What predicts policy outcomes for certain target groups, especially those with little political power? One proposed answer is social construction theory: that is, groups broadly viewed by the public as more deserving are rewarded with more favorable policies (Schneider and Ingram Reference Schneider and Ingram1993). This approach helps to explain biased outcomes in the policy-making process: political actors have incentives to disproportionately reward positively constructed (deserving) groups and disproportionately punish negatively constructed (undeserving) groups. Despite facing criticism on both epistemological and methodological grounds, social construction theory has gained popularity in the discipline and been subject to empirical testing (Pierce et al. Reference Pierce, Siddiki, Jones, Schumacher, Pattison and Peterson2014).
Recent methodological advances have allowed scholars to address one particularly important aspect of the theory: the measurement of social constructions. Evidence from a large-scale crowdsourcing task demonstrates that members of the public can differentiate target groups in terms of deservingness, as postulated by the original theory (Kreitzer and Smith Reference Kreitzer and Smith2018). At the same time, there is substantial disagreement among respondents about the relative positions of specific groups, meaning that social constructions are more contentious than initially supposed. These findings suggest that variation in the endorsement of social constructions at the individual level can impact policy opinions—and thus ultimately policy outcomes. For instance, voters who view union members as deserving should be more likely to support policies favoring organized labor, whereas those who view them as undeserving should oppose such policies.
Scholars also have suggested that the content of social constructions can be richer than the simple deserving-versus-undeserving dichotomy. Most important, there is evidence that members of policy-target populations often are imagined as disproportionately belonging to politically salient social groups. For instance, drug users have been represented as African American, which leads to skewed arrest statistics (Yates and Whitford Reference Yates and Whitford2009), and people with AIDS have been prominently portrayed as gay (Rollins Reference Rollins2002). However, the discipline lacked until recently a methodological tool that allowed researchers to assess the strength of these associations or their impact on policy opinions.
Our study continues this line of research and further explores individual-level variation in the endorsement of social constructions and its implications for relevant policy opinions. We extend the social construction approach by exploring the dimensionality of the “deservingness” attribute, which is used in the original formulation of the theory as shorthand for positive-versus-negative group image. We argue that the contents of social constructions often are more complicated than this simple binary: members of a specific group can be viewed as lazy, violent, not aligning with the American cultural mainstream, and so on. These are all negative constructions; however, they can be weighted unequally in the conceptions of different groups and, even more important, have disparate impacts on public support for policies benefiting (or harming) those groups. Because American public opinion is group-centric (Conover Reference Conover1988; Nelson and Kinder Reference Nelson and Kinder1996), policies benefiting negatively constructed groups should experience lower public support and vice versa.
We extend the social construction approach by exploring the dimensionality of the “deservingness” attribute, which is used in the original formulation of the theory as shorthand for positive-versus-negative group image.
Our empirical analysis focuses on a specific group that is both politically salient and a target of prominent (as well as contested) government policies: immigrants without legal status. Although the politics of immigration in the United States typically is not discussed in terms of social construction theory, much of the debate in the literature revolves around attributes that Americans may be associating with immigrants. Evidence suggests that anti-immigration attitudes are informed by anti-Hispanic prejudice (Perez Reference Perez2010; Valentino, Brader, and Jardina Reference Valentino, Brader and Jardina2013), which indicates that the US public imagines immigrants as Hispanics.Footnote 1 Moreover, Americans’ preferences for new immigrants seem to be dominated by sociotropic economic concerns (Hainmueller and Hopkins Reference Hainmueller and Hopkins2015). Our study contributes to this debate by assessing the relative weight of race/ethnicity and attributes that signal economic productivity in the social construction of immigrants without legal status and its political implications.
Specifically, we replicate and extend a conjoint experiment that measures social constructions by asking respondents to rate hypothetical immigrants in terms of suspected illegality (Flores and Schachter Reference Flores and Schachter2018). The results confirm that social constructions are multidimensional: perceptions about immigrants’ legal status are associated independently and significantly with attributes such as race/ethnicity, police record, English proficiency, and level of education. Extending these findings, we use the most recent advances in conjoint-experimental methodology to measure the individual-level endorsement of these constructions and relate them to opinions on immigration-enforcement policies. We find that associating the absence of legal status with receiving government benefits and, to a lesser extent, being Hispanic predicts support for stricter enforcement policies.
Overall, the results of our analysis confirm the applicability of the social construction approach in empirical research on politics and policy. We also demonstrate that social constructions of salient target groups are contentious and politically consequential. Our findings further suggest that future studies should measure all potentially relevant dimensions in the social constructions of target groups—including race/ethnicity for groups ostensibly defined in race-neutral terms—rather than focus on deservingness as a catch-all attribute.
SOCIAL CONSTRUCTION THEORY: PROMISE AND CONTROVERSY
Social construction theory in public policy studies has been proposed to explain why groups with equal levels of political power can be systematically advantaged or disadvantaged in terms of policy outcomes (Schneider and Ingram Reference Schneider and Ingram1993). The social construction approach is based on two postulates: (1) target groups are imagined as either deserving or undeserving; and (2) members of the public support policies that benefit positively constructed groups (or harm negatively constructed groups). Election-minded public officials enact policies that are popular with the public—that is, they enact policies that disproportionately benefit groups viewed as deserving and disproportionately harm groups viewed as undeserving. As a result, the policy process often fails to achieve its goals and produces ineffective or inefficient policy designs. It also creates a feedback loop: policies create groups by introducing legal boundaries that, in turn, contribute to the content of social constructions.
Over time, the social construction approach has gained popularity in the literature—one review counts 111 empirical applications from 1993 to 2013 (Pierce et al. Reference Pierce, Siddiki, Jones, Schumacher, Pattison and Peterson2014)—but has not been universally accepted in political science and policy studies. Specifically, social construction theory has been criticized for lacking a clear and testable causal argument as well as for ignoring the contested and dynamic nature of social constructions (Lieberman Reference Lieberman1995). It is telling that the social construction approach was explicitly excluded from the first edition of the influential volume, Theories of the Policy Process, on the grounds of being non-falsifiable (Sabatier Reference Sabatier and Sabatier1999).Footnote 2 The approach subsequently appeared in the second edition (Ingram, Schneider, and DeLeon Reference Ingram, Schneider, DeLeon and Sabatier2007) but was absent from the most recent fourth edition (Weible and Sabatier Reference Weible and Sabatier2017). Nevertheless, scholars in policy studies continue to use the social construction theory: it recently has been applied to cases including affirmative action (Bell Reference Bell2021), human trafficking (Blanton and Jones Reference Blanton and Jones2023), climate change (Koski and Manson Reference Koski and Manson2024), and gun control (Merry Reference Merry2018).
In another important development, researchers have begun to address the measurement of social constructions. A large-scale survey—in which respondents are asked to rate dozens of policy-relevant groups in terms of deservingness—demonstrates that members of the public indeed distinguish between more- and less-deserving groups (Kreitzer and Smith Reference Kreitzer and Smith2018). The revealed perceptions also align with theoretical expectations: for instance, veterans are viewed as deserving whereas criminals are not. This evidence suggests that social constructions in the mass public are measurable—as originally suggested by the theory—and contradicts the claims regarding its non-falsifiability.
However, recent studies that address the behavioral foundations of social construction leave two important questions unanswered. First, existing studies rely on the perceived deservingness of target populations, but is perceived deservingness different from simple positivity? Can scholars measure different facets of deservingness, such as being hardworking and law-abiding? Are some target groups racially constructed even if they are legally defined in race-neutral terms? Second, there is still little or no evidence on the relationship between social constructions and policy opinions. For example, do people who view unauthorized immigrants as deserving oppose strict immigration enforcement? Even if the answer is yes, which aspects of perceived deservingness are the most consequential? This study addresses these questions using conjoint-experimental methodology.
CONJOINT MEASUREMENT OF MULTIDIMENSIONAL CONSTRUCTIONS
Conjoint experiments are a class of online survey experiments designed to study choices that potentially are affected by multiple considerations (Hainmueller, Hopkins, and Yamamoto Reference Hainmueller, Hopkins and Yamamoto2014). In conjoint tasks, respondents are presented with hypothetical individuals or objects described in terms of several attributes with randomized values and asked to rate them or make a choice. This design allows researchers to assess which attributes are most consequential in respondents’ choices and which values make alternatives more (or less) attractive. The estimates of preferences obtained from conjoint experiments have been validated against both aggregate and individual behavioral benchmarks (Hainmueller, Hangartner, and Yamamoto Reference Hainmueller, Hangartner and Yamamoto2015; Jenke et al. Reference Jenke, Bansak, Hainmueller and Hangartner2021). Conjoint tasks also are resilient to some of the common problems of survey data collection such as satisficing and social-desirability bias (Bansak et al. Reference Bansak, Hainmueller, Hopkins and Yamamoto2018; Horiuchi, Markovich, and Yamamoto Reference Horiuchi, Markovich and Yamamoto2022).
Although conjoint experiments were developed originally to measure preferences, social scientists recently began using them to explore perceptions and stereotypes (Flores and Schachter Reference Flores and Schachter2018; Goggin, Henderson, and Theodoridis Reference Goggin, Henderson and Theodoridis2020; Myers Reference Myers2023). In these conjoint tasks, respondents are asked to classify hypothetical individuals into groups; researchers then use these classifications to explore which attributes members of the public associate with the groups in question. An example of this application is a recent study of perceptions about welfare recipients in the United States that has shown face, concurrent, and predictive validity of conjoint measures (Myers, Zhirkov, and Lunz Trujillo Reference Myers, Zhirkov and Trujillo2024).
An important methodological advance in conjoint-experimental literature concerns obtaining respondent-level estimates known as individual marginal component effects (IMCEs) (Zhirkov Reference Zhirkov2022). IMCEs are computed by running linear regressions that predict profile ratings with attribute values independently for each respondent, which is possible because each respondent rates multiple profiles in a standard conjoint experiment. Because IMCEs are individual-level estimates, they can be used as covariates in subsequent inferential analyses. This procedure does not involve any additional statistical assumptions beyond those already necessary to conduct classical conjoint analysis, although estimation of IMCEs requires several adjustments to the conjoint design compared with the standard setup. Specifically, current guidelines recommend using interval outcomes, minimizing the number of potential values per attribute and maximizing the number of rated profiles.
There are three benefits of using classification-based conjoint analysis for the measurement of social constructions. First, conjoint experiments are inherently multidimensional and therefore allow the exploration of various facets of social constructions beyond the simple dichotomy between deserving and undeserving groups. Second, its resilience to social-desirability bias makes conjoint experiments particularly useful to measure social constructions in sensitive policy areas, especially if they potentially involve important group identities (e.g., race and gender). Third, the ability to estimate the endorsement of social constructions by individual respondents allows the assessment of their implications for policy opinions in regression analysis.
MULTIDIMENSIONAL CONSTRUCTION OF IMMIGRANT (IL)LEGALITY
We measure multidimensional social constructions and assess their political implications using the case of immigrant (il)legality. This case is politically important because opinions on immigration profoundly impact mass partisanship and voting behavior among non-Hispanic white Americans (Abrajano and Hajnal Reference Abrajano and Hajnal2015). Members of the US public are particularly opposed to illegal immigration (Wright, Levy, and Citrin Reference Wright, Levy and Citrin2016), and negativity toward immigrants without legal status spills over to the broader category of immigrants (McCabe, Matos, and Walker Reference McCabe, Matos and Walker2021). Americans’ opinions on policies that target immigrants without legal status have been studied from the perspective of voters’ identities and media framing (Frasure-Yokley and Wilcox-Archuleta Reference Frasure-Yokley and Wilcox-Archuleta2019; Merolla, Ramakrishnan, and Haynes Reference Merolla, Ramakrishnan and Haynes2013). Our study approaches the same topic from the social-construction perspective.
Immigrants without legal status represent a good application of social construction theory for at least three reasons. First, the group in question is a product of legal construction—the category “illegal immigrants” simply would not exist without laws and regulations around border crossing and citizenship. Second, immigrants without legal status are the target of numerous federal, state, and local policies, including the Deferred Action for Childhood Arrivals (DACA) initiative; building the fence on the US–Mexican border; laws such as Arizona SB 1070 that require state law-enforcement officers to inquire about immigration status; and local-level sanctuary regulations. Third, Americans’ opinions about immigration policy are derived from a complex combination of beliefs that are not summarized easily in terms of positivity versus negativity.
Which dimensions of those beliefs are likely to be prominent in both the content of social constructions and their political implications? Existing literature suggests several potential answers to this question. Opposition to immigration among white Americans is predicted by both implicit and explicit anti-Hispanic prejudice (Perez Reference Perez2010; Valentino, Brader, and Jardina Reference Valentino, Brader and Jardina2013). This indicates the presence of a connection in people’s minds between being an immigrant and being Hispanic. Furthermore, whites who view immigrants as Hispanics may be more likely to support restrictive immigration policies due to perceived status threat (Abrajano and Hajnal Reference Abrajano and Hajnal2015). It is important to note that this evidence is only circumstantial, and the degree to which immigrants in the United States are viewed as Hispanics has not been measured directly.
On the other hand, Americans’ preferences for skilled and law-abiding immigrants are stronger than preferences for European origins, and this holds independently of respondents’ partisanship (Hainmueller and Hopkins Reference Hainmueller and Hopkins2015). There also is an argument in favor of value-based explanations: Americans are most welcoming to immigrants who show their readiness to assimilate by learning English and becoming self-sufficient (Levy and Wright Reference Levy and Wright2020). Assimilation turns out to be a more important factor in attitudes toward immigration than racial cues, although the latter still have non-trivial effects (Ostfeld Reference Ostfeld2017). Therefore, social constructions that link immigrants to education or the ethos of hard work may be consequential for immigration-policy opinions.
An interesting debate in the literature concerns the link between immigration and welfare. Scholars have recently noticed the increasing overlap between attitudes toward immigration and support for welfare spending (Garand, Xu, and Davis Reference Garand, Xu and Davis2017). The explanation for this relationship is the fear that immigrants may take advantage of the US welfare system, which has become known as the “immigrationization” of welfare attitudes. Other scholars challenge this interpretation and argue that both anti-welfare and anti-immigration attitudes stem from a broader conservative ideology (Levy Reference Levy2021). Recent findings partially support the immigrationization hypothesis by showing that stereotypes linking welfare and immigration are not widespread but nevertheless politically consequential (Myers, Zhirkov, and Lunz Trujillo Reference Myers, Zhirkov and Trujillo2024).
What role do attributes such as race/ethnicity, education, criminality, English proficiency, and welfare dependency have in the social construction of immigrant (il)legality? Sociology scholars have addressed this question using conjoint-experimental design and found that white Americans associate the absence of legal status with reliance on welfare, Hispanic origins, criminal record, poor English fluency, and low education (Flores and Schachter Reference Flores and Schachter2018). However, the existing research on the social construction of (il)legality is conducted only at the aggregate level. It describes the rich content of average perceptions about immigrants without legal status but does not measure these perceptions on the individual level. As a result, researchers cannot explore how images of target groups impact policy opinions—and therefore do not implement a full test of social construction theory. We use the most recent developments in the same method (i.e., conjoint analysis) to measure social constructions of immigrant (il)legality on the individual level and to assess their implications for preferences regarding immigration-enforcement policy.
POLITICAL IMPLICATIONS OF MULTIDIMENSIONAL CONSTRUCTIONS
As we have discussed, prior research has demonstrated that the social construction of immigrant (il)legality is multidimensional. It also suggests that by focusing on the deservingness attribute, the standard variant of the social-construction framework can miss important dimensions of group conceptions that are not easily reducible to the positive-versus-negative dichotomy (e.g., race and gender). Moreover, existing research describes social constructions only in the aggregate and therefore cannot assess the implications for opinions on relevant policies.Footnote 3 This study goes further by exploring the degree to which individuals endorse social constructions of target groups in a multidimensional framework using a conjoint experiment. We use individual-level variation in endorsement of social constructions to predict opinions on policies that implicate relevant target groups using immigrant (il)legality as our case study.
We use individual-level variation in endorsement of social constructions to predict opinions on policies that implicate relevant target groups using immigrant (il)legality as our case study.
Conceptually, this study extends the political science literature on the consequences of social constructions for policies that target implicated groups, aligning it with recent sociological literature on the measurement of social constructions. Our empirical analysis conceptually replicates and extends the conjoint experiment on the social construction of immigrant legal status (Flores and Schachter Reference Flores and Schachter2018), but it also makes important design adjustments that enable us to estimate individual-level variation in these constructions. To assess the implications of individual-level variation in endorsement of social constructions, respondents are asked about support for several policies related to immigration enforcement. We then use individual-level differences in social constructions of (il)legality to predict policy opinions.Footnote 4 This design is informed by an expectation that non-Hispanic whites who associate the absence of legal status with nonwhite ethnicity, fiscal burden, and criminal behavior would show support for strict immigration enforcement.
DATA AND METHODS
We recruited non-Hispanic white US adults for our online survey study fielded in December 2021 using the Lucid panel (Coppock and McClellan Reference Coppock and McClellan2019). The survey, including the conjoint task, was completed by 935 respondents.Footnote 5 The sample characteristics were as follows: mean age was 48.3 years; gender ratio was 49.1% male to 50.9% female; median income was $45,000 to $49,999; college education was reported by 41.7% of respondents; and 31.1% were Democrats, 36.8% were Republicans, and 32.1% were independents (Zhirkov and Van De Hey Reference Zhirkov and Van De Hey2025).
In the conjoint-experimental part of the survey, respondents rated hypothetical profiles by the likelihood of belonging to an “illegal/undocumented immigrant” group using a scale from 0 = Extremely unlikely to 10 = Extremely likely.Footnote 6 Each respondent was asked to rate 20 profiles.Footnote 7 Profiles were described in terms of eight attributes: race/ethnicity, gender, age, police record, receipt of government benefits, years stayed in the United States, English fluency, and level of education.Footnote 8 Attribute values were fully and independently randomized.Footnote 9 Table 1 presents the full list of attributes and potential values.
Table 1 Attributes for Profiles in the Conjoint Experiment
Notes: “Age” and “Stay in the United States” values (in years) were randomly chosen from the specified intervals. Collapsed values are in italics.
In conjoint profiles shown to the respondents, we used varying attribute labels to make them more realistic (e.g., using names of specific government programs). However, the analysis contrasted only dichotomized attribute values (except for the race/ethnicity attribute, which was included with its four randomized values). Dichotomization was necessary to reliably estimate IMCEs—according to current guidelines, the number of values per attribute should be minimized in either the design or the analysis stage (Zhirkov Reference Zhirkov2022). In our design and analysis, we implemented dichotomization in a way that maximized statistical power and thus the reliability of the resulting IMCE estimates.
For categorical attributes (i.e., police record, government benefits, and education), we collapsed values to ensure an approximate 50/50 split that maximized the respective variances. In the design stage, we assigned equal probabilities to “no” and “yes” categories for government-benefits and police-record attributes. Each specific welfare program and crime had an equal chance of being presented in the “yes” category. For the education attribute, we contrasted having (college or graduate degree) and not having (complete or incomplete high school) tertiary education. For numeric attributes (i.e., age and years of stay in the United States), we implemented uniform distributions in the design stage and divided the values by the median in the analysis stage. This ensured an approximate 50/50 ratio between the pairs of categories and maximized the variance. Figure 1 is an example of a conjoint profile.
Figure 1 Sample Screenshot from the Conjoint Task
Opinions about immigration enforcement were measured using a four-item battery that asked respondents about their support for specific policies implying both more-lenient and more-strict measures: (1) the DACA program; (2) bills requiring state law-enforcement officers to inquire about immigration status similar to Arizona SB 1070; (3) sanctuary initiatives; and (4) building the wall on the US southern border. Answers were rated on a 7-point Likert-type scale from 1 = Strongly oppose to 7 = Strongly support. We reversed items that asked about support for more-lenient policies and then computed an index of support for strict immigration enforcement by taking the average of the four opinions. See the online appendix for the exact question formulations.
RESULTS
We begin by implementing the standard procedure for conjoint experiments: estimating the effects of different profile attributes on the probability of being categorized as an immigrant without legal status. Results are presented in figure 2. Standard errors are clustered on the level of individual respondents. Estimates suggest that social construction of immigrant (il)legality among non-Hispanic white Americans is multidimensional such that individuals described as Hispanic or Asian, male, having a criminal record or shorter stays in the United States, speaking English poorly, or less educated are considered more likely to not have legal status. Immigrants who are described as Black and receiving government benefits are considered less likely to not have legal status. Age has no significant or sizable effect.
Figure 2 Conjoint Results: Effects of Different Attribute Values on a Profile’s Estimated Likelihood of Belonging to an Immigrant Without Legal Status
Point estimates with 95% confidence intervals.
Next, we estimate IMCEs as the measures of respondents’ endorsement of social constructions about immigrant (il)legality and use them to predict opinions on immigration-enforcement policies. IMCEs measure the direction and intensity of perceptions linking immigrants without legal status to the corresponding attributes for individual respondents. For instance, a negative IMCE value for the “receive benefits” attribute indicates that a respondent believes that people on welfare are less likely to not have legal status and vice versa. Greater IMCE magnitudes, in turn, indicate a stronger belief in these perceptions.
The results for models with and without controls (i.e., demographics and partisanship) are presented in figure 3. The dependent variable in this analysis is the index of support for stricter immigration policies constructed from the four policy opinions: opposition to DACA and sanctuary initiatives and support for border wall and state immigration enforcement. It is recoded to the same scale as IMCEs (from -10 to 10) so that coefficients can range from approximately -1 to 1. One dimension is consequential independent of the model specification: associating the absence of legal status with receiving government benefits significantly predicts support for stricter immigration enforcement at the 95% confidence level. The second strongest effect is exhibited by perceiving immigrants without legal status as Hispanics.Footnote 10
Figure 3 OLS Regression Results Predicting Support for Stricter Immigration Enforcement Using Conjoint IMCEs
Point estimates with 95% confidence intervals. Control variables (not presented due to space considerations): age, gender, education, income, and partisanship.
These results reveal the contested nature of multidimensional social constructions and the important political consequences of variation in the degree to which individuals endorse them. Whereas on average, white Americans do not associate the absence of legal status with receiving government benefits, those who believe this association express stronger support for stricter immigration enforcement.Footnote 11 This finding highlights the importance of exploring individual-level endorsements of social constructions and their implications instead of simply describing them in the aggregate. Even social constructions not endorsed by a majority of the public can have significant effects on policy opinions. This phenomenon is illustrated in figure 4, which presents the estimated distribution of perceptions linking the absence of legal status and welfare (with a negative mean) on the left side and the (positive) bivariate relationship of perceptions linking the absence of legal status and welfare with support for strict immigration enforcement on the right side.
Figure 4 Estimated Distribution of IMCEs for the “Receives Benefits” Attribute (Left) and its Relationship with Preferences for Stricter Immigration Enforcement (Right)
Positive IMCE=unauthorized immigrants are viewed as more likely to receive benefits. Negative IMCE=unauthorized immigrants are viewed as less likely to receive benefits. Estimated linear effects presented with a 95% confidence interval.
CONCLUSION
This article extends the existing research on the social construction of target groups and its consequences for politics and policy (Schneider and Ingram Reference Schneider and Ingram1993). Because recent findings demonstrate that social constructions are contentious rather than uniformly accepted (Kreitzer and Smith Reference Kreitzer and Smith2018), we suggest that individual differences in their endorsement can be consequential for policy opinions. Following recent research on the measurement of stereotypes (Myers, Zhirkov, and Lunz Trujillo Reference Myers, Zhirkov and Trujillo2024), we also argue that social constructions are multidimensional and that some of these dimensions (e.g., race and gender) may not be captured by the standard deservingness axis.
Our analysis explores the multidimensionality of social constructions and their consequences for policy opinions using the example of immigrant (il)legality. Specifically, we conceptually replicate a conjoint experiment that measures social construction of immigrant (il)legality among non-Hispanic whites in the United States (Flores and Schachter Reference Flores and Schachter2018), with necessary design adjustments. Corroborating previous findings, we show that the social construction of immigrant (il)legality is multidimensional: the absence of legal status is associated with Hispanic origins, police record, poor English fluency, and less education. We also extend these results by estimating IMCEs from the conjoint experiment to measure the degree to which individuals endorse the social constructions in question. Using IMCEs in regression analysis, we demonstrate that non-Hispanic white Americans who associate the absence of legal status with the receipt of government benefits or with Hispanic ethnicity support stricter immigration-enforcement policies.
These results make several contributions to the literature on social construction of target groups. On the one hand, we corroborate the social construction approach by demonstrating that members of the public possess perceptions about social categories targeted by government policies that can be measured by researchers. Moreover, going beyond recent empirical studies on social construction (Kreitzer and Smith Reference Kreitzer and Smith2018), we show that the degree to which individuals endorse these perceptions predicts opinions on relevant policies. This finding directly substantiates one of the core tenets of social construction theory, thereby undercutting previous criticism of the theory as being non-falsifiable (Sabatier Reference Sabatier and Sabatier1999). On the other hand, we show that previous conceptualizations focused on perceived deservingness of target groups may have overlooked the multidimensionality of social constructions. In addition to exploring different dimensions of deservingness (e.g., economic productivity versus criminal behavior), we highlight how race and ethnicity may be an important component of social constructions. Building on previous studies that describe multidimensional social constructions (Flores and Schachter Reference Flores and Schachter2018), we demonstrate that different dimensions have disparate political consequences. It is important to note that social constructions of target groups do not have to be universally or even broadly endorsed to have implications for policy opinions.
In addition to exploring different dimensions of deservingness (e.g., economic productivity versus criminal behavior), we highlight how race and ethnicity may be an important component of social constructions.
Because our study focuses on immigrants without legal status, it also addresses important debates in the literature on the politics of immigration. First, scholars have long noticed spillover effects from negative attitudes toward Hispanics to restrictive immigration preferences among white Americans (Perez Reference Perez2010; Valentino, Brader, and Jardina Reference Valentino, Brader and Jardina2013). Our study directly measures the degree to which respondents connect perceived legal status with race/ethnicity and confirms that immigrants described as Hispanic are more likely to be viewed by non-Hispanic white respondents as being in the country illegally. Second, the conjoint-experimental design enables us to directly compare the role of race/ethnicity and attributes related to economic productivity in both the social construction of immigrants without legal status and the political implications of that construction. Our results reveal the non-trivial importance of race/ethnicity in the social construction of immigrant (il)legality, thereby suggesting that anti-immigration politics in the United States stems at least in part from racial and ethnic prejudice.
Our findings have implications for the debate around immigrationization of welfare in American politics (Garand, Xu, and Davis Reference Garand, Xu and Davis2017; Levy Reference Levy2021). They suggest that a related but distinct phenomenon may be taking place: “welfarization” of immigration, or opposition to immigration informed by the perception that immigrants take advantage of welfare. The welfare argument against immigration has a history of being used in US politics: California Proposition 187 and Trump’s revision of the public-charge rule are examples. Therefore, it may not be surprising that beliefs about immigrants’ welfare dependency affect how the public views immigration.
Existing literature shows that policy shifts are possible following successful changes in the content of social constructions caused by rhetoric from politicians and the media. For instance, the passage of the Fair Housing Act in 1968 was made possible in part because its beneficiaries were constructed successfully as “deserving” members of the African American community (Sidney Reference Sidney2001). Can the same logic apply to immigrants without legal status and immigration-enforcement policies? Our results suggest that the two most consequential dimensions of the target group’s construction are Hispanic ethnicity and welfare dependency. Therefore, portraying immigrants as diverse and self-sufficient may shift opinions in favor of more-lenient policies. It is necessary to note, however, that our study has not addressed directly the origins and malleability of social constructions and therefore cannot assess the effort necessary to change the perceptions in question.
Methodologically, the conjoint-based approach to measuring social construction of target groups has several important advantages. It enables the simultaneous measurement of multiple dimensions of social construction, including attributes that cannot be easily reduced to the deserving-versus-underserving dichotomy (e.g., race and gender). Recent advances in conjoint methodology also make it possible to explore heterogeneity in social constructions as well as the political implications of this heterogeneity.
When using conjoint experiments to measure social constructions, scholars should be aware of an important tradeoff that has implications for research design. The two main benefits of conjoint analysis are the means to describe multiple dimensions of social constructions and to explore their implications for policy opinions. Unfortunately, these two goals imply somewhat different conjoint designs. If the goal is describing social constructions in detail, researchers must include many meaningful values per attribute (see, e.g., Flores and Schachter Reference Flores and Schachter2018). If, however, the goal is exploring the implications of social constructions for policy opinions, the number of values per attribute should be minimized (Zhirkov Reference Zhirkov2022), as done in the conjoint experiment presented in this article.
Some of the limitations of our design provide an opportunity for future work in this important area. For instance, it may be interesting to explore whether the social construction of immigrants without legal status is different among nonwhite Americans. Even if members of racial and ethnic minority groups also view immigrants as Hispanics, this perception may have political implications that are different from those observed among whites. Answering that question, however, would require oversamples of Asian, Black, and Hispanic respondents. Similarly, future studies may attempt to investigate additional dimensions of social construction, such as religion. Given the negative construction of Muslims in America (Kalkan, Layman, and Uslaner Reference Kalkan, Layman and Uslaner2009; Lajevardi Reference Lajevardi2020; Oskooii, Dana, and Barreto Reference Oskooii, Dana and Barreto2021), one may expect it to overlap with “illegality” in important ways.
Overall, our results continue the line of research that addresses the applicability of the social construction theory in empirical research to politics and policy. Our analysis based on the conjoint-experimental method demonstrates that researchers can measure multiple dimensions of social constructions and assess their political implications. We show that social constructions impact policy opinions—and, thus, potentially policy outcomes. This is yet another illustration of how combining insights from policy studies, public opinion, and political methodology can lead to advances in both theory and measurement.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit http://doi.org/10.1017/S1049096525000034.
ACKNOWLEDGMENTS
This research was supported by the Latin American Public Opinion Project Lab at Vanderbilt University and the Democracy Initiative at the University of Virginia.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the PS: Political Science & Politics Harvard Dataverse at https://doi.org/10.7910/DVN/HPFSZJ.
CONFLICTS OF INTEREST
The authors declare that there are no ethical issues or conflicts of interest in this research.
Open access
