Enforcing the Rule of Law in the EU: Effects on Public Opinion*

How does the European Union's enforcement of democracy and rule-of-law standards infuence the domestic public's attitudes toward backsliding governments? On the one hand, enforcement actions by international organizations may increase the costs of supporting backsliding governments and provide informational cues about non-compliance with international norms. On the other hand, scholars and practitionersworry that enforcement actions create “rally-around-the-flag” effects that inadvertently increase support for backsliding regimes. We report descriptive, experimental, and quasi-experimental results from a survey designed to assess the public-opinion effects of EU action in response to rule-of-law backsliding in Poland. The survey results suggest both that Polish citizens perceive the EU as a main critic of measures undermining judicial independence and that these perceptions are correlated with opposing the targeted measures. We find no evidence that additional information about EU actions turns public opinion against the targeted measures but there is also no evidence of any rally-around-the-flag effect. In diagnosing the null findings from our experimental and quasi-experimental designs, we also highlight important challenges associated with using survey-experimental methods to assess the effectiveness of EU interventions in ongoing and salient public debates. *


Introduction
Part of the mission of regional international organizationssuch as the European Union (EU)is to help safeguard democracy, the rule of law, and human rights in their member states (Pevehouse, 2002a;Manseld and Pevehouse, 2006).The desire to lock-in liberal democracy and strengthen and safeguard domestic institutions were important reasons for why states joined and were invited to join such international organizations (Pevehouse, 2002b;Moravcsik, 2000).After they joined, newly democratized states continued to use their participation in international organizations to signal their democratic commitments and entrench liberal democracy (Grewal and Voeten, 2015) and, particularly since the 1990s, regional international organizations have been empowered with new enforcement tools to combat democratic backsliding in their member states (Emmons, 2021).
Yet, the evidence for international organizations eectively countering democratic backsliding in their member states is mixed at best (Poast and Urpelainen, 2015).Perhaps most strikingly, the EU was unable to prevent or reverse the authoritarian measures under the Fidesz government in Hungary after 2010 and the Law and Justice party (PiS) government in Poland between 2015 and 2023 (Laurent and Scheppele, 2017).The advent of authoritarian governments within the EU has prompted an important strand of scholarship seeking to explain the absence of stronger enforcement actions against backsliding member states (Kelemen, 2020;Emmons and Pavone, 2021;Sedelmeier, 2014).Yet, a crucial question concerns the extent to which eorts by the EU to enforce obligations to uphold the rule of law and democracy actually have (and will have) the desired effects.Although some scholars and observers have lamented the EU's failure to use the tools at its disposal to forcefully counter backsliding (e.g., Kelemen, 2022), other scholars warn that EU interference in domestic politics risks rallying the public in support of their backsliding governments (Schlipphak and Treib, 2017;Schlipphak et al., 2022).
We investigate the public-opinion eects of EU actions in response to a series of measures by the PiS government targeting the judiciary and undermining judicial independence and the rule of law in Poland (Sadurski, 2019).We combine insights from two strands of scholarship.First, we build on the literature seeking to explain voters' responses to the institutional reforms at the heart of contemporary backsliding episodes.This literature highlights both how uncertainty concerning government intentions prevents voters from punishing autocratic politicians (Chiopris, Nalepa and Vanberg, 2021;Staton, Reenock and Holsinger, 2022) and how polarization aects the tradeos voters make between pursuing policy objectives and safeguarding democratic institutions (Svolik, 2020;Graham and Svolik, 2020).To be eective, attempts at countering backsliding need to alter these domestic political dynamics in the targeted states.
Second, we build on the international relations literature concerned with how inter-national institutions promote compliance with international agreements (Dai, 2005;Simmons, 2009) to theorize how EU enforcement actions will impact the domestic politics of backsliding.This literature suggests that actions by international organizations with only weak enforcement powers can still be eective precisely because they provide the public with information about the authoritarian or unlawful nature of targeted policies and governments (Chaudoin, 2022) and strengthen the position of those domestic actors with international law on their side in domestic debates (Alter, 2014).For the EU with its comparatively strong enforcement powers, including the ability to withhold important membership benets, the latter eect should arguably be even stronger than for other international institutions.Accordingly, we expect information about EU enforcement actions to be associated with increased opposition against authoritarian reforms in targeted states.
The ability of EU enforcement actions to inuence the views of ordinary citizens is likely to be conditioned by their pre-existing political views (e.g., Sejersen, 2021).Socalled political moderates are particularly consequential for electoral accountability as they are more prone to change their political choices in response to relevant information.(Fowler et al., 2023).EU enforcement actions are most likely to inuence the views of citizens that are not already strongly in favor or strongly opposed to the sitting government or to the opposition.Such citizens may be more likely to update their preferences in light of new information either because they are on the fenceand therefore do not experience high costs of switching their allegianceor because they are not as politically engagedand therefore genuinely uncertain about the nature of the changes introduced to the judiciary.By contrast, citizens with strong views about the main political parties or the judicial institutions are more likely to already have made up their minds about the changes to the judiciary and to align themselves with their preferred political camp when forming a view about these changes irrespective of what they learn about EU enforcement actions.
The ability of the EU to inuence domestic public opinion through enforcement actions is also likely to be moderated by counter-mobilization from supporters of the incumbent government and the contestation that ensues (Chaudoin, 2016(Chaudoin, , 2022)).Crucially, most citizens will learn about EU enforcement actions through how they are presented in domestic media and by important opinion leaders (Brutger and Strezhnev, 2022;Chaudoin, 2022;Brutger, Chaudoin and Kagan, 2022).Counter-narratives oered by the incumbent regime or its supporters may raise considerable doubt both concerning the value of the signal sent by the EU through its enforcement actions and concerning the likely eects of such enforcement actions among ordinary citizens.They may also lead citizens to deantly rally in support of the targeted government and its reforms (Galtung, 1967;Grossman, Manekin and Margalit, 2018;Efrat and Yair, 2022).Accordingly, we expect exposure to government counter-narratives to attenuate, and perhaps even reverse, the public-opinion eects of EU enforcement action.
To assess how EU enforcement actions intended to safeguard the rule-of-law impact on the domestic political contestation over backsliding measures, we conducted a public opinion survey with an embedded vignette experiment.The survey was conducted in June 2023 during the run-up to the legislative elections in Poland in October 2023 and in a context of ongoing bargaining between the EU and the Polish governments over changes to the judiciary in order to secure billions of euros in EU grants and loans.Our vignette experiment was designed to estimate the causal eects of information about the EU's decisions to continue to withhold funds as well as information about counter-narratives advanced by the Polish government on public opposition to the contested measures, including their willingness to turn against the incumbent government in the upcoming elections to achieve a reversal of the measures.Furthermore, on June 8, 2023three days after we started elding the surveythe European Commission opened an infringement procedure against Poland concerning the so-called Lex Tusk, a Polish law establishing a State Committee for the Examination of Russian inuence on the internal security of Poland between 2007 and 2022 widely seen as targeting the political opposition.This unexpected event received high levels of media attention and potentially introduced quasi-experimental variation in respondents' information about the EU enforcing democratic norms that were portrayed as being threatened by the Polish government.We exploit that potential variation by using an unexpected event during survey design (Muñoz, Falcó-Gimeno and Hernández, 2020) to test whether the launch of the infringement procedure impacted on respondents who received the survey after the event.
Observationally, our results show that the EU is perceived by many Polish citizens as a strong critic of the Polish government's measures related to the judiciary.They also show that there is a relatively high level of opposition against the measures and while a large group of respondents are uncertain about the nature of the measures, there are few enthusiastic supporters of the changes to the judiciary.Among opponents of PiS, a relatively high share of voters reported that they perceived democracy and the rule of law as among the most important issues in the October 2023 elections that would oust the PiS government.Furthermore, perceiving the EU as a critic of the measures targeting the judiciary is at least weakly correlated with opposing these measures, indicating that EU critique is more likely to be associated with domestic opposition to rule-of-law backsliding than to support of such autocratic measures.While these observational results do not provide strong evidence for the eectiveness of EU enforcement actions they suggest that concerns about rally-around-the-ag eects are likely to be overstated.
Our vignette experiment provides no evidence of a causal relationship between exposure to information about EU enforcement actions and opposition to the judicial reforms.
Although our experimental design is in theory well-powered, we caution against interpreting these null ndings as evidence of a null eect: First, high levels of inattention in the survey experiment severely reduced the number of respondents that may realistically be considered as treated in the experiment.Second, and more critically for scholars' abilities to use survey-experimental methods to learn about the eectiveness of international organizations' enforcement actions against backsliding member states, we nd that most respondents are already well aware of the measures targeting the judiciary and of EU criticism of these measures.Thus, the ability to manipulate the information available to respondents in a vignette experiment is limited.
We also nd no evidence that (potential) exposure to news about EU infringement action concerning Lex Tusk aected the approval ratings of domestic or EU actors and institutions.In sum, our ndings indicate that after several years of contestation, including the EU withholding large sums of funding and strongly criticizing the Polish government for not living up to the standards of EU membership, in a highly polarized domestic political context, information about (one more) event is unlikely to shift public opinion either against or in favor of the government and its backsliding policies.

EU Enforcement Actions and the Domestic Politics of Backsliding Member States
Eorts by democratically elected governments to undermine institutional checks and balances are central to contemporary episodes of democratic backsliding (e.g., Ginsburg and Huq, 2018).The Polish experience is instructive.After winning both the parliamentary and presidential elections in 2015, the newly formed PiS government enacted several measures to drastically undermine the independence of the Polish judiciary (Sadurski, 2019).
The newly elected president Andrzej Duda refused to accept the oaths of judges appointed to the Constitutional Tribunal at the end of the previous legislative term.In their place, the PiS government instead appointed its own slate of judges.A series of institutional changes then secured the government's capture of the Constitutional Tribunal: The quorum was raised so that the PiS appointed judges would have to be included in judicial panels, a two-thirds majority became necessary for declarations of unconstitutionality, and parliament was empowered to dismiss judges by majority vote (Chiopris, Nalepa and Vanberg, 2021).
After achieving control of the Constitutional Tribunal, the PiS government proceeded with a series of measures targeting the rest of the judiciary.In particular, these measures have included introducing a new disciplinary chamber of the Supreme Court with the power to sanction and remove judges deemed to have engaged in political activity, taking control over the National Council of the Judiciary in charge of appointment of judges to various courts, including the Supreme Court, by replacing the 15 judicial members of the National Council with members selected by the Sejm, and, in late 2019, enacting a socalled muzzle law prohibiting judges from criticizing the judicial reforms or questioning the legitimacy of judicial appointments.As illustrated in Figure 1  A growing literature investigates the domestic political conditions that allow governments to get away with this type of backsliding.First, partisan considerations may be more important than deep-seated commitments to the rule of law and democracy in shaping citizens' attitudes toward judicial institutions (Bartels and Kramon, 2020).While citizens value judicial constraints on their political adversaries, they may still fail to punish a government they support for undermining the rule of law.Instead, they may celebrate how a more unconstrained government is free to deliver on its campaign promises.Po-larization further exacerbates this problem: As the ideological distance to the opposition increases, the cost for pro-government voters to shift their support to the opposition to punish violations of the rule of law increases (Graham and Svolik, 2020;Svolik, 2020).
One consequence is that at least part of the electorate may be willing to sacrice democracy and the rule of law in order to promote their own policy objectives and to frustrate the policy objectives of their opponents.Indeed, recent survey-experimental work from Poland suggests that while no part of the electorate has been particularly enthusiastic about measuring undermining checks and balances, supporters of the PiS government were willing to tolerate such reforms (Mazepus, 2022).Furthermore, apart from in the youngest segment of the electorate, information about how such measures violated ruleof-law standards did not make PiS voters regret their vote choice (Chiopris, Nalepa and Vanberg, 2021).
Second, voters may simply be uncertain about the extent to which incumbents and the reforms they advance are in fact undermining democratic norms and the rule of law (Chiopris, Nalepa and Vanberg, 2021;Staton, Reenock and Holsinger, 2022, 21).Such uncertainty may lead even voters that are committed to the rule of law and democracy to go along with authoritarian measures.In the Polish case, it is not unreasonable to expect that many citizensand particularly those that are ideologically aligned with the PiS governmentmay believe that reforms undermining judicial constraints were enacted in good faith, for instance to combat ineciencies in the judiciary and to root out alleged communist-era judges, or at least that the opposition's critique of the measures has been exaggerated.Indeed, framing such measures in terms of reforms intended to make the judiciary more ecient often provides useful political cover for interventions that ultimately undermine judicial independence and power (Levitsky and Ziblatt, 2018).In sum, both polarization along partisan lines and uncertainty about the true nature of the government's actions are likely to condition public attitudes toward backsliding.
For EU member states, attacks on domestic judiciaries are not only a domestic concern.Judicial independence is a fundamental value of the EU and a key condition for membership (Article 2 of the Treaty of the European Union).Although the EU has been criticized for not acting more quickly and forcefully to counter backsliding (Kelemen, 2022)

2
Are such enforcement actions helpful in countering backsliding?Drawing both on the previously discussed literature on public attitudes toward backsliding and on a rich tradition of scholarship concerning how international institutions promote compliance with international obligations (Dai, 2005;Simmons, 2009;Alter, 2014;Chaudoin, 2022), we propose that EU enforcement actions may be successful in tilting the balance in the domestic politics of backsliding states in two ways.
First, addressing the question of uncertainty about the true nature of the government's measures, EU enforcement actions may send an important signal to the Polish public that its government's measures targeting the judiciary are in fact undermining rule-oflaw principles and provide a focal point for resistance against the measures (Dai, 2005;Manseld, Milner and Rosendor, 2002;Simmons, 2009;Chiopris, Nalepa and Vanberg, 2021;Chaudoin, 2016).As argued by Chaudoin (2022), international organizations can act like an alarm that provides information on government practices.This helps subnational actors, like interest groups or voters, better demand that politicians implement their preferred policies.Thus, enforcement actions by the EU may help alleviate some of the uncertainty concerning the authoritarian nature of the measures, which is one important condition enabling backsliding.
Second, addressing the question of partisan preferences, enforcement actions by the EU or other international organizations may directly inuence tradeos voters make when deciding whether to punish ideologically close authoritarians for rule of law violations or pursue their policy preferences.Making benets of membership conditional on reversing backsliding measures may make voters concerned about preserving these benets more likely to withdraw support for measures they would otherwise be willing to tolerate.
Such positive eects may need to be balanced against deant reactions from parts of the public.In other contexts, sanctions and shaming from international actors have backred and rallied the public in support of the targeted government and their polithe Court of Justice of the European Union, see https://euruleoflaw.eu/rule-of-law/rule-of-lawdashboard-overview/polish-cases-cjeu-ect.cies (Grossman, Manekin and Margalit, 2018;Efrat and Yair, 2022;Cope and Crabtree, 2020).Citing such rally-around-the-ag eects, some scholars have warned that the EU should be very cautious about intervening in backsliding states since they may easily strengthen anti-EU and illiberal political forces at the domestic level (Schlipphak and Treib, 2017).For the case of EU enforcement actions against backsliding member states, there is, however, weak evidence of a public backlash against enforcement actions.among the Polish public.Cheruvu and Krehbiel (2022) nd that the Hungarian public is evenly split in their response to information about a CJEU judgment allowing the EU to withhold funds from Hungary for outing democratic standards and that only a small share of respondents would support withdrawing from the jurisdiction of the CJEU.
Thus, although we acknowledge that EU enforcement actions may alienate parts of the Polish electorate, we expect such detrimental eects to primarily aect citizens already supportive of the government and skeptical of European integration (Cope and Crabtree, 2020).On average, we therefore expect that the main eect of EU enforcement action is to strengthen the public's belief that the measures violate the rule of law and make voters more likely to oppose measures that jeopardize parts of Poland's rewards from European integration.Accordingly, Polish citizens should be more likely to oppose judicial reforms if they learn or are reminded that the EU has taken action to compel the government to reverse such measures: Hypothesis 1 Opposition to measures undermining the rule of law is increased by information about EU action condemning the measures.
It should be noted that we do not intend to adjudicate between the two mechanisms that we have identied as possible drivers of Hypothesis 1, that is, less uncertainty about the (un)democratic nature of the measures targeting the judiciary, on the one hand, and the potential costs generated by a loss of EU funding, on the other hand.
We further expect the eect of learning or being reminded of EU enforcement action to be greater than the eect of simply learning that the measures have been criticized by domestic actors, such as lawyers and judges.Compared to domestic critics in a polarized partisan context, EU enforcement action provides a strong signal that also authoritative third parties perceive the measures as a threat to judicial independence.Moreover, enforcement actions involve considerably greater economic costs than domestic critics can invoke.
Hypothesis 2 Opposition to measures undermining the rule of law is greater if citizens are exposed to information about EU actions condemning the measures than if they are only exposed to domestic criticism of the measures.
While we are skeptical of claims that EU enforcement actions will make the public rally in support of the PiS government, we expect the impact of enforcement actions on public opinion to be mitigated by the contestation and counter-mobilization they are met with from supporters of the Polish government and its backsliding policies.Enforcement actions by international organizations may not only provide information to uncertain voters and leverage to opponents of the regime.They also trigger counter-eorts and blame avoidence by incumbents and their supporters (Schlipphak et al., 2022;Chaudoin, 2016).
Understanding the moderating eects of government counter-narratives is crucial because most citizens will learn about EU enforcement actions through how they are presented in domestic media or by domestic opinion leaders (Brutger and Strezhnev, 2022;Brutger, Chaudoin and Kagan, 2022).While its dicult to predict the eect of concurrent exposure to competing frames (Avdagic and Sedelmeier, 2023, 4), existing research suggests that contestation between pro-and anti-compliance actors attenuates the public-opinion eects of enforcement actions (Chaudoin, 2022).Such attenuation is likely also in the context of EU enforcement of rule-of-law obligations, where targeted governments claim that the EU is overstepping its mandate and is undermining national sovereignty.For some parts of the public, counter-narratives may even succeed in triggering the rallyaround-the-ag eect discussed earlier and accordingly reinforce support for government reforms.
Hypothesis 3 Opposition to measures undermining the rule of law is lower if citizens are exposed to information about EU enforcement actions together with government counternarratives than if they are only exposed to information about EU action condemning the measures.
As discussed above, the degree to which EU enforcement actions shifts public opinion is likely to be conditional on pre-existing political beliefs.In particular, the marginal eect of EU enforcement action is likely to depend on how strong citizens' pre-existing views are (Baum, 2002, 266).Citizens already vehemently opposed to the incumbent government and its reforms may perceive EU enforcement as additional conrmation of their existing views, but may not signicantly change their views on reforms they already oppose (Svolik, 2021).On the other side of the spectrum, it may be dicult for the EU to convince the most zealous supporters of the government to withdraw support from government-championed reforms.The ability of the EU to inuence this group of voters has likely been further weakened by PiS voters becoming increasingly hostile to the EU during the period of rule-of-law backsliding (Cinar and Nalepa, 2022).
The ability of EU enforcement actions to make a dierence may therefore be stronger among those citizens that do not yet hold strong favorable or unfavorable views of either the government or the judicial institutions.Such political moderates are known to be particularly consequential for holding governments accountable because they are more likely to change their political behavior in response to relevant information than more partisan voters are (Fowler et al., 2023).Moderates may be genuinely uncertain about the nature of the measures and therefore update their beliefs more when learning about EU enforcement actions.Their views on the changes imposed on the judiciary may also to a greater extent be driven by pragmatic considerations (Sejersen, 2021), for instance relating to the economic costs of EU enforcement actions than by pre-existing partisan allegiances.
Accordingly, we expect support for hypothesis 1 to be strongest for citizens without strong, pre-existing favorable or unfavorable views about the government, the political opposition, or the judiciary.
Hypothesis 4 Opposition to measures undermining the rule of law is increased more by information about EU action condemning the measures among respondents who do not hold strong pre-existing favorable or unfavorable views about the government, the political opposition, or the Polish judiciary than among respondents with very favorable or unfavorable views concerning these institutions.

Research Design
To assess these hypotheses, we investigate how Polish citizens' views concerning the measures introduced by the PiS government with regards to the judiciary are inuenced by information concerning the ongoing battle between the EU and the Polish government over these measures.Specically, we embedded a well-powered and pre-registered vignette experiment in the survey to a representative sample of Polish citizens in June 2023, which was also the year of parliamentary elections in Poland (held on October 15, 2023) in which the PiS government was defeated by the opposition.
Investigating the consequences of EU enforcement actions for public opinion concerning the measures directed toward the judiciary is challenging: Both the measures and the conict with the EU had lasted for multiple years by the time of our survey and have received considerable media coverage (although a survey conducted by Toshkov et al. (2022) in 2022 suggests relatively poor knowledge of the details of EU enforcement actions in the Polish public).By contrast, information provided to respondents in a survey experiment provides only a very modest intervention (Sniderman, 2018).Respondents' strong pre-existing awareness of questions concerning democratic and rule-of-law backsliding is  concerns over democracy and the rule of law was primarily important with opposition voters, the gure shows that this issue was already high on the agenda and that many respondents had already made up their minds prior to our survey.
Our research design aimed to alleviate these challenges by zooming in on ongoing debates concerning the future of the disciplinary regime for judges and the link to important EU funding.Even if we expected that many respondents would already be well aware both of the general rule-of-law issues and about the EU withholding funds, fewer respondents might be expected to have already made up their minds concerning a relatively recent set of developments.Moreover, connecting the withholding of funds to the specic issue of the disciplinary regime requires more detailed knowledge than simply knowing about the criticism against the government's measures concerning the judiciary.The vignettes were therefore designed to manipulate the extent to which respondents perceive change to the disciplinary regime and access to EU funds as directly related and thus inuence their attitudes to the measures introduced with regards to the judiciary.
Because previous exposure to the conict between the EU and the Polish government concerning the changes to the judiciary is likely to limit our ability to manipulate perceptions of this conict, nding an eect of exposure to information about EU actions would provide strong evidence that EU actions inuence public opinion concerning backsliding measures.

Sampling Strategy
The survey was elded by Kantar Public in June 2023 using their KANTAR Proles panel in Poland.The panel includes 1,879,680 unique respondents.Kantar Public recruited 5007 respondents from their pre-existing panel using stratied random sampling with the aim of achieving representation of the Polish adult population in terms of gender, age, geography, and education.The sample corresponds relatively well to the population in terms of gender, age and geography, but diers signicantly from the population with respect to education (see Table A1 in the Supplementary Materials for details).People with higher education and vocational education are overrepresented while people with no education, primary education, or secondary non-vocational education are underrepresented.Older cohorts are also somewhat underrepresented in our sample.In our analysis, we adjust for these dierences between the sampling and the target population by using sampling weights provided by Kantar Public.

Vignette Experiment
Respondents were randomly assigned to one of four vignettes.All respondents received a brief description of the government's measures realted to the judiciary.Respondents in the control group received only this information (vignette 1).
Because we are interested in investigating the eects of EU enforcement actions beyond simply learning or being reminded that the measures are criticized for undermining judicial independence, we also include a vignette (vignette 2) in which respondents were informed that critical voices, including some judges and lawyers have criticized the measures, but without any information about EU enforcement actions.
The remaining two vignettes are designed to expose respondents to information only about EU enforcement action (vignette 3) and information about EU enforcement actions in combination with government counter-narratives (vignette 4).
In order to realistically represent dierent views as they are presented to the Polish public, we conducted a search of coverage of EU enforcement actions against Poland in six dierent Polish media, including both government-friendly and oppositional media.We analyze these data systematically elsewhere.For the purposes of this article, we use the media data to ensure that we use arguments, terms, and phrases that are in fact invoked in Polish media.This includes using the term reforms to describe the government's measures targeting the judiciary. 3 Vignette 1 The control group receiving no information about EU enforcement actions or domestic criticism of the measures received a Polish translation of the following vignette: Vignette 2 The second group received information also about criticism of the measures by judges and lawyers, but not about EU enforcement actions.The purpose of this vignette is to help us distinguish the eect of EU enforcement action from the eect of simply learning or being reminded that the measures are criticized for undermining the rule of law.
After taking power in 2015, the PiS government initiated a number of judicial reforms.The Disciplinary Chamber is one of the elements of these reforms.
The Disciplinary Chamber may discipline judges for engaging in public activities, including criticizing and protesting the judicial reforms.Critical voices, including some judges and lawyers, claim that these changes violate the rule of law.
Vignette 3 A third group received information about EU enforcement actions (but not the counternarratives questioning the legitimacy of the EU's actions): After taking power in 2015, the PiS government initiated a number of judicial reforms.The Disciplinary Chamber is one of the elements of these reforms.
The Disciplinary Chamber may discipline judges for engaging in public activ- Vignettes 13 follow a similar structure, but dier concerning the information they provided about judicial independence and EU action in response to the changes of the judiciary.Vignette 4 is structured dierently to better reect how the information would be portrayed in government-friendly media and to avoid priming the respondents with a pro-EU narrative prior to being exposed to the counternarrative.

Outcome Variables
Our dependent variables seek to measure opposition to the measures taken by the incumbent government with regards to the Polish judiciary.After respondents were exposed to the vignettes, we asked them both about their previous awareness of the measures and about four outcome variables: (1) Their support for or opposition to the reforms, (2) whether they agree or disagree that the reforms undermine judicial independence, (3) whether they agree or disagree that the reforms ought to be reversed, and (4) whether the reforms make it more or less likely that they will support the incumbent government in the upcoming parliamentary elections.
Figure 3 reports the questions asked and the distribution of the responses conditional on reported awareness of the measures.
About two-thirds of respondents reported being somewhat aware or very aware of the changes to the judiciary.Among the respondents with a stated preference, there is more opposition than support for the measures.However, there is also a signicant share of respondents that indicate support for the measures andwith the exception of the question of whether the measures have made judges less independentthere are also many respondents indicating that they do not know or prefer not to disclose their opinion.
For the main analysis, we dichotomize each outcome measure and run linear probability models on whether respondents indicated opposition to the measures targeting the judiciary.Since a large share of the respondents answered that they do not know or prefer not to disclose and providing information about the nature of the reforms is part  4 As noted, in section C of the Supplementary materials, we report results of models in which do not know or prefer not to disclose is instead coded as missing.The supplementary materials also report results in which the outcome variables are treated as interval level measures, with and without do not know or prefer not to disclose as middle category.These choices do not aect the conclusions from our analysis.The results are very similar across these dierent specications.

Heterogenous Treatment Eects
To assess hypothesis 4, we need to measure the strength of pre-existing approval and disapproval of the government.In addition to measuring approval and disapproval of the government, we also measure approval and disapproval of the main opposition party, of the Polish courts, and of the EU and EU institutions as attitudes toward these actors and institutions may similarly moderate the eects of our treatments.
We measure existing views by asking respondents to rate their approval of various actors and institutions in Polish and EU politics on a -10 to 10 scale.The actors/institutions and histograms of their approval ratings are displayed in Figure 4.The gure suggests that our sample is relatively hostile toward PiS and the (at the time of the survey) incumbent government (perhaps reecting how highly educated and younger people are overrepresented).By contrast, opinion concerning the EU institutions is more evenly divided.
When including these measures in the regression models, we take the absolute distances and divide them by 10 so that one-unit dierence reects dierences between respondents with a neutral view on the actor or institution and actors with a completely favorable or unfavorable view.

Other Covariates
We run both bivariate models and models that control for other covariates that may predict resistance to the judicial reforms.When testing hypotheses 1, 2, and 3, the only eect of the included covariates should be to increase the eciency of the estimates.In the models testing hypothesis 4, the covariates also help address omitted variable bias for our measures of the strength of pre-existing political views.
Specically, we control for gender, age, level of education, employment status, whether the respondent if they had to choose would prefer PiS or the main coalition of opposition parties KO, which issue the respondent perceives as most important for his or her vote choice in the next parliamentary election, the degree of satisfaction with the way that democracy works in Poland, and the degree to which respondents think Polish membership in the European Union is a good or a bad thing.

Estimation Strategy
For the main analysis, we estimate separate weighted linear probability models with HC2 standard errors for the dierent outcome measures.Because we are interested in generalizing the estimated treatment eects to the population of Polish citizens, we estimate all

Descriptive Results
Before presenting the results from the experimental and quasi-experimental research designs, we present a descriptive analysis of the factors that correlate with opposition to the changes to the judiciary.In contrast to the vignette experiment, these analyses were not pre-registered and should therefore be treated as exploratory.
Figure 5 reports the results from a set of linear probability models for each of our binary outcome measures with indicators for whether a respondent perceived each of the actors we listed as an important critic of the judicial reforms.All models control for reported vote choice in 2019, background characteristics, and the vignette respondents were assigned.Since one important eect of EU enforcement actions may be to help citizens form an opinion about the reforms and many respondents answer do not know, we prefer to treat do not know as no on the binary outcome measures.Coecients from these models are reported as blue dots in Figure 5.However, we also report results from models in which respondents indicating that they do not know are dropped from the analysis.
When treating do not know as not indicating opposition to the reforms, perceiving the EU as an important critic is correlated with a somewhat greater likelihood of opposing the reforms.This may be interpreted as evidence that EU enforcement actions are more likely to mobilize respondents against the government's measures, rather than generating a rally-around-the-ag eect.When dropping those respondents that answer do not know from the analysis there is no signicant correlation in either direction.It should also be noted that perceiving civil society as a critic of the government measures is much more strongly correlated with opposition to the measures compared to all the other actors, including the EU.This may be interpreted as an indication of the important role of civil society in mobilizing against rule-of-law backsliding.

Experimental Results
We now turn to the pre-registered analysis of the vignette experiment.Before respondents could proceed to answering the outcome measures, they had to conrm whether they read the vignettes.Respondents that indicated that they did not read the vignettes were encouraged to read the vignette before proceeding.

5
We measure the time it took from respondents having received the vignette and having answered both of these questions. Figure A1 in the Supplementary Materials displays the number of seconds respondents spent on the vignette before proceeding with answering the remainder of the survey.Most respondents read the vignettes only quickly before proceeding with the survey (median time spent is 9 seconds).
We use this information to estimate models in which we drop respondents that did not seem to engage with the vignettes.In accordance with the pre-registration plan, we estimate both models in which we do not drop any respondents from the analysis, models in which respondents in the lowest 10 percent quantile (which is only 2 seconds across 5 About 7 percent of the respondents reported not reading the vignette. Reforms ought to be reversed  vignettes) are dropped, and models in which respondents spending less than 10 seconds are dropped.Our expectation is that dropping respondents that spent very little time reading the vignettes should increase the estimated treatment eects.
Figure 6 reports the estimated eects of being assigned to vignettes 24 (with vignette 1 serving as the reference category) from a set of weighted linear regression models using our binary outcome variables (treating don't know as not indicating opposition to the measures).As can be seen from the gure, we nd no evidence that any of the treatment vignettes inuence opposition to the measures relative to the control group.In most specications, there are also no discernable dierences between the eect of being assigned to vignette 3 relative to vignette 2 or to vignette 4 relative to vignette 3. Hypotheses 13 are thus not supported.As shown in section C, these (null) results are consistent to not collapsing the categories on the outcomes measures and to excluding respondents that answered do not know or prefer not to disclose.
To assess hypothesis 4 we estimate models in which treatment assignment is interacted with each of the dierent absolute (dis)approval measures.If hypothesis 4 is correct, we would expect negative interactions between receiving vignette 3 and these absolute approval measures (i.e., the eect of receiving vignette 3 should be weaker among respondents with strong pre-existing attitudes).
Figure 9 displays the interaction terms from models including each of the binary outcome measures and each of the dierent (dis)approval items.We report the results only for respondents that spent at least 10 seconds considering the vignette, which severely reduces statistical power to detect these interaction eects.As can be seen from the gure, we nd no consistent support for hypothesis 4.There is some evidence that the eect of vignette 3 is weaker among respondents with stronger views on the sitting government and various PiS actors with respect to whether the reforms undermine judicial independence.However, considering the number of comparisons made and that we did not anticipate dierences between the dependent variables in the pre-registration plan, caution is warranted before interpreting these results as evidence in favor of hypothesis 4.

Engagement with the Vignettes
In order to gauge whether the null ndings can be trusted to indicate that information about EU enforcement actions do not impact public opinion, we conduct additional analyses on whether the treatments in the vignettes worked as intended.Share of respondents perceiving the actor as an important critic of the judicial reforms Actor One plausible interpretation is that many respondents are already familiar with EU criticism of the measures targeting the judiciary and that mentioning criticism of the measures may be sucient for many of the respondents receiving vignette 2 to recall that the EU is one of the critics (even if the EU is not mentioned in the vignette).This interpretation seems to be supported also by the fact that more exposure to information about the changes of the judiciary is associated with a greater likelihood of reporting that the EU is an important critic (with the possible exception of respondents receiving vignette 1, which did not mention criticism of the measures at all).
Both the relatively low engagement with the vignettes among the respondents and the ability of most respondents to recall that the EU has been a main critic of the reform is likely to reduce the ability of our experimental design to detect treatment eects of 6 The infringement action received high coverage in Polish media.
This unexpected event gives us an opportunity to complement the vignette experiment with an actual intervention, to further test the extent to which Polish citizens take cues from EU enforcement actions.If such enforcement actions are informative to people with regards to the democratic deciencies of the government's policies we would expect it to increase the opposition to the sitting Polish government and PiS.If, on the other hand, such enforcement actions lead citizens to rally around the ag we might expect exposure to this event to reduce approval of the EU and the Commission.
To evaluate whether the infringement action inuenced public opinion, we subset the dataset to the 2878 respondents interviewed between June 5 and June 12 and estimate a regression discontinuity with days since the data collection started as the running variable and a dummy for whether respondents were interviewed after the EU Commission had issued its letter of notice as the treatment variable.We include survey weights and control for background characteristics of the respondents (Muñoz, Falcó-Gimeno and Hernández, 2020).
The results from the linear regression models are reported in Figure 10.We nd no evidence that exposure to the infringement action inuenced approval or dissaproval of any of the relevant institutions.This provides additional evidence, complementing the null ndings from the vignette experiment, that a single publically salient EU enforcement action is unlikely to have a strong impact on Polish public opinion.

Conclusions
Democratic backsliding poses a fundamental challenge to the EU.While scholars have extensively studied and debated the response (or lack of response) of EU institutions to backsliding tendencies among some of the EU's member states, in particular Poland and Hungary, few studies have focused on the impact of EU enforcement actions on the domestic politics in backsliding states.This is unfortunate, as in the end it is up to the electorates in backsliding states to determine whether they accept the changes introduced 6 see https://ec.europa.eu/commission/presscorner/detail/en/ip_23_3134Estimated effect of being interviewed on or after June 8, 2023 This study has found no evidence that a single EU enforcement action, like an infringement case or a decision to withhold funding, is likely to shift public opinion toward opposing the measures targeting the Polish judiciary or the sitting government that implemented those measures.It also nds no evidence that such enforcement actions trigger a rally-round-the-ag eect.Despite the well-developed theoretical mechanisms that exist in the literature, including non-partisan vetting of information and trading o policy preferences for democratic values, both the vignette experiment and the event-during-survey analysis report null ndings in this regard.
Importantly, these ndings should not be interpreted as indicating that EU enforcement actions are meaningless with regards to shaping public opinion in this area.The fact that EU institutions for several years have been critical of the measures directed toward the Polish judiciary is well known among many Poles.In fact, our survey indicates that the EU is perceived as a critic of the reforms more often than any other actor, including the political opposition and civil society.Exposure to one additional event is likely to bring little extra information to many people.We also nd only weak and inconsistent support for our hypothesis that this information is more likely to have an impact on people with less strong views about the reforms and the sitting government.
This study does not take into account the cumulative eect on public opinion of the long period of increasingly strong critique and enforcement actions by the EU against the backsliding measures.We can only speculate about how the support for the measures, and the government that has implemented them, would have developed had the EU not engaged in any critique or enforcement actions during this period.The observational data in our survey indicate that there is at least a weak correlation between perceiving the EU as a strong critic of the reforms and being critical oneself of these reforms.While this is fairly week evidence that EU action mobilizes public opinion against the reforms, at least it indicates that the opposite scenario a rally-round-the-ag eect is even less likely.
Overall, there is nothing in our study that supports the argument that the EU needs to tread carefully in its response to what it perceives as democratic backsliding out of fear for inadvertently strengthening the domestic support for such actions.
, which displays annual scores on the Variety of Democracy project's Judicial constraints on the executive measure for Poland, the PiS government has succeeded in drastically weakening judicial constraints on the government since assuming oce in 2015.

Figure 1 :
Figure 1: Annual levels of judicial constraints on the executive in Poland according to the Variety of Democracy project , dierent EU institutions have taken actions to combat rule-of-law backsliding in Poland (Blauberger and Sedelmeier 2023).For instance, following infringement proceedings by the Commission, the European Court of Justice has ordered Poland to disband the disciplinary chamber for judges and has imposed daily nes of 1 million euros until Poland complies.1 The Commission has been withholding EU funds to help EU members 1 See C-204/21, Commission v Poland.For an overview of rule-of-law cases against Poland before recover from the Covid-19 pandemic until the Polish government reaches a set of milestones toward restoring judicial independence, including complying with the European Court Justice judgment ordering the disbandment of the disciplinary chamber (Morijn and Scheppele 2023).The Commission further warned that it considered withholding the even greater sums that Poland is due to receive from the EU's cohesion funds if judicial independence is not restored.According to one Commission spokesperson, the EU would withhold virtually all funds for Poland until we repair the judiciary.
Toshkov et al. (2022) nd relatively high levels of acceptance of EU sanctions, including the daily one million euro nes imposed by the Court of Justice of the EU (CJEU),

Figure 2 :
Figure 2: Responses to the question what issue do you see as most important in inuencing your vote choice in the next parliamentary election?by preferred main political party or coalition ities, including criticizing and protesting the judicial reforms.The European Union claims that these changes violate the rule of law.The Court of Justice of the EU imposed nancial penalties on Poland until the liquidation of the Disciplinary Chamber.In April 2023, The Court of Justice of the EU decided to continue to impose nancial penalties against Poland as Poland has not yet fully implemented the measures requested by the Court.The European Union is also withholding further funding until Poland reaches specic milestones regarding the independence of the judiciary.Vignette 4 A nal group of respondent received information about the EU enforcement actions and about government counter-narratives: The European Union and the Polish government are in conict over the judiciary reforms initiated by PiS since 2015.The Disciplinary Chamber is one of the elements of these reforms.The Disciplinary Chamber may discipline judges for engaging in public activities, including criticizing and protesting the judicial reforms.The European Union claims that these changes violate the rule of law.PiS denies this criticism which it sees as politically motivated.The EU is also withholding funds until Poland reaches EU milestones in protecting the independence of the judiciary.The PiS government says it has already reached these milestones to satisfy the European Union and that the Polish judiciary requires their reforms to depoliticize the courts, increase their eciency, and eliminate corruption.The government emphasizes that in April 2023, the Court of Justice of the EU agreed to reduce the nancial penalties imposed against Poland due to adjustments the government made to the judicial reforms.

Figure 3 :
Figure 3: Distribution of responses on outcome variables and reported awareness of the measures taken with regard to the judiciary 4

Figure 5 :
Figure 5: Coecients show the eect of indicating whether the EU and other actors are important critics of the measures on each of the binary outcome variables of resistance to the measures.The coecients are from linear probability models with survey weights, control variables, and HC2 standard errors.Error bars indicate 95 percent condence intervals.

Figure 7 :
Figure 7: Shares of respondents perceiving each actor as an important critic of the PiS government's changes to the judiciary

Figure 10 :
Figure 10: Estimated eect of exposure to opening of infringement procedure concerning Lex Tusk on approval of Polish and EU actors and institutions.Results from linear regression models with survey weights and HC2 standard errors.Only respondents interviewed between June 5 and June 11, 2023 are included.All models control for days since the eldwork started (the running variable), gender, age, education, employment status, vote choice in 2019, residence in a city or the countryside, and region.Error bars indicate 95 percent condence intervals.
Approval and disapproval of actors and institutions in Poland and the EU models with survey weights, calculated by the survey company Kantar Public to achieve representativeness on gender, age, education, and geographical area.
interferes with the democratic process.The activities of the committee, e.g., investigations and public hearings, risking to create grave reputational damage for candidates in elections and, by nding that a person acted under Russian inuence, could limit the eectiveness of the political rights of persons elected in democratic elections.
Commission announced that it opened an infringement proceeding against Poland for violating EU law with respect to the so-called Lex Tusk, a Polish law that according to the Commission: unduly