1. Introduction
Politicians perceive public opinion as more conservative than it is both in the U.S. and in Western democracies (Broockman and Skovron, Reference Broockman and Skovron2018; Pilet et al., Reference Pilet, Sheffer, Helfer, Varone, Vliegenthart and Walgrave2023). The existence of “conservative bias” in the way political elites perceive constituency opinion has important implications for the quality of democratic representation. However, less is known about whether this case of pluralistic ignorance—i.e., incorrect beliefs about what others believe—is peculiar to political elites, or if it extends also to voters. Such second-order beliefs have been shown to exert influence on the way people process political information (Ahn et al., Reference Ahn, Huckfeldt and Barry Ryan2014), form and validate their own opinion (Festinger, Reference Festinger1950), vote (Huckfeldt and Sprague, Reference Huckfeldt and Sprague1995; Feddersen and Pesendorfer, Reference Feddersen and Pesendorfer1996), and behave. For instance, the perception that public opinion is more conservative than it actually is can trigger inaction on behalf of individuals holding progressive views through pressure to conform to status quo norms (or perceived inability to challenge them)—i.e., the “spiral of silence” (Taylor, Reference Taylor1982; Noelle-Neumann, Reference Noelle-Neumann1993). This tendency is also likely to affect politicians’ information environment, potentially impacting the policymaking process.
I examine whether public opinion exhibits conservative bias by using abortion rights in the U.S. as a case study. Abortion is the prototypical culture war issue and is argued to play a crucial role in shaping social identity (Goren and Chapp, Reference Goren and Chapp2017). Moreover, it gained renewed salience with the overturning of Roe v. Wade and the subsequent modifications in state-level abortion legislation. Salience and the consequent availability of public opinion poll data support the expectation that people would demonstrate a high degree of accuracy in their assessments of public opinion on the topic. However, the deeply morally rooted nature of beliefs about abortion rights coupled with a highly polarized political debate predict the presence of pluralistic ignorance. Previous findings indicate that public opinion was perceived as more supportive of abortion access immediately after the Dobbs decision compared to the previous period (Clark et al., Reference Clark, Levy Paluck, Westwood, Sen, Malhotra and Jessee2024), but it remains unclear if people hold accurate views about the beliefs of others on this issue.
Survey data collected approximately seven months after the overturning of Roe v. Wade from a sample representative of the U.S. adult population on key demographic characteristics (n = 1,536) indicates that individuals underestimate public support for abortion rights. To further explore this instance of pluralistic ignorance, two analyses were conducted: first, I looked at the factors that correlate with accuracy in the perception of public support for abortion access; second, I examined individuals’ perceptions of how different their own views are from the public’s view (“self-positioning”). The first analysis reveals that such underestimation is primarily driven by individuals identifying as “pro-life.” However, when looking at self-positioning, both subjects identifying as “pro-life” and “pro-choice” appear to be similarly aware that their opinions are more extreme compared to the average. This result sheds light on how egocentric bias—also referred to as “false consensus effect,” in which people overestimate the commonality of their views (Ross et al., Reference Ross, Greene and House1977; Krueger and Clement, Reference Krueger and Clement1994)—operates in practice. Indeed, it suggests that the false consensus effect may exist even when subjects are self-aware that their opinions are more extreme than the average. Moreover, these findings can deepen the comprehension of the mechanisms behind forms of conservative bias held by political elites. Prior research has shown that conservative voters tend to be more vocal with their representatives compared to liberal ones (Broockman and Skovron, Reference Broockman and Skovron2018; Pereira, Reference Pereira2021). My results suggest that conservative subjects are also the ones who are more likely to misperceive public opinion in a conservative direction. The combination of the two factors is likely to play a crucial role in creating a skewed information environment for politicians. Consequently, these findings point to practical implications for political developments surrounding abortion politics. Underestimation of public support for abortion access among voters could empower already antiabortion-oriented states to implement further restrictions and reduce legislative actions to protect abortion rights in supportive states.
2. Misperceptions of public opinion
Broockman and Skovron (Reference Broockman and Skovron2018) provided evidence of a systematic overestimation of how conservative constituency opinion is by elite politicians. Pilet and colleagues further investigated the topic outside the U.S., documenting consistent right-skewed misperceptions (2023). A similar bias was found among U.S. Members of Congress’ senior legislative staff (Hertel-Fernandez et al., Reference Hertel-Fernandez, Mildenberger and Stokes2019).
This tendency, often labeled as “conservative bias,” represents a case of pluralistic ignorance, defined in social psychology as a systematic misperception of public opinion (Allport, Reference Allport1924; Shamir and Shamir, Reference Shamir and Shamir1997). The mechanisms proposed to explain this phenomenon include conservative voters being more vocal with their elected representatives and elite politicians projecting their own preferences onto voters (i.e., egocentric bias or “false consensus effect”) (Broockman and Skovron, Reference Broockman and Skovron2018; Hertel-Fernandez et al., Reference Hertel-Fernandez, Mildenberger and Stokes2019; Pereira, Reference Pereira2021; Sevenans et al., Reference Sevenans, Walgrave, Jansen, Soontjens, Bailer, Brack and Breunig2023). These findings are crucial for the debate on the quality of democratic representation and responsiveness to constituencies’ policy preferences (Miller and Stokes, Reference Miller and Stokes1963; Page and Shapiro, Reference Page and Shapiro1983).
However, such second-order beliefs among ordinary citizens remain relatively understudied in political science. Some research has shown an underestimation of public support for policies aimed at mitigating climate change (e.g., Mildenberger and Tingley, Reference Mildenberger and Tingley2019; Sparkman et al., Reference Sparkman, Geiger and Weber2022). Scholars have also explored perceptions of voting behavior, finding that Americans perceive more polarization than actually exists (Ahler, Reference Ahler2014; Levendusky and Malhotra, Reference Levendusky and Malhotra2016) but also that citizens are able to correctly predict voting behavior based on demographic characteristics (Carlson and Hill, Reference Carlson and Hill2022; Titelman and Lauderdale, Reference Titelman and Lauderdale2023).Footnote 1 In general, however, there is little research that considers whether people consistently overestimate public support for conservative policy positions in the same way that politicians often do.Footnote 2
Investigating second-order beliefs is relevant to the understanding of public opinion dynamics, which, in turn, shape politicians’ information environment. Empirical evidence supports the claim that citizens consider peers’ positions when forming their own (Ahler, Reference Ahler2014) and this is theoretically explained as an inherent desire to avoid social rejection from reference groups (Cialdini and Goldstein, Reference Cialdini and Goldstein2004) and to seek social validation (Hardin and Tory Higgins, Reference Hardin and Tory Higgins1996). Indeed, second-order beliefs have been argued to shape behaviors (e.g., voting decisions, engagement in activist endeavors) and beliefs (Snyder and Swann, Reference Snyder and Swann1978; Durkheim, Reference Durkheim1982; Searle, Reference Searle1995). As a prominent example, second-order beliefs about climate change were shown to shape support for climate policy action (e.g., Ding et al., Reference Ding, Maibach, Zhao, Roser-Renouf and Leiserowitz2011) and personal engagement in environmentally-friendly behaviors (Jachimowicz et al., Reference Jachimowicz, Hauser, O’Brien, Sherman and Galinsky2018). Furthermore, mutations in perceived social norms were found to motivate changes in personal attitudes, e.g., on racial discrimination (Stangor et al., Reference Stangor, Sechrist and Jost2001). Nevertheless, it has also been acknowledged that individuals may perceive shifts in collective opinion without changing their own (Sherif, Reference Sherif1936; Paluck, Reference Paluck2009).
The objective of this study is to assess whether pluralistic ignorance in the form of conservative bias exists in how individuals perceive public opinion about salient political issues, focusing on abortion rights in the U.S. after the Dobbs decision.
Overall, two main elements influence accuracy in the perception of public opinionFootnote 3: information availability and the presence of factors that intensify behavioral biases in its absorption (from now on “bias-predisposing factors”). For example, a highly polarized political debate is likely to increase selective exposure and motivated reasoning when processing information (Taber and Lodge, Reference Taber and Lodge2006; Tesler, Reference Tesler2018). The two elements move accuracy in opposite directions: more information available is expected to increase accuracy, whereas the presence of “bias-predisposing factors” is expected to reduce it.
Abortion access, especially in the U.S., is regarded as the prototypical culture war issue (e.g., Castle, Reference Castle2019), mostly because it touches upon universal concerns about human sexuality and family organization (Jelen, Reference Jelen2009; Putnam and Campbell, Reference Putnam and Campbell2012). Culture war attitudes serve as social identities; therefore, individuals are likely to be alert in gauging opinion orientations on this type of issue. Additionally, abortion is widely regarded as a visible and important issue for voters and it can be framed in non-complex ways (Hare et al., Reference Hare, Armstrong, Bakker, Carroll and Poole2015; Kalla and Broockman, Reference Kalla and Broockman2018). The debate on abortion rights has regained major salience with the landmark 2022 Supreme Court decision that removed abortion from the constitutionally guaranteed rights and returned the issue to the states. Moreover, the subsequent months have witnessed a series of relevant events impacting abortion legislation, including total or early gestational age bans achieving enforceability in a set of states and the Kansas abortion referendum.
These characteristics are compatible with opposed expectations on accuracy in perception of public opinion: high visibility and renewed salience,Footnote 4 sustained also throughout the 2022 midterm elections campaign (Mutz and Mansfield, Reference Mutz and Mansfield2024), suggest accuracy in perceptions; however, the highly polarized debate and the role of abortion attitudes in shaping social identities are likely to increase biases in the way information is accessed and processed, predicting the presence of forms of pluralistic ignorance.
3. Data and methods
This study is based on original survey data collected in February 2023 from a sample (n = 1,536) representative of the U.S. adult population on key demographic characteristics.Footnote 5
Perceived public opinion and personal opinion about abortion rights were captured using survey questions borrowed from the General Social Survey.Footnote 6 Respondents were asked to mark on a seven-point Likert scale to what extent they think Americans oppose or support abortion access (perceived public opinion) under seven distinct circumstances encompassing elective reasons (i.e., any reason, the woman already has children and does not want any more, the household has a very low income, the woman is not married) and traumatic reasons (i.e., strong chance of serious defect in the fetus, carrying the pregnancy to term would jeopardize the woman’s health, rape). Respondents were subsequently asked to go through the same survey items, marking to what extent they themselves oppose or support abortion access (personal opinion).
Moreover, I collected information about demographic and ideological variables that reflect the main correlates of abortion attitudes identified in the literature (Jozkowski et al., Reference Jozkowski, Crawford, Turner and Wen-Juo2020; Osborne et al., Reference Osborne, Huang, Overall, Sutton, Petterson, Douglas, Davies and Sibley2022), trust in institutions, and knowledge about the Dobbs decision.
Data analysis was organized into two parts. The first one focused on “perception versus reality.” Here, I analyzed the accuracy in perceptions of public opinion and the factors associated with it. In this section, I compared the actual level of support for abortion access emerging from the survey with respondents’ perceptions of it to assess the presence of misperceptions. Furthermore, I estimated multivariable models using the absolute value of the difference between actual and perceived opinion to identify the characteristics that correlate with accuracy in the perception of public opinion. The second part of the analysis looked at “self-positioning” of the respondents’ level of support compared to what they perceived support to be among Americans. Here, both pieces of information are known by the respondent, while in the “perception versus reality” section respondents were not aware of the actual level of support emerging from the survey. Hence, this second analysis allows us to collect information about how close respondents perceive their position to be with respect to the average one. To do so, I estimated multivariable models using as a dependent variable the absolute value of the difference between a respondent’s personal opinion and their perception of public opinion.Footnote 7
4. Results
4.1. Perception versus reality
Figure 1Footnote 8 reports the actual support for abortion access emerging from the survey and perceived support, by abortion cause. Respondents significantly underestimated public support for abortion access for most abortion reasons, except for rape and the most generic cause. The largest discrepancy was found for the woman not being married (actual support [measured in −3 to 3 scale]: 0.1113, perceived support: −0.0723, p < 0.001), followed by the case in which the woman already has children and does not want any more (actual support: 0.0710, perceived support: −0.1205, p < 0.001) and the household cannot afford any more children (actual support: 0.2402, perceived support: 0.0788, p < 0.001).
Figure 1. Underestimation of public support for abortion rights. The bar charts display the sample average for actual support and perceived support, error bars, and p-value of the t-test on equality of means.
The analysis of the characteristics that correlate with accuracy in the perception of public opinion revealed that, compared to subjects identifying as “pro-choice,” those identifying as “pro-life” were systematically more wrong in their assessment of public support for abortion access, while subjects identifying as “neither pro-life nor pro-choice” or “equally both” were more accurate in their perceptions, limited to elective causes. A similar pattern, although less consistent, emerged for ideology, where both strongly liberal and strongly conservative respondents misperceived public opinion—especially when dealing with elective causes, while moderately conservative subjects were more accurate in their judgement with respect to traumatic causes, as reported in Figure 2.
Figure 2. Identifying as “pro-life” is associated with a significant misperception of public opinion, while individuals with neutral positions have more accurate perceptions. A similar pattern is found for ideology.
4.2. Self-positioning
The analysis on self-positioning revealed that both individuals identifying as “pro-life” and those identifying as “pro-choice” perceived their opinions as more extreme compared to those of the average American, as displayed in Figure 3.Footnote 9 On the contrary, subjects taking a neutral stance in the abortion debate perceived that the average opinion is close to their view. Multivariable analyses confirmed the finding (see Supplementary materials, Table A14).
Figure 3. Both individuals identifying as “pro-life” and “pro-choice” perceive their opinion as more extreme than the American public. The bar charts display personal support (darker) and perceived support (lighter) scores by respondent’s abortion identity label. The black solid line indicates the average actual support.
All results reported in this section were produced without using survey weights. Results do not change in sign and significance when performing weighted analyses. Full details on the post-stratification technique and results are available in Supplementary materials (Section 4, Tables A6, A11, A16).
5. Discussion and conclusion
Drawing on survey data collected after the 2022 landmark Supreme Court decision that removed abortion from the constitutionally protected rights in the U.S., this study found that individuals underestimate public support for abortion access. This result has important implications for the understanding public opinion dynamics, pluralistic ignorance, and the politics of abortion. First, it shows that systematic misperceptions can exist even in the context of a highly salient topic, with a large availability of public opinion poll data. Second, conservative biases in the perception of public support for abortion access among citizens can have practical consequences from the policymaking perspective. Although they may not alter the behavior of strongly committed pro-abortion access activists, they may discourage political action on behalf of individuals holding moderately liberal views on the topic, compatibly with the “spiral of silence” theory. This, in turn, may be reflected in politicians’ information environment, potentially encouraging further restrictions in antiabortion-oriented states and lessening the pressure to safeguard abortion access in protective states.
When analyzing the influence of demographic and ideological characteristics on accuracy in perceptions, identifying as “pro-life” was consistently associated with the most inaccurate estimation of public support for abortion access. This finding points in the direction of the presence of forms of egocentric biases such as the “false consensus effect,” with pro-life and conservative Americans especially likely to attribute their own views to others. It is worth noting that false consensus is also found for individuals taking liberal positions—although to a lesser extent (Figure 3). This result partly contrasts with the empirical regularity that associates liberals with the “illusion of uniqueness” effect, defined as judging own’s beliefs as more distinctive than they actually are (Suls et al., Reference Suls, Wan and Sanders1988; Stern et al., Reference Stern, West and Schmitt2014b).Footnote 10 Fully assessing the mechanisms that lead subjects identifying as “pro-life” to display the largest misperception is outside the scope of this research note. Nevertheless, it is possible to link these findings with existing contributions. It is known that conservatives possess a stronger desire to “share reality” (Stern et al., Reference Stern, West, Jost and Rule2014a) compared to liberals. Similarly to what has been documented in the climate change debate, selective exposure to information and motivated reasoning could be plausible mechanisms at play (Bayes and Druckman, Reference Bayes and Druckman2021); however, more research is needed to confirm this hypothesis.
As for self-positioning, both individuals identifying as “pro-life” and “pro-choice” appear comparably aware that their opinions are more extreme compared to the average. This result therefore suggests that, for individuals holding “extreme” positions, self-awareness that their stance is farther from the perceived average public opinion does not necessarily lead to accuracy in assessment.
Finally, given the specificity of the case study, caution is required when making claims about the results generalizability. Considering accuracy in the perception of public opinion as the net result of two countervailing forces—information availability and “bias-predisposing factors,” these results suggest that the latter may prevail in the case of abortion access. With abortion as the prototypical culture war issue, it is conceivable that the same effect is present for other issues that are strongly correlated with partisanship and social identity. On the contrary, “bias-predisposing factors” may be less present for topics that do not evoke comparably strong emotional responses. Therefore, in these cases, accuracy may primarily be a function of information availability.
Overall, these findings call for further research on second-order beliefs for different issues and contexts, with specific attention to the mechanisms driving results. Biases in perceptions of public opinion among citizens are interconnected with biases in political elites’ perceptions of preferences among their constituents and can impact agenda-setting and policymaking. Therefore, improving knowledge on this topic can substantially contribute to understanding patterns of democratic representation.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/psrm.2025.26. To obtain replication material for this article, https://doi.org/10.7910/DVN/SPEOL9.
Acknowledgements
The author acknowledges invaluable comments received at various stages of this manuscript from Scott Williamson, Letizia Mencarini, Kerim Can Kavakli, Aleksandra Torbica, Patrizio Armeni, Oriana Ciani, Nicolò Cavalli, Matthijs Rooduijn and the participants of the panel “Political Disagreement and Polarization” at the 2024 European Political Science Association Annual Meeting, the editorial team at PSRM, and the anonymous reviewers.
Competing Interests
The corresponding author declares that there are no conflicts of interest.
Open access
