English (pdf)
Article in xml format
Article references
Automatic translation
Permalink
Political polarization has been a phenomenon identified in different countries, characterized by the intensification of negative feelings between rival political groups (Fuks & Marques, 2022). This polarization manifests itself in a distinction between our own political group and the others. Various elements can contribute to these identity boundaries, such as gender (Jones & Brewer, 2019), ideology, religion, ethnicity, and skin color (Mason & Wronski, 2018; Mason,2023). In this research, we will specifically analyze the relationship between information about a person’s political preferences and their categorization into racialized social groups. Therefore, we will evaluate political orientation as a marker of racial perception.
The literature commonly distinguishes between two types of polarization: ideological polarization and affective polarization. Ideological polarization refers to increased ideological divergence and reduced dialogue between individuals with different political views. Affective polarization, on the other hand, involves the intensification of negative emotions between opposing groups, manifesting itself in attitudes of hostility or mistrust towards the other group, based more on social identity than on ideological differences (Fuks & Marques, 2022). Specifically in Brazil, the existence of greater affective polarization than ideological polarization has been identified, which seems to have negative consequences for democracy, such as the delegitimization of governments (Franco & Pound, 2022). It is also noteworthy that affective political polarization has had a negative impact on the challenges faced during the pandemic (Modesto et al., 2020), so it is understood that, in the Brazilian context, this phenomenon has been acknowledged and documented.
In this way, we understand polarization when differences in society align along a single dimension, leading to a convergence of positions and people increasingly perceive and describe politics and society in terms of “Us” versus “Them” (McCoy, 2024). In this sense, Social Identity Theory presents itself as a framework for understanding political phenomena.
Social Identity Theory (SIT) emphasizes that collective phenomena cannot be adequately explained by individual processes alone, and that it is essential to consider social identity, defined as the part of the self-concept derived from belonging to a group, the emotional value attributed to this belonging and the social value associated with the group (Hogg, 2006).
Group formation involves cognitive processes, such as categorization and social comparison, which distinguish one’s own group (in-group) from others (out-group) and influence individuals’ perceptions and attitudes towards other groups (Brown, 2020). These processes can lead to depersonalization and favoritism, exacerbating differences between groups (Hogg, 2006). It should be noted that identities are emergent and context-specific results of the interaction between the observer and social reality, i.e. the extent to which categorization becomes prominent and relevant in a given context (Oakes, 2002). Therefore, when considering political polarization, SIT provides a framework for understanding how belonging to social groups shapes political perceptions and attitudes.
In Brazil, different identity markers have been used to delimit the “borders” of political identities, which favors polarization. For example, adherence to religious values and valuing the family, distrust of institutions and a conservatism aimed at maintaining traditional values (Sousa et al., 2022).
Although the factors indicated as delimiting the boundaries for political polarization make an important contribution, in the Brazilian context, the issue of race is also relevant. It should be noted that racial classification in the country is not a consensus, depending very much on how we define who is black in Brazil (Silva & Leão, 2012). Traditionally, blacks have been understood as a group of browns (black and white mixed race) and blacks, and this categorization can occur through self-classification or heteroclassification (Petruccelli & Saboia, 2013). In addition, examining data from the Brazilian Institute of Geography and Statistics (IBGE) for 2022 shows that the population is more mixed, increasingly self-declaring as “non-white” compared to 2012. We should point out that, in 2022, 55.9% of the population declared themselves black, in other words, more than half of the Brazilian population (Brazilian Institute of Geography and Statistics [IBGE], 2023).
Despite being the majority of the population, we find an under-representation of blacks in both the Chamber of Deputies and the Chamber of Councilors, who represented 24.4% of federal deputies and 28.9% of state deputies in 2018, and 42.1% of councilors in 2016 (Supreme Electoral Tribunal [TSE], 2019). Furthermore, when we mention black and brown women, the proportion is even lower, making up 2.5% of federal deputies and 4.8% of state deputies in 2018, and 5.0% of councilors in 2016 (TSE, 2019).
In addition to racial inequality in the political sphere, it is interesting to note how racial perception has an impact on the discussion of politics. For example, there is evidence, according to studies in the United States, that we tend to perceive black candidates as more liberal than white candidates who adopt similar political positions, that is, belonging to certain racial groups affects how citizens evaluate the ideologies of candidates (Jacobsmeier, 2015). In addition, similar results were found by Fulton and Gershon (2018), suggesting that white voters perceive candidates from minority groups (blacks and Latinos) as more liberal and ideologically distant than non-minority candidates. In short, the literature indicates that the biased perception of political candidates, based on racial categorization, influences voting decisions and choices.
Beyond the evidence of empirical studies developed in other cultural contexts, specifically in the Brazilian context, it is worth noting that in the mid-1970s and 1980s, the activism of the black movement occupied spaces of representation through active participation in trade union organizations and center-left parties, and promoted a rapprochement with some parties, such as the Democratic Labor Party (PDT), the Workers’ Party (PT) and the Brazilian Democratic Movement Party (PMDB), which, through black commissions, contributed to the inclusion of the movement’s demands in political discussions (Rodrigues & Pereira,2022).
Thus, in addition to the findings of studies in other cultural contexts, specifically in Brazil, the most progressive positions are often linked to black communities that have emerged from the struggle for equality and social justice. However, it is important to note that these links are not absolute rules and that there are white and black politicians with different political orientations. It is also important to note that political parties and government institutions tend to incorporate only a portion of black leaders, i.e. those who are affiliated to or sympathetic to the parties in power, leading to the exclusion of both opposition leaders and those who have no party affiliation (Guimarães, 2001).
Despite this, not all people belong to easily discernible social categories. In addition, as already mentioned, in Brazil there is a growing self-declared brown population, and studies on the perception of brown or multiracial individuals, within psychology and in the Brazilian context, are scarce (Chen et al., 2018). Although people place others in social categories, such as belonging to racialized groups (e.g. skin color), because these categories help to organize and simplify the social environment, the categorization of multiracial individuals is not so simple, and the decision about this categorization depends on who perceives and various contextual factors (Chen, 2019; Pauker et al., 2018). Studies show that the characteristics of the person being evaluated, such as socioeconomic status (Pauker et al., 2018) or a caption that indicated words related to a certain racial group (MacLin & Malpass, 2001; Tskhay & Rule, 2015), alter how they are categorized, that is, they are perceived according to the marker or contextual clues. Therefore, this study investigated the impact of information about an individual’s political orientation on the categorization into racialized social groups that is attributed to them.
We believe that this study in the Brazilian context is highly relevant, mainly due to the lack of studies with non-WEIRD (Western, Educated, Industrialized, Rich, and Democratic) samples on the subject (Muthukrishna et al., 2020). Most of the studies reviewed come from an American context in which racial identity is understood as hypodescendant (Pauker et al., 2018), that is, it is understood that an individual’s ancestry is central to determining their racial group, that is, regardless of their other ancestries or their appearance (Chen et al., 2018). That said, it is necessary to explore other cultural contexts in which hypodescendence is not the norm for categorization into racialized groups, such as in Brazil.
Given the above, considering that the literature reviewed points to a tendency to perceive black candidates as liberal (Jacobsmeier, 2015), as well as that external variables can alter the racial perception of multiracial people (MacLin & Malpass, 2001), we proposed the current research. As mentioned above, this research aimed to investigate the impact of information about an individual’s political orientation on their categorization into racialized social groups. We therefore hypothesized that (H1) a brown person with a left-aligned statement will be perceived as blacker; and that (H2) a brown person with a right-aligned statement will be perceived as whiter. Our hypotheses were based on previous studies that point out that political ideology can alter the way people are racially categorized (Fulton & Gershon, 2018; Jacobsmeier, 2015), as well as the understanding that in Brazil left-wing parties have a closer relationship with the black movement (Rodrigues & Pereira, 2022).
Overview of Studies
To test the proposed hypotheses, two studies were conducted, each preceded by a preliminary study. The general aim of the studies was to test the impact of information about an individual’s political orientation on the categorization into racialized social groups that is attributed to them. In Study 1, a photo of a male face was used as the candidate, and in Study 2, a photo of a female face was used, seeking to broaden the generalizability of the results of Study 1. In both studies, participants were exposed to different scenarios in which the supposed candidate made right- or left-aligned statements on topics such as public health, education, and tax laws.
Method
Participants
A total of 576 people answered the questionnaire, and one person was removed for being under 18. To check the suitability of the sample for the study’s analyses, ana posteriori power analysis was carried out using the G*Power software version 3.1.9.7 (Faul et al., 2009). The sample has power = .80 to detect an effect equal to or greater than d = .17 with p < .05. In the left-handed experimental condition (N = 293), ages ranged from 18 to 74 years (M= 39.01; SD= 12.58), with the majority identifying as female (68.3%), white (60.1%). In terms of regions, the Southeast and South concentrated most of the participants, with 146 from the Southeast (49.8%) and 83 from the South (28.3%). The other regions had 22 participants from the Midwest (7.51%), 30 from the Northeast (10.2%) and 12 from the North (4.10%). The majority of participants declared themselves to be “Far-left”, 131 (44.79%), or “Slightly left-wing”, 115 (39.2%). A portion of 26 participants (8.47%) identified themselves as “Center”, while 15 participants (5.12%) positioned themselves as “Slightly right-wing” and 6 participants (2.05%) as “Far-right”. In the right-wing experimental condition (N = 282), ages ranged from 18 to 73 years (M= 39.05; SD = 12.6), with the majority identifying as female (70.6%) and white (59.2%). In terms of regions, the Southeast and South concentrated most of the participants, with 150 from the Southeast (53.2%) and 73 from the South (25.9%). The other regions had 17 participants from the Midwest (6.03%), 31 from the Northeast (11%) and 11 from the North (3.90%). The majority of participants declared themselves to be “Far-left”, 133 (47.2%), and “Slightly left wing”, 98 (34.8%). A portion of 29 participants (10.3%) identified themselves as “Center”, while 15 participants (5.32%) positioned themselves as “Slightly right-wing” and 7 (2.48%) as “Far-right”.
Instruments
Selected face: In Study 1, a photo of a male face was used as a candidate. We carried out a pre-test in which 15 faces were presented and only one was selected. To choose it, a one-sample t-test was carried out, using the midpoint (5.5) as a reference (for more information: https://osf.io/keunr). The face that differed the least from this point was selected (to view the images see: https://osf.io/yeuhs). The faces used, 100% synthetic, were provided by the Generated Photos academic database (https://generated.photos/datasets).
Scenarios: To expose the supposed candidate’s political statement, two scenarios were created, in which a supposed politician makes a statement aligned with the left and another with the right, the content of which dealt with the topics: public health, education, and tax laws.
Experimental manipulation check (Candidate’s Political Position): As a manipulation check item, after reading the scenario, participants were asked to indicate the candidate’s political position on a scale ranging from 1 (Far-left) to 5 (Far-right).
Categorization of Racial Belonging: As a measure of racial perception, the participants were asked to classify the politician on a scale ranging from 1 (he is definitely white) to 10 (he is definitely black).
Political identification: To assess their own political identity, participants answered a scale ranging from 1 (far-left) to 5 (far-right).
Sociodemographic data: The following sociodemographic data were collected: age, gender, racial self-categorization, and region of the country in which they live.
Procedures
Data collection. The questionnaire was promoted (WhatsApp and Instagram) and boosted (Facebook) on social networks via an electronic form. Before starting the test, participants were asked to accept the terms and conditions of the research. In this part, they were informed that the data collected was anonymous, the risks involved, and the voluntary nature of participation were indicated, and the person could withdraw at any time. The participants then provided their sociodemographic data. Next, participants were automatically allocated to one of the experimental conditions (candidate position: Right vs. Left-wing). Participants then examined a photo of a brown person next to a political statement and answered a sequence of items in which they had to indicate the political position of the candidate, a measure of racial perception, intention to vote for the candidate, and indicate their own political identity. The survey was conducted between 05/19/2022 and 06/15/2022.
Data analysis. The data obtained was analyzed using the R programming language software (R Core Team). Firstly, descriptive analyses were carried out on all the study variables. A Student’s t-test was used to check whether the manipulation of information on the candidate’s political position was adequate. To assess the mean difference in the categorization of the politician’s racial belonging, a linear regression was performed. The model included the categorization of the candidate’s racial belonging as the dependent variable, and the independent variables were the manipulation of information about the candidate’s political position (Left vs. Right), the participants’ racial self-categorization (coded as White and Non-White), the participants’ political identification (centered on the mean), the participants’ age (centered on the mean) and the participants’ gender (Female vs. Male).
In addition to the main effects, second and third order interactions were included in the model. Specifically, second-order interactions were tested between the manipulation of information about the candidate’s political position and the participants’ racial self-categorization, between the manipulation of information and the participants’ political identification, and between the participants’ racial self-categorization and political identification. A third-order interaction was also tested between the manipulation of information about political position, participants’ racial/ethnic identity and political identification. These interactions were included to explore whether the effect of the experimental conditions on the categorization of the candidate’s racial belonging varies as a function of the interaction between racial self-categorization and political identification, controlling for age and gender.
Ethical Considerations
This research was conducted under ethical principles, approved by the Research Ethics Committee of the Centro Universitário de Brasília - UNICEUB (CAAE No. 64092022.6.0000.0023) and all the necessary measures were taken to ensure respect for the rights and privacy of the participants. Informed consent was obtained and the data were treated confidentially.
Results
Firstly, a Student’s t-test was carried out to check whether the experimental manipulation was adequate. The two-sample t-test indicated that there is a difference in the perception of political positioning between the Right (M = 4.30, SD = 0.90) and Left (M = 1.98, SD = 0.82) conditions, which suggests that the effect is positive, statistically significant and large (difference = 2.33, 95% CI [2.18, 2.47], t (573) = 32.34, p< 0.001; Cohen’s d = 2.70, 95% CI [2.48, 2.93]). In other words, participants in the left-wing condition perceived the politician as having a more left-wing political stance and in the right-wing condition perceived the politician as having a more right-wing political stance. In other words, we can assume that the manipulation is adequate. Table 1 shows the parameters estimated by the regression model used to test the proposed hypotheses.
Table 1. Regression Models on Categorization of Racial Belonging
Note. The categorical variables were coded as follows: Manipulation (-.5 = Right-wing candidate; .5 = Left-wing candidate); gender (-.5 Female; .5 = Male); Racial self-categorization (-.5 = White; .5 = Non-white). The other variables were centered on their respective means. *p < .05. **p < .01. ***p <.001.
The results confirm the influence of the manipulation of the candidate’s political identification on their racial categorization. However, the participants’ political identification, age and the interactions between the variables did not show statistically significant main effects.
The adjusted means indicate that, controlling for other variables, participants in the Left-wing condition categorized the target candidate as being blacker (M = 6.37, EP = 0.146) than participants in the Right-wing condition (M = 5.22, EP = 0.147). This indicates that the politician presented in the Left-wing experimental condition is associated with a racial categorization as blacker.
The association between the gender of the participants and the racial categorization of the candidate was significant, indicating that participants who identified with the female gender had a racial perception of the candidate as blacker (M= 6.12, EP = 0.117) compared to men (M = 5.47, EP = 0.183).
Discussion
The aim of this research was to investigate the impact of information about an individual’s political orientation on categorization in racialized social groups. The hypotheses were that a brown person with a statement aligned to the left will be perceived as blacker; and that a brown person with a statement aligned to the right will be perceived as more white. The results showed that when classifying on a continuum between black and white, the same politician with different political orientation information (Left vs. Right) had their racial categorization altered. The brown politician with information about the left-wing political orientation was perceived as blacker than the one with information about the right-wing political orientation, pointing to the influence of the political context on racial categorization. This result is consistent with the literature in which, in the presence of contextual stimuli, identical faces can be perceived as faces from different racialized groups (MacLin & Malpass, 2001).
In addition to the effect of the experimental manipulation, the gender of the participants was also associated with the racial categorization of the candidate. Participants who identified with the female gender perceived the candidate as blacker compared to participants who identified with the male gender. This result can be understood by considering that the salience of group identity is not only a matter of the perceptual salience of relevant cues but is also related to the actual psychological meaning of belonging to a group (Oakes, 2002). This aspect is especially relevant when considering racial categorization associated with gender. Thus, female voters may perceive greater proximity or identification with candidates who represent or symbolize the fight against inequality, due to a stronger identification with groups that suffer discrimination. This identification could lead to a more pronounced perception of the racial characteristics of the candidates, especially in scenarios that reinforce these links, such as political alignment with the left (Rodrigues & Pereira, 2022).
Given the relevance of the findings of Study 1, especially the influence of the manipulation of information on political orientation and gender on the racial categorization of the brown political candidate, the question arises as to whether this effect persists when female faces are also analyzed. To address this question, Study 2 was conducted, which will follow the same methodological procedures as Study 1, including the preliminary analysis of the faces. However, unlike Study 1, images of female faces will be used. This additional step will allow us to investigate whether the effect found previously is generalizable to both genders or whether there are significant variations when it comes to the racial categorization of female candidates, especially considering that there is a lack of representation of black and brown women in Brazilian politics (Oliveira, 2021).
Method
Participants
The sample has power = .80 to detect an effect equal to or greater than d = .20 with p < .05. In the left-handed experimental condition (N = 211), ages ranged from 19 to 89 years (M = 57.6; SD = 15.5), with the majority identifying as female (71.6%), white (78.2%). In terms of regions, the Southeast and South concentrated most of the participants, with 94 from the Southeast (44.5%) and 66 from the South (31.3%). The other regions had 33 from the Midwest (15.6%), 16 from the Northeast (7.58%) and 2 from the North (0.94%). The participants self-declared politically as “Far-left” with 53 participants (25.1%) and “Slightly left-wing” with 45 participants (21.3%), 43 participants (20.9%) identified themselves as “Center”, while 21 participants (9.95%) positioned themselves as “Slightly right-wing” and “Far-right” with 48 participants (22.7%). In the right-wing experimental condition (N = 189), ages ranged from 18 to 86 years (M = 53.3; SD = 16.8), with the majority identifying as female (59.8%) and white (70.4%). In terms of regions, the Southeast and South concentrated most of the participants, with 82 from the Southeast (43.4%) and 55 from the South (29.1%). The other regions had 27 from the Midwest (14.3%), 20 from the Northeast (10.6%) and 5 from the North (2.65%). The participants self-declared themselves as “Far-left” with 43 participants (22.8%) and “Slightly left-wing” with 35 participants (18.5%), 31 participants (16.4%) identified themselves as “Center”, while 30 participants (15.9%) positioned themselves as “Slightly right-wing” and “Far-right” with 50 participants (26.5%).
Instruments
The same instruments were used as in Study 1, but the male items in the statements were changed to female. (The chosen face can be viewed at: https://osf.io/yeuhs).
Results
Firstly, as in Study 1, a Student’s t-test was carried out to check whether the experimental manipulation was adequate. Similar to Study 1, there was a difference in the perception of political positioning between the Right (M = 3.81, SD= 1.27) and Left (M = 2.09, SD = 1.04) conditions, which suggests that the effect is positive, statistically significant and large (difference = 1.72, 95% CI [1.50, 1.95], t(398) = 14.92, p < 0.001; Cohen’s d = 1.50, 95% CI [1.27, 1.72]). Thus, participants in the left-wing condition perceived the candidate with a more left-wing political leaning, while in the right-wing condition they perceived her with a more right-wing leaning. Therefore, we can consider the manipulation to be appropriate. To evaluate the main hypothesis, a linear regression was also conducted to examine the influence of the manipulation on racial categorization, controlling for political identification, racial self-categorization, age, gender and the respective interaction terms (Table 1).
According to the hypothesis and as in Study 1, there is a difference in the racial categorization of the candidate between Left and Right-wing. In other words, the candidate on the left is perceived as blacker than the candidate on the right. On the other hand, age was also significantly related to racial categorization, suggesting that older participants tend to perceive the candidate as whiter. Other variables did not show significant relationships in isolation or in interactions in the model. The adjusted means indicate that, controlling for other variables, participants in the Left condition categorized the target candidate as blacker (M = 6.00, EP = 0.223) than participants in the Right condition (M = 5.23, EP = 0.203). This indicates that the Left experimental condition is associated with a racial categorization of the candidate as blacker.
Discussion
Study 2 sought to investigate whether the findings of Study 1 were replicated in the racial categorization of female candidates. The results indicate that, by classifying on a continuum between black and white, the candidate with different political orientation information (Left vs. Right-wing) had her racial categorization altered, as hypothesized. In other words, the results suggest that the racial categorizations of brown individuals tend to vary according to political declaration for both male and female candidates.
Unlike Study 1, the age variable was associated with the candidate’s racial categorization. Specifically, the analysis revealed that as the age of the participants increases, there is a tendency to perceive the candidate as whiter. This result suggests that racial perception is sensitive to ageing, possibly due to generational differences in the way racial categorization and identity are understood and internalized. Younger participants, who grew up in a different socio-political environment, may be more aware of issues involving racialized groups and may thus be more inclined to perceive the candidate as blacker, especially in a left-wing context.
This research aimed to investigate whether political orientation influences the racial categorization of brown political candidates. The hypotheses were that a brown person with a left-aligned statement would be perceived as blacker and that a brown person with a right-aligned statement would be perceived as whiter. Studies 1 and 2 showed that information about political orientation influences the racial categorization of brown candidates, revealing that left-aligned candidates are perceived as blacker compared to those aligned to the right. In addition, Study 1 showed that the gender of the participants is also related to racial categorization, which was not observed in Study 2. On the other hand, in Study 2, the age variable showed an association, unlike Study 1, highlighting the complexity of the interactions between the variables studied.
These findings suggest that racial categorization is not just a matter of the candidates’ individual characteristics but is also shaped by the political position they express. In terms of social identity theory, political orientation functions as a “marker” that acquires psychological and behavioral meaning from the interaction between the social-political context and the dispositions of observers. According to Oakes (2002), the salience of a group identity is not a characteristic of the individual or the situation in isolation but emerges from this interaction. Political and racial markers become psychologically significant only when these designations have a behavioral and perceptual reality for observers, providing clues that reinforce stereotypes or narratives associated with political groups.
The political landscape, therefore, can act as a “marker” that influences the perception of candidates’ racial characteristics, depending on who is observing and the contextual factors (Chen, 2019; Pauker et al., 2018). Racial categorization is not static, but dynamic, reflecting the interaction between the situational context and the predispositions of the observers. When a candidate is identified with a political group, racial characteristics can be inferred in a way that reinforces stereotypes or narratives associated with that group.
That said, it is necessary to consider some limitations. Firstly, the racially ambiguous faces were created digitally. This method of creating racially ambiguous faces was chosen for the highest level of experimental control (Gaither et al., 2018), and the faces were pilot tested to ensure that they were ambiguous. However, future studies should consider using photographs of individuals who are truly multiracial, as well as using multiracial politicians.
To improve understanding of the interaction effects of gender and political orientation on racial perception, future research could benefit from adopting a more comprehensive experimental design. A 2x2 model would allow for a more detailed analysis of the interactions between these variables, gender (Male vs. Female) and political orientation (Left vs. Right) in a single model. In Study 2, we chose to focus exclusively on female candidates to generalize the findings of Study 1, which focused on male candidates. This decision was made to explore the applicability of the initial results to a new context. However, the absence of a full 2x2 design limits the ability to examine complex interactions between gender and political orientation in a single analysis.
Finally, we relied mostly on a white and female sample, so caution should be exercised when generalizing these findings. Future studies should investigate how members of other racial groups perceive these racial boundaries, as well as consider other ethnic-racial groups, such as indigenous people.
Despite the limitations indicated and the possibilities for future studies pointed out, we believe that this research makes a theoretical contribution by deepening the understanding of how information about political orientation influences racial categorization, especially in a context of racial diversity such as Brazil. This is a relevant advance, as it expands the scarce literature in social psychology on the perceptions of racialized groups in Brazil (Chen et al., 2018). In addition to the theoretical contribution, we highlight some practical implications. As already mentioned, the population is increasingly mixed and mostly self-declares as “non-white”, while the presence of black people in politics, although still underrepresented, has grown (IBGE, 2023). Considering affirmative policies, such as quotas and electoral funds for women and blacks, it is relevant to investigate who these blacks are in politics, especially given the discrepancy between self-declaration and racial heterodeclaration in elections. Future studies should explore these complex interactions, contributing to a broader understanding of racial perceptions and their implications for political representation, promoting greater equity in the public sphere.
