Testing Opt-Out Movements: Resistance, Disputes and Transformation education policy analysis archives Equality, Inclusion, and the Opt-Out Movement: Who Chooses to Opt Out?

: There is ongoing debate in the United States about just how diverse, inclusive, and equitable the opt-out movement — and grassroots education activism more broadly — has been over the past decade. Today, a prevailing stereotype holds that the opt-out movement predominantly mobilizes white, middle-class, and suburban parents and is therefore a force for educational privilege, not equity. However, this stereotype has never been rigorously tested and it is still an open question as to how opt-out movement participants differ from non-participants, especially along the lines of race and class. Leveraging a comprehensive quantitative dataset of New York school district opt-out rates and community characteristics as well as original case study data collected from four purposefully sampled New York school districts, this study uncovers evidence challenging the dominant narrative of opt-out participation. It finds that the opt-out movement has been active across a diversity of district contexts and, with only small qualifications, has simultaneously mobilized a diverse cross section of parents within districts. However, the most active and longest-tenured opt-out parents do appear to be whiter and wealthier than the movement as a whole. This paper concludes by discussing the implications of these findings for our understanding of the U.S. opt-out movement as well as the potential of grassroots educational activism to be a force for equity and inclusion in U.S. education politics more generally. Javier Campos Martínez is assistant professor at the Institute of Education Sciences of the Universidad Austral de Chile, researching educational inclusion in teacher training and higher education. He participates in the Working Group on Educational Policies and the Right to Education of the Latin American Council of Social Sciences (CLACSO), and is one of the national coordinators of the Teacher Work Studies Network (ESTRADO) in Chile. Co-founder of the campaign against Chilean standardized tests “Alto al SIMCE,” he has worked on designing and implementing experimental programs that foster inclusion and school success through assessment for learning at the local level.


Equality, Inclusion, and the Opt-Out Movement: Who Chooses to Opt Out?
While neoliberal reform ideas like standardized testing and accountability remain some of the most salient and entrenched features of today's education policy environment in the United States, this agenda has increasingly become the target of grassroots education protest (Ferman, 2017). Between 2012 and 2020, no grassroots education movement in the United States achieved greater salience or mobilized more people than the opt-out movement, in which millions of parents across the country kept their children at home on test day and engaged in various other protest activities like rallies, marches, canvassing drives, letter-writing operations, and social media campaigns. Originating in 2011 as a small, scattered, and mostly internet-driven protest in New York State, the opt-out movement exploded over the next several years, eventually engaging millions of parents, particularly in the American Northeast and West. The epicenter of the opt-out movement always remained New York, though, where a whopping 22% of students did not take their standardized tests during the movement's peak in 2016 and some districts experienced opt-out rates as high as 89% (NYSED, 2016).
Despite the explosive growth and widespread media coverage of the opt-out movement, however, very little systematic and rigorous research has been conducted into key questions about the opt-out movement's participants and actors on the ground. Who are the parents opting their children out of testing and how broad, diverse, and inclusive is this coalition of resistance? Such an examination of the opt-out movement is important not only for its own sake, but also for what it can reveal about the opportunities and challenges of grassroots education activism as a tool for rectifying chronic inequalities in American education politics in general. Additionally, public perceptions of the opt-out movement have been highly variable, with supporters often lauding it as an emblem of democracy in action and opponents often condemning it as a privilege-driven phenomenon undermining reforms needed to help marginalized students-leaving it an open question of just how diverse and inclusive the movement actually is.
This paper helps to address this gap in our understanding by investigating the following research questions about opt-out movement participation: Who participates in the opt-out movement and who does not? What demographic characteristics do opt-out participants share and how are opt-out participants different from non-participants? How equal and inclusive is this form of educational political participation in terms of race and class?

Guiding Theoretical and Empirical Literature
In answering these questions, this research is informed by a large body of theoretical and empirical literature from the fields of political participation (e.g., Verba et al., 1995), the sociology of social movements (e.g., della Porta & Diani, 2007), and public engagement with public education (e.g., Orr & Rogers, 2011), which together highlight and attempt to explain the chronic inequalities that exist today in U.S. political participation and educational engagement. By incorporating the central insights of this interdisciplinary set of literature in the design of this study, this analysis is able to not only unearth important insight about the nature of the U.S. opt-out movement, but also contribute to broader scholarly dialogues about participation in education politics and the potential of grassroots educational activism to equalize political and educational voice in ways that might promote more equitable outcomes for students and communities.
First, this study is informed by existing political science scholarship on the topic of political participation. Today, the most widely accepted model of political participation among political scientists is Verba et al.'s (1995) civic voluntarism model. The civic voluntarism model contends that participation in political activities is driven primarily by three factors: resources (i.e., money, time, education), engagement (i.e., interest in politics), and recruitment (i.e., being asked to participate). Because these factors are highly correlated with socioeconomic status, unequal patterns of participation are evident in society in ways that ultimately reinforce the privilege and political influence of high-social economic status (SES) individuals. Indeed, empirical evidence suggests that high-SES individuals are more likely to have the time and money needed to participate in politics, and they also are more likely to inhabit social environments that encourage political engagement, the acquisition of political knowledge, and the cultivation of civic skills (e.g., Putnam, 2000;Rosenstone & Hansen 1993;Verba et al., 1995). In addition to having more resources for participation in politics, high-SES individuals tend to have opportunities to participate in deeper and more substantive ways than low-SES individuals. They are able to (and often do) contribute more hours, donate more money, and organize more thoroughly in ways that further amplify their voice . Wealthy individuals also tend to care about different issues than lowincome individuals, and governmental agendas consequently tend to skew toward their concerns (Bartels, 2016;Gilens, 2012).
Beyond issues of SES, political scientists have also examined the participatory inequalities that exist across different racial and ethnic groups, with a large body of literature indicating lower rates of participation among people of color due to a variety of resource, institutional, and cultural barriers (e.g., Guterbock & London, 1983;Leighley & Vedlitz, 1999;Miller et al., 1981;Stokes, 2003;Verba et al., 1993;Verba & Nie, 1972;Welch et al., 1975). While some research has found that racial differences in participation disappear when controlling for SES (e.g., Leighley & Vedlitz, 1999;Verba et al., 1993), other research has found that racial differences exist independently of SES (e.g., Rosenstone & Hansen, 1993;Stokes, 2003). Principal among the non-SES factors that explain these differences are uneven feelings of group consciousness and social connectedness, which can promote participation by sparking political interest and establishing collective identities that encourage participation among people of color McClurg, 2003;Miller et al., 1981;Stokes, 2003).
These participation inequalities, present in American democracy writ large, are equally manifest in the field of education, where, due to unequal resources and opportunities, white, high-SES parents generally have a louder voice in the governance of their schools. As evidenced in literature about parent engagement in education-the second body of scholarship informing this study-high-SES parents experience greater opportunities to volunteer on local PTAs, attend school board meetings, vote in local elections, and talk about education issues with their peers and local educators (Orr & Rogers, 2011;Smrekar & Cohen-Vogel, 2001). These parents are, to borrow a word from McAdams (2000), "persistent" in advancing their preferences, and almost always these parents are treated with deference and respect by educational officials who embrace cultural frames that value their participation. In contrast, low-SES parents and parents of color often face cultural barriers to education participation beyond simple resource barriers, finding that their interactions with school officials are structured by race, class, culture, and language (Lareau, 2003;Smrekar & Cohen-Vogel, 2001). Indeed, when low-income parents or parents of color attempt to redress a problem pertaining to their child's educational experience, they are frequently ignored or treated with hostility by education officials who do not appreciate the cultural capital they possess. In her book Unequal Childhoods, Lareau (2003) writes that "when working-class and poor parents try to intervene in their children's educational experiences, they often feel ineffectual…bullied and powerless" (p. 7). de Carvalho (2000) uses the phrase "symbolic violence" to describe the alienation and discomfort historically marginalized parents feel when learning that their school community does not value their cultural background, and she notes that as parents collect negative experiences trying to participate, they become increasingly estranged and cynical about the power of their own voice (p. 12). These challenges are even thornier if the parents do not speak English, and it is a fact of practice that many school systems do not provide adequate language support services for Englishlanguage learner (ELL) parents.
In the face of these disparities in participation and opportunities to participate, some scholars have advocated for new models of participation with education, including grassroots social movement tactics like those seen in the opt-out movement. Advocates for education social movements argue that this mode of participation is necessary to equalize political voice because conventional participation fails to challenge the institutions and cultural logics that inherently entrench inequality (Anyon, 2005;Ferman, 2017;Orr & Rogers, 2011;Rogers, 2006). Additionally, social movements can offer an opportunity for people of different backgrounds to come together in common cause, share valuable political resources, create new networks, build collective identities, and deploy organizing strategies that maximize political voice among less privileged parents.
According to sociological research on social movements-the third body of research that informs this study-this egalitarian faith in grassroots social movement activism is not on its face misplaced, and it remains an important question if and how grassroots social movements might promote greater equality in education participation (della Porta & Diani, 2007). Indeed, grassroots movements usually mobilize diverse and subdominant memberships, and they tend to exhibit egalitarian, decentralized control structures that allow for intra-movement equality of voice (della Porta & Diani, 2007;Horton, 2013). They also tend to be geographically bounded in localities, which facilitates dense social networks that promote group consciousness and identity building.
On the other hand, political resources (i.e., time, money, and education) remain indispensable to social movement mobilization, and as long as resources remain unequally distributed within and across communities, grassroots social movement mobilizations may not necessarily be a force for political equality. As Edwards and McCarthy (2004) write: Middle-class groups remain privileged in their access to many kinds of resources, and, therefore, not surprisingly social movements that resonate with the concerns of relatively privileged social groups predominate and the mobilizations of the poor groups are quite rare in advanced industrial democracies. (p. 117) Illustrating the upper-class bent of social movements, cross-national research has found that social movements tend to be founded earlier and at faster rates in wealthy democracies (Smith & Wiest, 2005), and within those countries, movements are more likely to emerge in metropolitan and suburban areas with more privileged populations (McCarthy et al., 1988). In the United States, many of the most salient social movements in history-such as the abolitionist, temperance, feminist, conservation, and Tea Party movements-have tended to present the voice of white, middle-class people. And even the movements that have mobilized underprivileged populations have tended to have as their leaders people who come from middle-and upper-class backgrounds (Morris & Staggenborg, 2004). Thus, a gap remains in our understanding of just how inclusive and equitable grassroots social movements like the opt-out movement may be in education and whether these types of movements might actually help rectify longstanding inequalities in political and educational voice for parents.

The Opt-Out Stereotype: White, Wealthy Suburbanites
The question of who participates in the opt-out movement takes on special urgency when we consider the very limited state of existing knowledge on this question. Today, the prevailing stereotype of the opt-out movement is that it is a movement driven primarily by white, wealthy, suburban parents, and as a result, it is not a force for political or educational equity (e.g., Hairston, 2017). The most famous espousal of this stereotype came in 2013, when U.S. Secretary of Education Arne Duncan dismissed opt-out parents as "a bunch of white suburban moms who-all of the sudden [on the test]-[see] their child isn't as bright as they thought they were" (Strauss, 2013). Duncan, however, was not the only person to spread this belief, and this caricature of the opt-out movement has been widely promulgated throughout the news media and on social media sites like Twitter, where the hashtag #OptOutSoWhite began trending in 2015. A 2016 editorial by the Washington Post lamented that "white suburban parents, the driving force of the opt out movement" were hurting low-income children by undermining important test-based accountability reforms-a sentiment that was echoed repeatedly in other outlets like the New York Times and Education Week (Editorial Board, 2015. Robert Pondiscio (2015) of the Fordham Institute likewise predicted that opt-out parents were on a "collision course" with "low-income families of color who have been the primary beneficiaries of testing and accountability."

Figure 1
The Opt-out Movement Stereotype on Twitter The claim that the opt-out movement is predominantly a white, affluent phenomenon has obtained greater resonance by public disavowals of the movement from civil rights groups as well as national polling data that reveal racial cleavages in testing views. In the midst of the 2015 opt-out explosion, The Leadership Conference on Civil and Human Rights (LCCHR, 2015) released a statement signed by 12 national civil and human rights groups condemning the opt-out movement. Furthermore, a 2015 Phi Delta Kappa/Gallup poll found that 72% of Black parents and 61% of Hispanic parents considered test scores either "very" or "somewhat important" for measuring the effectiveness of schools, compared to 55% of white parents (Phi Delta Kappa, 2015). When asked about the opt-out movement specifically, the poll also found that 44% of white respondents supported the right of parents to opt out and 41% opposed it. In contrast, only 28% of Black parents and 35% of Hispanic parents supported the right of parents to opt out. Furthermore, when asked if they would opt out their own child from testing, 75% of Black parents, 65% of Hispanic parents, and 54% of white respondents said they would not.
While this narrative of the opt-out movement continues to hold sway, it has not gone uncontested. In 2015, the Seattle/King County chapter of the National Association for the Advancement of Colored People (NAACP) broke with national leadership and encouraged parents to opt out from the new Common Core tests, and in 2016, the New York Times remarked that the opt-out movement appeared to be diversifying. Ceresta Smith, an African American leader of the national opt-out organization United Opt Out, expressed frustration that testing supporters remained fixated on the white and suburban elements of the movement, even as the movement mobilized diverse parents: I think the black and brown voices have been silenced [by testing supporters]. When it comes to their participation in this movement you see just a handful of the same faces, and they're predominantly white. When you look at the face of the movement, it's the Long Island moms -and unfortunately the Philly moms, the Jersey moms, the black women in Miami, and Fort Lauderdale, and in Seattle, you had a lot of African Americans involved but you don't see it. It's not visible. It's not publicized (Quinlan, 2016).
According to Smith, the effort to "whitewash" the movement was an intentional effort to marginalize the movement from education reform discussions (Quinlan, 2016; see also Teague, 2016). Amid all of this media speculation about opt-out participants, scholarly research on the topic of opt-out participation has been quite limited and of variable rigor, but where it does exist, it has generally upheld this stereotype. Chingos (2015), for example, examined unofficial district-level optout data in New York gathered by the opt-out organization United to Counter the Core and found that districts with higher opt-out rates tended to be more affluent, serve fewer disadvantaged students, and have higher test scores. Similar district-level analyses conducted by Bennett (2016) and Supovitz et al. (2016) also concluded that the movement has been populated primarily by privileged parents. More recently, studies by Mitra et al. (2016) and Rivera-McCutchen (2021) reviewed documents and multimedia sources from high-profile opt-out websites and organizations and determined that the movement's policy views tended to reinforce the power and privilege of white, wealthy parents and were thus exclusionary from an equity perspective. The one study to date that has examined the individuals involved in the opt-out movement, Pizmony-Levy and Green Saraisky's (2016) unpublished national survey of 1,641 opt-out activists, reported that "the typical opt out activist is a highly educated, white, married, politically liberal parent whose children attend public school and whose household median income is well above the national average" (p. 6). However, the design of this study, in which activists were surveyed on nationally oriented opt-out social media pages, left it highly vulnerable to bias issues. (Namely, it privileges movement elites at the expense of rank-and-file members and only examines those parents who are on social media, possess the leisure time to actively engage on opt-out social media pages, have the social capital to be connected to national networks, and take on global and often extreme orientations to their activism). It also does not capture anything about parents who choose not to participate in the opt-out movement.
Altogether, while this work has provided a valuable starting point in understanding the optout movement, it also leaves many critical questions unanswered and many valuable data sources untapped. Currently, we know a lot about how the opt-out movement has been portrayed in the media and online, and we understand a good deal about the most visible and active parents found on opt-out social media pages and in high-profile opt-out organizations (i.e., movement elites). However, we do not have a clear understanding of who has participated in the movement on the ground and, just as importantly, who has not been participating. Who are the ground-level, rank-andfile members of the opt-out movement? Are these individuals different from the national activists frequently studied? And how do these parents differ from the parents around them who are not participating in the opt-out movement?

Method
To address the research questions above, this paper draws on a statewide quantitative dataset of New York school district opt out and community characteristics during the 2015-2016 school year as well as original case study data collected from four school districts in New York during summer 2017. The statewide dataset allowed for a rigorous examination of variation in opt-out movement strength across districts with different demographic contexts (i.e., race, income, education, urbanicity) while the case study data allowed for an examination of patterns of opt-out participation within districts, assessing if movement participants on the ground reflected the demographics of the community or were instead compositionally biased in favor of one group of parents over another. By triangulating these diverse data sources, I was able to achieve a view of optout participation patterns that is much more robust and multidimensional than other analyses to date.

Statewide Quantitative Dataset and Analysis
The first source of data used in this study is an original dataset of district opt-out rates in New York State during the opt-out movement's peak in the 2015-2016 school year. The dataset lists every New York school district except New York City (n=685) alongside its grade 3-8 opt-out rate and various demographic and political variables obtained from the New York State Education Department (NYSED), the American Community Survey (ACS), and the New York State Board of Elections. These demographic and political variables were derived largely from the theoretical and empirical literature described above pertaining to political participation, parent engagement with education, and social movement activism, and it was hypothesized that all of these variables would be associated (positively or negatively) with participation in the opt-out movement. A complete list of these variables and descriptive statistics can be found in Table 1 below. The statewide dataset was analyzed using descriptive statistics (cross-quartile comparisons) and Ordinary Least Squares (OLS) regression to assess differences in opt-out rates across districts and identify community factors most predictive of a district's opt-out rate. Additionally, I used GIS mapping software to produce a visualization of opt-out movement strength across the state. By employing these analysis tools, I was able to obtain a statewide perspective on the reach of New York's opt-out movement.

Case Study Data
In addition to the statewide dataset, this study also draws on original case study data collected from four New York school districts in summer 2017 to get a sense of the demographic trends in opt-out participation within districts (Yin, 2014). The four districts included in this case study were purposefully sampled using the matrix below to exploit variation in opt-out participation rates (high opt out or low opt out) and district demographic contexts (high racial diversity student populations or low racial diversity student populations) as reported by NYSED. 1 In the final sampling, districts were selected based on their desirability as a study site, attending to such concerns as district size, urbanicity, opt-out rates, and demographics (i.e., race, poverty, and ELL) so that my final four districts would be large enough to yield an adequate pool of potential parent participants and simultaneously control on characteristics other than the two sampling variables. Figure 1 reports the characteristics of the final four districts ultimately included in this study.
Within each of these districts, I collected five sources of original data during summer 2017 to obtain insights into the opt-out movement's composition, motivations, engagement with local education systems, and impact on local communities. These sources included: 1) an online survey of all grade 3-8 parents; 2) parent focus groups; 3) semi-structured interviews with district superintendents and school board members; 4) in-depth interviews with local opt-out activists; and 5) documentary artifacts (e.g., news reports, school board minutes, social media posts, etc.). While all of these sources of data were used in this present research, much of the discussion below draws on the parent survey and district elite interviews.

Figure 2
Sampling Matrix and Four Selected Districts (Pseudonyms)

Parent Survey
In summer 2017, I conducted an online (Qualtrics) survey of all grade 3-8 parents in each district which asked parents about a variety of opt-out topics, including their opt-out decisions, motivations, and protest activities; perceptions of the opt-out movement and its impact on their community; views on various education issues; political attitudes and participation habits; perceptions of district context; and demographics. The survey was developed and refined through an iterative process that included examination of relevant literature, examination of opt-out artifacts, consultation with survey development experts, and extensive piloting. The final survey that was administered to parents was 61 items long and required an average of 18 minutes to complete. All respondents were entered into districtwide raffles to win Amazon giftcards.
Within each district, parents were recruited to complete the survey through extensive district-wide communications campaigns. Parents were contacted using district-wide email and text message list-servs, robocalls, postings on the district homepages and social media outlets, and letters home with students. In total, I received a total of 570 complete survey responses. Of these, 271 were from opt-out parents (hereafter, OOPs) and 299 were from non-opt-out parents (hereafter, NOOPs), and district completion rates for the four districts ranged from 10% (Commonwealth) to 28% (Danville). N sizes ranged from 83 (Danville) to 190 (Easton). For the most part, the pool of survey respondents aligned with the opt-out and demographic composition of the districts, although the samples were heavily skewed toward females (mothers), and they tended to overrepresent wealthier and more highly educated parents across all four districts. Therefore, in any effort to generalize from a particular district sample to its district population, I weighted the sample using iterative proportional survey weights. After being collected, the survey data was analyzed quantitatively, including the use of descriptive statistics, data visualizations, parametric tests (i.e., ttests), and nonparametric tests (i.e., chi-squared tests of independence) examining differences in response patterns across districts and respondent groups to understand what types of parents were most likely to opt out of testing.

District Elite Interviews
In addition to obtaining the perspective of parents, I also sought to learn about the experiences of district leaders who have dealt with the opt-out movement in their professional practice and could potentially provide a countervailing perspective due to their position of leadership in the community, access to internal district opt-out data, and proximity to opt-out activities on the ground in their communities. On this front, I conducted in each district semistructured phone interviews with the superintendent and four school board members (for a total of 20 interviews) which were audio-recorded and later transcribed (Creswell, 2013). As part of these interviews, officials were asked about their perceptions of opt-out activism in their district and the types of parents who most frequently participated (or did not participate) in the opt-out movement. The superintendent interviews lasted an average of 45 minutes and the school board interviews lasted an average of 35 minutes, with officials in high opt-out districts generally having longer interviews. The protocols used in these interviews were standardized so that I could identify the degree of consensus that emerged among officials within and across districts.

Qualitative Data Analysis
The qualitative data from the interviews were uploaded to Dedoose qualitative analysis software and analyzed using simultaneous pattern coding to identify key linkages and themes within and across sources as well as within and across respondent groups. Coding initially followed a "ground up" analytic strategy (Creswell, 2013;Saldaña, 2009) and codes were developed through a multistage content exploration of the qualitative data that involved the assistance of an external partner to ensure reliability. A ground up approach was preferrable for initial coding given the dearth of existing scholarship on the opt-out movement as well as a commitment to prioritize the voices of individuals on the ground (i.e., parents, activists, and district leaders) who have historically been neglected from scholarship on the specific research question of opt-out participation patterns. The inductive codes derived from this initial process were then supplemented where appropriate with deductive codes derived from the theoretical and empirical literature described above regarding political participation, parent engagement in education, and social movement activism. In Dedoose, the data were examined visually and numerically in the form of code clouds, crosstabulations, and charts showing the frequency with which codes occurred as well as the presence or absence of codes in and across the sources and participant voices.
Throughout the data collection and analysis process, I adopted Miles and Huberman's (1994) approach to realism as my epistemological stance, which maintains that social phenomena exist both "in the mind" as well as "in the objective world" and that my goal as a researcher is to identify stable patterns of social phenomena in the real world while also attending to the "local and historical contingencies under which they occur" (p. 19). From this perspective, it is critical to assess any research question from many different vantage points (e.g., state-level, district-level) as well as from many different perspectives (e.g., state data, parent voices, district leader voices) to develop a sense of both general trends and local contingencies.

The Opt-Out Movement across New York School Districts
The results of the data analysis directly challenge the prevailing stereotype of the opt-out movement as being a predominantly white, wealthy, and suburban phenomenon. In fact, whether one looks at patterns in opt-out participation at the level of the district or the level of the individual, the opt-out movement appeared to mobilize (with only slight exceptions) a diverse coalition of parents across a diverse array of New York school districts. Figure 3 below illustrates the strength of the opt-out movement in school districts across New York State during the 2015-2016 school year. As the figure illustrates, the opt-out movement, while varying in strength across localities, impacted the majority of school districts in New York and did not seem confined to any particular geographic region (e.g., Long Island) or district type (e.g., suburban districts) as suggested by other reports. In fact, 94.5% of districts included in my dataset (648) had opt-out rates above the 5% "legal limit" for opting out.

Figure 3
District ELA Opt-Out Rates, by Quartile (2016) Note: Colors show the percentage of grade 3-8 students who opted out of the 2016 NYS ELA exam as reported by NYSED. The colors represent percentage quartiles of the total opt-out rate in the state with the addition of a separate color for 0-5% (the "legal limit" of opting out). For instance, the darkest blue represents districts that were in the 4th quartile of opting out and had opt-out rates over 41%. Dataset excludes New York City.
Cross-quartile comparisons of district opt-out rates likewise confirmed that the opt-out movement has not been confined to white, wealthy, or suburban districts (Table 2). While suburban districts did tend to have higher average opt-out rates (36.1%) than their urban (15.6%), rural (23.8%), and town (27.3%) counterparts, all four types of districts experienced high average opt-out rates. Moreover, the opt-out movement was a significant presence in racially diverse districts, and districts in the top half of nonwhite student enrollment actually had higher average opt-out rates than districts in the bottom half of nonwhite student enrollment. Mirroring the findings of previous literature (Bennett, 2016;Chingos, 2015;Supovitz et al., 2016), opt-out rates did seem positively correlated with district wealth, but it is important to highlight that even the least wealthy districts exhibited average opt-out rates over 20%. Note: ELA opt-out rates reported. Similar findings were produced using Math opt-out rates.
These descriptive statistics complicate existing narratives about the demographics of the districts impacted by the opt-out movement, but they are not able to predict which district characteristics are most powerfully associated with opt-out activism when controlling for other confounding variables. To address this limitation, I turned to Ordinary Least Squares (OLS) regression analysis of the statewide dataset. Table 3 reports the results of the OLS regression analysis in which the dependent variable is district opt-out rates on the 2016 New York State ELA and Math exams and the independent variables include various community-level demographic and political characteristics. First and foremost, the regression confirmed that suburban districts have significantly higher opt-out rates than urban, rural, and town districts. When controlling for various demographic and political variables, suburban districts exhibited average opt-out rates 13 percentage points higher than urban districts, 6 percentage points higher than town districts, and 9 percentage points higher than rural districts. Support for the notion that the opt-out movement is most prevalent in socioeconomically privileged districts, however, was more mixed. The results revealed that districts that serve a larger percentage of nonwhite and poor students do tend to have lower opt-out rates, and this relationship is statistically significant (p=0.000). Specifically, a 1 percentage point increase in nonwhite student enrollment or a 1 percentage point increase in poor student enrollment was associated with approximately a 0.2 percentage point decline in district opt-out rates. On the other hand, districts that enrolled a higher percentage of special needs students tended to have higher opt-out rates. While this finding may seem paradoxical (because special needs enrollment is positively correlated with nonwhite enrollment and poverty), it is in fact consonant with the finding elsewhere (Casalaspi, 2018) that parents who have special needs children were more likely to opt out due to concerns that the tests would pose a severe burden to their children. District educational performance likewise appeared to have at most a mixed relationship with opt-out rates. Districts that posted higher performance on the state math tests during the previous school year (2014-2015) had slightly lower math opt-out rates in 2015-2016 (p=0.001), but the relationship between ELA performance and ELA opt-out rates was not statistically significant (p=0.081). The regression further revealed that the proportion of the adult population who work in district schools was not significantly related to opt-out activism, casting doubt on the claim that the opt-out movement is spurred by educators and teachers' union actors. In fact, the relationship between those two variables was actually negative.
Finally, two political variables from the model are worth discussing. The literature on political participation generally concludes that participation in one form of political activity, such as voting, is positively correlated with participation in other forms of political participation, such as volunteering on a campaign or attending a demonstration. In other words, people who vote are also more likely to participate in other forms of political activity. This predicted positive relationship between voting and participation in the opt-out movement, however, was not borne out in my own analysis. ELA and Math opt-out participation rates were actually negatively correlated with voter turnout in the 2016 presidential election (r=-0.17 and r=-0.14 respectively), and in the regression model, the coefficients for these variables were near zero and not statistically significant (p>0.8). Moreover, my survey data reveal that OOPs did not appear any more or less likely than NOOPs to have voted in the 2016 election or engage in other forms of political participation over the previous 12 months. The political variable that appeared to be most closely associated with district opt-out rates was the percentage of voters in the Republican Primary who voted for Donald Trump (r=0.44). In fact, the regression model suggests that a 1 percentage point increase in support for Trump in the 2016 Republican Primary was associated with a 0.9 percentage point increase in district opt-out rate. This finding suggests that participation in the opt-out movement may be driven by anti-establishment political attitudes more than a particular political orientation or ideology (see also Casalaspi, 2018).
Taken together, the statewide data challenge the stereotype that the opt-out movement is confined to white, wealthy, and suburban districts. It is true that larger enrollments of nonwhite and poor students are negatively associated with opt-out rates and that suburban districts have much higher opt-out rates than their urban, rural, and town counterparts. However, elevated average optout rates appeared present in virtually all types of districts, and the most racially diverse districts in actuality exhibited opt-out rates on par with the least racially diverse districts. Furthermore, I found little evidence that the presence of educators in a district was associated with opt-out activism. -Overall these findings suggest that, at least at the district level, the opt-out movement is in fact much more diverse than previously reported, and it may in fact mobilize parents across the demographic spectrum.

The Opt-Out Movement within Districts
The findings above challenge the stereotype of the opt-out movement as white, wealthy, and suburban, at least when examining state-wide, district-level data. However, the preceding analyses are vulnerable to the same criticisms that can be leveled against other scholars who have previously studied the question of opt-out participation: they run the risk of committing an ecological fallacy by inferring that patterns found in the aggregate at the district level are also found within districts (see Bennett, 2016;Chingos, 2015;Supovitz et al., 2016). It is possible, in other words, that while the opt-out movement is present in a wide diversity of districts, the individuals participating in the movement are actually much more homogenous. A district may be racially diverse, but perhaps the parents opting out within that district are mostly white, wealthy, and unrepresentative of the district as a whole.
To address this question, I turn now to the data collected in my four case districts, focusing attention on the survey data and interviews with district elites. Using my survey data, I tested for demographic biases in the composition of the opt-out movement by comparing the percentage of total survey respondents in a particular demographic group to the percentage of OOPs and NOOPs in that same demographic group. For example, if white parents comprised 70% of the total survey sample in a particular district, but 90% of OOPs in that district, then the claim could be made that the opt-out movement is skewed toward the concerns of white parents in that district.  Table 4 above and Table 5 below report the results of these analyses for the characteristics of race and income respectively. Across the total pooled sample, no statistically significant differences emerged on either characteristic. Moreover, within each district subsample, no statistically significant differences emerged across income levels, and in only one district (Commonwealth) did there appear to be any racial differences in opt-out participation patterns. In Commonwealth, white parents comprised 84.6% of the total survey sample and 89.1% of OOPs while nonwhite parents comprised 15.4% of the total survey sample and 10.9% of OOPs (p=0.01). Additional analyses allowed me to test another characteristic of opt-out participants-their partisan affiliations and political ideologies. Both Pondiscio (2015) and Pizmony-Levy and Green Saraisky (2016) have suggested that OOPs are politically liberal and that the opt-out movement may therefore be a vehicle for leftist policy ideas. Contrary to this stereotype, I found in my own survey data that the opt-out movement appeared to mobilize Democrats (24.4%), Republicans (33.0%), and Independents (30.9%) across all four districts, and there were no significant differences in participation rates among parents in these groups. Furthermore, when asked about their political ideology on a 1-7 scale (with 1 being "extremely liberal", 4 being "middle of the road", and 7 being "extremely conservative"), both OOPs (4.18) and NOOPs (3.75) appeared to be relatively moderate in their political leanings, and no statistically significant differences emerged within any of the districts. Thus, the opt-out movement appeared to mobilize people across partisan and ideological divides.
Finally, the results of my survey produced mixed evidence regarding the prevalence of educators in the opt-out movement. In the total pooled sample, educators, who made up 22.9% of all survey respondents, did appear significantly overrepresented among OOPs as they comprised 28.6% of the total sample of OOPs and only 17.7% of NOOPs (p=0.01). In three of the districts, educators comprised a larger percentage of OOPs than their percentage of the district survey sample, but statistically significant differences were not evident in any of the four districts. These differences appeared larger in the two low opt-out districts (Greenville and Easton) than in the two high opt-out districts (Danville and Commonwealth). In Greenville, educators made up 16.8% of the total survey sample and 23.5% of OOPs, and in Easton educators made up 19.0% of the total survey sample and 26.5% of OOPs. In contrast, educators in Danville and Commonwealth exhibited greater parity between their percentage of the total survey sample and percentage of OOPs. Thus, in low opt-out districts, educators appeared to play a slightly more outsized role than non-educators in the opt-out movement, perhaps due to their professional knowledge of the movement or union mobilization.
The findings of limited demographic differences between OOPs and NOOPs within districts were generally confirmed by the interviews I conducted with district elites and activists, although there was greater consensus on this point in the high opt-out districts than in the low opt-out districts, and some important nuances became apparent, especially with regard to perceptions of Hispanic and Asian parent participation in the movement. Table 6 below shows the frequency with which the different participants believed the opt-out movement had mobilized a demographically balanced cross section of their district. It is important to note that given the triangulation approach in this study, no single participant's (or group of participants') views should be interpreted as dispositive or "more expert" than others on the research questions.

YES YES
Notes: Among other questions, participants were asked in their interview: "From your perspective, what types of people participate in opt-out activities in your community? Do people who opt out of testing represent a general cross section of your district? Or are opt-out participants different from non-participants?" * Two Commonwealth school board members stated that the movement mobilized a cross section of their community, but they sensed slightly lower participation among Hispanic and Asian parents. † The lead local activist in Commonwealth stated that the opt-out movement mobilized a general cross section of their community, but there seemed to be lower participation among Asian parents.
As Table 6 shows, in both of the high opt-out districts (Commonwealth and Danville), the superintendent and all four school board members reported that the opt-out movement seemed to mobilize a broad cross section of their community. One Commonwealth board member said it mobilized "pretty much everybody," and another directly challenged the white, wealthy caricature of the opt-out movement popular in the media: [Commonwealth] gives you a nice broad cross section of all different types of income levels. And I believe it was Arne Duncan who said it was a bunch of suburban soccer moms who are now just finding out their kids aren't as smart as they thought they were. But no, I think Commonwealth is a perfect example of it going across all different socioeconomic models. You have lower-income families that are just as invested in their kids' education in this district and participating in the refusals as you do the pockets of wealth that are in this district. I could say with confidence, that our district definitely isn't just a bunch of rich white people who don't want their kids taking these tests. We've got a nice cross section of people who are refusing.
Echoing the views of his board, the Commonwealth superintendent even expressed surprise that this was the case: "I would have thought it [the opt-out movement] would've [fit a particular demographic profile] but it seems to be a random cross section." District officials in Danville expressed similar views about the movement in their district. However, some district officials were careful to differentiate between the most active OOPs (i.e., the leaders of the movement) and the rank-and-file OOPs. According to the Danville superintendent, the movement in her district could be roughly divided up into three groups: a cadre of activists, a group that is socially connected to these activists but not as individually active, and another group of parents who were just mindlessly following the herd: The movement was actually started by two parents, who are sisters and former valedictorians at this school. The two women themselves are very educated, very bright, very passionate about the topic, and so they began informational meetings. They started a Facebook page, and that gathered a lot of steam. I'd say that the first group of people [is] the group of people who are very educated on the topic, embrace the opt-out [movement] completely, and are knowledgeable about it. They know their reasons why they're opting out. They know what they want to see change.
They are writing letters. They are making phone calls [and] emails to senators and different people about what they're dissatisfied about. So that's the one group. The second group, I'd say, are the people who are not completely aware of everything that's going on, but heavily influenced by the group that does know what's going on. So they're connecting through the Facebook page. They're connecting at their kid's sporting events and things like that, and they're talking, and they're having these conversations.
And then, there's another group, I think, that is completely clueless and are just along for the ride. I wanna be honest about that. I do. I would basically put them in those three categories.
When asked what seemed to differentiate the most active opt-out parents (group one) from the less active ones (groups two and three), she said the people in that category tended to be "more educated and, socioeconomically, are probably in a higher class as far as income is concerned…[but] level of education is more central than how much money they make." Contrary to the findings in the high opt-out districts, in the two low opt-out districts, the results were more equivocal regarding the diversity of the movement-complicating the survey findings of null demographic biases. In both Easton and Greenville, two of the four board members remarked that the opt-out movement represented a cross section of their community, although the superintendents tended to disagree with this assessment. The Easton superintendent remarked that he felt it was "the higher socioeconomic group" opting out, and the Greenville superintendent said that "educated parents" were the ones most likely to opt out. Furthermore, the Easton superintendent felt that people connected to the teaching force were more likely to opt out, and this sentiment was echoed by one of the Easton board members: "I guess the opt outs here I think are either related to teachers [or] teachers' kids-and that's not a small community here." This perception of outsized teacher presence in the Easton opt-out movement did seem to be borne out in the survey data (discussed above). Taken together, in the low opt-out districts, some district leaders still harbored perceptions that the movement was biased to some degree toward more privileged elements of the community, although this did not seem to be consistently supported by the other data sources.
Another qualification about the general diversity of the movement worth mentioning was the perception among a couple school board members in each racially diverse district (Commonwealth and Easton) that Hispanic and Asian parents were less likely to be involved in the opt-out movement-perceptions that were not corroborated by other evidence but which take on heightened significance in the context of previous research about parent engagement with education. Three out of these four board members felt that the barriers to participation among Hispanic and Asian parents had little to do with language differences or differences in political resources (e.g., time, money, and education) but were primarily cultural in nature. They sensed that parents in these communities were more reluctant to challenge the authority of educators or political officials when it comes to their child's education. 2 Illustrating this phenomenon, one Commonwealth school board member remarked: culturally, you tend to notice that the families that are generally Asian are gonna say, "No, these tests are tests that they're supposed to be taking. They're gonna take them." You definitely can draw comparisons or links to cultural backgrounds and the test refusals.
One of the Commonwealth activists I spoke to even laughed about the futility of attempting to organize the Asian community: "My husband is Chinese, and I wouldn't even think of trying to organize in the Chinese community or any place like that. You're talking about thousands of years of hierarchy!" Furthermore, with regard to the Hispanic community, a different Commonwealth board member claimed that the Hispanic community "tend[s] to opt out at slightly lower numbers [and] part of it is cultural. There's this sense that you don't tell your teacher no, and so if they give you a test, take the test." Another board member in Easton, who works as an ambulance paramedic, offered his own observation that the reluctance of Hispanic parents to opt out was related to a general feeling of uneasiness in that community. Thus, while evidence of lower participation rates among these groups was not corroborated in my other data sources, and the opt-out movement on the whole appeared to reflect the communities in which each operates, there did appear to be some compelling on-the-ground perspectives that suggested the opt-out movement exhibited slight deviations from demographic parity, particularly in the low opt-out districts (where higher educated parents tended to be a bit more visible in the movement) and in the high racial diversity districts (where Hispanic and Asian parents tended to be a bit less visible in the movement). This finding takes on added significance when considered in the context of existing literature on the culturally bounded nature of parenting and school-family relationships, suggesting that some parents may face similar cultural barriers when it comes to engaging in grassroots social movement protest activities.

The Most Active and Longest-Tenured OOPs
Overall, the balance of evidence suggests that in all four districts, the opt-out movement successfully mobilized an approximate cross section of parents, but a lingering question remains about potential demographic biases in terms of movement leadership, visibility, and activity ratessomething suggested by the interview participants. To further explore if the parents who were most active and visible in the opt-out movement tended to be whiter, wealthier, or more highly educated, I again turned to my survey data. On the survey, parents were asked if they had participated in any of 11 activities in support of the opt-out movement, such as attending a meeting, demonstrating, writing a letter, donating money, or posting on social media. The survey also asked parents how many times they had opted out their child during the past five years. Parents were asked "Have you ever participated in any of the following activities in support of opting out? Select all that apply." Eleven options were available for selection: attend a meeting; attend a demonstration; call in to a radio/TV show; contact an official; donate money; join an online group; post on social media; raise money; write a letter to the editor; convince others to opt out; or other. *p<0.05 **p<0.01 ***p<0.001 Table 7 above reveals the mean number of opt-out activities reported by OOPs broken down by district and demographics. Overall, white OOPs reported participating in a greater number of activities to support the opt-out movement than nonwhite OOPs in low opt-out districts. Additionally, college-educated OOPs appeared significantly more active than non-college-educated parents in all four districts as well as across the total pooled sample, although these differences were not always statistically significant. Wealthy OOPs were not only more active, but they also reported being involved in the movement for a longer period of time than poorer OOPs. The wealthiest OOPs (those making more than $100,000 per year) reported that they had opted out their children for an average of 3.3 years whereas the poorest OOPs (those making less than $50,000 per year) reported that they had opted out their children for an average of 2.5 years (p=0.01). Moreover, when examining just those OOPs who reported opting out their children for 4 or more years, 75% of them were from the highest income bracket and only 4% were from the lowest income bracket. Thus, while the opt-out movement has embraced parents of all backgrounds, the most invested and visible parents in the movement appeared to be more highly educated than the movement as a whole. At the same time, there is some evidence that the movement may have originally been comprised of high-SES parents, but over the years it has diversified and brought in parents of less privileged backgrounds. My data is not well-equipped to confirm this trend over time, however.

Discussion and Significance
As discussed at the beginning of this article, it has been an open question as to whether or not grassroots social movements like the opt-out movement can be an inclusive force for political and educational equity. The findings of this study suggest that the answer to that question is a qualified yes. Indeed, the most important discovery from the preceding discussion is the finding that the opt-out movement is in fact a diverse movement that generally mobilizes people across racial and socioeconomic divides. It does not appear to be the white, wealthy, suburban phenomenon it is widely stereotyped to be (Bennett, 2016;Chingos, 2015;Mitra et al., 2016;Pizmony-Levy & Green Saraisky, 2016;Rivera-McCutchen 2021;Supovitz et al., 2016). Moreover, unlike many social movements, it was not confined to wealthy metropolitan areas at its peak, but rather existed across all geographic regions of New York State. Taken together, these findings lend credence to the idea, espoused by some scholars, that social movement activism can help alleviate inequalities in education participation by providing a locally based and relatively accessible opportunity for parents to get involved, learn about education issues, and work coalitionally with others unlike them.
At the same time, though, it is important not to paint too sanguine of a portrait of the optout movement's diversity since a second important finding in this study is the double-edged nature of white, middle-class involvement. Scholars of social movement studies have emphasized the indispensability of middle-class involvement for movement success since middle-class participants are uniquely positioned to lend vital material resources, social networks, and cultural capital to collective action endeavors. Indeed, it is doubtful that the opt-out movement, which has always been dependent upon the voluntary contributions of time and energy from parents, would have ever taken off without the involvement of middle-class parents. In Danley and Rubin's (2017) comparative case study of grassroots protest against state control in Newark and Camden, the authors attribute the success of Newark activists and the failure of Camden activists to the presence of a robust, civically engaged middle class in the former city but not the latter. Furthermore, in a U.S. educational system that implicitly favors white, middle-class cultural values, the inclusion of parents who embody those values may be critical in ensuring a movement obtains recognition and legitimacy (Lightfoot, 1981). Whereas low-SES parents and parents of color are often dismissed as uninformed troublemakers when they advocate for educational change, white, middle-class parents are often treated with respect and deference by educators and policy elites (Lareau, 2003). In light of this fact, it seems unlikely that the opt-out movement would have obtained the same level of traction and public attention if it did not include a substantial number of white, middle-class parents alongside nonwhite and low-income parents.
While the presence of white, middle-class parents is an enormous benefit for the opt-out movement, it also appears in my data to pose some risk vis-à-vis participation equity. As Schlozman et al. (1999) demonstrate, white, middle-class individuals not only participate more frequently in politics, but they also participate in deeper and more substantial ways. My data suggest that this has remained the case in the opt-out movement in New York insofar as white, wealthy parents reported participating in a greater number of opt-out activities than low-income parents and parents of color. They also reported being involved in the opt-out movement for a longer period of time. Thus, while the rank-and-file membership of the opt-out movement may be diverse, the most active membersand the ones most likely to assume local leadership positions-have tended to be whiter, wealthier, and more educated than the movement as a whole. This uneven level of participation within the movement could have serious consequences for overall equity and inclusiveness. As Morris and Staggenborg (2006) write: "Social movement leaders are the actors whose hands and brains rest disproportionately on the throttles of social movements" (p. 191). It is the leaders of a movement who frame the movement's agenda, organize its activities, and serve as its public face in interactions with policy elites and members of the media. If the leadership does not reflect the diversity of the movement, then the more privileged elements of the movement could come to exert outsized influence over the movement's agenda, mobilization efforts, and public image (see Mitra et al., 2016;Rivera-McCutchen, 2021).
Additionally, the presence of easily spotlighted white, middle-class parents inevitably opens up a movement to stereotyping, which can marginalize the movement in the discussion of education reform. The allegation of Ceresta Smith that the movement was being intentionally "whitewashed" by testing supporters is an eminently reasonable one that takes on added severity because so much of today's education reform debate is imbued with racial and social justice overtones. If a grassroots education social movement can be raced and classed as white and wealthy, this may delegitimize its voice from the political discussion of education reforms purported to improve education for historically marginalized populations. Thus, grassroots activists must be careful to guard against this threat, and it would behoove them to make conscious efforts to foreground their movement's diversity, especially early in its lifecycle, so that stereotypes of privilege are less likely to stick. This task, however, is challenging, and one opt-out activist highlighted the frustration she felt as a highly educated leader of color: There's this persistent feeling that it's a white movement. I mean, even within our own ranks that happens. Sometimes I'm like, "Ugh!" because people will be like, "Well, we have to have parents of color say that," and I'm thinking, "I'm a parent of color! You don't even see me?" I know that's not what they mean, 'cause I'm a parent of color, but I went to Stanford. It's like I'm not the parent of color they're thinking about. We've got to dispel that mess. (Interview with author) Finally, the perception that Hispanic and Asian parents may face cultural barriers to participation is something that should be probed in further research. Research has definitively established that these parents face institutional barriers when it comes to participation in the life of their school communities, due in part to their distinctive orientations toward the school-family relationship. This study suggests that these cultural barriers may continue to be present when it comes to extrainstitutional forms of participation such as grassroots social movement protest. Given the deference of these parents to the wishes of educators and policy elites, it is less likely that these parents will be intrinsically motivated to challenge educational authority without receiving encouragement from those same authorities. In this way then, explicit requests to participate from teachers, educators, or other local officials could be especially critical in activating these parents, and grassroots activists would be wise to form alliances with sympathetic educators and then leverage the authority of those educators in their outreach to those parents. Activists working in the Hispanic community may face the additional challenge of reassuring parents who are already skittish about national political developments. The barriers to equity in participation not only exist at the level of the locality or the individual, but also remain invisibly embedded in national political developments-suggesting that activists will have to attend to national concerns even while laboring locally to organize their community. Readers are free to copy, display, distribute, and adapt this article, as long as the work is attributed to the author(s) and Education Policy Analysis Archives, the changes are identified, and the same license applies to the derivative work. More details of this Creative Commons license are available at https://creativecommons. About the Editorial Team: https://epaa.asu.edu/ojs/index.php/epaa/about/editorialTeam Please send errata notes to Audrey Amrein-Beardsley at audrey.beardsley@asu.edu Join EPAA's Facebook community at https://www.facebook.com/EPAAAAPE and Twitter feed @epaa_aape.