Ambivalent classism attitudes

Beliefs regarding poverty and the poor contribute to the maintenance and normalisation of socioeconomic differences (^{Bullock et al., 2003}; ^{Sainz, Martínez, et al., 2020}). For instance, previous research on the attributional process shows that people tend to ascribe poverty to the lack of abilities or the misbehaviours of poor people (i.e., internal attributions of poverty), more than they do to structural barriers, such as discrimination and the lack of educational or working opportunities (i.e., external attributions of poverty; ^{Cozzarelli et al., 2001}). This narrative regarding the causes of poverty frequently relies on news media coverage that mostly depicts poor people as substance abusers, practicing inappropriate consumption or taking advantage of social benefits (^{Bullock et al., 2003}; ^{Hayward & Yar, 2006}). Typical stereotypes of poor people are that they are incompetent (^{Durante & Fiske 2017}) or less evolved (i.e., animal-like) than other socioeconomic status groups (i.e., animalistic dehumanisation; ^{Loughnan et al., 2014}; ^{Sainz et al., 2019}), which contributes to the maintenance and perpetuation of poverty (^{Sainz, Loughnan, et al., 2020}).

Additionally, negative perceptions and beliefs regarding poverty are increasingly prominent among people who adhere to system justification ideologies (^{Hunt & Bullock, 2016}). In this regard, the present study acknowledges that an orientation towards social dominance (i.e., SDO) and meritocratic or other system justification beliefs are associated with the tendency to justify the wealth gap. In line with this, classism attitudes also play a role in the understanding and justification of poverty. However, existing studies mostly conceptualise classism as negative or hostile attitudes or behaviours directed at the poor (^{Cozzarelli et al., 2001}; ^{Liu, 2011}; ^{Yun & Weaver, 2010}) without considering other possible manifestations. In this regard, recent approaches broaden the study on classism attitudes by expanding its definition and measuring its manifestations. This case is true of the scale developed by ^{Colbow et al. (2016)}, which evaluated attitudes averse to the poor (i.e., downward classism) but is also an independent factor that evaluates the attitudes and discrimination towards individuals belonging to the higher social classes (i.e., upward classism; e.g., ^{Castillo & Rivera-Gutiérrez, 2018}). Additionally, ^{Jordan et al. (2021)} propose a recent development in the concept of classism that addresses how classism attitudes towards the poor can lead to ambivalent manifestations, ranging from a thorough and explicit derogation of the poor to subtle and paternalistic manifestations.

Ambivalent classism attitudes (^{Jordan et al., 2021}) rely on previous literature on the ambivalence of stereotype content (^{Fiske et al., 2002}) and sexist attitudes (^{Glick & Fiske, 2001}). On the one hand, the stereotype content model proposes that stereotypes typically include a combination of negative and positive traits (^{Fiske et al., 2002}). For instance, ^{Durante & Fiske (2017)} emphasize that the poor are frequently stereotyped as incompetent. At the same time, however, they are considered warm and sociable. On the other hand, the ambivalent sexism theory (^{Glick & Fiske, 2001}) posits that people can hold hostile or negative attitudes towards women (hostile sexism: the belief that women try to manipulate men). At the same time, they maintain more paternalistic and benevolent attitudes towards women (benevolent sexism: the belief that men should protect women). Thus, perceptions regarding disadvantaged groups rely on negative evaluations or beliefs and incorporate positive ones. As ^{Jordan et al. (2021)} point out, such ambivalence can be observed in people’s attitudes concerning poverty and the poor. In this sense, people can hold a negative view of the poor as incompetent or untrustworthy (i.e., hostile classism (HC)) aside from a positive or condescending view of the group as friendly or honest (i.e., benevolent classism). Nevertheless, individuals can hold both attitudes at the same time. Most important, together they help to reinforce the maintenance of the status quo.

Specifically, the factor structure of the ACI differentiates among three subfactors. In terms of HC, ^{Jordan et al. (2021)} include items that measure the extent to which people hold negative representations of the poor as being inferior, lacking traits, or intending to take advantage of others for profit (e.g., “poor people often take advantage of charitable individuals and organisations”). The authors differentiate between protective paternalism (PP) and complementary class differentiation (CCD) to measure benevolent attitudes. PP refers to the belief that the poor are less likely to behave in an evolved manner and require constant guidance and external supervision (e.g., “charitable organisations should guide poor people to make better life choices”). Lastly, CCD is related to the belief that the poor possess certain traits that render them more humble or honest than people from other socioeconomic groups (e.g., “poor people are often more humble than non-poor people”).

Identifying the different aspects of classism attitudes (i.e., from the hostile to the benevolent spectrum) may clarify the reason a proportion of the population tends to explicitly reject redistribution policies or welfare expenditure (i.e., autonomous help), yet simultaneously supports other assistance programmes that do not necessarily eradicate poverty and even promote the status quo (i.e., dependent help; ^{Nadler & Halabi, 2006}).

Correlates of Ambivalent Classism Attitudes

Ambivalent classism attitudes are related to social and psychological variables that, when combined, lay the foundations for the justification of the plight of the poor. As previously discussed, a belief exists that certain groups are naturally prone to occupy a dominant position (i.e., SDO) and this predicts discrimination and even dehumanisation of social minorities (^{Hodson & Dhont, 2015}; ^{Jordan et al., 2021}). This social dominance is frequently aligned with meritocratic beliefs with regards to how individuals and groups can overcome economic and social restraints only through their means to improve the quality of life (i.e., system justification beliefs; ^{Jost & Van der Toorn, 2012}). The ideological roots of people’s understanding of socioeconomic differences largely influence their willingness to help the poor (^{Bullock et al., 2003}).

Furthermore, such ideological positions are related to right-wing ideologies (^{Chambers et al., 2014}) that encourage the dismissal of the negative consequences of economic inequality and the tolerance of high levels of economic disparities (^{García-Castro et al., 2021}; ^{Wiwad et al., 2019}). In this regard, classism attitudes could increase tolerance of inequality by way of the perception that inequality is less harmful to those struggling their entire lives (^{Cheek & Shafir, 2020}). In addition to these variables, previous research also underscores that the tendency to dehumanise the poor triggers an internal attribution of poverty or perceptions of the poor as wasting their income (^{Sainz, Loughnan, et al., 2020}; ^{Sainz, Martínez, et al., 2020}). This tendency to dehumanise and blame the poor for their plight may be related to HC attitudes. Finally, previous research addresses how socioeconomic status may be related to classism attitudes (^{Jordan et al., 2021}). This evidence signalises that ideological positioning and socioeconomic status are not necessarily correlated because both processes are deemed independent.

Method

Adaptation of ACI items into Spanish

The study translated the original scale (ACI), which compromised 20 items along three subfactors (i.e., HC, PP, and CCD), using the back-translation design of ^{Hambleton (2005)}. Two translators with mastery of English and Spanish independently translated the original items from the source language (English) to the target language (Spanish). The authors then compared and discussed both versions to reach a consensual adaptation of each item. During this process, the authors acknowledged cultural discrepancies in the understanding of the item “By and large, if you give poor people an inch, they’ll take a mile.” This discrepancy was resolved by modifying the colloquial expression into an equivalent idiomatic expression in the source language: “En general, si le das la mano a las personas pobres, ellos te tomarán el brazo.” Afterwards, the two translators back-translated the items (from Spanish to English) and compared them with the original scale to ensure the quality and accuracy of the translation process. The authors made adjustments based on the consensual version of items when necessary.

Finally, a panel of experts composed of three scholars with expertise in scale construction and knowledge of the construct under evaluation (i.e., classism), evaluated the items (^{Carretero-Dios & Pérez, 2007}). The experts evaluated the wording for each item, the representation of the construct measured by the items, and the sub-factor to which each item belonged. The authors made final modifications based on expert evaluation (see Online supplementary information for a full disclosure of the materials). Table 1 presents the final items of the ACI in Spanish.

Table 1 Translated ACI ítems 

Results

Exploratory factor analysis

The study performed exploratory factor analysis on the first subsample of participants to test the scale’s factor structure. First, we confirmed the adequacy of the data. The results of the Kaiser-Meyer-Olkin sample adequacy test (KMO = .89) and Bartlett’s sphericity test (χ² = 69.86, p < .000) indicated that the data were well-suited for factor analysis. Moreover, parallel analysis (^{Horn, 1965}) indicated a three-factor solution as in the original scale.

Second, the study performed exploratory factor analysis using the maximum likelihood estimation with Promax rotation and selected an oblique rotation due to the expectation that the ACI factors would correlate. The results were similar to those of the original scale (^{Jordan et al., 2021}; Table 2) with three distinct factors that accounted for 58% of the total variance of data with HC (12 items), PP (4 items), CCD (4 items) explaining 32%, 12%, and 14%, respectively. Specifically, the descriptive statistics (mean, standard deviation, skewness, and kurtosis) fell within normal ranges. Moreover, the item-total correlation did not exceed 0.80. However, a very slight variation occurred in internal consistency reliability regarding the reliability of the total scale (α = . 88) when e ach item was dropped. Likewise, internal consistency reliability was acceptable for each subscale (HC: α = .93; PP: α = .87; CCD: α = .83). In succession, the coefficients exceeded 0.40 and loaded at least twice on the target factor because the next highest loading factor did not load more than 0.30 on multiple factors, whereas communalities exceeded 0.20 (^{Child, 2006}; ^{Tabachnick & Fidell, 2014}).

Table 2 Mean, standard deviation, item-total correlation, and Cronbach’s Alpha if the Item is eliminated, factor loading, and communality. 

Note. The highest loads in each factor are in bold.

Evidence of validity

The study collapsed both subsamples (N = 548) a nd computed Pearson’s bivariate correlations (Table 4) to test the ACI’s validity. In terms of intercorrelations between the inventory subfactors, PP is weakly in positive correlation with CCD and HC, whereas it did not display a significant correlation between CCD and HC. Moreover, the results highlighted that HC maintained a consistent and expected relationship with other ideological variables. In other words, higher HC was positively related to support for inequality, social dominance orientation, system justification, a lower ascription of humanity to poor people, and right-wing political orientation. However, PP exhibited a less consistent relationship with other ideological variables. Moreover, it showed only positive relationships with system justification and right-wing orientation but a negative relationship with tolerance to inequality. The scale is unrelated to the blatant dehumanisation scale. Finally, CCD was negatively related to many ideological variables, except for its positive relationship with system justification. The study further observed that neither SSC or OSC were related to the main variables.

Table 4 Descriptive statistics and Pearson’s bivariate correlations 

Note.*p < .05; **p < .01; ***p < .001.

Discussion

The project aims to adapt and validate the ACI developed by ^{Jordan et al. (2021)} to obtain an adequate tool for analysing classism attitudes and their consequences on the population of Spanish speakers. In general, the exploratory and confirmatory factor analyses illustrated that the psychometric properties of the adapted scale were acceptable and that the factor structure was similar to the original scale. In this sense, the analysis identified the originally proposed ACI factors, namely, HC (12 items), PP (4 items), and CCD (4 items). Further, the three-factor structure presented a better fit than other alternate models (one- and two-factor models). Thus, the study found that the proposed factor structure for native English speakers seems adequate for native speakers of Spanish. This finding indicated that classism attitudes differentiate between hostile and benevolent attitudes towards the poor. Moreover, the results demonstrated that the factors positively correlated with one another, except for HC and CCD. Compared with the original scale, the study considered that the strength of the correlations between variables was similar to those reported by English speakers. The biggest discrepancy between the scales then seemed to be the lack of correlation between HC and CCD, which also exhibited a weak and negative correlation in Jordan et al. (2021). This finding suggested that, at least for the Mexican sample, perceptions of complementary traits or behaviours between poor and rich groups (CCD) do not necessarily reflect a depiction of the poor (HC) but more of a paternalistic perception (e.g., poor but happy or honest; ^{Kay & Jost, 2003}). Thus, future research is needed to explore whether this difference is found only among Mexican individuals or whether it can be applied to other Spanish-speaking populations.

In addition, the ACI displayed a consistent pattern of relationships with the scales used by the present study to test the validity of the subfactors. Specifically, the results showed that system justification ideologies, the attribution of humanity to the poor, and the participants’ political orientation are more consistently related to HC than they are to other subfactors. This pattern of results may be due to the explicit and blatant nature of several variables, such as the Ascent of Man scale used to evaluate the humanity of the poor (^{Kteily et al., 2015}) even when ^{Sainz, Loughnan, et al. (2020)} and ^{Sainz, Martínez, et al. (2020)} demonstrated that the Ascent of Man scale is an adequate tool for understanding the tendency to dehumanise the poor and its consequences. However, this measure’s blatant and explicit nature was conceptually closer to hostile attitudes than to other dimensions of the scale. In this sense, future studies should aim to understand the possible relationships between subtle or implicit scales and hostile or paternalistic attitudes regarding poverty and the poor (^{Shor, 2019}).

Furthermore, the participants’ subjective and objective socioeconomic statuses failed to indicate any relationship with the ACI subfactors. This finding is in line with those of ^{Jordan et al. (2021)}, who found that the ACI is uniquely and weakly related to socioeconomic status. This notion underscores how the participants’ economic positioning and ideological standing do not necessarily relate to each other, considering the independence of both processes. Nevertheless, this hypothesis should be tested in-depth (e.g., increase the variability of the participants’ socioeconomic status in the sample). Previous research also identified a certain relationship between these factors (^{Horwitz & Dovidio, 2017}).

Limitations apply to the present study. First, the researchers carried out an adaptation among Spanish-speaking participants from one country. Given the diversity of the Spanish-speaking population across countries and continents, testing the performance and structure of the Spanish version of the ACI in samples from other countries may be interesting. Second, the study implemented the scale by focusing on classism attitudes toward the poor. ^{Jordan et al. (2021)} provided evidence of the cognition of the target population when answering the scale (e.g., homeless, afro-descendants and refugees), which indicates that lay representations of poverty include several groups. Thus, demonstrating the applicability of ambivalence classism attitudes to other groups, such as indigenous people, who are widely discriminated against in several countries in Latin America, could be another interesting avenue (^{Campos-Vazquez et al., 2019}). Finally, by implementing the Spanish adaptation of the scale in several countries, future research will test measurement invariance. The test results may verify whether the instrument measures the construct in the same manner across samples and corroborate whether possible differences are due to changes in the operationalisation of the construct within these samples instead of being due to differences in psychometric properties (^{Cheung & Rensvold, 2002}; ^{Lobato et al., 2020}).

Moreover, future studies might aid in the understanding of the antecedents of classism attitudes and how contextual factors shape this attitudinal variable. Hence, the possibility exists that the quality and quantity of contact with poor people, status anxiety, perception of social mobility, and other variables cited in the literature on prejudice will elucidate the appearance of classism (^{Hodson & Dhont, 2015}). Additionally, structural variables, such as economic inequality, the development of a country, or even poverty rates within countries, may influence classism attributes. Future cross-cultural studies should compare how country-level differences in these variables can modify the extent to which people develop more hostile or paternalistic attitudes towards the poor.

In other words, the current study provides evidence that the Spanish adaptation of the ACI can present adequate psychometric properties, which is a factor structure similar to the original scale and verifies convergent and divergent validity regarding related constructs in the literature. This ambivalent classism scale should contribute to understanding attitudes towards the poor and their consequences among Spanish speakers¹.