Action Civics for Promoting Civic Development: Main Effects of Program Participation and Differences by Project Characteristics

Abstract

Using both quantitative and qualitative data, this study examined the effect of participating in an action civics intervention, Generation Citizen (GC), on civic commitment, civic self-efficacy, and two forms of civic knowledge. The sample consisted of 617 middle and high schools students in 55 classrooms who participated, or were soon to participate, in Generation Citizen. Hierarchical linear models revealed that participating in Generation Citizen was associated with positive gains in action civics knowledge and civic self-efficacy. Qualitative coding identified three types of project characteristics that captured variability in the action projects student chose to complete: context, content, and contact with decision makers. Interactions between project characteristics and participation in GC revealed differences in civic outcomes depending on project characteristics.

Public schools in the Unites States (US) have a mandate to prepare students for participation in public life (Campaign for the Civic Mission of Schools, 2011; Dewey, 1916; Malin, Ballard, Attai, Colby, & Damon, 2014). Public schools aim to prepare students for college, careers, and civic life (Kaestle, 2011; National Council for the Social Studies, 2013). Arguably, the “third C” is especially important as it applies to all students regardless of their career paths. Or, as Ted McConnell memorably put it at the Ford Foundation convening on educating for democracy, “Not every student will be the next Steve Jobs, but every single dang student will be a citizen” (Jan 23, 2015, NYC).

Preparing Youth for Civic Life

Public schools in the US are currently not well-equipped to provide meaningful civic experiences and many students leave school unprepared for civic life (Generation Citizen, 2015). Schools largely prioritize subjects covered on standardized tests; civic learning is too often squeezed out (Campaign for the Civic Mission of Schools, 2011; Godsay, Henderson, Levine, & Littenberg-Tobias, 2012; Malin et al., 2014). Exacerbating the problem of too few civic opportunities, opportunities that do exist are unequally distributed across schools and neighborhoods, and race and socio-economic status are strongly predictive of civic preparedness and participation (Kahne & Mid-daugh, 2008; Levinson, 2010). While some schools and districts do prioritize civic learning, schools can do more to educate active citizens and inspire and support their civic interests. Importantly, schools must have support to do so as they are currently saddled with a broad and diverse set of expectations from multiple stakeholders.

Youth today exhibit mixed interest in civic life. In general, youth show low levels of participation in formal political activities such as voting and campaigning; however, many seem interested and involved in “apolitical” forms of civic action such as volunteerism as well as activist forms of civic participation, such as #BlackLivesMatter (Galston, 2001; Generation Citizen, 2015; Keeter, Zukin, Andolina, & Jenkins, 2002; Syvertsen, Wray-Lake, Flanagan, Osgood, & Briddell, 2011). Youth development researchers have shown that many youth are interested in social issues and that educators, both in and out of school, can capitalize on youth interests to inspire and prepare youth for effective civic participation (Ballard, Malin, Porter, Colby, & Damon, 2015; Kahne & Sporte, 2008; Zeldin, Larson, Camino, & O’Connor, 2005). Schools can play a role in closing civic development gaps by providing more civic opportunities, especially for those often left out of civic processes.

Promoting Civic Commitment, Efficacy, and Knowledge

Successful civic education should position students to leave high school with knowledge of civic structures, a belief in their own efficacy to engage meaningfully with those structures, and the intention to do so. Among the many aspects of civic education that are important to promote, three that educators and practitioners commonly strive for are promoting civic commitment (i.e., plans to engage in civic activities in the future), efficacy (i.e., the belief that one’s civic actions can lead to change), and knowledge (i.e., about political systems and processes; (Gingold, 2013; Kahne & Westheimer, 2006; Littenberg-Tobias & Cohen, 2016).

Classroom learning can support youth civic development (Eckstein & Noack, 2015; Wray-Lake & Sloper, 2015). Many specific practices and programs are connected with youth civic outcomes. For example, fostering a democratic and inclusive climate (Campbell, 2008; Flanagan, Cumsille, et al., 2007; Flanagan & Stout, 2010; Flanagan, Bowes, Jonsson, Csapo, & Sheblanova, 1998; Kahne & Sporte, 2008; Torney-Purta, 2002), prioritizing civic commitment as central to school identity (Ballard, Caccavale, & Buchanan, 2014) and providing civic opportunities through community service and especially service-learning (Hart, Donnelly, Youniss, & Atkins, 2007) seem to effectively support young people in civic development. In summary, there is an urgent need for young people to develop civic commitment, efficacy, and knowledge; schools are in a unique position to promote such civic learning; and there is evidence about how best to do so (Campaign for the Civic Mission of Schools, 2011; Malin et al., 2014). However, US public schools are also overburdened with high expectations for supporting youth development. We sorely need dynamic and effective approaches to civic education that prepare our young people for constructive participation in civic life and that work within the context of the US school system.

A New Approach: Action Civics

As athletes improve through practice, and performers through rehearsal, engaged citizens must hone their skills through active engagement in civic processes. A new model of civic education brings this premise to life through action civics (Generation Citizen, 2015; Gingold, 2013; Pope, Stolte, & Cohen, 2011). The National Action Civics Collaborative identified the guiding principles of action civics programs as: commitment to collective action, youth voice and agency, and reflection (Gingold, 2013).

Conceptually, action civics programs parallel programs rooted in community psychology such as youth participatory action research (YPAR). YPAR programs engage young people in identifying problems relevant to their own lives, conducting research to understand the problems, and advocating for changes based on research evidence (Kohfeldt, Chhun, Grace, & Langhout, 2011; Kornbluh, Neal, & Ozer, 2016; Ozer, 2015). Action civics programs include these steps within a curriculum aimed at civic learning. Action civics programs include elements from empowerment and critical consciousness theory and draw on many best practices from the robust literature on effective student-centered and project-based practices (Gingold, 2013; Levinson, 2012; Pope et al., 2011). Whether explicitly connected to empowerment theory or not, action civics programs are about empowering young people, who have traditionally been left out of formal civic settings, to engage in community development and improvement. Community psychologists define empowerment as the psychological aspects of processes through which people gain greater control over their lives, take a proactive approach in their communities, and develop critical understandings of their sociopolitical environments (Zimmerman, 1995). Many scholars describe its value for individuals and communities (Christens & Peterson, 2012; Wallerstein, 1992, 1993) and provide guidelines for effective youth empowerment work (Berg, Coman, & Schensul, 2009; Christens, 2012; Ozer, Newlan, Douglas, & Hubbard, 2013). The goals of action civics programs also align with the main principles of critical consciousness (CC) theory. CC is about how oppressed or marginalized people learn to critically analyze their social conditions and act to change them (Freire, 1973; Watts, Diemer, & Voight, 2011). Action civics programs tackle this through a focus on “root causes” of social problems that participants are taught to analyze. Students are expected to do critical analyses of why social problems exist and what actions might address them (Cipparone & Cohen, 2015; Pope et al., 2011).

Action civics programs capitalize on features that are evidenced to be effective from research on civic education, youth empowerment, and YPAR; for example, reflection, choice, engagement in interactive activities, and youth-adult interaction. Research on effective practices in service and service-learning courses provides evidence for the important role played by reflecting one’s experiences and connecting experience to classroom learning (van Goethem, Hoof, Orobio de Castro, van Aken, & Hart, 2014; Youniss, McLellan, & Yates, 1997). Furthermore, given the wide variety in the social issues that young people find important, a promising approach for effective civic education is to let students identify issues to pursue (Ballard et al., 2015; Morgan & Streb, 2001; Ozer et al., 2013). Action civics programs aim to help students learn civics by doing civics focused on personally relevant issues (Gingold, 2013). Civic learning might be especially effective in promoting civic knowledge, skills, and interest when students are engaged in interactive civic activities (Kahne, Chi, & Middaugh, 2006; Torney-Purta, 2002). Furthermore, effective civic education connects youth to others, especially to adults (Zeldin, Camino, & Mook, 2005) who are knowledgeable about civic and political life (McIntosh, Hart, & Youniss, 2007). The action civics model integrates these features in student-centered and interactive curricula.

Classroom-based action civics programs are fairly new and evaluation research is in its infancy. Thus far, evaluation is primarily based on self-reported student civic and academic outcomes (Gingold, 2013) and most often is conducted by action civics organizations themselves with the goal of self-understanding and improvement. There are many action civics models; the present study focuses on one called Generation Citizen, which integrates in-school with out-of-school approaches by partnering classroom teachers with college student volunteer “democracy coaches” to deliver an action civics curriculum (Pope et al., 2011). This approach seems promising because it integrates the knowledge and experience of classroom teachers with the energy and expertise of democracy coaches. In addition, this model brings students into communities and communities into schools, modeling interaction between the institutions of schools and communities. Initial evidence suggests that Generation Citizen led to some gains in civic knowledge and skills (Cohen, Littenberg-Tobias & Ridley-Kerr, 2014).

Are Action Civics Programs Equally Effective Across Classroom Projects?

Classroom-based civic programs have thus far been treated as homogenous interventions. In reality, this is far from true. Because these programs capitalize on student interest and choice (Gingold, 2013), the exact nature and goal of action civics projects vary widely. One group of students might choose an action project aimed at school lunch quality, while others may choose a project directed at community issues like gang violence or safe crosswalks. These projects differ along many dimensions such as the target of the project (school vs. broader community) and the type of issue they focus on. Importantly, some issues may be more likely to lead to civic growth. Action civics is a promising approach to civic education but evidence is needed to determine whether action civics programs deliver on their promise to help young people become informed, active citizens and if so, whether the nature of projects chosen modifies the effect of program participation.

The Present Study

Using both qualitative and quantitative data, we describe action civics classroom projects and test the effects of action civics projects on student civic outcomes. Through qualitative coding analyses we (a) examined the content of youth action civics projects from 1 year of Generation Citizen (GC) and created project characteristic codes to capture meaningful differences between action civics projects. Next we conducted quantitative analyses using hierarchical linear models to (b) test the main effects of GC involvement on four civic outcomes (civic commitment, civic efficacy, action civics knowledge, and local political knowledge) and (c) test whether the effects of GC on these outcomes differ by characteristics of action civics projects.

Method

Intervention: Generation Citizen

Generation Citizen offers a semester-long curriculum taught twice weekly by trained college student volunteer “democracy coaches” in partnership with classroom teachers. The Generation Citizen curriculum is student-centered and action-oriented, and works to promote a democratic classroom climate and empower youth (Stolte, Isenbarger, & Cohen, 2014). GC’s action civics process involves choosing a local issue to tackle collectively, learning strategies and skills for taking action, and developing and implementing an action plan accordingly (Pope et al., 2011).

Sample

Study participants were middle and high school students in schools participating in Generation Citizen during the 2013–2014 academic year. Data were collected in the winter of 2013. Schools were located in four metropolitan regions of the USA (the areas surrounding Providence, Rhode Island; Boston, Massachusetts; New York City, New York; and the San Francisco Bay Area, California). The sample for the current study consists of 617 students in 55 classrooms from 26 schools. All students from classrooms participating in GC were invited to participate in the study and parents and students who completed consent forms were included in study analyses. There are two groups of students: those who just finished Generation Citizen in the fall (the “post-GC” or treatment group) and those who are getting ready to participate in the program in the spring (the “pre-GC” or control group). The distribution of students in treatment and control groups can be considered quasi-random because GC program staff report that there were no systematic differences in when the program was delivered; rather, the semester of participation depended on scheduling logistics for schools and for Generation Citizen. Our analyses compare civic outcomes between the two groups and account for the fact that students were nested within classrooms. Participants were diverse with regards to ethnicity: 37% self-identified as Asian, 23% as Latino, 17% White, 12% Black, 11% other, and 53% of participants identified as female. Students attended schools where, on average, 62% of participants were eligible for free and reduced price lunch. Some demographic differences emerged between treatment and control groups. For example, a higher proportion of students in the treatment group were in high school, were female, identified as Black or Latino, and qualified for free or reduced price (FRPL) compared to students in the control group (see Table 1). To address these differences, we controlled for demographic variables in our analyses.

Table 1.

Sample demographics and action civics project characteristics by treatment and control group

	Overall (%)	Treatment (Post-GC; %)	Control (Pre-GC; %)	p-value
GC Participation	22	100	0	–
Demographic variables
Female	53	64	49	.010
White	17	15	18	.343
Asian	37	29	39	.042
Black	12	17	10	.040
Latino	23	28	21	.127
Other	11	11	11	.968
No absences	57	52	58	.302
1–5 absences	24	8	28	.010
More than 5 absences	19	40	13	<.001
% Free or reduced priced lunch	62	68	60	<.001
Class variables
High school	29	56	21	<.001
Number of democracy coaches	1.26	1.4	1.22	.003
Elective	13	15	13	.70
Civics	46	53	44	.27
History	41	33	43	.04
Action civics project characteristics
Easy/moderate access	38	32	39	.50
Difficult access	62	68	61	.50
School focus issue	45	69	39	<.001
Outside of school focus issue	55	31	61	<.001
Safety issue	41	53	38	.003
School environment issue	38	27	42	.003
Social issue	20	20	20	.95

Measures

We used two sources of data in the present study. The first comes from open-ended survey responses reported by democracy coaches who completed forms describing the action civics projects. Data from two questions are used for qualitative analysis in the present study: what issue their class focused on, and what the goal of the class’s action project was. The second data source is student-level data measured through self-report questionnaires.

Outcome Variables

Measures were created by GC or adapted from existing surveys such as the National Assessment of Educational Progress (NAEP) civics assessment, the IEA Civic Education Study, and the California Civic Index (Kahne, Middaugh, & Schutjer-Mance, 2005) and are described in detail elsewhere (Littenberg-Tobias & Cohen, 2016). See Table 2 for outcome variable descriptives. Future civic commitment was assessed through seven items in response to the question “When you think about life after high school, how likely do you think you will:” for example, “vote in every election,” “volunteer regularly,” and “stand up for your beliefs.” Participants rated items on a scale from 1 (definitely no) to 5 (definitely yes). Items were averaged to create a composite score. Civic self-efficacy was measured through five items in response to the question “How much do you agree with the following statements?” Statements included “I believe I can make a difference in my community” and “If I speak up about an issue my voice will be heard.” Participants rated items on a scale from 1 (definitely no) to 5 (definitely yes). Items were averaged to create a composite score. Action civics knowledge was rated through five multiple-choice items that were created by Generation Citizen and covered content from the GC curriculum. For example, “Imagine a neighborhood where many young people have been injured crossing the street on the way home from school. Which of the following is the biggest root cause of the problem?” with options: “(a) students don’t look both ways before crossing; (b) parents don’t pick up their children after work, (c) there are no cross-walks or safety precautions to slow cars down; (d) teachers let children walk home alone; (e) I don’t know” (correct answer is c). Items were coded as “correct/incorrect” and averaged to create a composite of correct answers out of five; the standardized score was used in analyses. Local political knowledge was rated through five open-ended items that covered local political content (measure created by Generation Citizen; Author 2). For example, “Today, who is the governor of your state?” Items were coded as “correct/incorrect” and averaged to create a composite of correct answers out of five; the standardized score was used in analyses.

Table 2.

Descriptives statistics for civic outcome variables

	Number of items	α	Min	Max	Full sample			Treatment (Post-GC)			Control (Pre-GC)
					N	Mean	SD	N	Mean	SD	N	Mean	SD
Future civic commitment	7	0.741	1	4.714	617	3.210	0.581	135	3.310	0.600	482	3.182	0.574
Civic self-efficacy	5	0.739	1	5.000	615	3.677	0.698	134	3.833	0.733	481	3.634	0.682
Action civics knowledge (% correct)	5	N/A	0	1.000	617	0.657	0.286	135	0.772	0.256	482	0.624	0.286
Local political knowledge (% correct)	5	N/A	0	1.000	617	0.324	0.249	135	0.379	0.267	482	0.309	0.242

Covariates

Student-level covariates included gender, race/ethnicity, and number of absences from school. Classroom level covariates included: the subject of the class (e.g., elective, civics, history) and the number of democracy coaches in classroom. School-level covariates were school level (e.g., high school, middle school), and percent of students at the school who are on free or reduced lunch, as reported by state departments of education.

Analysis

Qualitative Analyses

We first created a coding scheme to describe the GC action civics projects. Qualitative coding proceeded through an iterative process of open coding, creating a codebook, and refining codes (Boyatzis, 1998; MacQueen, McLellan, Kay, & Milstein, 1998; Miles & Huberman, 1994). Three coders were involved in the process: Author 1 and two research assistants (RAs). In the first phase, an RA open coded data provided by democracy coaches and created a list of 26 possible codes. This extensive list contained many codes that were too specific, so the first RA and Author 1 pared down the list into nine codes that captured meaningful aspects of action projects and ensured variability within codes, and together drafted a codebook. At this point, we also drew on Thapa, Cohen, Guffey, and Higgins-D’Alessandro’s (2013) analysis of the five dimensions of school climate (Thapa et al., 2013). Next, Author 1 trained a second RA on the nine codes, both coded 20 cases and met to discuss discrepancies and refine the codebook. The RA coded the remaining data and Author 1 coded 20% in order to check reliability. After coding the project characteristics, we reduced the number of codes for analysis: three final codes were chosen for quantitative analyses that captured unique information about action civics projects (to cut redundant codes, based on cross tabulations), were coded reliably (according to kappa reliabilities and percent agreement between coders), and had adequate variability across classrooms.

Table 3 shows the frequencies for the classroom action civics project characteristics. Kappa reliability analyses across two coders on 20% of the data for the three analysis codes ranged from good to excellent (Table 3). Coders resolved all discrepancies and refined the codebook and the second RA independently coded all remaining cases (codebook available from first author upon request).

Table 3.

Frequencies and reliabilities for action civics classroom project characteristics

	N students (%)	N classrooms (%)	Number of classrooms double-coded	Kappa	Number of cases disagreement	% agreement
Context
School	328 (53)	28 (50)	14	1.00	0	100
Outside of school	278 (45)	22 (40)
Content
Safety	254 (41)	19 (35)	17	0.911	1	94
School environment	236 (38)	20 (36)
Social issues	124 (20)	16 (29)
Contact with decision makers
Easy/moderate	294 (48)	27 (59)	11	1.00	0	100
Difficult	209 (34)	19 (41)

Context

This variable captures whether action civics projects address student concerns at proximal or distal levels. We coded two categories: projects that focused on issues in school and those that focused on projects outside of school such as local, city, and state level projects. Examples of projects at the school level were: school food; lack of community in school; current courses in school not reflecting students’ backgrounds and histories; and bullying at school. Examples of projects outside of school were: car safety; improvement of bike and pedestrian safety along [specific street]; [specific park] is in bad shape and students find a lot of used needles in the playground; and lack of mental health service for homeless in community.

Content

This variable represents the primary issue that the classroom projects focused on addressing. We coded three categories. The first is safety, which includes issues pertaining to personal and community safety at the school and community level. Examples of this are: bullying; theft on campus; conflict mediation; lack of information about safety concerns, safe driving habits; safety in community parks; safety on public transportation. The second category is school environment. School environment issues were focused on the institutional environment, resources, and relationships between students and teachers at school. These projects were about affecting school policies and practices. Examples of this are: lack of trust between students and teachers; lack of support for students; lack of college/career counselors or college prep courses; not enough educational programs; limited course selection; low quality substitute teachers; poor quality of facilities; lack of technology or basic supplies; poor school lunches; improper waste treatment at school; lack of after school options; school policies (timing of bathroom breaks; tardy-policy; reinstate study hall/recess). The third category, social issues, characterizes projects focused on issues in local community or society more broadly. Examples of this are: public transportation schedules; urban gardens; homelessness; criminal justice issues; animal rights; environmental issues; domestic violence; sexual violence.

Contact with decision makers

This code captured how easily the class members would be able to come in contact with the people most influential in making decisions about the community issues they identified. We coded two categories. The first category includes projects where students have direct, easy access to the decision makers they identified for their project goal: examples include teachers, other students, or parents. The second category includes projects where it is difficult for students to reach the decision makers they identified for their projects: examples include local government officials or corporate CEOs.

Quantitative Analyses

To determine the main effects of participating in GC on three civic outcomes, we ran hierarchical linear models with students nested within classrooms (Raudenbush & Bryk, 2002). While classrooms were also nested within schools, a three-level model was not feasible because of the small number of schools (n = 26) and average number of classrooms (M = 2.11) per school (Bell, Morgan, Kromrey, & Ferron, 2010; Clarke & Wheaton, 2007). Missing data were handled through five sets of multiple imputation analyses.

We estimated the moderating effects of program characteristics (context, content, and contact with decision makers) on the effectiveness of Generation Citizen by adding interactions terms between GC participation and the three project characteristics to each hierarchical regression main effect model (Baron & Kenny, 1986; Frazier, Tix, & Barron, 2004). This allowed us to discern whether the effect of participating in GC was different for students who participated in different types of projects. We conducted an additional set of analyses to further probe the magnitude of the difference in treatment effects across program characteristics. Using a simple slopes analysis, we estimated the expected marginal mean for each program characteristic for both the treatment and comparison groups and calculated the treatment effect for each program characteristic by subtracting the estimated outcome for the comparison from the treatment group. By examining the difference in treatment effects, we compared how the effect of GC differed across different types of action projects.

Results

To test the main effects of GC involvement on four civic outcomes (future civic commitment, civic self-efficacy, and two forms of civic knowledge), we analyzed the data using a two-level hierarchical linear models with students nested within classrooms. The analysis of the unconditional models indicated that there was significant variation between classrooms in student outcomes (p < .001). The intraclass correlations coefficients were 0.103 for future civic commitment, 0.079 for civic self-efficacy, 0.175 for action civic knowledge, and 0.323 for political knowledge. We then added student-level and classroom-level covariates to the model. Adding covariates improved the fit of the model, explaining 20% of the variance between classrooms in future civic commitment, 27% of the variance in civic self-efficacy, 52% of the variance in action civics knowledge, and 19% of the variance in political knowledge.

We ran a total of 12 separate main effect models (three classroom project characteristics on four civic outcomes). We then ran an additional 12 models (each interaction between GC participation and each of the three classroom characteristics for each of the four outcomes was run separately). Table 4 shows the results for all HLMs including effects of demographic control variables on outcomes, main effects of GC participation, and interactions between GC and project characteristics.

Table 4.

Hierarchical models predicting civic outcomes from participation in generation citizen and classroom project characteristics

	Civic outcome variables
	Future civic commitment	Civic self-efficacy	Action civics knowledge	Political knowledge
	β (SE) p	β (SE) p	β (SE) p	β (SE) p
Control variables
Model 1
Individual level
Female	−.005 (.086)	−.004 (.086)	.196 (.103)†	.086 (.075)
Asian	−.145 (.129)	.080 (.139)	−.120 (.120)	−.405 (.115)***
Black	−.144 (.166)	−.078 (.163)	−.292 (.155)†	−.207 (.150)
Latino	−.039 (.156)†	.095 (.177)	−.279 (.149)†	−.450 (.139)***
Other	.026 (.166)	.063 (.166)	−.287 (.156)†	−.190 (.145)
One to five absences	.010 (.107)	−.051 (.108)	−.152 (.102)	−.190 (.145)*
More than five absences	−.032 (.129)	−.072 (.129)	−.153 (.111)	−.140 (.107)
Classroom level
Elective	.131 (.251)	.040 (.230)	−.296 (.247)	−.210 (.093)
Civics	.282 (.201)	.185 (.195)	−.200 (.201)	−.140 (.107)
Number of DCs	−.050 (.111)	−.173 (.104)	.225 (.111)*	.002 (.341)
HS	−.146 (.140)	.040 (.132)	.455 (.139)***	.181 (.200)
FRPL	−.002 (.004)	−.179 (.262)	−.003 (.004)	−.005 (.006)
Main effects
Model 2
Generation citizen	.273 (.150)†	.318 (.138)*	.522 (.144)***	.360 (.212)†
Model 3
Context	.161 (.117)	−.009 (.113)	.022 (.116)	.274 (.210)
Model 4
Content (safety)	−1.09 (.158)	.063 (.144)	.091 (.158)	−.111 (.2334)
Content (school env.)	−.219 (.153)	−.090 (.142)	.020 (.152)	.022 (.224)
Model 5
Contact with decision makers	.066 (.144)	.059 (.119)	.100 (.123)	−.035 (.210)
Interactions
Model 7
Interaction GCXContext	.310 (.283)	.213 (.268)	−.360 (279)	−.762 (.396)
Model 8
Interaction GCXSafety	.997 (.333)**	.556 (.337)	.689 (.337)*	.550 (.531)
Interaction GCXSchEnv	.231 (.349)	.087 (.365)	.699 (.357)*	.677 (.532)
Model 9
Interaction GCXContact	−.200 (.463)	−.226 (.354)	−.749 (.299)**	−.400 (.539)

DCs, democracy coaches; HS, high school; FRPL, free and reduced price lunch.

^†p < .10,

^*p < .05,

^**p < .01,

^***p < .001

Future Civic Commitment

There was a marginally significant main effect of participating in Generation Citizen on future civic commitment (β = .273, SE = .150, p = .070). Participation in Generation Citizen explained an additional 20% of the variance in future civic commitment between classrooms compared to the model with covariates only. There were no main effects of project context, content, or contact with decision makers. This suggests that particular action civics projects did not differentiate levels of civic commitment.

One significant interaction emerged between participating in GC and projects with a safety theme (β = .997, SE = .333, p = .003). Including the interaction in the model explained an additional 6% of the variance in future civic commitment between classrooms. Probing the interaction further by examining the difference in expected marginal means revealed that students who chose a “safety” themed project were significantly higher on future civic commitments compared to those who did yet complete GC (Table 5). This suggests that action civics projects focused on safety issues had the most robust effect on civic commitment for students participating in GC compared to those projects focused on other issues.

Table 5.

Treatment effects of classroom project characteristics from simple slopes analysis

Outcome	Mean	SE	95% Confidence Interval		p-value
Future civic engagement
Generation citizen	0.273	.151	−0.022	0.569	.070
Context
School	0.030	.237	−0.434	0.494	.060
Outside of school	0.341	.181	−0.015	0.697	.898
Content
Safety	0.764	.195	0.381	1.147	<.001
School environment	−0.001	.229	−0.451	0.448	.995
Social issue	−0.233	.265	−0.752	0.286	.379
Contact with DM
Easy/Moderate	0.337	.248	−0.178	0.852	.174
Difficult	0.114	.302	−0.484	0.759	.114
Civic self-efficacy
Generation Citizen	0.318	.138	0.047	0.588	.021
Context
School	0.183	.225	−0.259	0.624	.019
Outside of School	0.395	.169	0.065	0.726	.418
Content
Safety	0.563	.192	0.186	0.940	.003
School environment	0.095	.230	−0.356	0.547	.680
Social issue	0.008	.277	−0.538	0.553	.978
Contact with DM
Easy/Moderate	0.394	.199	−0.002	0.790	.048
Difficult	0.168	.249	−0.327	0.663	.499
Action civics knowledge
Generation Citizen	0.522	.144	0.239	0.805	<.001
Context
School	0.746	.233	0.289	1.202	.001
Outside of School	0.385	.177	0.038	0.732	.029
Content
Safety	0.675	.204	0.275	1.075	<.001
School environment	0.685	.237	0.220	1.149	.004
Social issue	−0.015	.270	−0.545	0.516	.957
Contact with DM
Easy/Moderate	0.778	.177	0.431	1.125	<.001
Difficult	0.030	.234	−0.433	0.492	.899
Local political knowledge
Generation Citizen	0.356	.212	−0.059	0.771	.093
Context
School	0.751	.318	0.128	1.374	.018
Outside of School	−0.011	.256	−0.512	0.491	.967
Content
Safety	0.495	.324	−0.140	1.131	.127
School environment	0.624	.349	−0.060	1.309	.074
Social issue	−0.053	.408	−0.853	0.748	.898
Contact with DM
Easy/Moderate	0.512	.302	−0.086	1.111	.090
Difficult	0.112	.385	−0.659	0.883	.772

Civic Self-efficacy

There was a significant main effect of participating in GC on civic self-efficacy (β = .318, SE = .138, p = .021). Participation in Generation Citizen explained an additional 26% of the variance in civic self-efficacy between classrooms compared to the model with covariates only. There were no main effects of classroom project characteristics and the effect of program participation was not moderated by any of the classroom project characteristics. Upon examining the differences in expected marginal means, several differences in treatment effects emerged. Students who chose projects with easy access to decision makers, were focused on a “safety” theme, or were focused on issues outside of school reported significantly higher self-efficacy compared to those who had not yet completed GC (Table 5).

Action Civics Knowledge

There was a main effect of participating in GC on action civics knowledge (β = .522, SE = .144, p = .000). Participation in Generation Citizen explained an additional 14% of the variance in action civics knowledge between classrooms compared to the model with covariates only. There were no main effects of classroom project characteristics, but three significant interactions emerged between program participation and classroom characteristics. Interactions emerged between participating in GC and projects with a safety theme (β = .689, SE = .337, p = .041) and between participating in GC and projects with a “school environment” theme (β = .699, SE = .357, p = .050). This interaction term accounted for an additional 9% of the variance in action civics knowledge between classrooms. Probing the interactions further, students in classrooms that did projects focused on “safety” and “school environment” issues had higher action civics knowledge after participating in GC compared to students who had not yet participated in GC. However, those who did projects focused on social issues did not differ in action civics knowledge from those who had not yet participated in GC (Table 5).

A significant interaction also emerged between participating in GC and decision maker accessibility (β = −.749, SE = .299, p = .013). The interaction term explained an additional 15% of the variance in action civics knowledge between classrooms. Compared to students who had not yet participated in GC, students who chose projects coded as involving “easy to reach” decision makers had higher action civics knowledge whereas those who chose projects coded as involving difficult to reach decision makers did not significantly differ on action civics knowledge from those who had not yet participated in GC (Table 5).

Although the overall interaction was non-significant, different treatment effects emerged indicating that students who did projects focused on issues in school and projects focused on issues outside of school reported were higher on action civics knowledge compared to those who had not yet done GC (Table 5).

Political Knowledge

There was a marginal main effect of participating in Generation Citizen on general political knowledge (β = .360, SE = .212, p = .093). Participation in Generation Citizen explained an additional 4% of the variance in political knowledge between classrooms compared to the model with covariates only. There were no main effects of classroom project level, project themes, or accessibility of decision makers and the marginal main effect of program participation was not moderated by any of the classroom project characteristics. When examining the differences in expected marginal means, a difference in treatment effects emerged. Students in classrooms that did projects focused on issues in school had higher local political knowledge after participating in GC compared to students who had not yet participated in GC. However, those who did projects focused on issues outside of school did not differ in local political knowledge from those who had not yet participated in GC (Table 5).

Discussion

Generation Citizen appears to have had an overall positive impact on civic outcomes. Action civics projects varied on dimensions of context, content, and level of decision-maker contact. Certain types of projects appeared to affect civic outcomes more than others.

Effect of GC Participation on Civic Outcomes

Participating in GC had an overall positive effect on civic outcomes. Participation in GC has the strongest effect on action civics knowledge and civic self-efficacy and marginal effects on future civic commitment and general political knowledge. We interpret this in light of the fact that civic self-efficacy and action civics knowledge seem to be the outcomes most closely tied to program goals and content. GC seems to empower young people to see themselves as effective civic actors; such opportunities for civic identity development at the impressionable time of middle and high school can be critical for setting young people on positive civic trajectories (Hart et al., 2007; Watts & Flanagan, 2007). The effect of GC on feeling committed to future civic work was positive but weaker; it may be that the emphasis on local issues and on completing a project led students to feel satisfied with their participation and did not necessarily translate into a broader commitment to civic work. This may be an area that GC can strengthen if democracy coaches and teachers explicitly connect the classroom projects to ongoing social issues and encourage sustained civic participation.

In terms of knowledge, GC had a stronger effect on action civics knowledge than general political knowledge. Action civics knowledge is a specific subset of civic information that draws on GC vocabulary (such as “root cause” and “decision maker”) that is emphasized throughout the program. Therefore, it makes sense that program participation strongly increases this type of knowledge. The gain in action civics knowledge indicates that students are paying attention to GC’s framework for solving social problems. Students made positive gains in general political knowledge, but this finding was weaker, suggesting that exposure to GC may, but does not necessarily, encourage student to seek or retain information about politics more broadly.

It must be noted that in another study of the impact of GC, there was a significant effect of the program on action civics knowledge but not other outcomes measured. The effect varied by school (middle vs. high school) and class (subject that GC was embedded in) characteristics (Cohen, Littenberg-Tobias & Ridley-Kerr, 2014)). The authors conclude that GC is having a positive impact but also note “substantial room for improvement as Generation Citizen attempts to move the needle to positively impact more general measures of civic knowledge, skills, attitudes, and behaviors.” Our findings concur but suggest a slightly more positive overall picture of effects of GC. Differences in data sources (our data are limited to one school year and to students in classrooms with the most highly engaged democracy coaches who completed questionnaires about class projects) and modeling (ours were two-level models whereas previously published findings use three-level models) might explain slightly different findings and underscore a need for further study of GC in particular and action civics program in general.

Program Characteristics: Context, Content, and Contact with Decision Makers

Given that GC is not a homogeneous treatment, a major contribution of this study was to understand what types of projects students chose to address and to identify characteristics that captured meaningful variability across projects. As expected from research on youth community engagement (Ballard, 2014; Rubin, 2007) students in the present study chose action civics project that addressed a wide variety of issues ranging from gang violence to cleaning community parks to improving school lunches. There are many ways action projects could be coded in order to examine whether program effects vary across differences in projects. In the present study, we identified three characteristics that differentiated projects: context, content, and contact with decision makers.

One characteristic that emerged was the location of the social issue that the projects sought to address, which we conceptualized as the context of the issue. The two common contexts for action civics projects were focused in school and out of school. Conceptually, in-school issues are potentially very powerful because they take place in the context most proximal to students and students are direct stakeholders in school issues. On the other hand, issues taking place outside of school might provide students with powerful opportunities to connect with a wide set of community stakeholders and have experiences with local politics (Kirshner, 2007; Ozer & Wright, 2012). We found that doing projects in both contexts increased action civics knowledge compared to not having participated in GC. This suggests that the context of action civics projects does not affect the opportunity to gain the specific process-related knowledge about action civics. Counterintuitively, we found that projects focused on issues in school were associated with higher general political knowledge through participation in GC. It’s possible that teachers and democracy coaches working on in-school issues were cognizant to include broader political lessons for students; further research about how civic teaching differs depending on projects would help make sense of this finding. Doing projects focused on issues outside of school was associated with more gain in civic self-efficacy; perhaps exposure to community political processes provides unique opportunity to feel efficacious in civic life.

A second characteristic across action civics projects was the focus of the issue that the action civics project sought to address, which we conceptualized as content. The three most common types of projects in our study focused on safety, the school environment, and societal social issues. Doing projects focused on safety issues resulted in more gains in future civic commitments, civic self-efficacy, and action civics knowledge. This was the most robust finding regarding project characteristics. What might be special about doing projects that address safety concerns? Two prominent safety concerns across action civics projects were bullying in school and community violence. These strike us as issues that might feel both important and also addressable for students in our study. Feeling safe is a fundamental need for adolescents (Maslow, 1943; National Middle School Association, 2003). Perhaps being given the space to identify things that make them feel safe in their schools and communities is an empowering experience. At the same time, safety issues take place in the peer context more than school environment or broader social issues; it might be that students perceive that they are working on issues that primarily involve peers rather than primarily involving adults.

Through GC, students identified the decision makers that were powerful in relation to their chosen action civics projects. A third characteristic that emerged across action civics projects had to do with how accessible these decision makers were. We conceptualized this as ease of contact with decision makers (i.e., powerful people in the specific context of the civic project). The present study shows that classroom action civics projects that can be addressed with easily accessible decision makers led to gains in action civics knowledge and civic self-efficacy. Students who chose projects that required contact with hard to reach people, for example, government officials or local business owners, likely became frustrated when they reached barriers in access. This is a common barrier in youth community work; social change is difficult for youth who are embedded in systems where they lack power and access to resources for social change (Ozer & Wright, 2012; Sarason, 1996). This is an important finding for civic intervention programs; projects that give students access to people in power might be especially effective in teaching knowledge about action civics and fostering a sense of efficacy. This aligns with theoretical frameworks that explicitly emphasize understanding, and accessing, power through civic work (Christens, Winn, & Duke, 2016). We believe that part of effective civic work is knowing how to address local issues and who to go to for support; GC formalizes this in their model and our findings reiterate that projects that put students into contact with decision makers are especially tied to gains in civic outcomes.

Considerations for Practice and Policy

Beyond providing evidence that action civics interventions can positively affect civic outcomes, this study examines an important practical question for civic programs: When civic programs afford youth the opportunity for self-driven civic projects, how does the choice of project influence civic outcomes? We conclude that project characteristics differentially affect civic outcomes. Given that there are many worthwhile civic outcomes that programs aim to influence, we suggest that civic programs should carefully prioritize the outcomes of interest (e.g., civic efficacy or general political knowledge) and scaffold the intervention to support those outcomes. It is important to let youth chose personally meaningful projects (Ballard, 2014; Morgan & Streb, 2001; Ozer et al., 2013) so civic programs face the tricky task of balancing youth choice while maximizing civic growth. One suggestion is for program leaders to put parameters around project choices. Programs prioritizing civic self-efficacy might encourage students to choose projects that afford easy access to decision makers. Alternatively, program leaders can forge relationships with community partners to help gain access to decision makers that are harder to reach. Another suggestion is for civic programs to tailor program delivery in specific ways. For example, students may be allowed to conduct projects on issues either in- or outside-of-school, but can be encouraged to reflect on the connections across contexts. Recent evidence points out that adults may facilitate youth-led efforts by providing structural support to youth in forming networks to share their efforts (Kornbluh et al., 2016; Lenzi et al., 2015). Although we specifically examine questions of youth choice and developmental outcomes in the context of civic development, our findings contribute more generally to community psychology literature examining these issues in youth-led work more broadly (e.g. YPAR). Many others have written about tensions between youth choice on the one hand and maximizing youth development and social change on the other in civic programs (Larson & Walker, 2010; Larson, Walker, & Pearce, 2005; Ozer et al., 2013).

In terms of policy, the present findings suggest that action civics interventions are a promising approach to promote youth civic development. In the current US educational climate, where schools are over-burdened, the GC model is appealing. By partnering teachers and democracy coaches, GC can reduce teacher burden without displacing teachers. GC leverages the expertise of both rather than outsourcing civic education. By keeping civic education in the classroom, GC sends the message that this content is as important as other subject areas and by integrating an action component, GC capitalizes on best practices for learning while drawing on students’ expertise about their own lives and communities.

Limitations and Future Work

Several limitations of this study must be noted. One limitation is the cross-sectional and non-randomized nature of these data. Longitudinal data would allow us to measure the lasting impact of GC across development. Pre and post-GC groups also differed on demographic characteristics in the present study; the pre-GC group appeared to be more advantaged than the post-GC group suggesting that our estimate of the impact of Generation Citizen might be an underestimate. Although we controlled for demographic characteristics in our analyses, randomization would allow us to better account for demographic differences between treatment and control classrooms. The data in the present study are nested by classrooms, schools, and regions. Although our analyses allowed us to account for nesting by classrooms, we were not able to account for nesting by schools or region. It is possible that the implementation of GC across schools or regions varies in systematic ways that affect program impacts. This is an important issue when considering scaling school-based programs.

In coding the characteristics of GC action civics projects, we were limited by the data GC collected. There are many aspects of projects for which we did not have information. In particular, the process of action civics programs may be just as, or more, important than content. What might matter most is how skilled the teachers and democracy coaches are at guiding the civic learning process rather than the focus of the projects themselves. Beyond the limits of the data, project characteristics can be coded many ways. Our choices of coding reflect what emerged as prominent themes in our analysis but there are likely other important themes differentiating action civics projects.

Another limitation is that codes are provided at the classroom level, given that classes perform action civics projects together. This does not allow us to account for the fact that there are especially influential individuals in each classroom as well as students who are not as invested in civic project decisions. This approach masks variability in the effectiveness of GC across individuals, who likely differ on many dimensions, such as level of engagement, that have been shown to moderate the affect of youth organizational participation on outcomes (e.g., Lynch et al., 2016). Future work examining how classroom projects get decided on and how individuals invest in project participation would further clarify how the fit between individual interests, classroom project selection, and project characteristics affect civic gains across GC program participation.

Conclusion

This study contributes two major findings to literature in community psychology examining how best to support youth empowerment in civic life. First, participating in an action civics program, Generation Citizen, was associated with positive gains in action civics knowledge and civic self-efficacy. This supports and extends initial research on Generation Citizen finding it to be one effective way to promote youth civic development. This is especially important given the target of Generation Citizen is schools serving youth from low- and middle-income backgrounds, populations exposed to few high quality civic opportunities. A second important insight from this study is that the characteristics of the project that students choose to complete also affect their civic development. The three characteristics that emerged as important in the present study were the project context and content as well as the contact with decision makers that students have in their action civics projects. The empowering practice of student choice (having classrooms collectively research and choose a civic issue important to them) must be balanced against the reality that students will choose very different sorts of projects that vary in ways that can affect their civic growth. This adds to community psychology literature on the tensions that arise in implementing high quality youth programs.