Universal basic income (UBI) has been proposed as a policy solution to the rising inequality and job automation in the contemporary world (Bidadanure, 2019; Hoynes & Rothstein, 2019). Correspondingly, pilot schemes were implemented around the globe, including Brazil, Canada, Finland, India, Netherlands, Spain, and the US. In Switzerland, a national referendum on UBI was already held in 2016. In America, UBI was the centerpiece of Andrew Yang’s presidential campaign in 2020.

But similar to other welfare proposals, UBI faces strong opposition from conservatives. Survey responses from the European Social Survey reveal that right-leaning individuals across 23 European countries were much more opposed to UBI (Parolin & Siöland, 2020; Roosma & van Oorschot, 2020). Recent data from the Pew Research Center (2020) show that most American conservatives opposed UBI: while 72% of liberals favored it, 84% of conservatives opposed it. Such opposition, however, is not necessarily rigid because political elites can often deploy frames to sway public opinion in specific ideological directions (Lahav & Courtemanche, 2012). Given the power of issue framing in welfare politics (Avdagic & Savage, 2021; Brooks, 2012; Jacoby, 2000; Nelson, 2011; Nelson & Kinder, 1996; Winter, 2006), can UBI be framed to overcome the strong opposition from conservatives?

This question is politically relevant. First, the political prospect of UBI hinges on public support for the policy. In democracies, public opinion plays an important role in shaping policy outcomes, particularly in the realm of welfare (Soroka & Wlezien, 2010). This is especially true in the US, where policy changes often require strong backing of the public (Burstein, 2003). Although elites may exert disproportionate influence (Bartels, 2008; Gilens, 2012), policy attitudes among the masses—and across the ideological spectrum (Caughey & Warshaw, 2018; Erikson et al., 1993; Lax & Phillips, 2012)—also matter (Canes-Wrone, 2015; Kelly & Enns, 2010). Thus, the political feasibility of UBI is a function of political support from the masses, including conservatives.Footnote 1

Second, UBI has been endorsed by various conservative figures, especially in the US. Historically, the conservative economist Milton Friedman and former President Richard Nixon were strong proponents of a basic income or negative income tax. More recently, conservative Senators Marco Rubio and Mitt Romney, as well as former Congressman Paul Ryan, proposed welfare policies that are very close to UBI. Additionally, the Alaska Permanent Dividend—an unconditional cash transfer paid to Alaska residents annually—was introduced by Republican Governor Jay Hammond and has been championed by conservative governors since then.

UBI receives political endorsement from conservative figures because it has unique features that conform to the conservative ideology. A nonmeans-tested policy, UBI replaces other existing welfare programs that typically require more targeting efforts. Thus, it contrasts with other large-scale, means-tested welfare policies such as public housing and temporary assistance for needy families (TANF), which typically receive meager support from conservatives due to their opposition to big government (McClosky & John, 1984). To the extent that UBI reduces bureaucracy and limits government by replacing administratively costly welfare programs, it fits the conservative ideal of laissez-faire and individualism (Feldman & Zaller, 1992).

Leveraging such unique features, I design a framing experiment that highlights the key components of UBI that fit the conservative ideology. Specifically, I randomly expose respondents to one of two frames: (1) an equalizing-opportunity frame which emphasizes that UBI creates a level playing field and promotes self-responsibility, or (2) a limiting-government frame which highlights UBI as a policy that limits government and reduces bureaucracy. The frames are carefully designed to omit several variables, particularly partisanship and race. This helps to preempt partisan bias (Bullock & Lenz, 2019; Jerit & Barabas, 2012) and racial priming (Hutchings & Jardina, 2009; Mendelberg, 2008), which may confound framing effects. Thus, my design allows me to zero in on the impact of framing. Using these carefully devised, ideologically based frames, I conduct an experiment in the US to investigate how they shape conservatives’ support for UBI.

I focus on the US for two reasons. First, due to racialized politics, the negative stereotypes of welfare recipients are especially strong among American conservatives. As many conservatives believe that welfare programs disproportionately benefit Black Americans, they regard welfare with fundamental disdain (Gilens, 1999). Yet, the universality and unconditionality of UBI allow working classes of all races to benefit from the policy, thereby blurring racial lines. As such, UBI could potentially neutralize racial animus among conservative opponents, forming a basis for conservatives to be persuadable in reducing their opposition to UBI.

Second, while there is a burgeoning literature on UBI in the European context (e.g., Legein et al. 2018; Parolin & Siöland, 2020; Roosma & van Oorschot, 2020; Rossetti et al., 2020; Stadelmann-Steffen & Dermont, 2020), little research has been done to understand UBI public opinion in the US. Yet, American attitudes toward UBI may be systematically different. For instance, a cross-country study by Ipsos (2017) has shown that Americans—compared to citizens from 11 other countries—are most likely to think that UBI will make people reliant on the state for income and that it will discourage people from working. Such nuances call for more scholarly attention to the study of American public opinion on UBI.

My experimental results suggest a null framing effect. Conservatives—identified by using 10 policy statements—remain strongly opposed to UBI even after they are presented with the equalizing-opportunity or limiting-government frame. There is also no evidence that they are more responsive to such frames than liberals are, even when these frames are particularly designed to highlight the compatibility of UBI with conservative ideals. A systematic analysis of open-ended responses shows that, despite the framing, how conservatives explain their opposition to UBI remains unchanged. To probe the mechanism further, I provide suggestive evidence that conservatives’ rigid opposition to UBI is complicated by partisanship: most conservatives, who are also Republicans, naturally associate UBI with the Democratic Party—despite the carefully designed vignettes that do not explicitly prime their partisanship. Interestingly, this is unlikely due to their prior political exposure to UBI: many of them have not even heard of UBI before. UBI is already politicized on conservatives’ mind before politicians politicize the policy to them.

Taken together, the findings suggest that even if UBI has unique features that fit the conservative ideology, exploiting such features to make ideological appeals is still unlikely to be effective. An important implication is that, unlike many past instances where issue framing has proven effective in shaping welfare attitudes (Brooks, 2012; Jacoby, 2000; Nelson, 2011; Nelson & Kinder, 1996; Winter, 2006), pro-UBI politicians would need to go beyond such tactics to persuade their electorate. Even if UBI is considered by many as a timely policy response to technological advances and structural inequality, its political feasibility may be much more limited than previously thought.

Welfare Attitudes and Prospects of Framing UBI in the US

That conservatives dislike welfare is a stylized fact in American politics. Racial attitudes (e.g., DeSante, 2013; Gilens, 1999; Winter, 2006), opposition to big government (e.g., Feldman & Zaller, 1992; Free & Cantril, 1968; McClosky & John, 1984), and ideals of individualism (e.g., Hasenfeld & Rafferty, 1989; Kluegel & Smith, 1986) each play an important role in shaping anti-welfare attitudes among conservatives. At the same time, American welfare policies are characterized by these sensitive features (see discussions below), since most programs are means-tested and require substantial targeting efforts by the government.

UBI: A Distinctive Welfare Policy

UBI, however, is distinct from any welfare programs in the US. Although UBI is also a welfare policy, two characteristics of UBI may make conservatives’ dislike of it substantially less rigid compared to other welfare programs. First, UBI is universal. The unconditionality of UBI makes it different from most other means-tested welfare programs in the US, such as Medicaid, Supplemental Nutrition Assistance Program, and TANF. One major reason leading to white conservatives’ disdain for welfare policy is their perception that targeted welfare programs disproportionately benefit blacks (Gilens, 1999; see also the discussions on the racialized stereotype of the “Welfare Queen” in Hancock, 2004). UBI, however, effectively blurs racial lines by allowing both black and white working classes to benefit from the policy. This important implication of the universality of UBI, therefore, removes the racial prime that often makes American conservatives unpersuadable in welfare policy debate (Bridges, 2017; Katz, 2013).

Second, UBI replaces existing welfare programs. Since most existing welfare programs are means-tested and thus require substantial amount of administrative efforts, UBI reduces bureaucracy and limits government. Another major reason leading to American conservatives’ disdain for welfare policy is their laissez-faire ideal (Feldman & Zaller, 1992; Free & Cantril, 1968; McClosky & John, 1984). Although Zaller (1992) argues that individuals’ ideologies are sometimes inconsistent with their policy preferences, he and Feldman find that American conservatives are in fact highly consistent in their opposition to welfare (Feldman & Zaller, 1992). Specifically, they find that their “antisocial welfare arguments” almost always involve clear discussions of their “opposition to big government (laissez-faire and bureaucracy)” and “individualism” (Feldman & Zaller, 1992, 281). To the extent that UBI reduces bureaucracy and limits government, it fits the laissez-faire ideal among American conservatives.

In sum, these distinctive features of UBI make it possible that conservatives can be persuaded to increase their support for—or at least reduce their opposition to—UBI. If conservatives’ disdain for welfare is shaped by the features of the welfare program in concern, then their support for UBI will likely be increased by frames that highlight the compatibility of UBI with conservative principles. This claim not only builds on but also speaks to the limits of previous research on American welfare attitudes. On one hand, the literature has suggested that individual welfare attitudes are dependent on features of the welfare policy vis-à-vis the values that shape an individual’s outlook. On the other, such dynamics appear to be neglected by previous work when it makes general statements about liberal and conservative welfare attitudes, regardless of the specific features of the welfare policy. This is understandable because various welfare programs in the US often overlap in terms of their general features, and such features are usually not aligned with conservatives’ outlook. Thus, UBI—as a distinctive welfare policy with features that fit conservatives’ outlook—provides a rare and interesting case for the study of welfare attitudes among conservatives.

While UBI has features that uniquely conform to conservative principles, it is also important to note that the public may not hold principled ideology that orients their policy attitudes. Seminal work by Converse (1964) and the recent contribution by Kinder and Kalmoe (2017), for example, argue that only few Americans could develop—and strictly adhere to—ideological principles (cf. Jost, 2021). Among those who are able to do so, they tend to be more politically knowledgeable (Kalmoe, 2020; Zaller, 1992). Thus, instead of ideological principles that structure public opinion among the politically sophisticated, opinion leadership (Barber & Pope, 2019; Berinsky, 2009; Lenz, 2012) and group-centrism (Gilens, 1999; Nelson & Kinder, 1996) may play an even more important role in shaping policy views among the less sophisticated masses. Bearing such nuances in mind, the next subsection proceeds to explain why, if conservatives’ welfare attitudes are conditional on specific policy features, issue framing can shape their support for UBI.

Issue Framing in Welfare Politics

Frames matter in politics. According to Entman (1993, 52), “[t]o frame is to select some aspects of a perceived reality and make them more salient in a communicating text, in such a way as to promote a particular problem definition, causal interpretation, moral evaluation, and/or treatment recommendation for the item described.” Similarly, Slothuus and de Vreese (2010, 631) define issue framing as follows:

Building on prior work, we understand issue framing as a process in which a communicator “defines and constructs a political issue or public controversy” (Nelson et al., 1997, 567) by emphasizing “a subset of potentially relevant considerations” (Druckman & Nelson, 2003, 730) and thereby pointing the receiver to “the essence of the issue” (Gamson & Modigliani, 1987, 143). A framing effect occurs when such “frames in communication” subsequently affect the “frames in thought” of the receivers, that is, their cognitive understanding of a given situation and/or their opinion (Chong & Druckman, 2007b).

The key of framing, therefore, lies in selection and salience.Footnote 2 By mastering both of them, political elites could effectively sway public opinion. Hence, “public opinion often depends on how elites choose to frame issues” (Chong & Druckman, 2007a, 99; see also Zaller, 1992).Footnote 3 Although the power of framing could be limited at times (especially when competing frames are present), the latest meta-analysis reveals that “elites can substantially influence citizens’ support for (and evaluation of) a policy, at least in the short term” (Amsalem & Zoizner, 2022, 233).

Importantly, past research has shown that Americans’ preferences for government spending can be shaped by issue framing.Footnote 4 Issue frames encompass two important characteristics: first, they involve assertions of facts and “go beyond the facts to offer broader interpretations and characterizations of the issue”; second, they “support a distinct position on a (typically) controversial issue” (Nelson 2011, 190). Indeed, it has been shown that framing American welfare spending as giving away money to undeserving people makes individuals who hold negative attitudes toward the poor less likely to support government programs for the poor (Nelson & Kinder, 1996). Using survey questions that closely mimic the rhetoric of the 1992 US presidential campaign, Jacoby (2000, 763) relatedly shows that issue framing is “extremely powerful” in shaping American public opinion on government spending. In addition, Winter (2006, 400) argues that since American politicians use issue frames to associate Social Security symbolically with whiteness instead of blackness, “racially conservative whites feel more positively about Social Security than do racial liberals.” In short, past research has demonstrated that issue framing can effectively shape American welfare attitudes.

Crucially, the extant literature suggests that ideological predispositions matter in framing. Lahav and Courtemanche (2012) find that while framing immigrants as a security threat—instead of an economic or cultural threat—increases liberals’ support for immigration restrictions, it does not affect conservatives’ opinion. They conclude that “framing matters most for those with certain predispositions” (494) and that “[i]deology plays an important role in moderating the effect of frames on individuals” (497). In the realm of welfare politics, Brooks (2012) similarly shows that “[w]hen framing matters for social policy-related attitudes in the U.S. context, it may tend to do so by getting individuals to rethink welfare issues using ideas to which they already are likely to subscribe” (214).

Framing UBI: Two Possibilities

Given the characteristics of UBI and the ideological predispositions of conservatives, what frames may be effective in increasing conservatives’ support for UBI? I argue that one possibility is to frame UBI as a policy that equalizes opportunity.

Conservatives generally emphasize equality of opportunity over equality of outcome. They tend to favor self-responsibility, believing that people are rich because they worked harder, while people are poor because they lacked effort (Chow & Galak, 2012; Feather, 1984; see also Appendix A). Since UBI is not means-tested, it helps to preempt conservatives from perceiving it as a welfare policy that only benefits the “undeserving poor” (Bridges, 2017; Katz, 2013).

Conservatives are also much more optimistic about their prospect of achieving upward mobility than liberals are (Alesina et al., 2018; Davidai & Gilovich, 2018; see also Appendix A). According to the prospect of upward mobility (POUM) hypothesis (Benabou & Ok, 2001, 447), as these individuals believe that they or their offspring can achieve upward mobility, they would be adverse to outcome-equalizing policies.Footnote 5 Therefore, if framing UBI as an opportunity-equalizing policy helps to preempt conservatives from perceiving UBI as an outcome-equalizing policy, then such a frame may increase their support for UBI. This leads to the first preregistered hypothesis:

H1a (equalizing opportunity—framing effect on conservatives): Framing UBI as an opportunity-equalizing policy will increase conservatives’ support for UBI.

Yet, this frame may not have the same effect on liberals because they are generally more favorable of equality of outcome than conservatives are (see Appendix A). The frame may even backfire if liberals are already predisposed to favor equality of outcome (Galston, 1986). This leads to the following preregistered hypothesis:

H1b (equalizing opportunity—heterogeneous framing effect): Framing UBI as an opportunity-equalizing policy will have a larger positive effect on conservatives’ support than on liberals’ support for UBI.

I argue that another possibility is to frame UBI as a policy that limits government. This is because conservatives generally dislike bureaucracy and emphasize effective government. Data from the Pew Research Center show that they are much more likely to think that the government is wasteful and inefficient than liberals are (see Appendix A). Recent research suggests that conservatives have a strong desire for better—but not necessarily smaller—government, as they believe the highest priority for improving government is to improve efficiency and reduce waste (Lerman, 2019).Footnote 6

Anecdotal evidence also suggests that American conservatives tend to favor reducing bureaucracy and limiting government. For example, in the 2016 Republican Party presidential primaries, abolishment of the Internal Revenue Service was one of the major policies proposed by Ted Cruz—a policy that was highly welcomed by his conservative supporters. Further back in the past, President Reagan—the conservative figure of the US—famously asserted in his first inaugural address that “government is not the solution to our problems; government is the problem.” If conservatives already think that the US government is wasteful and should be restrained, then framing UBI as a policy that limits government should appeal to them. This leads to the following preregistered hypothesis:

H2a (limiting government—framing effect on conservatives): Framing UBI as a policy that limits government will increase conservatives’ support for UBI.

However, this frame may not have the same effect on liberals because they typically attach less importance to bureaucratic efficiency, and generally do not believe that the government is wasteful (see Appendix A). The frame may even backfire if they are predisposed to support active government (Pew Research Center, 2019). This then leads to the following preregistered hypothesis:

H2b (limiting government—heterogeneous framing effect): Framing UBI as a policy that limits government will have a larger positive effect on conservatives’ support than on liberals’ support for UBI.

Experimental Design

To test these hypotheses, I conducted a preregistered survey experiment in September 2021 via Lucid.Footnote 7 Replicating multiple classic experiments with Lucid, Coppock and McClellan (2019, 1) “conclude that subjects recruited from the Lucid platform constitute a sample that is suitable for evaluating many social scientific theories.” For this reason, many political scientists have turned to Lucid to conduct online experiments (e.g., Guay & Johnston, 2022; Wood & Porter, 2019). My sample consists of 2530 American adults.

Treatments

The survey experiment adopts a between-subjects design. There are three experimental groups: (1) a control group; (2) a treatment group which receives an equalizing-opportunity frame; and (3) a treatment group which receives a limiting-government frame. The treatment is imposed through vignettes. All three groups first read the following vignette that describes UBI:

Universal Basic Income in the U.S.

In the United States, some cities are considering—or have already started—to implement pilot programs for universal basic income (UBI). In a UBI program, the government pays everyone a monthly income to cover essential living costs. It replaces many other social benefits. The purpose is to guarantee everyone a minimum standard of living. Everyone receives the same amount regardless of whether or not they are working, and people also keep the money they earn from work or other sources. This program is paid for by taxes.

The descriptions of UBI are adapted from Round 8 of the European Social Survey. The main advantage of this vignette is that it impartially describes all basic elements of UBI. For instance, two unwelcome but necessary features of UBI found in the European context—that is, replacement of other social benefits and universality of the program whose beneficiaries include nonworkers (Rossetti et al., 2020; Stadelmann-Steffen & Dermont, 2020)—are mentioned in the text. This ensures that respondents have the essential information to form an opinion on UBI, even if they may not have heard of it before.Footnote 8

Next, respondents from treatment groups read an additional vignette that immediately follows the common vignette. Here is the equalizing-opportunity frame:

An important feature of UBI is that it will create a level playing field because poor households will also be financially empowered. This encourages people to be more self-reliant and self-responsible.

The limiting-government frame is as follows:

An important feature of UBI is that it will limit government because some existing government programs for the poor will be cut or replaced by UBI. This helps to minimize government bureaucracy.

Six nuances of the treatment vignettes are useful to note:

  1. 1.

    Explicit mention of key terms: On one hand, “level playing field,” “self-reliant,” and “self-responsible” are mentioned in the equalizing-opportunity frame. On the other, “limit government” and “government bureaucracy” are mentioned in the limiting-government frame. As discussed, these are the concepts that American conservatives tend to emphasize and, consequently, are more likely to respond to.

  2. 2.

    Real-world relevance to right-wing rhetoric: Relatedly, these key concepts have been stressed by pro-UBI right-wing think tanks in reality (Lewis & Stronge, 2018). For instance, the Adam Smith Institute (2018)—a leading right-wing think tank in the UK—stated the following: “A UBI streamlines the provision of welfare services and improves the autonomy and incentives of individuals. Allowing poor people to spend their money as they see fit stimulates bottom-up market solutions and cuts down on bureaucratic red tape.”

  3. 3.

    Similar length and structure of the two frames: In the case where one frame works while another not, the design allows me to rule out the explanation that it is the different lengths and/or structures of frames—rather than the contents—that lead to the different framing effects.

  4. 4.

    Vague content: I deliberately keep the content vague because this allows me to zero in on framing effects by distinguishing them from information effects. If the frames introduced new, concrete information to respondents, any treatment effects found in the study could be attributed solely or partially to information effects (Leeper & Slothuus, 2018).

  5. 5.

    No political endorsement: I do not specify the messenger of the frames in order to isolate framing effects from endorsement effects. If elite cues were introduced, politically attached respondents would be less likely to form their opinion based on ideological principles but more likely to base their views on group loyalty (Barber & Pope, 2019). Another advantage of not specifying the messenger is that it enhances the credibility of the message by avoiding partisan perceptual bias (Jerit & Barabas, 2012). Such abstraction in the vignettes—without necessarily tarnishing the external validity of results (Brutger et al., 2022)—is therefore empirically important.

  6. 6.

    No mention of race: I intentionally omit race from my vignettes in order to avoid racial priming. A long line of survey-experimental research has shown that racial cues can effectively shape white Americans’ policy preferences (Hutchings & Jardina, 2009; Mendelberg, 2008), be they implicit or not (Tesler, 2017). It is thus important to preempt any implicit or explicit priming of race in the vignettes, which could contaminate the treatments and confound the framing effects.

Survey Flow and Measurement

Figure 1 illustrates the survey flow. I begin the survey with a set of basic demographic questions. Two screening questions for attention check are also added in between.

Fig. 1
figure 1

Flow of the survey

Full size image

Subsequently, I randomly assign each respondent to one of three experimental groups (see Figure A4 for the balance on demographic variables across experimental groups). The control group is only given the vignette that describes UBI. In addition to this vignette, one treatment group receives the equalizing-opportunity frame, while another receives the limiting-government frame. After reading the vignette, respondents are asked the following: “Overall, would you be against or in favor of having universal basic income in the United States?” They choose from 0 (strongly against) to 6 (strongly in favor). This seven-point variable is the main dependent variable indicating support for UBI. Respondents are then asked to provide an open-ended response to briefly explain why they hold such views. Additionally, they are asked whether they think Democratic or Republican politicians are more likely to support UBI, with the same question also asked about TANF.

Next, I measure the political ideology of respondents. I follow Barber and Pope (2019) and Thal (2020) who use policy statements to gauge individual ideology. Specifically, respondents are asked whether they support each of the following statements:

  1. 1.

    To increase the minimum wage to over $10 an hour.

  2. 2.

    To allow teachers to carry guns on school property.

  3. 3.

    Acknowledging that humans are the largest contributing factor in global climate change.

  4. 4.

    To increase the amount of taxes paid by the wealthy.

  5. 5.

    To enforce penalties on women who obtain abortions.

  6. 6.

    Putting in place a health care system that covers all individuals under a government plan.

  7. 7.

    To allow illegal immigrants to the United States to obtain legal status.

  8. 8.

    To increase government funding for public education.

  9. 9.

    Strengthening government regulation of business.

  10. 10.

    Expanding public housing programs for low-income households.

These 10 items allow me to construct a conservatism score—my measure of political ideology—ranging from 0 (most liberal) to 10 (most conservative).Footnote 9 For each respondent, their conservatism score increases by one as they give a conservative answer to a policy statement (see Appendix C for how conservative answers are coded). Following the preanalysis plan, I define respondents whose conservatism scores are 5 or above as conservatives. Appendix C shows the distribution of conservatism scores (Figure A5).

This measure of ideology has several methodological advantages over a conventional, self-reported scale:

  1. 1.

    The self-reported scale assumes that ideology is unidimensional. Recent research, however, suggests that it is not. Political psychologists argue that ideology should cover both economic and social dimensions (e.g., Carmines & D’Amico, 2015; Feldman & Johnston, 2014). The 10 policy statements allow me to capture both dimensions.Footnote 10

  2. 2.

    The self-reported scale also critically assumes that respondents can identify their ideology accurately. Yet, previous research suggests that they cannot. For instance, Americans systematically overestimate their political conservatism (Zell & Bernstein, 2014). Recent research also shows that many Americans—especially Black Americans—are unfamiliar with the terms “liberal” and “conservatives,” and thus the self-reported scale fails to reflect the true ideology of specific populations (Jefferson, 2021).

  3. 3.

    Measuring ideology using multiple questions improves precision. Ansolabehere et al.’s (2008) seminal study argues that measuring important concepts using single items has fundamental flaws as it yields measurement error, which can be easily reduced by simply asking more questions.

Lastly, the survey ends with another set of demographic and attitudinal questions.

Estimation Strategies

To test my hypotheses, I use OLS to estimate the following equation:

$$\begin{aligned} \text {Support}_{i}&=\beta _{0}+\beta _{1}\text {EO-Frame}_{i}+\beta _{2}\text {LG-Frame}_{i}+\beta _{3}\text {Conservative}_{i} \nonumber \\&\quad +\beta _{4}(\text {EO-Frame}_{i}\times \text {Conservative}_{i})+\beta _{5}(\text {LG-Frame}_{i}\times \text {Conservative}_{i}) \nonumber \\&\quad +\varvec{\eta }\varvec{\Phi }_{i}+\epsilon _{i}, \end{aligned}$$
(1)

where i indexes a respondent and \(\varvec{\Phi }_{i}\) is a vector of individual characteristics. \(\text {EO-Frame}_{i}\) and \(\text {LG-Frame}_{i}\) indicate whether a respondent received the equalizing-opportunity frame and the limiting-government frame, respectively. \(\text {Conservative}_{i}\) indicates a conservative respondent. Thus, the baseline is non-conservatives in the control group.

For H1a and H2a, the estimands are \(\beta _{1}+\beta _{4}\) and \(\beta _{2}+\beta _{5}\): conditional average treatment effects (CATEs). Specifically, \(\beta _{1}+\beta _{4}\) is the CATE of the equalizing-opportunity frame on conservatives, and \(\beta _{2}+\beta _{5}\) is the CATE of the limiting-government frame on conservatives. Thus, the data will provide support for H1a and H2a if \(\hat{\beta }_{1}+\hat{\beta }_{4}>0\) and \(\hat{\beta }_{2}+\hat{\beta }_{5}>0\), respectively.

For H1b and H2b, the estimands are \(\beta _{4}\) and \(\beta _{5}\): heterogeneous treatment effects (HTEs). Specifically, \(\beta _{4}\) is the HTE of the equalizing-opportunity frame, and \(\beta _{5}\) is the HTE of the limiting-government frame. The data will provide support for H1b and H2b if \(\hat{\beta }_{4}>0\) and \(\hat{\beta }_{5}>0\), respectively.

Experimental Results

The results do not support H1a and H2a: there is no evidence that issue framing has a positive impact on conservatives’ support for UBI (Table 1).Footnote 11 Figure 2 shows the distribution of responses from conservatives and their average support for UBI across the three experimental groups. The average support among conservatives is 1.28 in the control group (\(n=191\), \(\text {SD}=1.63\)); 1.36 in Treatment Group 1 (\(n=218\), \(\text {SD}=1.67\)); and 1.39 in Treatment Group 2 (\(n=223\), \(\text {SD}=1.62\)). Applying covariate adjustment to increase the precision of estimates (i.e., controlling for \(\varvec{\Phi }_{i}\) in Equation 1; see Gerber and Green 2012) does not meaningfully change the results. Based on the regression estimates in Table 1, Panel A of Fig. 3 shows the CATEs on conservatives, which suggest null framing effects.

Fig. 2
figure 2

Average Support for UBI among conservatives across experimental groups. Error bars represent 95% confidence intervals. The jittered point clouds illustrate the distribution of responses. Most conservative respondents are strongly against UBI regardless of framing

Full size image
Table 1 OLS Regression Corresponding to Equation 1
Full size table
Fig. 3
figure 3

Conditional average treatment effects on conservatives corresponding to hypotheses 1a and 2a. The sample with “fully and partly attentive respondents only” eliminates respondents who fail both attention checks. The sample with “fully attentive respondents only” keeps respondents who pass both attention checks. Estimates are based on Model 3 in Table 1. Error bars represent 95% confidence intervals

Full size image

I take five extra steps to verify the null framing effects. First, I stratify the analysis by the attentiveness of respondents. One potential concern is that the null effects are masked by the fact that some respondents in the treatment group remain “untreated,” as they did not pay attention to the vignettes embedded in the survey. If that is the case, we should expect to see larger (and positive) treatment effects among respondents who are more attentive. My analysis, however, does not support this claim (Fig. 3).

Second, I reoperationalize the dependent variable. I convert the seven-point UBI support to a binary variable, which takes the value of 1 if the respondent is “somewhat in favor,” “in favor,” or “strongly in favor” of UBI (and 0 otherwise). Using the new outcome variable to reanalyze the data, I find that the null framing effects on conservatives remain robust (Fig. 3).

Third, I redefine the conservatism score threshold. In previous analyses, I defined respondents whose conservatism scores are 5 or above as conservatives. Now I raise the threshold to 6 or 7. That is, respondents are identified as conservatives only if they give 6 or 7 conservative answers to the 10 policy questions. Following the new operationalization, I continue to find no evidence that framing increases conservatives’ support for UBI (Fig. 4).

Fig. 4
figure 4

Conditional average treatment effects on conservatives after redefining the conservatism score threshold. Definitions of fully and partly attentive respondents follow those stated under the note for Figure 3. Estimates are based on Model 3 in Table 1. Error bars represent 95% confidence intervals

Full size image

Fourth, I focus independently on economic and social conservatives. I identify economic conservatives as those who give at least three conservative responses to the six economic policy statements, and social conservatives as those who give at least two conservative responses to the four social policy statements. I find no evidence that UBI support among economic or social conservatives is changed by framing (Fig. 5).

Fig. 5
figure 5

Conditional average treatment effects on economic and social conservatives. Definitions of fully and partly attentive respondents follow those stated under the note for Figure 3. Estimates are based on Model 3 in Table 1. Error bars represent 95% confidence intervals

Full size image

Lastly, I analyze self-reported conservatives. One potential criticism of my measure of conservatism is that it conceptualizes ideology only through the lens of policy preferences, while it is possible that the label of conservatism itself matters more in the context of policy persuasion. Another potential criticism of my approach is that my policy-based measure of ideology is obtained posttreatment, while it is possible that the treatment vignettes somehow biased individual responses to the 10 policy statements, thereby introducing posttreatment bias (Montgomery et al., 2018; cf. Klar et al., 2020).Footnote 12 To address these concerns, I rely on a pretreatment, self-reported measure of conservatism to identify conservatives. Reanalyzing all data using the new measure, I continue to find no evidence that framing reduces conservatives’ opposition to UBI (Appendix D).

The results also do not support H1b and H2b: there is no evidence that issue framing has a larger positive effect on conservatives’ support than on liberals’ support for UBI. All models in Table 1 suggest that the estimates of \(\beta _{4}\) and \(\beta _{5}\) in Eq. 1 are close to zero. Even if the issue frames are tailor-made for conservatives—and may even have a backfire effect on liberals—conservatives are still not more responsive to such frames than liberals are.Footnote 13 Their opposition to UBI remains rigid regardless of framing.

Rigid Attitudes Toward UBI Among American Conservatives

The null framing effects suggest a pessimistic view on the political feasibility of persuading conservatives to support UBI in the US. Their opposition to UBI appears to be fundamentally rigid. In the experiment, even if the key components of UBI that fit the conservative ideology are highlighted and brought to their attention, conservatives remain unpersuaded. Their views on UBI remain strongly negative.

Analyses of open-ended responses reinforce this finding. I use structural topic modeling (STM) (Roberts et al., 2014)—a semi-automated content analysis technique—to analyze the text responses (see Appendix F for technical details).Footnote 14 Estimating a four-topic model,Footnote 15 I identify the following topics raised by respondents: policy analysis (Topic 1), helping the needy (Topic 2), poverty and inequality (Topic 3), and discouraging work (Topic 4).Footnote 16 The prevalence of each topic across experimental groups, alongside its relationship with the ideology of respondents, is shown in Fig. 6.

Fig. 6
figure 6

Expected topic proportions across experimental groups and their relationship with political ideology. Error bars represent 95% confidence intervals. The overlapping confidence intervals—in each topic and across all experimental groups—suggest that, regardless of treatment status, how respondents explain their views on UBI remains unchanged. While Topic 4 is labeled as “Discouraging Work,” another common theme in Topic 4 is concerns over tax. See Appendix F for the word clouds and representative responses of each topic

Full size image

The analysis shows that how conservatives explain their views on UBI remains unchanged regardless of framing. Despite being made aware of the equalizing-opportunity or limiting-government features of UBI, conservatives sidestep these frames. On one hand, most of them continue to explain their opposition to UBI by raising concerns of work disincentives: regardless of treatment status, nearly half of conservative respondents highlight such concerns (see Topic 4 in Fig. 6).Footnote 17 On the other, they are no more likely to engage in other topics despite the framing: regardless of treatment status, their likelihood of discussing other policy implications of UBI—such as its impact on inequality and the poor—does not change (see Topics 1 to 3 in Fig. 6). The results thus shed further light on the rigidity of conservatives’ attitudes toward UBI.

To probe further, I read all open-ended responses from conservatives in the treatment groups. I find little evidence that they disagree with the frames introduced to them. Instead, most of them express their opposition to UBI by raising concerns over tax and work disincentives. This corroborates my findings from the STM analysis.Footnote 18

One potential factor reinforcing the rigidity of conservatives’ attitudes is partisanship. As welfare is politicized in the US, conservatives—who are most likely to be RepublicansFootnote 19 under the partisan-ideological sorting in American politics (Abramowitz, 2010; Mason, 2015)—may naturally associate UBI with the Democratic Party. Partisan loyalty may reinforce their opposition to UBI in turn, keeping them hardly persuaded by nonpartisans to support the policy. Figure 7 provides suggestive evidence that conservatives’ opposition to UBI may be complicated by partisanship. Even when there are no explicit cues in the experimental vignettes that prime partisanship, most conservative respondents (75%) still associate UBI with the Democratic Party, believing that Democratic politicians are more likely to support UBI.Footnote 20 Importantly, this is unlikely due to respondents’ prior political exposure to UBI, as my pilot survey suggests that (1) only half of self-reported conservatives (23 out of 47) had heard of UBI, and (2) none of them associated the policy with Andrew Yang when reading the experimental vignettes.Footnote 21

Fig. 7
figure 7

Individual ideology and association of UBI and TANF with politicians from the democratic party. Error bars represent 95% confidence intervals, calculated by using Yates’ continuity correction

Full size image

I also examine whether the strong and rigid opposition to UBI among conservatives is driven by their recent political views on the COVID-19 stimulus package.Footnote 22 If some respondents confused UBI with the stimulus package, then my experimental findings would be less attributable to the unconditional nature of conservatives’ opposition to welfare. To address this concern, I conduct a dictionary-based content analysis by counting how many respondents mentioned either the word “stimulus,” “package,” or “check” in their open-ended responses. The analysis reveals that only two out of 632 conservative respondents mentioned such terms. Hence, it is unlikely that the strong and rigid opposition to UBI among conservatives can be explained by their recent political views on the stimulus checks.

Limitations and Discussion

This study provides strong evidence that American conservatives remain unpersuaded by two ideological frames, although a few limitations must be noted. The main limitation stems from the country- and time-specific operationalization of conservatism adopted by this study. While I follow the best practices by measuring ideology using policy statements that cover both economic and social dimensions, an obvious trade-off is the reduced generalizability of results to other contexts. The findings here may not be directly applicable to other countries, where conservatives may differ in their ideological beliefs from those in the US. The results may also be less applicable to countries where partisan-ideological sorting is less prevalent, as one of the speculated mechanisms here is the salience of party identification in American politics. Probing the generalizability of this study would therefore be an important avenue for future research.

In addition, one cannot conclude from this study that conservatives can never be persuaded to support UBI. There are many possibilities of framing UBI, be they ideologically or non-ideologically oriented. Legein et al. (2018), for instance, find that framing UBI using metaphors can affect Belgian students’ opinion formation toward UBI. To tackle the negative connotation of UBI as a work-discouraging welfare policy, framing UBI as “an earned right” (Winter, 2006, 404) could prove useful—but only to the extent that it is true and believable to conservatives. Providing facts about UBI, such as its limited impact on work incentives based on pilot studies, to correct conservatives’ misconceptions about UBI could be effective (cf. Wood & Porter, 2019). Correcting the common misperception that most American welfare recipients are black may also bear fruit (see Abrajano & Lajevardi, 2021).

While the null findings may undermine some of Feldman and Zaller’s (1992) claims about the ideological principles underlying conservatives’ opposition to welfare policy,Footnote 23 this study alone does not allow me to conclude that opinion leadership—compared to ideology—matters more in shaping UBI support. Thus, future research should test whether conservatives’ opposition to UBI still remains rigid when they are given partisan cues. If such cues prove more effective than ideological frames, it will strengthen Barber and Pope’s (2019) finding that party trumps ideology through the lens of welfare politics.

The final limitation concerns the role of political sophistication, which remains unexplored by the current study.Footnote 24 On one hand, the more sophisticated may be more susceptible to framing because they have a better understanding of how the issue frames connect their ideological predispositions with UBI (Nelson et al., 1997). On the other, the less sophisticated may be more pliable because they are generally less stable in their political views (Zaller, 1992).Footnote 25 Future research would benefit from investigating more carefully how political sophistication moderates framing effects—given that the empirical relationship is still ambiguous in the literature (see Lee & Chang, 2010).

Conclusion

This study shows that conservatives’ opposition to UBI remains rigid, even after its features of equalizing opportunity and limiting government—principles that are highly compatible with the conservative ideology—are explicitly highlighted. Conservatives remain unpersuaded by frames that are particularly designed to persuade them in light of their ideological predispositions.

This study makes three contributions. First, it contributes to the burgeoning literature on UBI. While much research has been done in Europe to identify factors that shape individual support for UBI, we still know relatively little about American public opinion on UBI. This gap in the literature is surprising, given that both political scientists and economists have called for more scholarly attention to the political economy of UBI (Bidadanure, 2019; Hoynes & Rothstein, 2019). This study fills the void by joining recent work to study Americans’ attitudes toward UBI (Jordan et al., 2022), while paying specific attention to the political feasibility of framing UBI to persuade conservatives—the unlikely supporters—in the US.Footnote 26

Second, it advances our understanding of American public opinion on welfare. A long line of scholarship has shown that race and ideology are central to explaining conservatives’ disdain for welfare. What remains unclear in the literature, however, is whether conservatives can be persuaded to support welfare if the policy not only blurs racial lines but also fits their ideology. By designing frames that highlight the compatibility of UBI with conservative principles and by documenting the null effects of such frames on conservatives, this study shows that even if these important criteria are met, conservatives still remain strongly opposed to the welfare policy in concern. This finding is especially interesting in light of Sniderman et al. (1996), which shows that conservatives are strongly opposed to welfare programs regardless of whether the justification and targeting of such programs are race-specific or race-neutral. Their work, as noted by DeSante (2013), has been interpreted as evidence that conservatives’ disdain for welfare is shaped more by principled values than by racial animus. Yet, my study suggests a new interpretation of their finding: the null effects of racial frames, as documented by Sniderman and colleagues, do not necessarily imply that conservatives’ opposition to welfare is deeply rooted in ideological principles. Their null finding, combined with the null effects of the ideological frames in my study, might instead suggest that conservatives’ disdain for welfare is more intrinsic—in a rigid way that is unnecessarily associated with racial animus and ideological principles. The fact that my null was documented in the context where citizens had little prior exposure to the welfare policy—and where conservatives had nearly no exposure to elite cues—further sheds light on the rigidity of welfare attitudes among American conservatives.Footnote 27

Lastly, this study enriches the literature on political framing. While previous research has shown that American welfare attitudes could be effectively shaped by issue framing (Brooks, 2012; Jacoby, 2000; Nelson, 2011; Nelson & Kinder, 1996; Winter, 2006), my framing experiment documents unmoved welfare attitudes across the ideological spectrum. This is thought-provoking, given that the frames are carefully devised such that they not only remove racial and partisan primes, but also make explicit ideological appeals. The results thus offer a sharp contrast to Lahav and Courtemanche (2012), which documents the moderating role of ideology in framing. What, then, may account for such discrepancy? Recent scholarship suggests that, in a polarized era, ideology may have become less important among the American populace, whereas group loyalty dominates (Barber & Pope, 2019). If this is true, then political scientists may need to reconsider the role of ideology in political framing in contemporary US politics. This study provides a step forward in this regard, and future research should continue to investigate how framing and ideology interact in the post-Trump era.