students who had participated in Greek-life and/or lived on-campus were found to be more likely to participate in democratic awareness activities Students attending highly residential institutions were shown to be less likely to engage in democratic awareness activities than their peers at non-residential institutions, when we held constant other factors. Selectivity was positively correlated with democratic awareness activities, and students attending Midwest institutions were found to be less engaged in democratic awareness activities than students in the Northeast, after we controlled for other characteristics. Participation in the five HIPs examined was positively correlated with democratic awareness, after we held other factors constant. The largest estimated effect size was for learning communities (0.23), followed by study abroad (0.17) and research with faculty (0.15). The estimated effect sizes for servicelearning (0.10) and senior capstone projects (0.08) were lower. Table 2 compares the high impact practice participation estimates from Model 2 to the estimated effect of participating in service-learning. Learning communities and study abroad both had estimated effects greater than service-learning. The effect size differences were 0.13 and 0.07, respectively. The estimated effects of undergraduate research and senior capstone participation were statistically equivalent to service-learning.

Democratic Participation
Like the analyses for democratic awareness, we began by comparing the means for democratic participation between participants and non-participants in selected HIPs using t-tests. The mean differences were all significant. The mean differences (in SDs) were 0.14 for senior capstones, 0.30 for study abroad, 0.31 for service-learning, 0.31 for research with faculty, and 0.47 for learning communities. The multivariate results for democratic participation can be found in Table 1. The second model shows that, after we controlled for other characteristics, foreign, Asian, and Black students were found to be more likely to participate in democratic participation activities than Whites. Males, greek-life members, and on-campus residents were shown to have higher levels of democratic participation than their peers, when we held constant other characteristics. Apart from communications, media, and public relations, the social service professions, and the biological sciences, agriculture, and natural resources, where the differences were non-significant, social science majors on average had significantly higher levels of democratic participation than their peers in other major fields. Students who earned mostly B's showed slightly higher levels of democratic participation than students who earn mostly A's, after we controlled for other characteristics. After we controlled for participation in HIPs and other characteristics, students with a parental education level of high school or less were found to be more likely to participate in democratic activities than students with parental education level of bachelor's.
Students who attended master's-granting institutions were shown to be more likely to participate in democratic activities than their peers at doctoral institutions, when we held constant other characteristics. Students at private institutions exhibited higher levels of democratic participation than students at public institutions, when we controlled for other factors. Additionally, students attending colleges located in the Midwest or Southeast were less likely to engage in democratic activities than students in the Northeast. Other institutional characteristics were nonsignificant.
Participation in HIPs appears to be one of the best predictors of democratic participation. The estimated magnitude of participating in a learning community on democratic participation is 0.36 SDs, after we controlled for other characteristics. The estimated effects of participating in a service-learning course, research with faculty, and Learning community 0.13 *** 0.14 *** Study abroad 0.07 * −0.06 Research w/faculty 0.05 −0.02 Senior capstone −0.02 −0.24 *** * p < 0.05, ** p < 0.01, *** p < 0.001 study abroad were smaller but still sizable. However, working on a senior capstone project was not associated with democratic participation, when we controlled for other factors. Table 2 shows the results when the HIP coefficients from the second model are compared to the service-learning estimates. Participation in a learning community was associated with greater gains in democratic participation than taking a service-learning course. The estimates for study abroad and participating in research with a faculty member were equivalent to service-learning. However, service-learning appears to have a significantly stronger relationship to democratic participation than participating in a senior capstone project.

Discussion
Creating active and able citizens for our democracy is a cornerstone of U.S. higher education. However, throughout the 20th century, this outcome was deemphasized due to the increasing importance society placed on the private benefits of higher education (Astin et al., 2002;Boyte & Hollander, 1999;Eagan et al., 2015;Ehrlich, 2000).
Institutions have gradually recognized the problematic nature of this trend and responded by integrating service activities into their curricula. This response has assumed a strong linkage between community service activities and participation in democratic activities. While there is evidence that service-learning improves democratic outcomes (Conway et al., 2009), leaders such as Bok (2001) have cautioned against overemphasizing this connection. In this study, we examined if alternative educational practices have stronger relationships with two types of democratic behaviors: democratic awareness and participation. Based on data from a large multi-institutional sample of college seniors, our results comport with Bok's (2001) suspicions. We found that learning community and study abroad participation had a stronger relationship with democratic awareness than participating in a service-learning course, after we controlled for other student and institutional characteristics. We found a similar relationship for democratic participation because the estimated coefficient for learning community participation was significantly greater than service-learning. Additionally, the estimates for undergraduate research and study abroad were statistically equivalent to the estimated effect of service-learning.
However, despite the aforementioned findings, we must acknowledge that service-learning was significantly and positively related to both democratic awareness and participation. Thus, integrating service-learning into the curriculum helps improve undergraduates' democratic engagement and is not an activity that should be avoided. However, it may be possible to improve the efficacy of service-learning on democratic outcomes. As Finley (2011) noted, "although service-learning by definition engages students' in a community, that engagement may or may not [italics original] be politically-oriented or intentionally structured to deepen the specific knowledge or skills associated with developing democratic participation or citizenship" (p. 3). Accordingly, institutions may be able to improve the effectiveness of service-learning programs by taking steps to further connect the service activities to students' role in their communities and/ or focusing on the community dialogue required for democratic governance.
We believe our most important finding is the relationship between learning communities and democratic engagement. Our models indicate that, after controlling for other factors, learning community participation increases democratic awareness and participation by roughly a quarter and third of a standard deviation, respectively. These effect sizes are not trivial in education research (Lipsey et al., 2012), and we did not observe substantial reductions in the estimates after controlling for a variety of characteristics.
While the statistical results highlight the role of learning communities in promoting democratic engagement, the real world connection is not immediately clear. Learning communities are programs where a group of students take two or more classes (usually organized around a theme or common interest) together, typically in their first year. We posit learning communities help build bonding social capital because they bring students with a common interest or shared trait together for a sustained period of time in courses emphasizing group work. They also help integrate ideas and learning across the disciplines, frequently connect this learning to societal issues, and promote involvement in academic and non-academic activities outside of the classroom.
When considering these features, we believe learning communities most likely influence democratic engagement by creating mini-democracies. Frequent interactions among the students over a sustained period of time build trust and community among the members, which when combined with their shared interest, form sub-communities within the school. The trust built within communities allows for more engaging interactions and a free and open dialogue. This information sharing allows the learning communities to identify problems in their local community and beyond. Learning communities are natural locations to facilitate change because students use the teamwork skills they acquired through their collaborative-centric course work and social networks to identify allies and build collations that advocate for common goals.
Unlike the other practices examined, learning communities are the high impact practice best positioned to create bonding social capital. Therefore, the strong relationships built within learning communities may be the genesis of our results. Bonding social capital also may explain the duration of the effects. As learning communities are typically a first-year program, we would have expected their effects to dissipate over time because our sample is comprised of seniors. Rather, the results suggest that the effects of learning communities linger over time, which has been found by other researchers (Kilgo, Sheets, & Pascarella, 2015;Zhao & Kuh, 2004).
In addition to learning communities and service-learning, other HIPs appear to improve democratic engagement outcomes. Study abroad participation was associated with a non-trivial increase in both of our outcomes after controlling for other factors. Thus, it appears that study abroad promotes learning about, discussing with, and enlightening others about issues. We observed a similar result for undergraduate research with faculty because this practice was associated with higher levels of democratic activities holding constant other characteristics. The least effective of the HIPs studied in improving democratic engagement appears to be senior capstone projects. This activity was associated with a relatively trivial increase in democratic awareness and had no significant relationship to democratic participation.

Implications for Research
A primary goal of this study was to identify educational practices other than service-learning that may increase students' democratic engagement. By focusing solely on service-learning programs, researchers appear to have missed more organic episodes of democratic awareness and participation. Thus, looking beyond explicitly identified civic engagement programs may enable researchers to identify other areas where students are democratically engaged during their collegiate years. Our results suggest that multiple practices may be correlated with democratic engagement. Further research should confirm our results and design studies that are better equipped to estimate the unbiased effects of these programs.
Additionally, our interpretations of the learning community results indicate that friendship networks and other out of class activities influence democratic engagement. While democratic engagement is an educational pedagogy, the results of this education are likely to emerge and manifest themselves outside the classroom. This process is where student agency regarding issues of democracy is less constrained and can be put into action. More research should examine how extracurricular activities foster democratic engagement.

Implications for Practice
All of the high-impact practices studied had positive effects on democratic awareness, and all, with the exception of senior capstone projects, had positive effects on democratic participation. Therefore, our results show that multiple activities can improve undergraduates' democratic engagement. As a result, institutions should be cautious about viewing service-learning as the only or primary means for providing the civic knowledge and tools necessary for democratic engagement. Institutions should emphasize dialogue and relationship building for their students. During that process, students should be exposed to difficult ideas and controversial problems. As Kuh (2008) indicated, exposing students to controversial and difficult ideas within the context of a trusted group like a learning community can result in a wide range of positive outcomes. Furthermore, our service-learning findings may indicate that many service-learning courses are not well structured and implemented. Institutions should seek to advise and assist faculty on how to structure their courses so students receive the maximum possible benefit.

Conclusion
Undergraduates are the future leaders and inheritors of our democratic experiment. In this study, we attempted to address some of the literature gaps and investigate how postsecondary institutions can improve student outcomes related to democratic engagement. Identifying these practices is critical because as Boyte (2008) asserted, institutions must both create an environment and develop students' skills to cultivate democratic engagement. Using data from a variety of institution types, we found a stronger relationship between learning community participation and democratic engagement than for service-learning. Thus, it appears that multiple programs and practices may foster civic engagement and that postsecondary institutions have been overlooking these alternative practices to improving students' democratic engagement.