Public Opinion and the Legitimacy of Global Private Environmental Governance

The rise of global private environmental governance has inspired substantial research assessing whether organizations like the International Organization for Standardization (ISO) and the Forest Stewardship Council (FSC) are legitimate. These organizations address global challenges and help overcome collective action problems, but public opposition can severely curb their effectiveness. Yet, we do not know whether the public supports such organizations and perceives them as legitimate. This article draws on diverse political science literatures to outline why a focus on public opinion is important. The article tests two competing arguments explaining potential opposition toward organizations like the ISO and the FSC: accounts centered on the role of sincere preferences over the legitimate locus of authority and on the influence of domestic elite rhetoric. Results suggest that public opinion is generally positive and that elite rhetoric about a potential democratic deficit rather than simple information about the bodies’ governance structures decreases favorability.

The rise of private authority in global environmental governance has received considerable scholarly attention in political science and international relations (e.g., Cutler et al. 1999;Green 2014). Organizations such as the International Organization for Standardization (ISO) and the Forest Stewardship Council (FSC) perform important tasks overcoming collective action problems but have also been shown to produce environmental regulations with apparent distributive consequences (Büthe and Mattli 2011;Clapp 1998). Notably, these bodies' decision-making organs can easily be captured by industry interests and the private good and the public good are not necessarily equivalent (Taylor 2005). Hence, a central question in this line of research asks, How legitimate are the institutions involved in this new system of polycentric governance? This question has led scholars to scrutinize these governance actors' design and performance on a number of dimensions, including their compatibility with the ideals of deliberative democracy, accountability, and the openness of their participatory structures (e.g., Auld and Gulbrandsen 2010;Chan and Pattberg 2008;Gulbrandsen 2004).
This article goes beyond this normative debate by stressing the need to consider sociological legitimacy, that is, whether the public perceives the organizations as legitimate. To date, we know little about the public as a legitimating audience and whether there is public support for bodies such as the ISO and the FSC. My research asks, What are people's attitudes about organizations such as the ISO and the FSC developing global environmental standards? Do they perceive these bodies as legitimate? Furthermore, what factors shape their attitudes? To address these questions, I draw on lessons from the trajectory of public opinion toward the European Union (EU). I contend that private actors now play a significant role in global environmental governance, but critiques about their legitimacy have by and large been academic. Thus, bodies like the ISO and the FSC are not politically contested by the public. Consequently, we have not paid much attention to the public's attitudes toward these bodies or to implications for legitimation of the organizations. But public opinion can become central to the survival and effectiveness of actors involved in global governance once these bodies become politicized.
Acknowledging that public opinion matters invariably leads to the question of what might drive public opposition. I contrast two accounts of why these bodies become contested. On one hand, global nongovernmental actors may become contested by the public when it recognizes how much authority these organizations wield despite not being beholden to traditional democratic control. On the other hand, I hypothesize that the public does not hold such sincere preferences and that public backlash occurs when national elites frame this same information by adding affective tags suggesting that these bodies are undemocratic and illegitimate. Importantly, departing from partisan source cue (Mondak 1993;Zaller 1992) or motivated reasoning (Lodge and Taber 2013) accounts, my argument contends that people will be receptive to these messages irrespective of the partisan identity of the elite. Adjudicating between these competing conjectures helps shed light on how public opinion might affect the legitimation of global private environmental governance. The purpose of the empirical analyses in this article is thus threefold. First, I examine the extent to which people support bodies like the ISO and the FSC and whether they perceive them as legitimate actors in global environmental governance. Second, I am interested in whether it is basic information about the bodies' governance arrangements or frames highlighting a potential democratic/legitimacy deficit that alter support and perceptions. Third, I seek to identify correlates of this support.

The Legitimacy of Global Private Environmental Governance
The emergence of polycentric governance in global environmental regulation has led to the creation of a number of diverse private bodies that share an important characteristic: their roles are not prescribed or mandated by law (Black 2008). Nonetheless, they develop impactful rules and regulations through nongovernmental channels that often lack traditional democratic public participatory structures (Davis 2012), leading them to sometimes be labeled as nonstate market-driven governance systems (Cashore 2002). Organizations that can be categorized in this manner come in a variety of forms, but generally, they develop and implement standards and rules through the involvement of multiple stakeholders and often have a membership-based character, composed of representatives from national standard-setting organizations.
In this article, I explore attitudes toward global private environmental governance in general terms. However, to ensure that survey respondents are aware of these types of organizations, I provide them with information about ISO 14001 and FSC standards. While it is important to acknowledge that there are key differences between the ISO and the FSC, chiefly that the former is nonmarket based, whereas the latter is market based, both organizations are classified as private rule-making organizations in Büthe and Mattli's (2011) typology of international standard setting. Admittedly, the ISO and FSC may follow different public legitimation logics given that the FSC will be more reliant on consumer trust, whereas the ISO may be more reliant on expert and national stakeholder support. However, the commonality that I seek to leverage in this article is that they are both private organizations creating standards. Moreover, both organizations have received considerable scholarly attention (e.g., Auld and Gulbrandsen 2010;Cashore 2002;Clapp 1998;Gulbrandsen 2004).
Importantly, the rules and standards that such bodies develop and implement can have significant distributive consequences (Clapp 1998;Büthe and Mattli 2011). Unsurprisingly, this has led to scholarly interest in the legitimacy of such forms of governance. The scholarly debate on the legitimacy of such bodies can broadly be divided into two strands, one normative and one empirical or sociological in nature. The normative strand focuses on delineating the characteristics that an organization ought to have to be considered legitimate (Buchanan and Keohane 2006). Here institutions are evaluated on whether they adhere to some commonly embraced ideal, such as deliberative democracy (Bäckstrand 2006;Bäckstrand et al. 2010). By contrast, sociological or empirical legitimacy is attained when an organization's audience believes it has the right to rule (Buchanan and Keohane 2006). Naturally, there are also intersections between these two debates (see, e.g., Quack 2010). For example, transparency can increase perceptions of legitimacy if the underlying decision-making structures it shines light on approximate notions of deliberative democracy (De Fine Licht et al. 2014), and thus sociological legitimacy may follow from normative legitimacy.
However, my central concern is not whether bodies like the ISO and the FSC are legitimate in terms of achieving some normative ideal but rather whether they are considered legitimate by the public. This stands in contrast to the extant empirical literature that has focused on whether organizations are considered legitimate by those most directly affected, such as individual firms, industries, or interest groups (cf. Bernstein 2011; Bernstein and Cashore 2007;Cashore 2002;Falkner 2003).
The notion of the audience is crucial here. It makes a difference if one is focusing on whether actors are legitimate because states or other international organizations adopt their rules or whether firms do, for instance. Scholars have rightly pointed out that the legitimacy requirements will vary over time as audiences change (Bernstein 2011;Bernstein and Cashore 2007). Indeed, the audiences in most of these cases are still emerging, and as audiences increase in variety and number, disagreement over legitimacy will inevitably grow (Quack 2010).

The Importance of Studying Public Opinion
The lack of a global "public" has always posed a problem for legitimacy in global governance, and thus the audience for which legitimation is construed is usually not the public but rather more proximate actors, such as NGOs, firms, or specific industries (cf. Bernstein 2011;Bernstein and Cashore 2007).
Against this backdrop, I posit that a comprehensive empirical assessment of sociological legitimacy should include consideration of both public support for the organizations or system in question and public perceptions of their legitimacy. Factoring in public opinion dynamics is best motivated by an examination of the trajectory of public support for the EU. Prior to the 1990s, public opinion toward the EU was characterized as dormant and inconsequential. Due to its focus on supranational regulation and low political contestation among mainstream national political elites, which facilitated a period of "permissive consensus" (Lindberg and Scheingold 1970), elites were able to advance European integration without paying much attention to the public. Thus, public opinion was not consequential for elites' assessments of the EU's legitimacy.
Times have radically changed. Nowadays, the EU is in an era of "constraining dissensus" (Hooghe and Marks 2005), where elites cannot afford to ignore public opinion. Explanations for the rise of public opposition differ and often include economic considerations, identity-based explanations, or the types of institutional considerations that will be discussed later in this article. But what does the rise in public contestation over the EU have to do with the legitimacy of global private environmental governance? Clearly, the ISO and the FSC are quite distinct in institutional design and scope from the EU. More importantly, to date, the ISO and FSC are largely uncontested by the broader public.
However, the same argument about an absence of public contestation was made regarding the EU in the 1970s (Lindberg and Scheingold 1970). The early academic debate surrounding the EU's normative legitimacy was uncannily similar to the current discourse on the legitimacy of global private governance. For instance, there were extensive discussions regarding legitimation based on sound institutional features or assessments of organizational performance. Of course, the EU's institutional design, scope, and need for public legitimation is very different from the case of the ISO and the FSC. I am not suggesting that private environmental governance will follow the EU's trajectory but rather that legitimating audiences can change and that-as the results will show-support can be affected by elite rhetoric. It is possible that domestic elites seek to politicize and undermine organizations for domestic political gain or as part of populist and anti-elitist rhetoric. While the public may never be the central audience for the legitimation of private environmental governance, loss of public support could decrease the reputational value that firms derive from participation in standards. There is thus value to examining under what conditions people will support private governance in order to understand whether public perceptions are based on sincere preferences regarding institutional design or whether they are simply driven by elite rhetoric.
Indeed, this article does not seek to compare private environmental governance to the EU but rather it asks whether lessons can be drawn from the EU case. I draw on two potentially competing arguments: the authority transfer argument (e.g., Zürn 2018) and the elite framing perspective (e.g., Marks 2005, 2009).
On one hand, scholars have expanded the study of public opinion dynamics vis-à-vis supranational or transnational organizations by shifting the focus from the sui generis experience of Euroscepticism to the broader concept of "politicization" (e.g., Zürn 2016). I conceptualize politicization to mean when an issue becomes contested in the public domain and people believe that it should be subject to public deliberation. The idea then is to examine how bodies involved in global governance become politically contested. This literature posits that increased international authority leads to politicization. I thus test whether public support for organizations will decrease when the public notices that substantial authority has been transferred from the nation-state to global organizations. If the public has sincere preferences over the legitimate locus of political authority, then increasing awareness of authority transfer and loss of national democratic control should decrease support for bodies such as the ISO and the FSC, which in turn will raise new questions about their democratic and participatory credentials as well their legitimacy. This leads to the following conjecture: H1: Information about supranational organizations' authority and decision-making structure will decrease support for these organizations and increase perceptions that they are illegitimate and undemocratic.
On the other hand, Marks' (2005, 2009) "postfunctionalist" theory of European integration posits that political entrepreneurs played a central role in mobilizing public tension toward the EU. Their argument implies that people may not have had strong opinions about European integration-as opposed to rejecting any supranational authority on sincere grounds-but that political elites stoked rising antipathy toward the EU for political gain.
Moreover, evidence suggests that people do not universally reject expert or industry-involved decision-making in favor of popular democratic arrangements. In their book Stealth Democracy, Hibbing and Theiss-Morse (2002) find, contrary to the popularized view of citizens as eager to be more involved in politics, that Americans are quite content for nondemocratic actors to be involved in policy making. Citizens often yearn for more efficient policy making, including business and expert involvement instead of the traditional partisan conflict they have become accustomed to at the national level.
These insights lead me to posit a rather different expectation about the trajectory of public support toward global private environmental governance.
Here the public might be positively predisposed toward bodies such as the ISO or the FSC, yet such support could easily be undermined if people are exposed to frames suggesting that they are undemocratic or illegitimate. Thus, people, when exposed to the same information about institutional arrangements discussed earlier, will decrease support not based on this information but because of the affective tags provided by elites, which suggest that the organizations are undemocratic or illegitimate. Moreover, these cues can come from any domestic elite because the purported conflict is between the domestic and the supranational level rather than between partisan factions, which should reduce partisan motivated reasoning. That said, when these cues are explicitly linked to a partisan or otherwise trusted source, the expectation is that cues from copartisans/trusted sources will be especially effective (cf. Mondak 1993;Zaller 1992). In general, I thus hypothesize the following: H2: Elite frames claiming the existence of a democratic or legitimacy deficit will decrease support for these organizations and increase perceptions that they are illegitimate and undemocratic.
Besides these expectations about how information will affect attitudes, there are individual-level differences that should be associated with higher levels of support. Centrally, the stealth democracy rationale suggests the following: H3: Respondents who support expert and industry involvement in decision-making will have more favorable attitudes toward these organizations and perceive them as more legitimate.
Before examining attitudes toward global private environmental governance, an important question regarding nonattitudes must be addressed: do people know enough about the institutions involved to have formed opinions? Simply asking respondents their opinions of the ISO and the FSC could yield unstable or even invalid results given that these bodies do not enjoy widespread public recognition. The next section therefore outlines an experimental design that seeks to overcome this problem by providing respondents with sufficient information about the organizations in a baseline condition and then comparing responses in that condition with responses to information that was explicitly developed to operationalize the competing hypotheses.
To test these hypotheses, I conducted two experiments embedded in original surveys on US samples, which I discuss in subsequent sections. I then turn to observational analyses to further unpack the public opinion dynamics regarding private environmental governance.

Sample
The first study was fielded in April 2016 using respondents (n = 503) recruited through the crowdsourcing platform Amazon Mechanical Turk (MTurk). The survey was restricted to US-based adults, yielding a convenience sample that was 43 percent female, 17 percent nonwhite, and 20 percent Republican (48 percent independent and 32 percent Democrat). Research has validated MTurk as a subject pool that has a reasonable correspondence to the US population and is more diverse than traditional samples of college students (Berinsky et al. 2012). Moreover, research shows that experimental effects across twelve studies generalized from the original sample to MTurk samples (Coppock 2019).

Experimental Design
In the survey, respondents, first, answered a host of pretest questions, including demographic information and some political variables, ranging from trust in government to attitudes about climate change. Note that there was a split-split sample manipulation such that half the respondents answered questions about "stealth democracy" (details in the correlates section) in the pretest and half the respondents answered those questions in the post-test. 1 Then, respondents were randomly assigned to read a short vignette about global private environmental governance. After exposure to the treatment, respondents answered post-treatment questions before being provided with debriefing and compensation information.
The experimental component of the study seeks to reduce the problem of measuring nonattitudes by providing participants with varying amounts of information about the ISO and the FSC. The experiment consisted of four conditions, crafted to isolate pieces of information that were theorized to affect dynamics of support. First, participants in the No Information control condition did not receive any information about the ISO 14001 environmental management standards or the FSC. These respondents were told simply that they would answer questions about global governance. This condition allows examination of reactions in the absence of information.
Second, participants in the Basic condition read a brief text introducing two organizations involved in global nongovernmental environmental standard setting: The rise of global non-governmental environmental standard-setting There has recently been a rise in non-governmental organizations setting standards in the area of environmental regulation. For instance, the International Standards Organization's (ISO) 14001 environmental management standards provide a framework for companies to develop and implement effective environmental management systems.
Another organization, the Forest Stewardship Council (FSC) was founded in 1993 to provide a forest certification and labeling system. This non-governmental organization produces certification to show whether forest products come from well-managed forests.
The main purpose of this condition was to provide participants with a simple introduction to the ISO and the FSC. While these bodies are undoubtedly much more complex than presented and the vignette does not highlight institutional features that distinguish the two bodies, the intention was to highlight the bodies' nongovernmental nature as well as a sense of the scope of their functions. This condition will be used as the baseline in subsequent models.
A key question is whether citizens hold sincere preferences favoring domestic governments over unelected organizations and thus whether mere awareness of organizations' nongovernmental, stakeholder-driven decision-making is consequential for public opinion. Thus, the Governance condition augments the Basic condition by providing additional detail about the ISO and the FSC. Specifically, it adds the following text operationalizing authority transfer: Both the ISO and the FSC are non-governmental private standard setting bodies who develop and monitor standards based on industry and NGO input. Thus, decisions are largely made without formal government participation or interference.
This additional text reiterates that the ISO and the FSC are private bodies with decision-making structures that involve stakeholders, such as industry groups and NGOs, as opposed to decisions being made by national governments. While this is a very stylized and incomplete view of their decision-making structures, it operationalizes authority transfer and, according to H1, should depress popular support.
By contrast, the competing hypothesis states that information about governance arrangements, by itself, will not powerfully affect public support. Rather, elite affective cues about the legitimacy or illegitimacy of these bodies will have a greater effect. Thus, the Politicized Governance condition augments the text by introducing a possible critique of private governance: Some national politicians and activist groups have raised concerns about these organizations, suggesting that these private bodies are developing rules and regulations through non-governmental channels, thus raising serious questions about their democratic legitimacy. This manipulation was intended to add a politicization cue, thereby providing participants with the heuristic that there are concerns about a democratic or legitimacy deficit. While the claim is not particularly substantive in nature, it establishes a connection between private rule making and potentially lacking legitimacy.
The aim of the experiment is to ascertain whether exposure to any of these frames alters support for organizations that the public knows little about but that exert significant economic and political authority. Thus, while the information provided to respondents is incomplete and stylized, the distinctions between the conditions are grounded in theoretical predictions. To reiterate, examining differences between the Basic condition and the Governance condition operationalizes H1, whereas the differences between the Basic condition and the Politicized Governance condition operationalize H2. The No Information condition is not used to assess the core hypotheses but rather provides leverage on the non-attitudes question by enabling a comparison with respondents in the Basic information.

Measures
The key dependent variables capturing attitudes toward global private environmental governance were measured with three questions. For the first variable, Favorability, respondents were asked "How strongly do you favor or oppose global non-governmental bodies setting environmental standards?" Response options ranged from "strongly oppose" to "strongly favor" on a 5-point scale with a neutral midpoint. The second outcome, Legitimacy, was based on the question "How legitimate do you think global non-governmental environmental standard-setting bodies are?"-again measured on a 5-point scale with response options ranging from "very illegitimate" to "very legitimate." The variable Democracy was measured by asking "How democratic do you think global non-governmental environmental standard-setting bodies are?" Again, responses were measured on a 5-point scale ranging from "very undemocratic" to "very democratic." There was no "don't know" option for any of these questions.
These three variables are conceptualized as tapping into the broader concept of public support toward bodies like the ISO or the FSC. Factor analysis revealed that the three variables load onto one common factor and that combining them into a single, additive measure, the Favorability Index, produced a reliable scale (α = 0.80). Note, however, that because the order of these variables was not randomized, it is possible that there are order effects whereby responses to the Favorability item affect the subsequent items and thus general support for the organizations may anchor responses regarding Legitimacy and Democracy. While I consider this index a general measure of support, I also present results for the individual items and discuss any differences (all variables are recoded 0-1).

Results
Given the lack of prior research on the organizations studied, before examining the experimental results, some descriptive questions merit discussion: Does the public support bodies like the ISO or the FSC setting environmental standards? Does it perceive them to be legitimate and democratic? The overarching takeaway is that public opinion appears to be quite positive. I first concentrate only on respondents in the Basic condition. Here, 61 percent of respondents have either a very favorable or somewhat favorable opinion of global nongovernmental bodies setting environmental standards, compared to only 12 percent who have a negative opinion. Similarly, 72 percent of respondents perceive the organizations as legitimate, while only 12 percent find them illegitimate. Moreover, 54 percent of respondents perceive the bodies as either very or somewhat democratic, compared to 17 percent who perceive them as undemocratic. Note, however, that MTurk samples are more liberal, younger, and more politically engaged than the general population and thus these point estimates should not be interpreted as population estimates. I discuss this limitation in the concluding section.
Turning to the experimental results, I first examine effects vis-à-vis the Basic condition by regressing the dependent variables on an indicator for the experimental conditions where the excluded category is the Basic condition. Note that the model also includes covariates to improve efficiency. 2 These results are presented in Figure 1.
The first hypothesis posited that respondents in the Governance condition would be less supportive than respondents in the Basic condition. This conjecture captured the argument that individuals would reject global private governance simply as a result of learning about the governance structure (i.e., private bodies through stakeholder input making decisions as opposed to governments). Hence, we would expect a negative and significant coefficient for the Governance condition. However, this conjecture is not supported by the data, as the difference between the Governance condition and the Basic condition is not statistically significant for any of the four dependent variables. Thus, individuals do not appear to reject bodies like the ISO or the FSC simply on the grounds that they are nongovernmental and market driven.
The second hypothesis stated that cues about a democratic deficit would lead to lower levels of support. Here, I first compare the Politicized Governance condition to the Basic condition. For each dependent variable, there is a clear statistically significant decline in positivity when comparing the Politicized Governance condition to the Basic condition (all at p < 0.05). While I argue that the central comparison is between the Basic condition and the Politicized Governance condition, I also examine the difference between the Governance condition and the Politicized Governance condition to see whether these are actually significantly different from one another. Here, the results are weaker and the difference between the Politicized Governance condition and the Governance condition is 2. See Table A1 in the appendix for full model specifications and models without covariate adjustment.
only statistically significant (at p < 0.1) for the Democracy and Favorability variables (see Table A2 in the appendix, https://www.mitpressjournals.org/doi/ suppl/10.1162/glep_a_00539). Lastly, there was no explicit expectation about the difference between the Basic condition and the No Information condition. The results are nonetheless of some interest. While the means in the Basic condition are higher across all four variables, only the difference for the index and democratic evaluation reaches statistical significance ( p < 0.05). This may suggest that people's conceptions of the bodies are generally positive, and thus they align very closely in the condition where people base their evaluations on no information and the condition where people receive basic information. However, this could also be construed as evidence that people do not have fully formed opinions and that the information provided in the Basic condition does not really augment understanding. I return to this question in the discussion.

Study 2
The study 1 results provide initial evidence that information about organizations' governance structures does not diminish public support. Rather, elite Ordinary least squares (OLS) regression coefficients and 95 percent confidence intervals. Baseline category: Basic condition; underlying models can be found in Table A1 in the appendix. rhetoric claiming that these organizations are illegitimate depresses support. While informative, the first study has some limitations that I seek to address in a second study. First, it is unclear whether the findings are generalizable beyond the MTurk sample. Second, there is a need to unpack whether it matters which domestic elites are politicizing private governance organizations. In addition, I seek to increase internal validity by reducing imbalances in vignette length and more precisely operationalizing authority transfer.

Sample
For the second study, 3 I contracted with Survey Sampling International (SSI) for a sample of US adults balanced by age, sex, ethnicity, and census region. The survey was fielded in February-March 2018 and yielded a sample of 1,012 respondents. The sample was 51 percent female and 73 percent white. The median age was forty-two years, and 64 percent of respondents reported having at least some college education. Including leaners with partisans, the partisan makeup of the sample was 45 percent Democrat, 39 percent Republican, and 16 percent independent. These respondents may still be more highly educated and more politically engaged than the general population; therefore, I discuss sample limitations and implications for results in the conclusions.

Experimental Design and Measures
The basic experimental setup mirrors that of the first study. The No Information and Basic conditions are largely equivalent to the ones used in the first study (full vignettes are provided in the appendix). 4 The Governance condition was modified to more precisely operationalize authority transfer by highlighting that "standards are largely negotiated and implemented without formal participation or ratification by national governments." The vignette thus more explicitly emphasizes loss of governmental authority in the domains of negotiation, ratification, and implementation.
The most significant change concerns the inclusion of a broader variety of conditions containing elite rhetoric. Specifically, I operationalize politicization with three different conditions whereby the politicization cue is attributed to either journalists (Politicized Media condition), Democratic politicians (Politicized Democrats condition), or Republican politicians (Politicized Republicans condition). Compared to the first study, the length of these conditions was significantly reduced, and the cue explicitly stated that the cue giver "condemn[ed] the organizations as undemocratic and illegitimate." These politicization conditions enable testing of whether any main effects differ depending on who is delegitimating private governance organizations and whether reactions are responsive to messages from any domestic elite or whether respondents are simply reacting in ways consistent with work on partisan identities and motivated reasoning (cf. Lodge and Taber 2013;Zaller 1992).
The dependent variables are the same as in the first study. These variables again load on a single factor, and combining the Favorability, Legitimacy, and Democracy questions into an additive index yields a reliable scale (α = 0.78). For analyses broken down by partisanship, leaners are included with partisans.

Results
First, the descriptive results in this more representative sample largely confirm those of the MTurk sample. Focusing on the Basic condition, we see that opinion is generally favorable, with 60 percent of respondents having a very favorable or somewhat favorable opinion compared to 10 percent of respondents who hold an unfavorable view. Figure 2 displays the coefficients on the indicators for the experimental conditions compared to the Basic condition (see Table A3 in the appendix for full model specification and models without covariate adjustment).
Recall that H1 suggested that support in the Governance condition would be lower than in the Basic condition. Replicating the results from the first study, I find no support for this conjecture. Across all four dependent variables, I find OLS regression coefficients and 95 percent confidence intervals. Baseline category: Basic condition; underlying models can be found in Table A3 in the appendix. no evidence that providing respondents with information suggesting a lack of governmental control over negotiation, ratification, and implementation of environmental standards depresses support.
By contrast, Figure 2 does provide support for H2 and the conjecture that elite cues affect support for global private environmental governance. Specifically, respondents in the Politicized Media and Politicized Democrats conditions reported significantly lower favorability than respondents in the Basic condition (this effect is significant at p < 0.05 for all outcomes apart from the Democratic outcome in the Politicized Democrats condition, in which it is only significant at p < 0.1). Interestingly, contrary to expectations, this result was not replicated for respondents in the Politicized Republicans condition. I discuss potential reasons for this finding later. Moreover, the Politicized Media and Politicized Democrats conditions also differ significantly from the Governance condition for two and three of the dependent variables, respectively (albeit at p < 0.1 in two cases; see Figure A1 and Table A4 in the appendix).
Although my main conjecture was that the three politicization conditions should yield significant negative main effects, 5 I also suggested that there may be heterogeneity in the effects by respondents' partisanship based on research on source cues and partisan motivated reasoning (e.g., Mondak 1993; Lodge and Taber 2013). 6 Therefore, in Figure 3, I estimate treatment effects among Democrats and Republicans.
The left panel of Figure 3 shows that Democratic respondents exhibit decreased levels of support in both the Politicized Media and the Politicized Democrats conditions compared to the Basic condition. There is no effect of the Politicized Republicans condition for these respondents, suggesting that while partisans respond most strongly to copartisan cues, there is no evidence of a backlash effect whereby support increases in response to a delegitimating cue from the out-party. Turning to Republican respondents, first it is noteworthy that providing Basic information actually increases support compared to the No Information condition. While none of the other conditions affects support among Republican respondents at the p < 0.05 level, there is suggestive evidence, at the p < 0.1 level, that the Politicized Republicans condition leads to lower support measured by the index and lower perceptions of Democracy. One reason for this weaker effect may be the fact that the domain of regulation is the environment, which is a more liberal issue, thereby making a Republican cue less credible. Or it may be that this information is already "priced in" by Republican respondents given their more negative baseline attitudes toward environmental regulation. Moreover, I cannot reject the null hypothesis that the coefficients for Democrats are statistically distinguishable from the coefficients for Republicans. 5. When the three politicization conditions are pooled, I also observe significant differences between this pooled condition and the Basic condition for two of the four dependent variables. 6. There also are no significant differences between the Basic condition and the Politicized conditions among independents, but I caution against interpreting this result given the small sample size (n = 161).
Taken together, these results indicate that while respondents are most receptive to cues from trusted sources, out-party cues do not produce opinion backlash. In sum, across two samples and different operationalizations of both authority transfer and elite rhetoric, no evidence was uncovered to support the hypothesis that information about private environmental governance organizations' authority is sufficient to decrease public support. By contrast, the results highlight the power that domestic elite rhetoric plays in delegitimating private environmental governance.

Correlates of Support for Private Environmental Governance
This section complements the above experimental analyses by examining correlates of support. The purpose of this section is twofold. First, it allows testing of H3, which posited that stealth democracy attitudes are related to support. Second, given the lack of prior empirical work on public support for bodies like the ISO and the FSC, this section provides additional insight into the structure of these attitudes.
Specifically, I examine the correlates of the Favor Bodies Index. The regression models include basic demographic variables such as age (in years), dummy variables denoting nonwhite respondents and sex, and education level. Moreover, the models include factors that should arguably affect support for private environmental regulatory bodies based on existing work on determinants of domestic institutional trust and global governance more broadly. These variables include party identification on a 7-point scale ranging from "Strong Republican" to "Strong Democrat," support for limited government, trust in government, and a variable measuring how concerned respondents are about climate change. All continuous variables (with the exception of age) are rescaled to run from 0 to 1. Importantly, my theoretical account for why respondents might not inherently reject this form of governance was informed by the work on stealth democracy in the domestic US context. Therefore, I include a variable to test whether respondents holding such attitudes are indeed more favorably predisposed toward bodies like the ISO and the FSC. The variable, Stealth Democracy, is operationalized using the four-item agree-disagree battery originally developed by Hibbing and Theiss-Morse (2002). The four items tap into a set of beliefs associated with the desire for efficient, noninvolved government. The items are (1) "Elected Officials would help the country more if they would stop talking and just take action on important problems," (2) "What people call 'compromise' in politics is really just selling out on one's principles," (3) "Our government would run better if decisions were left up to successful business people," and (4) "Our government would run better if decisions were left up to non-elected, independent experts rather than politicians or the people." Unfortunately, although these items are generally positively correlated, the scale reliability coefficients for this index were quite low (Cronbach's α = 0.42 in study 1 and α = 0.56 in study 2). 7 This low reliability is consistent with work showing that people can selectively embrace stealth democracy traits and that support for the expert and support for the business components of the scale do not necessarily go hand in hand (Fernández-Martínez and Fábregas 2018), which is why some work has sought to examine support for expert decisionmaking and business involvement independently. Therefore, results are also discussed in disaggregated form to facilitate interpretation.
The results of two OLS regression models are presented in Table 1, with MTurk (study 1) results presented in the first column and SSI (study 2) results presented in the second column. 8 The first set of coefficients simply reproduces the effects of the experimental conditions on the Favor Bodies Index discussed previously.
Turning first to the demographic factors, we see slight differences between the two models. While age, education, race, and sex do not appear to be correlated in the MTurk study, we do see that younger and more highly educated 7. Responses to these questions were measured on 4-point scales in study 1 and 5-point scales in study 2. 8. The numbers of observations in these models drop by nine observations in the MTurk sample and seven observations in the SSI sample due to nonresponse on covariates (chiefly partisanship). respondents have more favorable attitudes toward bodies like the ISO and the FSC. These results also speak to the generalizability of the experimental point estimates and highlight that the slightly younger and more educated sample may be inflating levels of support. While this poses no threat to the internal validity of the experiments, I return to the question of external validity in the conclusions.
Partisanship is associated with favorability, with Democrats being more supportive than Republicans, ceteris paribus. Somewhat surprisingly, the Limited Government variable, which captures agreement with the belief that the government is taking on responsibilities that should be the purview of individuals and businesses, is not significant. Moreover, Trust in Government is positively associated with support. This latter finding indicates that, contrary to some suggestions, there may not necessarily be a tension between accepting global nongovernmental regulation and support for national governmental political institutions. Unsurprisingly, concern about climate change is positively correlated with support for global private environmental governance.
Notably, as conjectured in H3, there is a strong and statistically significant positive relationship between holding views that are indicative of stealth democracy and favorability toward bodies like the ISO and the FSC setting environmental standards. Owing to the low scale reliability of the Stealth Democracy Index, this model was also run with the four distinct variables that make up the index. In both studies, the results of the index are driven by the "take action" question and the "experts" question. Specifically, believing that elected officials "should stop talking and just take action on important issues" is associated with greater favorability, as is the desire to leave decisions up to independent experts. Conversely, there is no relationship between holding the belief that businesspeople should be entrusted with decisions and favorability.

Discussion
What do these results imply? Primarily, they suggest that people will not reject global bodies simply after learning about the nongovernmental nature of their authority. Rather, it seems that attitudes turn more negative when people are exposed to frames providing affective cues suggesting that the bodies are not democratic or legitimate. More broadly, these findings suggest that awareness of organizations' increased authority is not a sufficient condition for declining public support. Indeed, while the transfer of authority may be an issue that can be made salient and consequential through elite frames, the results here suggest that mere awareness of authority transfer does not, by itself, depress support, thus highlighting the role that political entrepreneurs play in driving opinion dynamics about global governance organizations (cf. Zürn 2018).
The results also provide some preliminary insight into the question of nonattitudes. In general, support for organizations was higher when respondents were provided with some basic information compared to when they were simply asked to render an opinion about the ISO and the FSC. Moreover, the observational results show that attitudes toward global private environmental organizations are related to other attitudes in meaningful ways. In particular, the research suggests that there is value in applying the concept of stealth democracy to attitudes about global governance.
The partisan dynamics uncovered in this article raise interesting questions for future research about the power of elite frames conditional on underlying partisan attitudes. Republicans are less supportive of global private governance than Democrats, and we do not see any significant experimental results for this subset of respondents. One reason for this may be that for Republicans this negative information is already "priced in," thus frames providing affective cues are less likely to move opinion.

Conclusions
As private actors have become important players in global environmental governance, questions of legitimacy have been central. This article has suggested that both the normative and the empirical literatures on the legitimacy of global governance need to pay more attention to public attitudes toward a broader set of organizations. While some private standard-setting bodies may never be dependent on public legitimation, the EU example demonstrates that metrics of legitimation can easily change when bodies become politicized. More specifically, assessments of legitimacy that rely on an audience legitimation component have to take into account that organizations' audiences are not set in stone and can easily expand to include the general public.
Descriptively, results from two surveys suggested that the public is generally supportive of bodies like the ISO and the FSC being involved in setting environmental standards. These findings are noteworthy because they provide initial evidence that, in the absence of politicization, there is no inherent opposition toward this form of governance. Moreover, results from two experiments suggest that providing people with information about organizations' nongovernmental, industry-involved decision-making structures does not reduce support. By contrast, the results highlight the role that elites can play in delegitimating such bodies.
Nonetheless, it is important to acknowledge some limitations. First, the samples are not representative, and while the SSI sample is a better match to the US population in terms of partisanship than the MTurk sample, both samples are likely to include people who are more politically engaged/aware than the general population. Given that these samples arguably skew liberal and more educated, the point estimates may overstate support and potentially provide an upper bound of support. Second, this article solely examined US public opinion dynamics. Thus, future work using both more representative and crossnational samples is needed to make more precise and reliable inferences about levels of support and to test whether the experimental effects generalize beyond the American case.
Furthermore, as I tested opinion toward the FSC and ISO jointly, the treatments did not provide information regarding variation between the two organizations, which also could affect legitimation. While the FSC is a market-based, nongovernmental organization producing voluntary certification through its eco-labeling scheme, the ISO is a nonmarket-based organization whose membership includes bodies that have ties to domestic governing institutions. This variation also means that legitimation audiences and the level of authority transfer invariably differ between the two organizations, thus the authority transfer argument may be more relevant in the case of the ISO compared to the FSC. While this article examined attitudes toward private governance in general terms, future work should further leverage this variation and seek to tease out whether and how such variation in institutional design is consequential for public support. Such research could, for instance, unpack whether market-based (e.g., FSC) and expert-based (e.g., ISO) notions of authority produce different levels of support.
What are the implications of these findings for the debate surrounding the legitimacy of global private environmental regulation? First, the results suggest that, at present, public opinion may provide a "permissive consensus" to the actors involved in private environmental governance. Moreover, this support may not necessarily decline when people learn more about the organizations' decision-making structures and authority. Second, support could decline if national elites seek to increase contestation. Either way, empirical accounts of legitimacy should be cognizant of the role that the public could yet come to play in legitimating global private governance.
In conclusion, an increasingly interconnected world and global challenges require coordination beyond nation-states, and the rise of private governancewhether in the environmental domain or more generally-is merely one facet of contemporary global governance. The myriad organizations involved in global governance, from the bodies discussed in this article to more established international actors such as the EU, increasingly need to be viewed as legitimate by the public to remain effective in addressing complex global challenges (Buchanan and Keohane 2006;Tallberg and Zürn, 2019). Understanding when institutional arrangements will be supported and perceived as legitimate by the public as well as whether delegitimating elite rhetoric can be counteracted presents an important avenue for future research.