Proving I’m Right: Charter School Policy and Selective Exposure to Information

Abstract

Research suggests that individuals may make choices about the information they consume that are influenced by what they already believe. In this study, I investigate this process in a particularly contentious policy arena: charter schools. What kinds of information are important to people as they evaluate charter school policy? Are their choices shaped by their prior beliefs? Overall, I find that indeed people tend to seek out information that aligns with their prior beliefs. Altogether, the results of this study suggest that the dynamics of selective exposure present in the political arena more broadly also exist within education policy.

Keywords

politics of education charter schools information political science school choice

The Center for American Progress (CAP), self-described as an independent nonpartisan policy institute, published The Progressive Case for Charter Schools in October 2017. In this discussion, the authors presented a variety of pieces of evidence in support of charter schools that, as a whole, were meant to align with progressive values such as economic mobility. These pieces of evidence included overall comparative success of charter schools as per the popular 2015 CREDO (Center for Research on Education Outcomes) study, data from the74million.org on higher graduation rates for charter school students, increased teacher diversity in charter schools based on data from the National Center for Education Statistics, and movements of several major charter organizations toward disciplinary practices that align with principles of restorative justice (Roth, McDaniels, Brown, & Campbell, 2017).

Six days later, a rebuttal was posted through Truthout, which describes itself as a source of independent news and commentary. In short, the author of this piece strongly repudiated the idea of any progressive case for charter schools and claimed that charter schools, while possibly created with progressive ideals, have evolved in such a way that they have not met any of their progressive promises. To support this argument, the author, like the CAP report, cites multiple sources of evidence. Among other things, the author cites relatively small effect sizes of charter school studies (including the CREDO study). In addition, they highlight alternative types of information about the same items that CAP covered, all of which might persuade one to think otherwise about charter schools. For example, they include evidence about the Noble charter system, cited heavily in the CAP report, that concerned harsher discipline policies and weeding out of students with disabilities. In response to the assertion that charter schools recruit more diverse teachers, the Truthout post points to evidence that Teach for America teachers, who play a major role in charter school teacher recruitment, are much more likely to leave schools after being there for only a short period of time (Bryant, 2017).

Who are we, as citizens, supposed to believe? An interesting problem for this debate is that for the most part, both posts—from CAP and Truthout—cite important sources of empirical evidence for the charter school debate and, for the most part, the debate does not seem to be about the veracity of the evidence itself. Instead, one of the clearest divisions between the two arguments seems to be what evidence is important or not for determining whether charter schools are indeed an important policy for a progressive political agenda. Beyond questions about the interpretation of specific pieces of evidence (e.g., a discussion about practical significance vs. statistical significance), there were also different types of information being presented. Assuming that the original research that produced this information was done with fidelity and rigor (a discussion saved for another time), this kind of discursive problem may contribute to an oft-held belief that one can find research to support any argument.

This problem clouds the landscape when it comes to questions about the role of research and researchers in policy making. What evidence is important to people? For better or worse, existing research in political psychology suggests that the answer to this question varies by the individual. With these findings in mind, it is not evidently clear what interventions are available to change the policy debate environment, nor do we know how people may interact with the interventions. Thinking about it, we might consider two major types of problem sets in the politics of education that can be targeted in the development of an intervention: those dealing with people’s actual beliefs and those dealing with the relational politics of those involved. While the latter likely rests within the realms of diplomacy and conflict management, the first notably sits within the purview of education, research, and information. While there are certainly aspects of beliefs that are questions of morality and deeply held ethical ideologies, an understanding of the role of information in people’s opinion processes could greatly inform steps forward. In particular, approaches to study that involve direct intervention instead of observation could be important.

First, however, we must understand information-seeking behaviors. In this study, I explore the extent to which different types of information about charter schools are more or less important to people when they are evaluating policy. Using established methods in political psychology, I investigate the tendencies of individuals to make deliberate choices about the information they seek. These tendencies rest within the theoretical realm of selective exposure and motivated reasoning. In this theoretical framework, people are not simple evaluators of information but rather evaluate information in a way that is biased by prior beliefs, attitudes, and states of mind. When I refer to information types, I am referring to categories of information based on the concerns to which they speak (e.g., equity, innovation). Through this approach, I investigate three research questions:

Research Question 1: What characteristics of information (i.e., pro-/anti-charter, information type) are more or less interesting to people overall?

Research Question 2: To what extent do prior attitudes about charter schools influence the extent to which the above characteristics are differentially interesting?

Research Question 3: Does the influence of prior attitudes on item interest vary across people of different political and educational demographic groups?

This article proceeds as follows. First, I provide an overview of the information that does exist today on charter schools. Following a question about what information people do/should pay attention to, I then discuss the concept of motivated reasoning in political psychology as a potential source of insight. Next, I review the methods and data source for the exercise I conducted. I follow with a discussion of the results, organized by research question. Finally, I conclude with implications for practice and research.

Mixed Information on Charter Schools

Up until the 1990s, research around charter schools was mostly theoretical and/or rhetorical. Over time, as data become more freely available and as the implementation of charter policy became more stable, this research shifted to be more evaluative (Miron, 2010). How were charter schools serving the needs of students? Building on the motivations for charters discussed earlier, there are three ways in which we might consider evaluating charter schools outside of their direct impact on the achievement of their students: equity, innovation, and competition (Weitzel & Lubienski, 2010). First, equity motivations focus on the posited ability of charter school policy to offer underserved student populations opportunities that are usually afforded to privileged populations. Second, charters were seen as a way to give educators more flexibility for serving their students’ unique needs. Finally, through market-based mechanisms, charters (and school choice) ostensibly introduce competition into the education system, forcing all players, traditional public schools included, to perform better to compete. On all of these points, the evidence thus far has been mixed.

Equity

Regarding the ability of charters to uniquely serve the needs of historically underserved populations, the evidence has been mixed as to whether they are doing so (Weitzel & Lubienski, 2010). One common concern about the equity outcome of charters schools is that when coupled with school choice (as they often are), they contribute to a resegregation of schools as parents choose to send their students to schools with more people that look like them, especially among White parents (“white flight”; Eidelson, 2014). In addition, inequities in families’ realistic ability to choose as limited by transportation and information (Teske, Fitzpatrick, & Brien, 2009) cause some to wonder whether only the most advantaged and/or highest performing students are able to move from traditional public schools to charters, leaving less advantaged and lower performing students in traditional public schools in a phenomenon called “cream skimming” (Dean, 2014). The research on these resegregation points has not reached a consensus, and all conclusions are vulnerable to critique. Some research does indicate that white flight occurred, while some does not and points to possible self-segregation on the part of parents from underrepresented minorities (Garcia, 2010). However, this does not mean that charters have played no part in this resegregation. In addition, there is little consistent evidence of cream skimming, as charters and traditional public schools have been shown to serve comparable numbers of low performing, disadvantaged students. However, there is still concern that there are unmeasured characteristics that distinguish families that switch from those who do not, including involved parents (Garcia, 2010). Overall, there is some evidence of resegregation and cream skimming in limited cases, but neither of those phenomena have been found to be particularly widespread (Wohlstetter, Smith, & Farrell, 2013).

Innovation

The evidence on the extent to which charters are producing more innovative practices is also characterized by weak, inconsistent evidence. Overall, there has been little evidence of new programs or innovations in charter schools writ large, except in the increased use of educational technology in charter classrooms. There are reports that educators in charters often feel that there is more autonomy, but there is little evidence that these feelings are translating into innovative practice (Wohlstetter et al., 2013). Over the last decade, charter schools have been increasingly held to federal and state agency standards, and it may be the case that these regulations are limiting the innovation envisioned in their inception (Miron, 2010).

Competition

One of the challenges of studying the competitive effect of charter schools on traditional public schools is that charters are not randomly located in districts and neighborhoods. Not only that, but students and families self-select into charter schools, and there are often limits to how much families can actually “choose.” Finally, competition is hard to measure, and the distributional effects of competition on different types of schools are difficult to summarize (Lubienski & Weitzel, 2010a; Ni & Arsen, 2010). Still, the market-based argument here is that if traditional public schools are forced to compete with charter schools for students, they will expend more effort and energy to raise the quality of their own educational programs. There are reasons, however, to think that this market-based mechanism might not work. For example, the turbulence of constant student switching and moving may be detrimental for schools, and competitive marketing may create conflict between traditional public schools and charters.

Finally, there is some practical concern that when more students go to charters, traditional public schools will lose funding that they need to serve their students (Ni & Arsen, 2010). A typical example of this concern is articulated by Caref, Hainds, Hilgendorf, Jankov, and Russell (2012):

The supposed “bureaucracy” that holds back innovation in neighborhood schools is not an inherent feature of public neighborhood schools but an intentional policy of disinvestment that withholds resources for innovation in neighborhood schools and gives additional funding and autonomy for charters. (p. 4)

There is currently a growing literature on this issue that investigates whether this is indeed the case. So far, research finds that traditional public schools do lose revenue when charters exist (Xing, Maugeri, Pierson, & Reitano, 2015), but that this is not necessarily through the disinvestment of the public itself (Honey, Blissett, & Woo, 2016). In this way, traditional public schools may become less efficient in that they may not reduce their revenue or expenditures until they know that reduced enrollments are permanent (Ni & Arsen, 2010). However, there is little evidence that the competition created by the presence of charter schools improves the achievement of traditional publics (Wohlstetter et al., 2013). If anything, these schools are not necessarily getting better, but are instead shifting their resources toward new programs, changing leadership, making magnets, or making programs consistent with parent preferences, but without the increase in achievement that market advocates might expect to see (Ni & Arsen, 2010).

Student Achievement

Beyond the innovation argument, charter advocates believe that charters will perform better than traditional public schools because the ability of charters to be run by nongovernment managers, which allows them to rid themselves of what is perceived as government inefficacy. Conceptually, charters also have an additional source of accountability that traditional public schools do not have: the ability to fail. They have to be accountable to their own consumers (families and students) because in the event that they are not serving their consumers’ needs, they risk being shut down (Miron, 2010). As such, there has been a robust literature on the extent to which charters produce higher achievement gains than traditional public schools.

Overall, as has been the theme with charter school research, results have been mixed. In early years of the charter movement, most evidence showed lower or equal performance by charter schools as compared with traditional publics (Lubienski & Weitzel, 2010b). Berends, Watral, Teasley, and Nicotera (2008) provide a broad overview of reviews of charter school effects throughout time, ranging from a review by Goldhaber in 1999 to a review in 2006 by Hill et al. This review of reviews summarized the mixed effects of charter studies across time, leaving us with few concrete answers about the relative benefits of charter schooling. There is some evidence that more recently, charters have gained ground (Wohlstetter et al., 2013). Using meta-analyses, Betts and Tang (2011) find consistent positive effects. However, these aggregated effect sizes, though statistically significant, were relatively small, hovering at about 0.05. Cremata et al. (2013) find similar results, with no evidence of effects of math scores, and general positive results for reading, but only amounting to eight additional days of instruction. Finally, a review by Berends (2015) again emphasized that there have been mixed results in terms of the effect of charter schools on student achievement, plus positive results in terms of educational attainment (e.g., graduation, college attendance, college persistence).

Why would there be so much mixed evidence? Beyond typical research heterogeneity regarding differing methodologies, a major reason for this is simply that charter schools are also heterogeneous. Across the nation, there are diverse laws, diverse models, and diverse students being served by charter schools (Wohlstetter et al., 2013). As such, it is difficult to pin down one “policy” that is being evaluated. As such, Berends et al. (2008) propose that research on charter schools should not be asking whether charter schools work as a whole, but rather under what conditions they work. Still, this information exists in the world. What are people to pay attention to? What information do people find compelling when evaluating charter schools?

Selective Exposure and Motivated Reasoning

Historically, the study of selective exposure has focused on the tendencies of individuals to systematically expose themselves to information consistent with their prior beliefs and avoid inconsistent information (e.g., Arceneaux & Johnson, 2013). Here, however, I use the term selective exposure more broadly to encompass also those decisions to select information along lines other than belief congruence. Some types of information and arguments may be differentially compelling to different individuals based on their own moral foundations. For example, Peralta, Wojcieszak, Lelkes, and de Vreese (2017) find that numerical evidence, for populations with stronger beliefs about climate change and health care, tended to be more important than narrative evidence. In short, the amount of information available often exceeds the intake capacity for individuals, and as such, people make nonrandom choices about what information to consume. These choices are important for policy makers and researchers to understand to inform the kinds of evidence and arguments that should be presented if the goal is to shift the landscape of policy debate.

The typical understanding of selective exposure ties into the proposed defense motivation that people hold in information selection. Here, people have the goal of protecting their existing beliefs and attitudes. By contrast, it is also suggested that people sometimes are driven by an accuracy motivation, whereby they select information in a way that, to them, optimizes the chance that they will be “correct” in their attitudes and beliefs at the end (Kruglanski & Klar, 1987). Both of these concepts are covered in the study of motivated reasoning. Motivated reasoning is the concept that the rational considerations that people make in their evaluations of political objects are influenced by relatively unconscious affective biases that are triggered automatically. In other words, the processes people go through, including the acquisition and appraisal of new information, in evaluating a political object are themselves influenced by automatic emotions that exist in that context (Lodge & Taber, 2013).

Given these theoretical frames, what might we expect to see? The first, most obvious conclusion based on the existing evidence is that people select pro-attitudinal information. People who already support charter schools will be more likely to consume information supporting charter schools, and vice versa. The evidence thus far, however, does not speak much to the types of information that people with different prior attitudes will find to be differentially compelling. This study is one of the first to explore this topic in education politics. To an extent, this lack of evidence makes sense, as the relevant information for any particular policy arena is likely to be very arena-specific. For example, while the extent to which a policy supports increased student achievement is likely to be highly discussed in conversations about teacher evaluation, this same evidence type may be more absent from discussions about suspension policy. As such, any attempt to influence the policy discussion that is empathetic to meeting people where they are may need to consider the different kinds of information that people find to be compelling.

Methods in Political Psychology and Motivated Reasoning

Methods used to understand motivated reasoning as introduced by Lodge and Taber (2013) may help us get closer to an answer. Lodge and Taber’s (2013) The Rationalizing Voter stands today as one of the most important works on public opinion. In short, the authors present the results of a variety of political psychology experiments that establish a model of political cognition in which people’s evaluations of political objects are influenced by not only rational processes, but also rationalizations and motivated reasoning.

The experiment that the authors conducted is fairly expansive as it asked multiple research questions at once, and I only describe part of it here. The main experimental tool was the use of an information board. In this device, experiment participants were shown a static grid of possible choices for information they could investigate about a topic. They chose a topic, and then the technology logged the information they chose. The computer recorded the order and viewing time for the arguments selected. Consistent with their hypotheses, the authors find evidence that despite asking people to view information even-handedly with the objective of being able to explain the issue to others, people still tended to seek out pro-attitudinal information consistent with their prior beliefs.

A limitation of the information board approach is that it assumes that the information available to citizens is static and always-available. In many contexts, such as the political campaign environment, there is often more information than can be digested easily by citizens, and the information available is always changing. To address this concern, instead of a static information board, researchers have more recently used the Dynamic Process Tracing Environment (DPTE). The DPTE is a web-based interface available from researchers at the University of Iowa and funded by the National Science Foundation. Instead of having access to a constant set of information, participants are presented with scrolling labels of information (e.g., “NBC/Wall Street Journal Poll, early February”), and they can click on any label to gain more information. Scrolling continues throughout the entire process such that people can read two to three of the pieces of information before position changes, and there is a cost to accessing any information while the information continues to scroll. Data gathered from this environment include what item was accessed, time of access, and length of access (Redlawsk & Lau, 2009).

I propose in this study that the information searching mechanisms that are present in the political context are also present in the charter school context. As reviewed, there is a large and evolving base of information available about charter schools. People must make choices about the information they consume, and those choices may be motivated by prior beliefs. What do these choices look like? In this study, I first classify information about charter schools in terms of whether it is pro- or anti-charter schools. This does not mean that the original work was written with the explicit intent to help or harm the charter school movement, but rather that the presented narrative could be seen as either supporting or conflicting with the notion that charter schools are good for American education. Item “type” was delineated by whether the item primarily concerned itself with issues of student achievement, equity, no excuses, or financial resources. These items types were derived from both extant literature as presented above as well as qualitative analysis I have conducted in a parallel study.

Participants and Methods for This Study

A total number of 400 participants for this study were recruited via Amazon Mechanical Turk (MTurk) on two dates: May 5 and 23, 2017. Especially in recent years, MTurk has been used frequently in political science literature (e.g., Gerber, Huber, Doherty, & Dowling, 2011; Huber, Hill, & Lenz, 2012). In comparing MTurk participants with traditional collegiate samples used in political science on the extent to which they performed on a measure of attentiveness to instructions, Hauser and Schwarz (2016) found that MTurkers were more attentive. One common concern with MTurk samples, however, is the external validity of the subject pool, as people differently select into participation. To test this, Berinsky, Huber, and Lenz (2012) compare an MTurk sample with data from the American National Election Panel Survey (ANEPS), Current Population Survey (CPS), and American National Election Studies (ANES). On most characteristics tested, the MTurk sample was comparable with the ANEPS sample, while both exhibited similar distortions from highly rigorous CPS and ANES samples. In particular, MTurk participants tended to be younger, more liberal, and more educated. This is important to note in the analysis of my results here. However, studies comparing results using traditional samples and MTurk samples have shown consistent results across these groups. Mullinix, Leeper, Druckman, and Freese (2015) found comparable results in treatments effects compared across MTurk samples, other convenience samples, and a population-based sample. Paolacci, Chandler, and Ipeirotis (2010) found similar comparability when comparing an MTurk sample with an online discussion board sample and a collegiate sample at a large Midwestern United States university.

People received the exercise in several steps. First, after an introduction to the full exercise, they were given an attitude battery and attitude strength battery.¹ Second, they practiced using the DPTE system using a series of headlines on arguments pertaining to a topic unrelated to charter schools (whether or not penguins are in fact a type of duck). Third, they conducted the actual DPTE exercise for 2 min. In this exercise, eight different headlines scrolled through, and each was shown around 6 times. Content from original sources was edited to be approximately the same length for each item. After the conclusion of the DPTE exercise, participants were shown the attitude and attitude strength items again.² Finally, they received a demographic questionnaire and then were asked to briefly describe their attitudes toward charter schools. This last item was included—and they were told about it beforehand—to incentivize individuals to take the information search seriously. In my study, participants were paid US$2.00 for their time.

Overall, I have a design that looks like the diagram in Figure 1, with each dotted line area indicating the data that were used to answer each of my three research questions. With the attitude direction items, to use them as aggregates in the analysis, I extracted principal components factors from factor analyses. I used a polychoric correlation matrix for the categorical attitude items. A scree plot of the eigenvalues is shown in Figure 3. This factor analysis was done using all the data, pre- and post-questionnaire data combined, to capture the consistent factor. This factor accounted for 75% of the variance in the items.

Figure 1.

Experimental design and research questions.

I model the various research questions using multilevel logistic regression,³ where data were analyzed at the item-within-person level, and individual characteristics were included later. To examine the extent to which different characteristics influence item selection, I begin with the following model at the item level:

\Pr (o p e n e d_{i j}) = α_{j} + β_{1 j} p r o c h a r t e r_{i j} + β_{2 j} i t e m t y p e_{i j} + ε_{i j} .

In this model, opened is a binary variable indicating whether a specific person opened a specific item in the DPTE exercise, procharter is a binary variable for whether the item was explicitly pro-charter, and itemtype is a vector of three binary variables for whether the item was about equity, no excuses schools, or resources (with academic items being the reference group). First, I estimate random intercept models to account to likely random variation in item selection rates in general across individuals using Level 2 specifications as shown below.

α_{j} = γ_{00} + u_{0 j},

β_{1 j} = γ_{10},

β_{2 j} = γ_{20} .

Next, to investigate the variation in information searching behavior across individuals, I include person-level error terms in the slope specifications to allow for random intercepts, as shown below. For all models, the correlations of random effects were left unconstrained.

α_{j} = γ_{00} + u_{0 j},

β_{1 j} = γ_{10} + u_{1 j},

β_{2 j} = γ_{20} + u_{2 j} .

As posited by theories of selective exposure and motivated reasoning more generally, we might expect the extent to which different items are interesting to people (operationalized by the $β$ s) to vary according to prior beliefs. As such I test this assertion in two steps. First, I modify the intercept and $β_{1}$ models to include a person-level measure of prior attitudes (the attitude direction factor as discussed above).

α_{j} = γ_{00} + γ_{01} a t t i t u d e_{j} + u_{0 j} .

β_{1 j} = γ_{10} + γ_{11} a t t i t u d e_{j} + u_{1 j} .

This allows me to test what I will call the “pro-attitudinal hypothesis,” which is measured by the extent to which prior attitudes about charter schools predict the relationship between the valence of the item and whether the item is opened by a person ( $γ_{11}$ ). However, this is only one specific facet of how prior attitudes might influence information search behaviors. We might also expect people who are more pro- or anti-charter schools to be interested in different types of information. To test this hypothesis, which I will call the “general motivated searching” hypothesis, I instead modify the intercept and $β_{2}$ models.

α_{j} = γ_{00} + γ_{01} a t t i t u d e_{j} + u_{0 j} .

β_{2 j} = γ_{20} + γ_{21} a t t i t u d e_{j} + u_{2 j} .

Finally, I am interested in the extent to which the influence of attitudes on information search behaviors varies across several important groups: party identification (Democrat, Independent, or Republican), ideological identification (liberal, moderate, or conservative), schooling background (whether or not the individual graduated from a traditional public school), and educational background (whether or not the individual has earned a bachelor’s degree). To test these influences, I include interaction terms in the Level 2 models above as shown below. X_j is a placeholder representing a demographic characteristic, which is either party identification, ideological identification, the type of school the participant attended, or the participant’s educational background.

α_{j} = γ_{00} + γ_{01} a t t i t u d e_{j} + γ_{02} X_{j} + γ_{03} (a t t i t u d e_{j} \times X_{j}) + u_{0 j} .

β_{1 j} = γ_{10} + γ_{11} a t t i t u d e_{j} + γ_{12} X_{j} + γ_{13} (a t t i t u d e_{j} \times X_{j}) + u_{1 j} .

β_{2 j} = γ_{20} + γ_{21} a t t i t u d e_{j} + γ_{22} X_{j} + γ_{23} (a t t i t u d e_{j} \times X_{j}) + u_{2 j} .

Results

Results below are organized by research question. Overall, I find evidence of the pro-attitudinal hypothesis. I do not, however, find evidence of the general motivated search hypothesis, though there is indeed variation in the extent to which different types of items are interesting to individuals.

Research Question 1: What Information Was Most Important?

In Table 1, I show the percentages of individuals who selected each of the eight items. Across the board, all items were selected approximately half of the time. There was not very much variation in the selection of items, with the pro-charter, no excuses item being chosen least frequently (by 45% of people) and the pro-charter, resources item being chosen most frequently. An initial glance at the percentages reveals no significant patterns of pro- or anti-charter material or any specific type of material being selected more frequently. It does appear, however, that the no excuses item was selected less frequently, whether it was pro- or anti-charter. This is confirmed by evidence in Table 2, which shows the results of a logistic regression analysis of item selection on item valence and type. From these results, I find that a no excuses item has 17% lower odds of being opened compared with an achievement item. Importantly, the estimated standard deviations of the item type coefficients across groups were of a notable size. These standard deviations are reported in logged odds units, so the value of 0.13 for the equity item coefficient can be translated to infer that there exists a 0.85 to 1.10 one-standard-deviation band around the estimated overall odds ratio of 0.97. This indicates that there is a significant amount of variation across individuals in the extent to which equity items were of interest relative to achievement items.

Table 1.

Percent of People Who Opened Each Item.

Valence	Type	Percent opened
Anti-charter	Achievement	50.65
Pro-charter	Achievement	49.35
Anti-charter	Equity	51.68
Pro-charter	Equity	46.77
Anti-charter	No excuses	47.29
Pro-charter	No excuses	44.70
Anti-charter	Resources	47.29
Pro-charter	Resources	53.75

Table 2.

Predicting Item Selection From Item Valence and Type.

	(1)		(2)
	γ	τ	γ	τ
Is pro-charter	0.97	—	0.97	0.00
Is pro-charter	(–0.36)		(–0.35)
Equity	0.96	—	0.97	0.13
Equity	(–0.33)		(–0.30)
No excuses	0.82*	—	0.83*	0.14
No excuses	(–1.73)		(–1.66)
Resources	1.03	—	1.03	0.02
Resources	(0.22)		(0.22)
Intercept	1.00	1.00	1.00	1.03
Intercept	(0.02)		(0.01)
Random Intercept	Yes		Yes
Random Slopes	No		Yes
Observations	3,096		3,096
Groups	387		387

Note. The Z statistics are shown in parentheses. Coefficients are reported as odds ratios. Results shown here are from multilevel logistic regression models. Fixed estimates are shown in the γ columns, and standard deviations of the between-group estimates are in the τ columns.

p < .10. **p < .05. ***p < .01.

Research Question 2: How Do Previous Attitudes Predict Information Selection?

Before looking at the effects of prior attitudes, I first review their distribution. In Figure 2, I show a stacked bar plot showing the percentages of individuals who responded in each response category for each item. There is substantial variation among respondents in their answers. The item for which there was the most agreement by participants was that charter schools allow for more opportunities for teachers to implement innovative practices, with 72% of individuals at least somewhat agreeing with this statement. Conversely, the item for which there was the most disagreement was that charter schools take away resources from schools that need them, with 52% of individuals at least somewhat disagreeing with this statement. It is also important to note that responses to a couple of the items exhibited a substantial amount of neutrality. In all, 28% of individuals had a neutral/no opinion about whether charter schools provide a better education than traditional public schools, and 32% held a neutral/no opinion about whether charter schools use practices that align with their values more than traditional public schools.

Figure 2.

Pre-questionnaire attitudes toward charter schools.

Factor analysis of the attitude items, as described in the methods section, revealed the presence of only one underlying factor, as shown by the eigenvalues in Figure 3. Logistic regression models including these standardized factors are shown in Table 3. These models, as discussed, predict the probability of any one item being opened by an individual. First, in column 1, the exponentiated coefficient of 0.75 suggests that on average, those with more pro-charter attitudes were less likely to select an item. In other words, higher pro-charter scores are associated with having selected fewer items in general. To observe the extent to which different types of items were more or less important to people with different prior beliefs, I included interaction effects of the prior belief factors with the various item characteristics. I find a statistically significant coefficient of 1.24 on the interaction between prior attitude and the valence (pro- or anti-charter) of the item. This coefficient above 1 (and significantly so) indicates that the more pro-charter a person was, the more interesting a pro-charter item was. This result is in line with previous research on selective exposure. I do not, however, find any evidence that people with varying prior attitudes about charter schools weighed different item types differently in their information searches.

Figure 3.

Scree plot for attitude factor analysis.

Table 3.

Conditioning Item Selection on Prior Attitudes.

	(1)		(2)
	γ	τ	γ	τ
Is pro-charter	1.08	0.03	1.09	0.05
Is pro-charter	(0.70)		(0.77)
Equity	0.95	0.43	0.96	0.42
Equity	(–0.29)		(–0.27)
No excuses	0.79	0.31	0.79	0.31
No excuses	(–1.47)		(–1.48)
Resources	0.98	0.44	0.97	0.43
Resources	(–0.15)		(–0.16)
Attitude	0.75***	—	0.83	—
Attitude	(–2.76)		(–1.28)
Pro-Charter × Attitude	1.24*	—
Pro-Charter × Attitude	(1.87)
Equity × Attitude			0.91	—
Equity × Attitude			(–0.60)
No Excuses × Attitude			1.08	—
No Excuses × Attitude			(0.49)
Resources × Attitude			1.05	—
Resources × Attitude			(0.27)
Intercept	0.94	1.22	0.94	1.21
Intercept	(–0.40)		(–0.42)
Observations	1,544		1,544
Groups	193		193

p < .10. **p < .05. ***p < .01.

Research Question 3: How Did Information Selection Differ by Political and Educational Identification?

While the previous research question identified the presence of a pro-attitudinal type of information selection, I was interested in whether this type of motivated search behavior might differ across people of different identifications. The different categorical characteristics I tested were partisan identification, ideological identification, schooling background, and educational attainment. Overall, my sample (true to existent research on MTurk) leaned more Democratic, liberal, and was generally more educated that what we might expect from a random sample of the population. Given the fact that I only initially found evidence of the pro-attitudinal hypothesis, I only test for variation in that relationship across groups. These results are shown in Table 4.⁴ Here, I find little to no evidence of differential information search behaviors, in terms of the pro-attitudinal hypothesis, across people of different backgrounds. The largest association was found along ideological lines. In this model, those identifying as conservative had greater odds of exhibiting pro-attitudinal search behavior than those identifying as liberal, and the same was true for those identifying as independent, as compared with liberal. Interestingly, the magnitudes of these relationships far exceeded the magnitudes of what we might have expected to be similar relationships with party identification, given possible assumptions that Democrats largely identify as liberal and Republicans largely identify as conservative. In the descriptive data, while this is somewhat true, I find that in my sample, 15% of Democrats identified as moderate or conservative, 54% of Independents identified as liberal or conservative, and 13% of Republicans identified as liberal or moderate. That all said, the standard errors for the log odds estimates suggest that there should be some caution about interpreting the odds ratios, while large, as conclusive.

Table 4.

Item Selection by Political Identification.

Party		Ideology
Independent	0.92	Moderate	0.82
Independent	(–0.42)	Moderate	(–0.86)
Republican	0.93	Conservative	1.02
Republican	(–0.27)	Conservative	(0.07)
Pro-Charter × Attitude × Independent	0.97	Pro-Charter × Attitude × Moderate	1.19
Pro-Charter × Attitude × Independent	(–0.13)	Pro-Charter × Attitude × Moderate	(0.59)
Pro-Charter × Attitude × Republican	1.01	Pro-Charter × Attitude × Conservative	1.51
Pro-Charter × Attitude × Republican	(0.04)	Pro-Charter × Attitude × Conservative	(1.41)
Observations	1,504	Observations	1,544
Groups	188	Groups	193
Schooling		Education
Traditional public	1.14	Bachelors	1.50**
Traditional public	(0.52)	Bachelors	−2.31
Pro-Charter × Attitude × Traditional Public	1.08	Pro-Charter × Attitude × Bachelors	0.91
Pro-Charter × Attitude × Traditional Public	(0.28)	Pro-Charter × Attitude × Bachelors	(–0.50)
Observations	1,544	Observations	1,544
Groups	193	Groups	193

Note. The Z statistics are shown in parentheses. Coefficients are reported as odds ratios. Results shown here are from multilevel logistic regression models. All models include the same terms as shown in Table 3, column 1.

p < .10. **p < .05. ***p < .01.

Discussion

In this article, I investigated information search behavior and how it differed across individuals. Overall, I found that compared with achievement, equity, and resources, no excuses items tended to be less interesting to people. In addition, different subgroups of participants did differentially search for information. People tended to seek out pro-attitudinal information. However, I did not find evidence that this pro-attitudinal search behavior differed significantly across people with different party identifications, ideological identifications, experience with traditional public schools, or educational backgrounds. I also did not find evidence that people with different prior attitudes found different types of information about charter schools to be interesting. That said, the variation of interest in those different items (as indicated by the standard deviations of group slopes) seems large enough to suggest that at the least, different people do indeed find different types of information interesting, even if we cannot particularly ascertain from this research why this is the case.

There are several aspects of this study that are worth keeping in mind for the interpretation of the broader implications of the results. First, the focus here was specifically on charter school policy, which has been a relatively high-salience issue in recent American history. The dynamics of charter school political reasoning may not repeat themselves in more low-salience subsystems, and future research to investigate the ubiquity of pro-attitudinal thinking across the education policy sphere will be important. In addition, this study was conducted in a relatively controlled environment, not situated within people’s lived experiences with local communities. The results of this research should be coupled with endeavors using other approaches and methods to fully understand the dynamics of charter school and school choice policy debate.

Consistent with hypotheses from motivated reasoning, people tended to seek out pro-attitudinal information. This conclusion, by itself, has important implications. First, this result conflicts with normative ideas about how people should seek out information; should we want people to learn about opinions that differ from their own? Indeed, as researchers, we might hope that the information we provide the public helps people update their beliefs. The capacity of our work to do this, however, is limited if people tend to seek out information that they already agree with. As such, it is important in the framing of research and its dissemination to think critically about the audiences one wants to reach and perhaps do more to present information in a way that, at first glance, is more congruous with people’s priors. Second, some may conclude that this result is troubling. While this research does not investigate the reasons why people are dedicated to their priors, it may be good for researchers and policy makers to be sensitive to this and even seek ways to make people more amenable to investigating perspectives different from theirs.

Implications for Research and Policy

Extant theories of selective exposure do not particularly provide us much direction in knowing what kind of evidence is important to people. As noted, what information is compelling, outside of congruence with prior beliefs, is likely to be subject-specific. The result that different people seek information differently is also important. The conclusions here are similar to those just stated: Any dissemination of information and its framing needs to be sensitive to the ways in which target audiences may digest or seek the information. Not all information reaches all people, and the results of this study provide direct implications to researchers as a field. The future of the relationship between empirical science and policy may depend on the extent to which we as a research community can adjust to the information-seeking behavior of the public and policy makers. This study only begins this discussion and opens up further questions that the research community should reflect upon. In particular, are we answering the questions that are important to people in their decision-making processes? What do we have to do to make our work change the hearts and minds of those who are making policy decisions? Future work should continue to investigate the kinds of information that is most important for the people that matter.

Finally, those within and outside of the policy-making arena should keep in mind that policies are not introduced into a neutral atmosphere where people have no prior beliefs. This research emphasizes an important reason for remembering this reality, in that people’s evaluations of the information they are given may be colored by what they already believe. Whether the imperative is that advocates and activists cater their messaging more toward people’s beliefs, or rather that they work to change the extent to which people are able to be more open about their beliefs is beyond the scope of this study. However, a more person-centered approach to the work of information-based policy making may yield results that are better aligned with our senses of what is right for policy and/or right for democracy.

Supplemental Material

DPTE_appendices – Supplemental material for Proving I’m Right: Charter School Policy and Selective Exposure to Information

Supplemental material, DPTE_appendices for Proving I’m Right: Charter School Policy and Selective Exposure to Information by Richard S. L. Blissett in Educational Policy

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Supplemental Material

Supplementary material is available online with this article.

Author Biography

Richard S. L. Blissett is an Assistant Professor of quantitative methods and education policy at Seton Hall University. His research primarily focuses on the attitudinal and behavioral dimensions of the politics of education policymaking. His research agenda is comprised of three overlapping fields of study: (1) information, values, and attitudes, (2) democratic structures in education governance (particularly, school boards), and (3) advocacy and social movements.

References

Arceneaux

Johnson

(2013). Changing minds or changing channels? Partisan news in an age of choice. Chicago, IL: The University of Chicago Press.

Berends

(2015). Sociology and school choice: What we know after two decades of charter schools. Annual Review of Sociology, 41, 159-180.

Berends

Watral

Teasley

Nicotera

(2008). Charter school effects on achievements: Where we are and where we are going. In Berends

Springer

M. G.

Walberg

H. J.

(Eds.), Charter school outcomes. New York, NY: Taylor & Francis.

Berinsky

A. J.

Huber

G. A.

Lenz

G. S.

(2012). Evaluating online labor markets for experimental research: Amazon.com’s Mechanical Turk. Political Analysis, 20, 351-368.

Betts

J. R.

Tang

Y. E.

(2011). The effect of charter schools on student achievement: A meta-analysis of the literature. Seattle, WA: Center for Reinventing Public Education.

Bryant

(2017, October 30). There is no “progressive case” for charter schools. Truthout. Retrieved from http://www.truth-out.org/news/item/42416-there-is-no-progressive-case-for-charter-schools

Caref

Hainds

Hilgendorf

Jankov

Russell

(2012). The black and white of education in Chicago’s public schools. Class, charters & chaos: A hard look at privatization schemes masquerading as education policy. Chicago, IL: Chicago Teachers Union.

Cremata

Davis

Dickey

Lawyer

Negassi

Raymond

M. E.

Woodworth

J. L.

(2013). National charter school study. Stanford, CA: Center for Research on Education Outcomes.

Dean

A. B.

(2014, January 13). Charter schools should educate, not discriminate. Al Jazeera. Retrieved from http://america.aljazeera.com/opinions/2014/1/charter-schools-shouldeducatenotdiscriminate.html

10.

Eidelson

(2014). Christie’s charter school nightmare: “White flight, and they’re bankrupting us.” Salon. Retrieved from https://www.salon.com/2014/03/13/christies_charter_school_nightmare_white_flight_and_they%E2%80%99re_bankrupting_us/

11.

Garcia

D. R.

(2010). Charter schools challenging traditional notions of segregation. In Lubienski

C. A.

Weitzel

P. C.

(Eds.), The charter school experiment: Expectations, evidence, and implications (pp. 33-50). Cambridge, MA: Harvard Education Press.

12.

Gerber

A. S.

Huber

G. A.

Doherty

Dowling

C. M.

(2011). Citizens’ policy confidence and electoral punishment: A neglected dimension of electoral accountability. The Journal of Politics, 73, 1206-1224.

13.

Hauser

D. J.

Schwarz

(2016). Attentive Turkers: MTurk participants perform better on online attention checks than do subject pool participants. Behavior Research Methods, 48, 400-407.

14.

Honey

Blissett

R. S. L.

Woo

(2016, November). Charter schools and local support for public school funding ballot measures. Association for Public Policy Analysis and Management Annual Meeting, Washington, DC.

15.

Huber

G. A.

Hill

S. J.

Lenz

G. S.

(2012). Sources of bias in retrospective decision making: Experimental evidence on voters’ limitations in controlling incumbents. American Political Science Review, 106, 720-741.

16.

Kruglanski

A. W.

Klar

(1987). A view from a bridge: Synthesizing the consistency and attribution paradigms from a lay epistemic perspective. European Journal of Social Psychology, 17, 211-241.

17.

Lodge

Taber

C. S.

(2013). The rationalizing voter. Cambridge, MA: Cambridge University Press.

18.

Lubienski

C. A.

Weitzel

P. C.

(2010a). Information use and epidemics in charter school policy. In Lubienski

C. A.

Weitzel

P. C.

(Eds.), The charter school experiment: Expectations, evidence, and implications (pp. 197-218). Cambridge, MA: Harvard Education Press.

19.

Lubienski

C. A.

Weitzel

P. C.

(2010b). Two decades of charter schools: Shifting expectations, partners, and policies. In Lubienski

C. A.

Weitzel

P. C.

(Eds.), The charter school experiment: Expectations, evidence, and implications (pp. 1-14). Cambridge, MA: Harvard Education Press.

20.

Miron

(2010). Performance of charter schools and implications for policy makers. In Lubienski

C. A.

Weitzel

P. C.

(Eds.), The charter school experiment: Expectations, evidence, and implications (pp. 73-92). Cambridge, MA: Harvard Education Press.

21.

Mullinix

K. J.

Leeper

T. J.

Druckman

J. N.

Freese

(2015). The generalizability of survey experiments. Journal of Experimental Political Science, 2, 109-138.

22.

Arsen

(2010). The competitive effects of charter schools on public school districts. In Lubienski

C. A.

Weitzel

P. C.

(Eds.), The charter school experiment: Expectations, evidence, and implications (pp. 93-120). Cambridge, MA: Harvard Education Press.

23.

Paolacci

Chandler

Ipeirotis

P. G.

(2010). Running experiments on Amazon Mechanical Turk. Judgment and Decision Making, 5, 411-419.

24.

Peralta

C. B.

Wojcieszak

Lelkes

de Vreese

(2017). Selective exposure to balanced content and evidence type: The case of issue and non-issue publics about climate change and health care. Journalism & Mass Communication Quarterly, 94, 833-861.

25.

Raudenbush

S. W.

Bryk

A. S.

(2002). Hierarchical linear models: Applications and data analysis methods. Thousand Oaks, CA: SAGE.

26.

Redlawsk

D. P.

Lau

R. R.

(2009, September 10-12). Understanding individual decision making using process tracing. Paper presented at the General Conference of the European Consortium for Political Research, Potsdam, Germany.

27.

Roth

McDaniels

Brown

Campbell

(2017, October 24). The progressive case for charter schools. Center for American Progress. Retrieved from https://www.americanprogress.org/issues/education-k-12/news/2017/10/24/440833/the-progressive-case-for-charter-schools/

28.

Teske

’Fitzpatrick

J. O.

Brien

(2009). Drivers of choice: Parents, transportation, and school choice. Seattle, WA: Center for Reinventing Public Education.

29.

Weitzel

P. C.

Lubienski

C. A.

(2010). Grading charter schools: Access, innovation, and competition. In Lubienski

C. A.

Weitzel

P. C.

(Eds.), The charter school experiment: Expectations, evidence, and implications (pp. 219-230). Cambridge, MA: Harvard Education Press.

30.

Wohlstetter

Smith

Farrell

C. C.

(2013). Choices and challenges: Charter school performance in perspective. Cambridge, MA: Harvard Education Press.

31.

Xing

Q. W.

Maugeri

Pierson

J. B.

Reitano

(2015). Financial impact of charter schools on traditional public schools in North Carolina: A cross-districts comparison. Raleigh: North Carolina Department of Public Instruction.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.14 MB