Bolstering Policy Support for Disadvantaged Groups Through Humanization

Abstract

Cooley et al. and Hodson and Doucher show that individuals, individuals within groups, and groups evoke different levels of perceived humanity, and that these differences affect sympathy and willingness to help. In three preregistered experiments, we successfully replicate these findings in a different cultural context (Slovakia). We then test whether manipulating these depictions also affects support for policies that benefit the target. We focus on a disadvantaged ethnic minority (the Roma). Finally, we investigate whether internal (under the beneficiary’s control) versus external attribution (outside of the beneficiary’s control) is a mitigating factor. We confirm individuals and group-compositions evoke higher levels of policy support than groups through increases in perceived humanity. However, this relationship only holds under conditions of external attribution. To humanize disadvantaged groups and bolster policy support, advocates should center their communicative messages around individuals rather than unitary groups and avoid stereotype-enforcing internal attributions.

Keywords

policy support dehumanization attribution Roma replication

Bolstering majority support for policies that benefit disadvantaged groups can be difficult (Harell et al., 2016), especially when these groups are dehumanized and deemed responsible for their ordeal. Given the negative effects of dehumanization, scholars have developed strategies for out-group humanization (Vezzali et al., 2022). However, this work offers little evidence of how humanization improves prosocial behavior toward disadvantaged groups and of how humanization can bolster support for policies intended to benefit them. Cooley et al. (2017) and Hodson and Doucher (2020) introduced an effective approach to humanization: describing targets as individuals, individuals within a group (group-compositions), or groups led to different levels of perceived humanity, and these shifts had positive downstream consequences on sympathy and willingness to help. We replicate and extend these findings by examining whether manipulating target group depictions also has a positive effect on support for policies intended to benefit the disadvantaged group. Importantly, given that previous studies focused on situations of external attribution—that is, when the target was not personally responsible for their circumstance—we ask whether the identified relationships hold under situations of internal attribution—that is, when the target was responsible.

We conduct the replication and extension in the Slovak Republic. The substantive differences between samples (previous studies employed American MTurkers) render our analysis a tough test of cross-cultural external validity. Next, given that previous studies on humanization (e.g., Au & Ng, 2021; Cooley et al., 2017; Hodson & Doucher, 2020; Rai & Diermeier, 2015; Tang & Gray, 2021) have not focused on ethnic or racial minorities, we ask whether the positive effects of humanization hold when beneficiaries are members of a disadvantaged ethnic group: the Roma. The discrimination, stereotypes, and dehumanization faced by Roma have important implications for their effort attributions and perceived moral character (Kende et al., 2020; Kteily et al., 2015, Study 4). Thus, we consider our design a tough test of humanization. In sum, this article tests for the possible boundary conditions under which humanization occurs (Nosek & Lakens, 2014). Doing so provides tangible advice that policymakers, philanthropists, and advocates can use to craft communicative messages that effectively garner support for policies intended to benefit disadvantaged groups. We start by discussing the components of our theory.

Dehumanization

Dehumanization—“perceiving a person or group as lacking humanness” (Haslam & Loughnan, 2014, p. 401)—diminishes prosocial, helping behaviors (Andrighetto et al., 2014; Cuddy et al., 2007; Sainz, Loughnan, et al., 2020; Sainz, Martínez, et al., 2020; Zebel et al., 2008) and can even create support for instrumental violence (Rai et al., 2017). Dehumanized targets are denied experience (capacity to feel) and agency (capacity for thought) (Gray et al., 2007; Waytz et al., 2010), and are thus perceived as animalistic—lacking in human uniqueness (e.g., rationality, morality, civility)—and mechanistic—lacking in human nature (e.g., emotionality, warmth) (Haslam et al., 2008).

Disadvantaged groups are often subjected to animalistic dehumanization due to their immigration status (Esses et al., 2008), racial categorization (Goff et al., 2008), or socioeconomic status (SES) (Loughnan et al., 2014; Sainz, Martínez, Rodríguez-Bailón, & Moya, 2019). The animalistic dehumanization of low-SES groups, that is, considering them as “primitive, bestial, and incompletely human” (Loughnan et al., 2014, p. 54), “may contribute to justifying income inequality by considering poverty as a natural outcome of poor people being less evolved” (Sainz, Martínez, Rodríguez-Bailón, & Moya, 2019, p. 2). Bain et al. (2009) showed that animalistic and mechanistic forms of dehumanization reflect cultural stereotypes about national or ethnic groups. In addition, Andrighetto et al. (2014) found that the negative impact of dehumanization on empathy and on helping the victims of natural disasters aligns with prevalent stereotypes about out-groups (p. 573).

Internal and External Causal Attributions and Stereotypes

Individuals are more likely to support government assistance when they perceive the situation warranting aid as outside of the beneficiary’s control (external attribution) rather than under their control (internal attribution) (Bullock et al., 2003; Krijnen et al., 2022; Petersen, 2012; Petersen et al., 2012). By answering the question “who is to blame,” attributions signal recipients’ moral character, and thus humanity attributes like agency (control over and effort put into improving the situation) and patiency (is the situation a coincidence or long term) (Celniker et al., 2023). For example, Zagefka et al. (2011) found that participants were less willing to donate to victims of humanly caused than naturally caused disasters because they perceived the former as more worthy of blame for their plight and as making less “of an effort to help themselves” (p. 353).

Internal and external causal attributions, morality, humanity, and deservingness perceptions closely relate to socioeconomic and racial/ethnic stereotypes. Individuals often blame socioeconomically disadvantaged groups for their poverty (Cozzarelli et al., 2001; Weiner et al., 2011), and stereotypical perceptions of African Americans as lazy lead to low support for American welfare policies (Gilens, 1999). Moreover, stereotypes are more readily applied to groups than to individuals—a finding especially relevant for this study, which focuses on the heavily stereotyped Roma community (Cooley & Payne, 2019). We proceed by discussing the relationship between target group depictions and humanization.

Humanization, Prosocial Behavior, and Policy Support

Humanization—“promotion of humanity attribution to outgroups” (Vezzali et al., 2022, p. 215)—has been extensively studied as an outcome of intergroup help (Davies et al., 2018; Saguy et al., 2015), multiple categorization (Albarello & Rubini, 2012; Prati et al., 2016), intergroup contact (Capozza et al., 2013), animal–human similarity (Costello & Hodson, 2010), and emotional similarity (McDonald et al., 2017). Yet, to the best of our knowledge, we are not aware of any evidence supporting the impact of humanization on individual support for policies that benefit members of disadvantaged groups.¹

To fill this gap, we test whether previous findings about the prosocial impact of changes in target group depictions also extend to policy support. Specifically, we focus on the humanization effect of group-composition identified by Cooley et al. (2017) and replicated by Hodson and Doucher (2020). These studies found that group-composition depictions shifted perceptions of a target’s ability to possess experience and agency, which in turn mediated a shift in sympathy toward the target. Groups evoked the least sympathy and perceived agency and experience, while individuals evoked the most. Yet, referring to targets as individuals within a group led to near-equivalent levels of perceived agency, perceived experience, sympathy (Cooley et al., 2017), and willingness to help (Hodson & Doucher, 2020) as in the individual condition.

Purpose of the Present Research

We investigate whether internal and external attributions condition the humanization effect. Cooley et al. (2017) and Hodson and Doucher (2020) used a vignette that describes the target as one man, 20 people, or a company whose situations worsen as a result of a malevolent action by an agent (external causal attribution) (see Cooley et al., 2017, pp. 695–696; Hodson & Doucher, 2020, p. 1604, original vignette from Rai & Diermeier, 2015). Whether a different scenario describing the targets’ situations as a consequence of internal factors (e.g., their negligence or ineffectiveness) supports the humanization effect remains to be seen. Second, we investigate the cross-cultural applicability of the humanization effect by conducting our replications and extensions in the Slovak Republic (vs. with Amazon MTurkers). Finally, we examine whether the humanization effect holds when there are salient identity differences between respondents and the target group. Specifically, we focus on the Roma, a disadvantaged, heavily stereotyped minority ethnic group.

Study 1 attempts to replicate the humanization effect of group-composition depictions with a quota-representative sample of Slovak citizens.² Study 2 extends Cooley et al. (2017) and Hodson and Doucher (2020) and tests whether the humanization effect of group-composition extends to policy support and to the Roma. Finally, given that Studies 1 and 2 focused on situations of external attribution, Study 3 turns to a situation of internal attribution—that is, when the target can be perceived as personally responsible for the situation warranting aid.

Study 1

Study 1 replicates Cooley et al. (2017, Study 3) and Hodson and Doucher’s (2020) findings about the humanization effect of group-composition depictions on perceptions of a target’s experience and agency, and participants’ sympathy for and willingness to help the target.

Method

Participants

We recruited a quota-representative sample of Slovaks—1,014 participants (M_age = 44.5, SD_age = 15.8, 52% female)—from an online panel managed by the Slovak market research firm, 2muse. We used quotas on age, education, gender, region, and size of settlement. Post hoc analysis for global effects showed that a multivariate analysis of variance (MANOVA) with 1,014 participants across three groups would be sensitive to effects as small as f ² = 0.008 with 80% power (α = .05).

Procedure

Following the previous studies, we used a between-subject design, and randomly assigned participants to three conditions (manipulations in brackets and bold):

Take a moment to imagine [a man] [20 employees who compose a small company] [a small company] who was [were] quite successful. Now imagine that, recently, the [man’s] [20 employee’s] [company’s] electronic security firewalls were breached, and [his] [their] [their] private accounts were hacked, and as a result [he] [the company] [the company] went bankrupt.

Measures

After reading one of the vignettes, participants responded to single-item measures adopted from Cooley et al. (2017) (Study 3), assessing the target’s capacity for experience (“to what extent is the [target] capable of feeling pain and suffering”) and agency (“. . . the [target] capable of having intentions and goals”) on 0–100 scales. Both scales were anchored at 0 (“they are not at all capable”) and 100 (“they are very capable”). Participants then rated their sympathy for the target (0–100), again anchored at 0 (“they are not at all sympathetic”) and 100 (“they are very sympathetic”). Following Hodson and Doucher (2020), participants also indicated their willingness to help the target: “if you had the time and capacity, how likely or unlikely is it that you would help [target] . . . where 0 indicates that it is not likely and 100 indicates that it is very likely.” Finally, participants answered a manipulation check about the vignette.

Analytic Strategy

To analyze differences in outcomes across groups, we used a resampling-based MANOVA with 1,000 iterations (Friedrich et al., 2019). Specifically, we calculated the Wald-type statistic (WTS) and the modified analysis of variance (ANOVA)-type statistic (MATS), which provide multivariate ANOVA-like results without assumptions about multivariate normality or covariance structures. We report test statistics with resampled p values calculated using a parametric bootstrap method. Individual ANOVAs for each outcome complement the multivariate analysis.

To test the mediation hypotheses—that shifts in perceptions of experience and agency explain sympathy and willingness to help—we used a robust bootstrap test for mediation from the ROBMED package for R (Alfons et al., 2022),³ which allows us to calculate regression-based bootstrap tests for mediation even when data deviate from normality or include outliers. We used the default 5,000 bootstrap replications for all mediation analyses.

Results and Discussion

Across all mediators and outcomes (see Figure 1 and Table 1), participants evaluated the individual and group-composition conditions similarly,⁴ and assigned lower scores to the group condition. The results of the MANOVA analysis confirmed an overall effect of the manipulation on the outcome variables: WTS(8) = 53.2, resampled p < .001; MATS = 86.2, resampled p < .001. Multivariate post hoc comparisons with Tukey contrasts showed a significant difference between the group and individual conditions (p = .004, estimated summary effect across mediators and outcomes = −19.5), as well as a significant difference between the group and group-composition conditions (p < .001, estimated summary effect across mediators and outcomes = −32.9) (Table 2).

Figure 1

Distributions, Means, and 95% Confidence Intervals of Outcomes Across Experimental Conditions (Study 1)

Table 1

Descriptive Statistics Across Outcomes and Experimental Groups (Study 1)

Mediators and Outcomes	M	SD	Confidence Interval	N
Experience
Individual	78.9	21.9	2.35	337
Group-composition	79.2	21.1	2.25	340
Group	73.1	24.4	2.61	337
Agency
Individual	67.0	22.6	2.42	337
Group-composition	69.9	22.8	2.43	340
Group	62.6	22.5	2.41	337
Sympathy
Individual	64.6	21.3	2.28	337
Group-composition	69.0	21.9	2.33	340
Group	61.6	23.4	2.51	337
Willingness to help
Individual	64.0	23.5	2.52	337
Group-composition	69.8	24.8	2.65	340
Group	57.8	28.1	3.01	337

Table 2

Differences Between Experimental Conditions (Study 1, Summary Effects)

Contrast	Summary effect estimate	95% CI lower limit	95% CI upper limit	p value
Group-composition–Individual	13.4	−0.58	27.41	.064
Group–Individual	−19.5	−34.13	−4.78	.005
Group–Group-composition	−32.9	−47.62	−18.12	<.001

Note. Summary effects are averaged over all dimensions, CIs are based on the bootstrap version of the sum statistic, and CIs and p values maintain a given level of alpha. CI = confidence interval.

Individual analyses of variance for the mediators and outcomes showed a significant effect of condition on experience (F(2, 1011) = 7.87, p < .001, ω² = 0.013); agency (F(2, 1011) = 8.93, p < .001, ω² = 0.015); sympathy (F(2, 1011) = 9.53, p < .001, ω² = 0.034); and willingness to help (F(2, 1011) = 18.92, p < .001, ω² = 0.013). Post hoc comparisons showed significant differences between the individual and group conditions across all mediators and outcomes except sympathy, and between the group-composition and group conditions across all mediators and outcomes. We report further details in Supplemental Material (SM) (Tables 4–7).⁵

We created four contrasts for the mediation analysis: (a) group and group-composition versus individual; (b) group versus group-composition; (c) group versus individual; and (d) group-composition versus individual. We then compared whether the outcome variables differ across these contrasts, and whether shifts in perceived experience and agency explain the differences.

Figure 2 shows that the combined group condition, when compared with the individual condition, has no association with perceived experience or agency, or with sympathy or willingness to help. Similarly, we observed little evidence of differences between the group-composition and individual conditions. However, when comparing the group condition to the group-composition condition, the group depiction was associated with less perceived experience and agency, and consequently less sympathy for the target and lower willingness to help. Indirect effects were present via both experience (b = −1.55, 95% confidence interval [CI] = [−3.31, −0.49]) and agency (b = −3.08, 95% CI = [−5.48, −1.52]) for both sympathy and willingness to help: experience b = −2.3, 95% CI = [−4.14, −0.79]; agency b = −2.44, 95% CI = [−4.23, −1.18].

Figure 2

Indirect Effects on Sympathy and Help Via Experience and Agency (Study 1)

A similar pattern arose when comparing the group condition with the individual condition. Respondents in the group condition perceived the target to be less experienced and have less agency, which in turn led to less sympathy and willingness to help. There was no significant direct effect, underlining the influence of the mediators on the outcomes: sympathy with the target (indirect effect of experience b = −2.08, 95% CI = [−3.6, −0.8], indirect effect of agency b = −1.37, 95% CI = [−2.6, −0.53]), and willingness to help the target (indirect effect of experience b = −1.58, 95% CI = [−3.3, −0.59], indirect effect of agency b = −1.53, 95% CI = [−3.19, −0.53]). For both outcomes, we also found significant total and direct effects of the group depiction relative to the group-composition depiction. The SM presents coefficients and standard errors (Table 25).

These results largely support Cooley et al. (2017) and Hodson and Doucher (2020). Individuals and group-compositions evoke similar levels of sympathy and willingness to help, and these levels are higher than those evoked by groups. Finally, agency and experience are significant mediators for sympathy and willingness to help in the group versus group-composition and group versus individual comparisons.

Study 2

Study 2 identifies whether previous findings extend to policy support. We predict that increases in agency and experience associated with the individual and group-composition conditions will translate to increased levels of policy support relative to the group condition, and that humanization will be the relevant mechanism. Studies 2 and 3 also investigate two additional, potentially mitigating factors. First, the vignettes focus on a stigmatized ethnic out-group in Slovakia: the Roma. Second, the studies contrast internal and external causal attribution. In Study 2, the vignettes focus on a natural disaster, indicating an external causal attribution, while in Study 3, they focus on a personal inability to find work, indicating an internal causal attribution.

Method

Participants

Following Study 1, we recruited a sample of 1,008 participants (M_age = 44.3, SD_age = 15.7, 52% female, Study 1 participants ineligible). A post hoc analysis for global effects showed that a MANOVA with 1,008 participants across three groups would be sensitive to effects as small as f ² = 0.008, with 80% power (α = .05).

Procedure

We reran Study 1 with vignettes that maintained a situation of external attribution (a natural disaster), but that focused on the Roma (manipulations in brackets and bold):

Take a moment to imagine [a man who lives in a marginalized Roma community] [twenty people who live in a marginalized Roma community] [a marginalized Roma community]. Now imagine that, recently, [the man] [these twenty people] [this community] lost [his housing] [their homes] [their housing]. A landslide destroyed [his] [their][their] home and ensured that [the man] [these twenty people] [this community] was left without a roof over [his] [their] [their] head.

Measures

We used identical measures as in Study 1 but added policy support as an outcome: “if there existed the necessary financial resources for building replacement housing, to what extent would you agree or disagree with financial aid to the [target].” Responses were measured on a 0 (“completely disagree”) to 100 (“completely agree”) scale.

Analytic Strategy

We used the same analytic strategy as in Study 1.

Results and Discussion

Across both mediators and the three outcomes, the individual condition received the highest scores from participants, the group condition received the lowest, and the group-composition condition fell somewhere between the two (see Figure 3 and Table 3). Importantly, across mediators and outcomes, the group condition evoked significantly lower evaluations than the other two conditions.

Figure 3

Distributions, Means, and 95% Confidence Intervals of Outcomes Across Experimental Conditions (Study 2)

Table 3

Descriptive Statistics Across Outcomes and Experimental Groups (Study 2)

Mediators and Outcomes	M	SD	Confidence Interval	N
Experience
Individual	74.4	26.1	2.81	334
Group-composition	71.2	28.1	3.01	337
Group	66.7	29.5	3.16	337
Agency
Individual	57.6	27.7	2.98	334
Group-composition	51.0	29.5	3.17	337
Group	44.7	29.3	3.14	337
Sympathy
Individual	48.6	25.9	2.79	334
Group-composition	42.1	27.5	2.94	337
Group	31.7	25.2	2.70	337
Willingness to help
Individual	43.9	28.3	3.04	334
Group-composition	42.7	29.0	3.10	337
Group	33.7	27.1	2.90	337
Policy support
Individual	52.9	32.6	3.51	334
Group-composition	48.8	31.2	3.35	337
Group	38.6	30.9	3.31	337

We estimated the overall effects of experimental conditions across mediators and outcomes using the same process as in Study 1 (Table 4). There was a significant effect of condition across all mediators and outcomes (WTS(10) = 85.25, resampled p < .001; MATS = 187, resampled p < .001). Multivariate post hoc comparisons with Tukey contrasts showed a significant difference between the group and individual conditions (p < .001, estimated summary effect across mediators and outcomes = −62.1), as well as a difference between the group and group-composition conditions (p < .001, estimated summary effect across mediators and outcomes = −40.4). Again, the group condition received lower scores than either the individual or group-composition conditions.⁶

Table 4

Differences Between Experimental Conditions (Study 2, Summary Effects)

Contrast	Summary effect estimate	95% CI lower limit	95% CI upper limit	p value
Group-composition–Individual	−21.7	−46.00	2.64	.089
Group–Individual	−62.1	−86.10	−38.06	<.001
Group–Group-composition	−40.4	−64.80	−16.08	<.001

Note. Summary effects are averaged over all dimensions, CIs are based on the bootstrap version of the sum statistic, CIs and p values maintain a given level of alpha. CI = confidence interval.

Individual analyses of variance for each mediator and outcome showed a significant effect of target group condition on experience (F(2, 1005) = 6.46, p = .002, ω² = 0.011); agency (F(2, 1005) = 16.96, p < .001, ω² = 0.031); sympathy (F(2, 1005) = 35.51, p < .001, ω² = 0.064); willingness to help (F(2, 1005) = 13.27, p < .001, ω² = 0.024); and policy support (F(2, 1005) = 18.24, p < .001, ω² = 0.033). The shift from individual to group led to significant post hoc differences across all mediators and outcomes, and the shift from group-composition to group led to significant differences in experience (though not agency), and all three outcomes. Finally, there was a significant difference between the individual and group-composition conditions for agency and sympathy. For further details, see the SM (Tables 11–14).

For the mediation analysis, we created the same four contrasts as in Study 1. We then tested whether the different conditions led to differences in the three outcomes—sympathy, willingness to help, and policy support—and whether shifts in perceptions of experience and agency were relevant mediators.

When comparing the combined group condition with the individual condition and the group condition with the individual condition, both mediators showed significant path effects across all three outcomes (see Figure 4). This was not the case when comparing groups with group-compositions or group-compositions with individuals. Lower scores on the outcome variables were present in the combined group condition relative to individual condition due to lower perceived agency (sympathy b = −6.67, 95% CI = [−9.49, −4.16], willingness to help b = −5.81, 95% CI = [−8.46, −3.63], and policy support b = −7.29, 95% CI = [−10.54, −4.45]), and, to a lesser extent, lower perceived capacity to experience (sympathy b = −0.40, 95% CI = [−1.04, −0.071], willingness to help b = −0.46, 95% CI = [−1.18, −0.08], and policy support b = −1.02, 95% CI = [−2.14, −0.24]).

Figure 4

Indirect Effects on Sympathy, Help, and Policy Support Via Experience and Agency (Study 2)

As in Study 1, we found significant effects for both mediation paths in the group versus individual comparison, although there was also a significant direct effect in the case of sympathy (indirect effect of experience b = −0.6, 95% CI = [−1.6, −0.07], indirect effect of agency b = −8.65, 95% CI = [−12.1, −5.59]) and policy support (indirect effect of experience b = −2.1, 95% CI = [−4.13, −0.73], indirect effect of agency b = −8.72, 95% CI = [−12.72,−5.53]). There was no direct effect in the case of willingness to help (indirect effect of experience b = −0.94, 95% CI = [−2.14, −0.2], indirect effect of agency b = −6.77, 95% CI = [−10, −4.15]). We report further information on partially significant mediation relationships in the SM (Table 26).

Study 2 finds that (a) existing results replicate to a different sample and when the target group is a marginalized ethnic minority and that (b) the humanization effect of group-composition depictions extends to policy support in the form of disaster relief funds.

Study 3

Study 3 identifies whether the results of Study 2—where the vignette indicated external attribution—hold if the vignettes suggests internal attribution instead. Accordingly, the vignettes focus on a marginalized Roma target living in substandard housing due to a personal inability to find work. The vignette reflects the precarious housing situation of the Roma across many European countries (Anthonj et al., 2020, p. 1). We expect the indication of internal attribution for the housing situation to be a tougher test for the humanization effect across all three outcomes.

Method

Participants

We recruited a quota-representative sample of 1,212 participants (M_age = 44.7, SD_age = 16, 52% female, Study 1 and 2 participants ineligible). A post hoc analysis for global effects showed that a MANOVA with 1,212 participants across three groups would be sensitive to effects as small as f ² = 0.007, with 80% power (α = .05).

Procedure

We reran Studies 1 and 2 with vignettes indicating internal attribution:

Take a moment to imagine [a man who lives in a marginalized Roma community] [20 people who live in a marginalized Roma community] [a marginalized Roma community]. Now imagine that [this man] [these 20 people] [this community] live[s] in precarious housing conditions, which do not fulfill fundamental hygienic and sanitary standards, and is a threat to [his] [their] [their] health. Because [this man who lives in a marginalized Roma community] [these 20 people who live in a marginalized Roma community] [this marginalized Roma community] has [have] been without work for a long time, [he] [they] [it] is poor and did not have a way to save money to buy or rent better housing.

Measures

We used the same measures as in Study 2.

Analytic Strategy

We used the same analytic strategy as in Studies 1 and 2.

Results and Discussion

Descriptively (see Table 5 and Figure 5), we found a similar pattern to the previous studies: participants assigned the highest scores to individuals, then group-compositions, and then groups.

Table 5

Descriptive Statistics Across Outcomes and Experimental Groups (Study 3)

Mediators and Outcomes	M	SD	Confidence Interval	N
Experience
Individual	65.9	27.5	2.69	405
Group-composition	62.3	28.6	2.80	403
Group	60.1	29.2	2.85	404
Agency
Individual	49.1	29.3	2.87	405
Group-composition	45.7	28.5	2.79	403
Group	42.8	28.4	2.78	404
Sympathy
Individual	37.4	25.5	2.49	405
Group-composition	36.5	26.0	2.55	403
Group	35.2	27.3	2.67	404
Willingness to help
Individual	44.6	27.8	2.71	405
Group-composition	40.4	29.6	2.90	403
Group	38.7	29.2	2.86	404
Policy support
Individual	48.0	31.0	3.03	405
Group-composition	42.5	32.6	3.19	403
Group	39.0	32.2	3.15	404

Figure 5

Distributions, Means, and 95% Confidence Interval of Outcomes in Respective Conditions (Study 3)

We found significant effects across all mediators and outcomes (results in Table 6): WTS(10) = 26.7, resampled p = .002; MATS = 46.2, resampled p < 001. Multivariate post hoc comparisons with Tukey contrasts showed a significant difference between the group and individual conditions (p = .014, estimated summary effect across all outcomes = −29.3).⁷

Table 6

Differences Between Experimental Conditions (Study 3, Summary Effects)

Contrast	Summary effect estimate	95% CI lower limit	95% CI upper limit	p value
Group-composition–Individual	−17.8	−42.10	6.58	.195
Group–Individual	−29.3	−53.70	−4.95	.014
Group–Group-composition	−11.6	−36.40	13.20	.493

Note. Summary effects are averaged over all dimensions, CIs are based on the bootstrap version of the sum statistic, CIs and p values maintain a given level of alpha. CI = confidence interval.

Individual analyses of variance for each mediator and outcome showed a significant effect of condition on experience (F(2, 1209) = 4.31, p = .014, ω² = 0.005); agency (F(2, 1209) = 4.92, p = .007, ω² = 0.006); willingness to help (F(2, 1209) = 4.52, p = .011, ω² = 0.006); and policy support (F(2, 1209) = 8.22, p < .001, ω² = 0.012); but not sympathy (F(2, 1209) = 0.74, p = .476). The shift from individual to group led to significant post hoc differences in experience, agency, willingness to help, and policy support. Finally, there was a significant difference between individual and group-composition depictions for policy support. For further details, see the SM (Tables 18–21).

We followed the procedures of Studies 1 and 2 for the mediation analysis. Path coefficients are reported in Figure 6. Hypothesized indirect mediation effects were, as in Study 2, consistently present when comparing the combined group condition with the individual condition and the group-composition condition with the individual condition. Contrary to Study 1, these effects were absent when comparing group and group-composition depictions. Participants scored lower across the three outcome variables when presented with groups (compared to individuals) because of groups’ lower agency (sympathy b = −2.833, 95% CI = [−5.19, −0.711], willingness to help b = −2.66, 95% CI = [−5, −0.692], and policy support b = −2.6, 95% CI = [−4.9, −0.624]) and, to a lesser extent, lower capacity to experience (sympathy b = −0.661, 95% CI = [−1.38, −0.17], willingness to help b = −1.02, 95% CI = [−2.19, −0.26], and policy support b = −1.66, 95% CI = [−3.15, −0.443]). Similarly, relative to the individual condition, participants in the group condition showed lower sympathy, willingness to help, and policy support as a result of perceived lower experience and agency (for sympathy: indirect effect of experience b = −0.9, 95% CI = [−1.94, −0.22], indirect effect of agency b = −3.48, 95% CI = [−6.04, −1.21]; for willingness to help: indirect effect of experience b = −1.1, 95% CI = [2.54, −0.27], indirect effect of agency b = −3.5, 95% CI = [−6.03, −1.19]; for policy support: indirect effect of experience b = −1.98, 95% CI = [−3.85, −0.53], indirect effect of agency b = −3.29, 95% CI = [−5.87, −1.25]). Tables with coefficients and standard errors are reported in the SM (Table 27).

Figure 6

Indirect Effects on Sympathy, Help, and Policy Support Via Experience and Agency (Study 3)

Study 3 suggests boundary conditions (Nosek & Lakens, 2014) for the humanization effect of group-composition depictions. As in Studies 1 and 2, participants evaluated individuals (relative to groups and group-compositions) as more able to possess experience and agency. These higher evaluations lead to higher levels of intention to help, sympathy, and policy support—despite the vignette indicating internal attribution. Yet, unlike in the previous studies, these relationships did not manifest when comparing the group-composition depiction with the group depiction. Thus, group-composition depictions may not have the same humanization effect in situations of internal attribution as they do in situations of external attribution.

General Discussion

The language used to describe those in need matters. Much of the previous literature on policy support for disadvantaged groups has focused on cues and heuristics that signal group differences—for example, race or immigration status (e.g., Ford, 2016; Harell et al., 2016; Nelson & Kinder, 1996). A complementary body of work focuses on deservingness evaluations and attributions (Findor et al., 2022; Petersen, 2012; Weiner et al., 2011). Yet, as this paper shows, more fundamental and subtle changes in depictions of beneficiaries matter as well.

In our replication-extension, we first confirm the findings of Cooley et al. (2017) and Hodson and Doucher (2020). Referring to beneficiaries as individuals, members of a group, or as a group evokes different levels of sympathy and willingness to help. Shifts in perceptions of experience and agency explain these differences. Second, we show that similar relationships apply when participants are asked about their willingness to support both disaster relief and subsidized housing policies benefiting the targets. Third, our focus on Slovakia and the Roma shows that the humanization effect of group-composition holds for stigmatized and dehumanized out-groups and in different cultural contexts. Fourth, an additional extension—focusing on internal attribution—yielded results that fail to support the humanization effect of group-composition. Thus, the humanization effect of group-composition depictions may only apply when the situation warranting aid is a result of external circumstance. Internal attribution limits the humanization effect.

These findings align with previous research on the impact of internal and external attributions on prosocial behavior and policy support (Bullock et al., 2003; Krijnen et al., 2022; Zagefka et al., 2011), and add to existing literature on humanization (Vezzali et al., 2022) by specifying the boundary conditions under which subtle changes in group depictions can humanize out-groups. More specifically, our findings highlight the importance of accounting for internal and external causal attributions as an intergroup evaluation bias in perceptions of disadvantaged people as human beings (through their moral character) and as (un)deserving policy beneficiaries—especially when these attributions align with prevalent negative stereotypes. One way to counter stereotype-induced dehumanization is to counter social stereotypes directly (Prati et al., 2015) by, for example, describing welfare recipients as hardworking or financially independent (Brown-Iannuzzi et al., 2021; Cooley et al., 2019).

Limitations

The study conditions set certain limits on comparability. As the size of the target differed between conditions, participants may have assigned fault based on numbers alone. Thus, future research, using stricter framing that “keeps the number of people constant across conditions,” would allow for greater equivalence and experimental control (Tang & Gray, 2021, p. 2). Future research should also consider the role gender plays in humanization. The studies used genderless representations of target groups. However, genderless or nongendered social policy beneficiaries may be perceived as less human than those who are represented as gendered (Martin & Mason, 2022).

Conclusions and Practical Implications

Our findings offer a nuanced perspective on target group dehumanization. Even subtle forms of dehumanization, encoded in inconspicuous ways to refer to beneficiaries, can have considerable effects on support for both disaster relief policy and subsidized housing policy. Seeing social targets as more or less human is a function of not only stereotypes and prejudice but also of target group depictions, which are often used indiscriminately.

These findings offer tangible recommendations for how policymakers and advocates should communicate about programs that benefit disadvantaged groups. Specifically, we show that majorities are more likely to be receptive to programs and policies when beneficiaries are humanized, and we show that humanization can be achieved through changes in target group depictions. Accordingly, when advocates are designing campaigns, they should pay careful attention to focus on individual stories rather than aggregate group-level claims. And, when language concerning groups cannot be avoided, campaigns would do well to remind the audience that those groups are, in fact, composed of individuals. We also reaffirm previous findings by showing individuals are more supportive of assistance when beneficiaries are not personally responsible for their circumstances. Accordingly, communication campaigns should convey that the situations of disadvantaged minority groups largely result from factors outside of their control.

Supplemental Material

sj-docx-1-spp-10.1177_19485506231167494 – Supplemental material for Bolstering Policy Support for Disadvantaged Groups Through Humanization

Supplemental material, sj-docx-1-spp-10.1177_19485506231167494 for Bolstering Policy Support for Disadvantaged Groups Through Humanization by Andrej Findor, Matej Hruška, Roman Hlatky, Alexa Dvorská, Tomáš Hrustič, Zuzana BošeI’ová and Ondrej Buchel in Social Psychological and Personality Science

Footnotes

Acknowledgements

We would like to thank the students of the Social Science Research Methods class at the Faculty of Social and Economic Sciences, Comenius University Bratislava, and Stano Daniel from Porticus Vienna for helpful feedback.

Handling Editor: Robyn Mallett.

Authors’ Note

The data, replication instructions, the data’s codebook, questionnaires, and preregistration protocols can be found at .

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Our work was funded by the European Union’s Rights, Equality and Citizenship Programme (Grant No. 809869—PERCOM) and by the Slovak Research and Development Agency under contract no. APVV-17-0596. The funding bodies were not involved in any stage of the research process. The content of this article represents the views of the authors only and is their sole responsibility. The European Commission does not accept any responsibility for use that may be made of the information it contains.

ORCID iDs

Andrej Findor

Roman Hlatky

Supplemental Material

The supplemental material is available in the online version of the article.

Notes

Author Biographies

Andrej Findor is interested in how interventions and communication based on moral, emotional, and identity appeals can lead to change in attitudes and behaviors related to contested and polarizing political issues such as immigration, human and minority rights, as well as policies helping marginalized Roma communities.

Matej Hruška works with survey and experimental data in academic, research, and policy contexts.

Roman Hlatky is an Assistant Professor in the Department of Political Science at the University of North Texas. His work has appeared in the American Political Science Review, East European Politics, and Political Research Quarterly.

Alexa Dvorská focuses on a variety of social challenges, including the marginalization of Roma communities, gender equality, and environmental injustice. Her main aim is to contribute to research on sustainable communities.

Tomáš Hrustič conducts research in the field of Romani studies. Besides his academic interests in Roma, he is also engaged in networking local Roma elected representatives and Roma youth leaders, mainly by facilitating workshops and training on various topics.

Zuzana BošeI’ová, as a nongovernmental organization (NGO) worker, is interested in the application of research in providing support to municipalities and national institutions that seek to build an inclusive environment and promote inclusive policies.

Ondrej Buchel focuses mainly on persistent economic inequalities and related ideologies, perceptions, and beliefs about social norms. His research also addresses policies aimed at reduction of inequalities and poverty alleviation.

References

Albarello

Rubini

(2012). Reducing dehumanisation outcomes towards Blacks: The role of multiple categorisation and of human identity: Reducing dehumanisation outcomes. European Journal of Social Psychology, 42(7), 875–882. https://doi.org/10.1002/ejsp.1902

Alfons

Ateş

N. Y.

Groenen

P. J. F.

(2022). A robust bootstrap test for mediation analysis. Organizational Research Methods, 25(3), 591–617. https://doi.org/10.1177/1094428121999096

Andrighetto

Baldissarri

Lattanzio

Loughnan

Volpato

(2014). Human-itarian aid? Two forms of dehumanization and willingness to help after natural disasters. British Journal of Social Psychology, 53(3), 573–584. https://doi.org/10.1111/bjso.12066

Anthonj

Setty

K. E.

Ezbakhe

Manga

Hoeser

(2020). A systematic review of water, sanitation and hygiene among Roma communities in Europe: Situation analysis, cultural context, and obstacles to improvement. International Journal of Hygiene and Environmental Health, 226, Article 113506. https://doi.org/10.1016/j.ijheh.2020.113506

Aronow

P. M.

Baron

Pinson

(2019). A note on dropping experimental subjects who fail a manipulation check. Political Analysis, 27(4), 572–589. https://doi.org/10.1017/pan.2019.5

R. H. Y.

G. T. T.

(2021). Mind perception and stereotype attribution of corporations and charities. British Journal of Social Psychology, 60(1), 271–293. https://doi.org/10.1111/bjso.12377

Bain

Park

Kwok

Haslam

(2009). Attributing human uniqueness and human nature to cultural groups: Distinct forms of subtle dehumanization. Group Processes & Intergroup Relations, 12(6), 789–805. https://doi.org/10.1177/1368430209340415

Brown-Iannuzzi

J. L.

Cooley

Marshburn

C. K.

McKee

S. E.

Lei

R. F.

(2021). Investigating the interplay between race, work ethic stereotypes, and attitudes toward welfare recipients and policies. Social Psychological and Personality Science, 12, 1155–1164. https://doi.org/10.1177/1948550620983051

Bullock

H. E.

Williams

W. R.

Limbert

W. M.

(2003). Predicting support for welfare policies: The impact of attributions and beliefs about inequality. Journal of Poverty, 7(3), 35–56. https://doi.org/10.1300/J134v07n03_03

10.

Capozza

Trifiletti

Vezzali

Favara

(2013). Can intergroup contact improve humanity attributions? International Journal of Psychology, 48(4), 527–541. https://doi.org/10.1080/00207594.2012.688132

11.

Celniker

J. B.

Gregory

Koo

H. J.

Piff

P. K.

Ditto

P. H.

Shariff

A. F.

(2023). The moralization of effort. Journal of Experimental Psychology: General, 152(1), 60–79. https://doi.org/10.1037/xge0001259

12.

Cooley

Brown-Iannuzzi

J. L.

Boudreau

(2019). Shifting stereotypes of welfare recipients can reverse racial biases in support for wealth redistribution. Social Psychological and Personality Science, 10(8), 1065–1074. https://doi.org/10.1177/1948550619829062

13.

Cooley

Payne

B. K.

(2019). A group is more than the average of its parts: Why existing stereotypes are applied more to the same individuals when viewed in groups than when viewed alone. Group Processes & Intergroup Relations, 22(5), 673–687. https://doi.org/10.1177/1368430218756494

14.

Cooley

Payne

B. K.

Cipolli

Cameron

C. D.

Berger

Gray

(2017). The paradox of group mind: “People in a group” have more mind than “a group of people.” Journal of Experimental Psychology: General, 146(5), 691–699. https://doi.org/10.1037/xge0000293

15.

Costello

Hodson

(2010). Exploring the roots of dehumanization: The role of animal—Human similarity in promoting immigrant humanization. Group Processes & Intergroup Relations, 13(1), 3–22. https://doi.org/10.1177/1368430209347725

16.

Cozzarelli

Wilkinson

A. V.

Tagler

M. J.

(2001). Attitudes toward the poor and attributions for poverty. Journal of Social Issues, 57(2), 207–227. https://doi.org/10.1111/0022-4537.00209

17.

Cuddy

A. J. C.

Rock

M. S.

Norton

M. I.

(2007). Aid in the aftermath of Hurricane Katrina: Inferences of secondary emotions and intergroup helping. Group Processes & Intergroup Relations, 10(1), 107–118. https://doi.org/10.1177/1368430207071344

18.

Davies

Yogeeswaran

Verkuyten

Loughnan

(2018). From humanitarian aid to humanization: When outgroup, but not ingroup, helping increases humanization. PLOS ONE, 13(11), Article e0207343. https://doi.org/10.1371/journal.pone.0207343

19.

Esses

V. M.

Veenvliet

Hodson

Mihic

(2008). Justice, morality, and the dehumanization of refugees. Social Justice Research, 21(1), 4–25. https://doi.org/10.1007/s11211-007-0058-4

20.

Findor

Hruška

Hlatky

Hrustič

Bošel’ová

(2022). Equality, reciprocity, or need? Bolstering welfare policy support for marginalized groups with distributive fairness. American Political Science Review, 1–17. https://doi.org/10.1017/S0003055422001046

21.

Ford

(2016). Who should we help? An experimental test of discrimination in the British welfare state. Political Studies, 64(3), 630–650. https://doi.org/10.1111/1467-9248.12194

22.

Friedrich

Konietschke

Pauly

(2019). Resampling-based analysis of multivariate data and repeated measures designs with the R package MANOVA.RM. The R Journal, 11(2), 380–400.

23.

Gilens

(1999). Why Americans hate welfare: Race, media, and the politics of antipoverty policy. The University of Chicago Press.

24.

Goff

P. A.

Eberhardt

J. L.

Williams

M. J.

Jackson

M. C.

(2008). Not yet human: Implicit knowledge, historical dehumanization, and contemporary consequences. Journal of Personality and Social Psychology, 94(2), 292–306. https://doi.org/10.1037/0022-3514.94.2.292

25.

Gray

H. M.

Gray

Wegner

D. M.

(2007). Dimensions of mind perception. Science, 315(5812), 619–619. https://doi.org/10.1126/science.1134475

26.

Harell

Soroka

Iyengar

(2016). Race, prejudice and attitudes toward redistribution: A comparative experimental approach. European Journal of Political Research, 55(4), 723–744. https://doi.org/10.1111/1475-6765.12158

27.

Haslam

Loughnan

(2014). Dehumanization and infrahumanization. Annual Review of Psychology, 65(1), 399–423. https://doi.org/10.1146/annurev-psych-010213-115045

28.

Haslam

Loughnan

Kashima

Bain

(2008). Attributing and denying humanness to others. European Review of Social Psychology, 19(1), 55–85. https://doi.org/10.1080/10463280801981645

29.

Hodson

Doucher

(2020). Language framing shapes dehumanization of groups: A successful replication and extension of Cooley et al. (2017). Journal of Experimental Psychology: General, 149(8), 1603–1607. https://doi.org/10.1037/xge0000723

30.

Kende

Hadarics

Bigazzi

Boza

Kunst

J. R.

Lantos

N. A.

Lášticová

Minescu

Pivetti

Urbiola

(2020). The last acceptable prejudice in Europe? Anti-Gypsyism as the obstacle to Roma inclusion. Group Processes & Intergroup Relations, 24, 388–410. https://doi.org/10.1177/1368430220907701

31.

Krijnen

J. M. T.

Ülkümen

Bogard

J. E.

Fox

C. R.

(2022). Lay theories of financial well-being predict political and policy message preferences. Journal of Personality and Social Psychology, 122(2), 310–336. https://doi.org/10.1037/pspp0000392

32.

Kteily

Bruneau

Waytz

Cotterill

(2015). The ascent of man: Theoretical and empirical evidence for blatant dehumanization. Journal of Personality and Social Psychology, 109(5), 901–931. https://doi.org/10.1037/pspp0000048

33.

Loughnan

Haslam

Sutton

R. M.

Spencer

(2014). Dehumanization and social class: Animality in the stereotypes of “white trash,” “chavs,” and “bogans.” Social Psychology, 45(1), 54–61. https://doi.org/10.1027/1864-9335/a000159

34.

Martin

A. E.

Mason

M. F.

(2022). What does it mean to be (seen as) human? The importance of gender in humanization. Journal of Personality and Social Psychology, 123, 292–315. https://doi.org/10.1037/pspa0000293

35.

McDonald

Porat

Yarkoney

Reifen Tagar

Kimel

Saguy

Halperin

(2017). Intergroup emotional similarity reduces dehumanization and promotes conciliatory attitudes in prolonged conflict. Group Processes & Intergroup Relations, 20(1), 125–136. https://doi.org/10.1177/1368430215595107

36.

Nelson

T. E.

Kinder

D. R.

(1996). Issue frames and group-centrism in American public opinion. The Journal of Politics, 58(4), 1055–1078. https://doi.org/10.2307/2960149

37.

Nosek

B. A.

Lakens

(2014). Registered reports: A method to increase the credibility of published results. Social Psychology, 45(3), 137–141. https://doi.org/10.1027/1864-9335/a000192

38.

Petersen

M. B.

(2012). Social welfare as small-scale help: Evolutionary psychology and the deservingness heuristic. American Journal of Political Science, 56(1), 1–16. https://doi.org/10.1111/j.1540-5907.2011.00545.x

39.

Petersen

M. B.

Sznycer

Cosmides

Tooby

(2012). Who deserves help? Evolutionary psychology, social emotions, and public opinion about welfare. Political Psychology, 33(3), 395–418. https://doi.org/10.1111/j.1467-9221.2012.00883.x

40.

Prati

Crisp

R. J.

Meleady

Rubini

(2016). Humanizing outgroups through multiple categorization: The roles of individuation and threat. Personality and Social Psychology Bulletin, 42(4), 526–539. https://doi.org/10.1177/0146167216636624

41.

Prati

Vasiljevic

Crisp

R. J.

Rubini

(2015). Some extended psychological benefits of challenging social stereotypes: Decreased dehumanization and a reduced reliance on heuristic thinking. Group Processes & Intergroup Relations, 18(6), 801–816. https://doi.org/10.1177/1368430214567762

42.

Rai

T. S.

Diermeier

(2015). Corporations are cyborgs: Organizations elicit anger but not sympathy when they can think but cannot feel. Organizational Behavior and Human Decision Processes, 126, 18–26. https://doi.org/10.1016/j.obhdp.2014.10.001

43.

Rai

T. S.

Valdesolo

Graham

(2017). Dehumanization increases instrumental violence, but not moral violence. Proceedings of the National Academy of Sciences, 114(32), 8511–8516. https://doi.org/10.1073/pnas.1705238114

44.

Saguy

Szekeres

Nouri

Goldenberg

Doron

Dovidio

J. F.

Yunger

Halperin

(2015). Awareness of intergroup help can rehumanize the out-group. Social Psychological and Personality Science, 6(5), 551–558. https://doi.org/10.1177/1948550615574748

45.

Sainz

Loughnan

Martínez

Moya

Rodríguez-Bailón

(2020). Dehumanization of socioeconomically disadvantaged groups decreases support for welfare policies via perceived wastefulness. International Review of Social Psychology, 33(1), Article 12. https://doi.org/10.5334/irsp.414

46.

Sainz

Martínez

Moya

Rodríguez-Bailón

(2019). Animalizing the disadvantaged, mechanizing the wealthy: The convergence of socio-economic status and attribution of humanity. International Journal of Psychology, 54(4), 423–430. https://doi.org/10.1002/ijop.12485

47.

Sainz

Martínez

Rodríguez-Bailón

Moya

(2019). Where does the money come from? Humanizing high socioeconomic status groups undermines attitudes toward redistribution. Frontiers in Psychology, 10, Article 771. https://doi.org/10.3389/fpsyg.2019.00771

48.

Sainz

Martínez

Sutton

R. M.

Rodríguez-Bailón

Moya

(2020). Less human, more to blame: Animalizing poor people increases blame and decreases support for wealth redistribution. Group Processes & Intergroup Relations, 23(4), 546–559. https://doi.org/10.1177/1368430219841135

49.

Tang

Gray

(2021). Feeling empathy for organizations: Moral consequences, mechanisms, and the power of framing. Journal of Experimental Social Psychology, 96, Article 104147. https://doi.org/10.1016/j.jesp.2021.104147

50.

Vezzali

Capozza

Di Bernardo

G. A.

Falvo

Stathi

Hodson

(2022). Strategies for the promotion of humanity attribution to outgroups. European Review of Social Psychology, 33(2), 215–254. https://doi.org/10.1080/10463283.2021.1963156

51.

Waytz

Gray

Epley

Wegner

D. M.

(2010). Causes and consequences of mind perception. Trends in Cognitive Sciences, 14(8), 383–388. https://doi.org/10.1016/j.tics.2010.05.006

52.

Weiner

Osborne

Rudolph

(2011). An attributional analysis of reactions to poverty: The political ideology of the giver and the perceived morality of the receiver. Personality and Social Psychology Review, 15(2), 199–213. https://doi.org/10.1177/1088868310387615

53.

Zagefka

Noor

Brown

de Moura

G. R.

Hopthrow

(2011). Donating to disaster victims: Responses to natural and humanly caused events. European Journal of Social Psychology, 41(3), 353–363. https://doi.org/10.1002/ejsp.781

54.

Zebel

Zimmermann

Viki

G. T.

Doosje

(2008). Dehumanization and guilt as distinct but related predictors of support for reparation policies. Political Psychology, 29(2), 193–219. https://doi.org/10.1111/j.1467-9221.2008.00623.x

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.35 MB