Introduction
The unabated growth of economic inequality in the United States (Solt Reference Solt2020) accompanied by flattening rates of intergenerational mobility (Chetty et al. Reference Chetty, Grusky, Hell, Hendren, Manduca and Narang2017) has led observers to question the vitality of the ‘American dream’. The American dream is a central feature of the American ethos (McClosky and Zaller Reference McClosky and Zaller1984), with a key pillar of the dream being the meritocratic ideal that anyone who works hard can achieve economic success. Popular media in the United States is replete with declarations that the American dream is dead or dying (Burtka Reference Burtka2024; Ferguson Reference Ferguson2023) and polling data suggest increasing pessimism among the American public about the attainability of the dream (Zitner Reference Zitner2023). Data collected in the Spring of 2024 indicates that nearly half of surveyed Americans (47 per cent) believe the dream is either no longer possible to achieve or was never really possible in the first place (Borelli Reference Borelli2024). In the lead-up to the 2024 US presidential election, former President and candidate Donald Trump solemnly declared in a televised speech in the battleground state of Pennsylvania that ‘the American dream is dead’.Footnote 1 This claim was echoed by his opponent, Vice President Kamala Harris, who stated in a televised interview with NBC News that, unlike past generations of Americans, current generations cannot count on ‘the idea of the American dream’ (Altus Reference Altus2024).
Against this backdrop, a growing vein of social science research explores whether exposure to economic inequality is a source of pessimism about the American dream (Davidai Reference Davidai2018; McCall et al. Reference McCall, Burk, Laperrière and Richeson2017; Mijs Reference Mijs2023; Newman Reference Newman2022). This work theorizes that exposure to inequality may erode belief in the American dream by shifting people’s attributions for wealth and poverty towards structuralist views emphasizing unfair practices that short-change the poor, working, and middle classes while ‘rigging’ the system in favor of the wealthy. While the mass media is one powerful source of exposure to meritocratic ideals in America (Carbone and Mijs Reference Carbone and Mijs2022; Kim Reference Kim2023), another important source is people’s local residential environment, which can vary considerably in terms of the amount of economic inequality (Sommeiller et al. Reference Sommeiller, Price and Wazeter2016) and presence of affluent persons living alongside the economically struggling (García-Castro et al. Reference García-Castro, Willis and Rodríguez-Bailón2019). In response to this reality, an expanding vein of scholarship examines the effect of local exposure to inequality on public opinion and political behavior (Davidai et al. Reference Davidai, Goya-Tocchetto and Lawson2024; Franko and Livingston Reference Franko and Livingston2022; Han and Kwon Reference Han and Kwon2023; Johnston and Newman Reference Johnston and Newman2016; Newman Reference Newman2020; Phillips Reference Phillips2017; Sands Reference Sands2017; Sands and Kadt Reference Sands and Kadt2020; Szewczyk and Crowder-Meyer Reference Szewczyk and Crowder-Meyer2022), with several studies focusing specifically on whether quotidian exposure to inequality dampens belief in meritocracy (Ellis Reference Ellis2017; Mijs Reference Mijs2023; Newman et al. Reference Newman, Johnston and Lown2015; Solt et al. Reference Solt, Hu, Hudson, Song and Yu2016). Understanding the sources of mass belief in the American dream is important given that variation in belief in meritocratic ideals is predictive of support for government redistribution (Alesina and La Ferrara Reference Alesina and La Ferrara2005; Fong Reference Fong2001; McCall et al. Reference McCall, Burk, Laperrière and Richeson2017; Newman Reference Newman2022).
This letter contributes to this growing scientific literature with a novel study: we explore the relationship of routine exposure to economic inequality in Americans’ adolescent residential context to their belief in the American dream upon entering adulthood. In contrast to prior research relying on samples ranging from roughly 6,000 to 35,000 Americans (Newman et al. Reference Newman, Johnston and Lown2015; Solt et al. Reference Solt, Hu, Hudson, Song and Yu2016), the data we use include information about pre-college home zip codes and a measure of belief in the American dream for roughly 1.3 million late-adolescent (17–19 years) Americans entering their first year of post-secondary education. What is more, these 1.3 million late-adolescents are dispersed across 27,695 unique zip codes, which encompasses 86.5 per cent of the roughly 32,000 zip code tabulation areas in the nation. This immense dataset affords a unique and unrivaled look at how quotidian exposure to economic inequality shapes belief in a key pillar of the American dream – namely, the meritocratic ideal that ‘Through hard work, everybody can succeed in American society’.
There is a longstanding literature on political socialization exploring the factors influencing the formation of political attitudes and behaviors (Jennings and Niemi Reference Jennings and Niemi1981; Sears and C Brown Reference Sears, Brown, Huddy, Sears, Levy and Jerit2023). The theory of ‘symbolic politics’, for example, contends that orientations formed during childhood and adolescence can become highly entrenched and powerful in steering adult partisan attachments, policy preferences, and vote choices (Sears and Funk Reference Sears and Funk1999; Sears Reference Sears, Iyengar and McGuire1993). One theorized mechanism through which orientations formed early in life have persistent influences on adult political attitudes and behavior is motivated reasoning (Lodge et al. Reference Lodge and Taber2000), which involves automatic (that is, uncontrollable) affect towards known attitude objects (Lodge and Taber Reference Lodge and Taber2005) and biased processing of new information (Taber and Lodge Reference Taber and Lodge2006) that functions to preserve preexisting attitudes and beliefs. Complementing this is a growing vein of evidence that events occurring early in life can have long-term effects on political attitudes and behavior (Erikson and Stoker Reference Erikson and Stoker2011; Holbein et al. Reference Holbein, Bradshaw, Munis, Rabinowitz and Ialongo2022; Komisarchik et al. Reference Komisarchik, Sen and Velez2022; Micatka Reference Micatka2025). Several of these studies focus on experiences tied to one’s childhood or adolescent residential context, such as being exposed to racially targeted violence (Antman and Duncan Reference Antman and Duncan2024), having different-race neighbors (JR Brown et al. Reference Brown, Enos, Feigenbaum and Mazumder2021), and living around larger racial outgroup populations (Goldman and Hopkins Reference Goldman and Hopkins2020).
While the literature is replete with studies exploring socialization experiences related to partisanship (Boonen Reference Boonen2019; Ojeda and Hatemi Reference Ojeda and Hatemi2015), race and ethnicity (Brown et al. Reference Brown, Enos, Feigenbaum and Mazumder2021; Goldman and Hopkins Reference Goldman and Hopkins2020), and gender (Bos et al. Reference Bos, Greenlee, Holman, Oxley and Lay2022; Healy and Malhotra Reference Healy and Malhotra2013), there is a paucity of political socialization research focusing on childhood economic context, with a notable absence of work investigating the effect of residing in economically unequal settings. To be sure, a considerable body of work explores the effect of growing up in economically disadvantaged settings on future health, educational, and economic outcomes (Chetty et al. Reference Chetty, Hendren and Katz2016; Sharkey and Faber Reference Sharkey and Faber2014), but this work rarely focuses on adult political attitudes and behavior as the outcome of study (cf. McNeil et al. Reference McNeil, Luca and Lee2023). Moreover, while scholars have been exploring adult Americans’ belief in meritocracy for decades (Kluegel and Smith Reference Kluegel and Smith1986), one gap in this literature involves examination of childhood or adolescent socialization experiences as potential sources of variation in adherence to the American dream. In sum, an important contribution to multiple veins of scholarship can be made with an empirical study of the relationship of adolescent exposure to economic inequality to belief in meritocracy. If, as theorized in prior research, routine exposure to inequality reduces belief in meritocracy, the political socialization literature suggests that strong doses of exposure during adolescence may exert long-term influences on Americans’ fidelity to meritocratic ideology.
Turning to analytical considerations, our focus on late-adolescents entering adulthood offers a valuable look at the relationship between residential context and ideological optimism early in life before the multiple changes in residence that typically occur throughout adulthoodFootnote 2 and the gamut of adult life experiences (for example, job loss, divorce, illness) that affect people’s optimism and political attitudes (Hacker et al. Reference Hacker, Rehm and Schlesinger2013; Starks Reference Starks2003). Moreover, our analysis of adolescent context mitigates concern over residential self-selection and reverse causality typically plaguing research investigating the effect of people’s context on their attitudes (Enos Reference Enos2016). Our focus on residential context during a life stage where location choices are typically made by parents and caregivers – not our respondents – lessens concerns that our adolescent respondents selected into high (versus low) inequality environments as a function of their economic outlook. In other words, we can reasonably assume that variation in residential exposure to inequality during adolescence is plausibly exogenous to belief in meritocracy in late-adolescence. Complementing this, the data enable us to control for parental characteristics (for example, income, education, and employment in a conservative-leaning occupation) potentially predictive of residing in a high- (versus low-) inequality setting and the economic outlook of their offspring.
Data and Methods
Our analysis uses the 2005–2008 waves of The Freshman Survey (TFS) conducted by the Cooperative Institutional Research Program. The TFS data, codebooks, and information about sampling and methods are archived at the Higher Education Research Institute (HERI).Footnote 3 The TFS is administered to incoming first-year college students during the summer or orientation before the start of classesFootnote 4 and the response rates in each wave are typically above 75 per cent (Mendelberg et al. Reference Mendelberg, McCabe and Thal2017). Each wave of the TFS contains a module of questions soliciting respondents’ economic, social, and political views, and the 2005–2008 waves include a question measuring belief in the American dream. Combined, these four waves include n = 1,399,081 late-adolescents (ages 17–19) collected across 839 post-secondary institutions.
The TFS is not a representative sample of late-adolescents, which limits our ability to generalize findings to all late-adolescent Americans. This said, the survey is administered during orientation before the respondents are ‘treated’ with any college education, which means that findings from the TFS are not limited in their relevance to only Americans holding four-year degrees. Critically, the late-adolescents in the 2005–2008 TFS are similar on key characteristics (for example, race/ethnicity, gender, political ideology) to other college and non-college samples of late-adolescents collected during these survey years (see Appendix A–B for sample comparisons and descriptive statistics). Moreover, the Bureau of Labor Statistics estimated that roughly 69 per cent of high school graduates in 2005 enrolled in colleges or universities.Footnote 5 The high rate of college entry implies that findings from analysis of the TFS speak to a large segment of the late-adolescent population. Even so, it is important to acknowledge the uncertainty related to the bearing of findings from the TFS on the segment of late-adolescent Americans that did not enter college during these years.
We measure belief in the American dream with respondents’ level of agreement with the statement: ‘Through hard work, everybody can succeed in American society’. The response options were (1) ‘Disagree strongly’, (2) ‘Disagree somewhat’, (3) ‘Agree somewhat’, and (4) ‘Agree strongly’. This item is consistent with measures of belief in the American dream used by the Pew Research Center and Gallup. At the beginning of the survey, TFS respondents are asked to report their pre-college ‘permanent/home address’ and the publicly available files provide the five-digit home zip code for all respondents.Footnote 6 Prior research establishes that zip code is an effective measure of the spatial plane envisioned by Americans when asked about their local residential context (YR Velez and Wong Reference Velez and Wong2017) and that Americans are aware of economic conditions in their local context (Newman et al. Reference Newman, Velez, Hartman and Bankert2015), including the amount of economic disparity (Minkoff and Lyons Reference Minkoff and Lyons2019; Newman et al. Reference Newman, Shah and Lauterbach2018). We merge with the TFS zip code data from the 2000 Decennial Census,Footnote 7 which provides raw household income category data used to calculate measures of economic inequality. To ensure any findings are not confined to any specific measure, we use three distinct measures of inequality: (1) the Gini Coefficient capturing income concentration, the 80:20 Ratio capturing the size of the gap between upper (80th percentile) and lower (20th percentile) income households, and (3) the interaction between the percentage of households with annual incomes below $25,000 (labeled %Below $25K) and those with annual incomes above $75,000 (labeled %Above $75K), which is used to capture the joint prevalence of ‘haves’ and ‘have-nots’ (Franko and Livingston Reference Franko and Livingston2022; Johnston and Newman Reference Johnston and Newman2016) and is a prepotent predictor of perceived local inequality (Newman Reference Newman2025).
We estimate multilevel models controlling for contextual and individual variables potentially correlated with zip code economic inequality and belief in the American dream. We control for zip code percentage college educated, median income, percentage unemployed, percentage Non-Latino White, population density, and the percentage of votes cast for George W. Bush in the 2000 presidential election.Footnote 8 We control for respondent race/ethnicity, gender, citizenship status, religious affiliation, religious attendance, left–right ideological self-identification, parental household income, father’s education, mother’s education, and parental employment in a profession associated with holding politically conservative views (for example, protective services, military, business manager or executive, clergy; see Appendix F). Controlling for parental characteristics is particularly important given these variables may shape both adolescent residential exposure to inequality and economic outlook upon entering adulthood. For ease of interpretation, we rescaled all non-binary independent variables to range between 0 and 1. For example, this rescaling of the Gini Coefficient rendered a one-unit change corresponding to a shift from zip codes with the minimum observed value of Gini to those with the maximum observed value. All models include random effects for survey year and unique post-secondary institution.
Results
Figure 1 depicts the relationship of the amount of economic inequality in respondents’ adolescent home zip code to their level of belief in the American dream upon entering adulthood. The left graph illustrates that moving from late-adolescents from home zip codes with the lowest values of the Gini Coefficient to those hailing from home zip codes with the highest values of Gini is associated with a significant decrease (β = −0.04, p < 0.001) in the belief that hard work will translate to economic success. Similar findings to the left graph are observed in the center graph when using the 80:20 Ratio measuring variation across home zip codes in the size of the income gap. The right graph presents the marginal effects of the increasing prevalence of low-income households (%Below $25K) across levels of prevalence of high-income households (%Above $75K) in respondents’ adolescent home zip code. We use the interflex package (version: 1.2.6) (Hainmueller et al. Reference Hainmueller, Mummolo and Xu2019) to ensure sufficient ‘common support’ in estimating conditional marginal effects and to avoid misleading model-induced linearity. This package estimates the marginal effect of predictor variables across binned values (low, middle, high) of moderator variables. The right graph reveals that zip code exposure to low-income people (%Below $25K) increasingly erodes belief in the American dream when it co-occurs with greater exposure to high-income people (%Above $75K). In sum, across three distinct measures, we find that Americans raised in areas with higher inequality enter adulthood with significantly lower levels of belief in the American dream than their peers raised in environments with lower inequality.
Adolescent residential exposure to economic inequality and belief in the American dream in late-adolescence.
Note: left and center graphs present the expected value of belief in the American dream across pre-college home zip code 0–1 rescaled values of the Gini Coefficient and 80:20 Ratio. Both include rug plots showing the distribution of observations across the x-axis. Right graph presents the marginal effects of %Below $25K on belief in the American dream across binned values of %Above $75K. See Appendix C for full model results.
The results in Figure 1 hold when estimating bivariate models that maximize sample size by minimizing data loss due to missing data on covariates (Table S11). Additionally, the results hold when excluding outlying values of inequality, using data from different census years, using a multilevel model with random intercepts at the zip code and county level, and controlling for institutional selectivity (Appendix D). To provide a greater sense of the size and substantive significance of the results in Figure 1, we (1) benchmark them to the coefficients observed for statistically significant model covariates, and (2) place them in the context of the size of effects reported in other studies using the TFS. To facilitate coefficient comparison, we present fully standardized coefficients in Table S13.
First, the standardized coefficient for Gini (−0.011) is larger than those observed for zip code median income (0.008) and population density (−0.006) and is nearly comparable to several arguably pretreatment individual-level covariates, such as gender (−0.015), citizenship (−0.012), and parental income (0.020). The standardized coefficient for %Below $25K in the third bin of %Above $75K is 0.129, which is larger than the standardized coefficients observed for all statistically significant model covariates except liberal-conservative ideology (see Table S14) and is roughly 73.3 per cent the size of the standardized coefficient for ideology. Thus, adolescent residence in areas with a pronounced joint prevalence of ‘haves’ and ‘have-nots’ is associated with a substantively meaningful decrease in belief in the American dream upon entering college. This finding aligns with those reported by Newman, (Reference Newman2025), who reveals that this interactive measure more strongly correlates with perceived local inequality than measures of income concentration (for example, Gini) or gap size (for example, 80:20). Second, we compare the standardized coefficients for our inequality measures in Figure 1 to standardized coefficients for key predictors in four published studies using the TFS (Appendix H, Table S23). These comparison studies explore the effect of having a different-race roommate during college on belief in meritocracy (Mijs Reference Mijs2023), the effect of college cohort parental affluence on preferences over taxing the rich (Mendelberg et al. Reference Mendelberg, McCabe and Thal2017), the effect of enrolling in an ethnic studies course on perceptions of racial discrimination (Chan & Raychaudhuri, Reference Chan and Raychaudhuri2025), and participation in a study abroad program on attitudes towards undocumented immigrants (Herrera et al. Reference Herrera, Garibay, Garcia and Johnston2013). Across these four published studies, the standardized effect for predictors ranges from 0.059 to 0.17, with an average standardized effect of 0.10. The standardized coefficients for Gini and the 80:20 Ratio are roughly 10 per cent and 4 per cent of the average effect observed for predictors in these other studies. In contrast, the standardized coefficient for %Below $25K in the third bin of %Above $75K exceeds the average effects observed in these comparison studies, providing further illustration of its substantive significance.
Heterogeneity by Childhood Household Income
Prior research theorizes that a person’s economic position may shape their reaction to income inequality. Some of this work contends that exposure to inequality will trigger ideological disillusionment and support for redistribution the most for lower income individuals (Meltzer and Richard Reference Meltzer and Richard1981; Newman et al. Reference Newman, Johnston and Lown2015; Sands and Kadt Reference Sands and Kadt2020), with the theorized mechanism being reminders of their own lower status in conjunction with unattainable wealth. This work also suggests that the economically privileged may experience guilt in response to inequality and may reduce feelings of guilt with heightened subscription to agency-based explanations for wealth and poverty (Newman et al. Reference Newman, Johnston and Lown2015). Other work, however, posits the opposite – that exposure to inequality will augment acknowledgment of economic unfairness and support for redistribution the most for higher income individuals (Franko and Livingston Reference Franko and Livingston2022; Rao Reference Rao2019; Suss Reference Suss2023), with the theorized mechanism being reminders of their own privileged status and increased empathy for the poor. Complementing this, system justification theory argues that, when confronted with inequality, lower status groups experience more ideological dissonance and are thus more motivated to reduce such dissonance with redoubled fidelity to inequality-legitimating belief systems like meritocratic ideology (Jost et al. Reference Jost, Pelham, Sheldon and Ni Sullivan2003). This framework suggests that exposure to inequality may dampen belief in meritocracy the most among higher status groups who are less motivated to justify the status quo.
We explore heterogeneity in the relationship of adolescent home zip code inequality to belief in the American dream by respondents’ adolescent household economic position – measured by parental income. We use the Gini Coefficient and 80:20 Ratio for this analysis because they are single-parameter measures and can be easily interacted with parental income. Figure 2 presents the estimated marginal effects of the Gini Coefficient (Panel A) and 80:20 Ratio (Panel B) among respondents raised in low-, middle-, and high-income households (that is, the three ‘bins’ of parental income) using the interflex package. The marginal effect estimates for Gini at the low-, middle-, and high-binned values of parental income indicate that, across all economic backgrounds, higher levels of exposure to inequality in one’s adolescent residential area are associated with significantly lower belief in the American dream. Turning to the 80:20 Ratio, we see that the marginal effect estimate is nearly zero among those raised in lower income households but negative and statistically significant for those raised in middle and higher income households. These results hold when estimating marginal effects across four or five bins of parental income (see Appendix E).
Adolescent residential exposure to economic inequality and belief in the American dream by parental income.
Note: figure plots the marginal effects of adolescent home zip code Gini by parental income (Panel A) and zip code 80:20 by parental income (Panel B) for the full sample
Figure 2 reveals that the negative relationships of these two local inequality measures to belief in the American dream are most pronounced among Americans raised in higher income households. While the estimated coefficient for Gini in the full sample is β = −0.04, (p < 0.001), the estimate more than triples in size among those in the highest bin of parental income (β = −0.14, p < 0.001). Furthermore, the estimated coefficient for the 80:20 Ratio in the full sample is β = −0.03, (p < 0.01) but nearly nonuples in size among those in the highest bin of parental income (β = −0.27, p < 0.001). This aligns with prior research finding that contact with poor peers can lead more affluent teens to be more prosocial, generous, and egalitarian (Rao Reference Rao2019), that attending non-affluent colleges leads to greater support for progressive taxation among American college students from affluent backgrounds (Mendelberg et al. Reference Mendelberg, McCabe and Thal2017), that residing near lower income households leads affluent Americans to be more aware of social problems (Thal Reference Thal2017), that zip code exposure to inequality is most strongly associated with greater support for redistributive policies among higher income Americans (Franko and Livingston Reference Franko and Livingston2022), and that Americans with higher incomes engage in more charitable giving when residing in higher inequality zip codes (Suss Reference Suss2023). A primary theorized factor facilitating the responsiveness of higher income people to poverty and inequality is empathy (Franko and Livingston Reference Franko and Livingston2022; Suss Reference Suss2023). We conducted ancillary analyses exploring the role of empathy in conditioning the findings reported in Figure 2. These analyses are presented in Figure S4 and S5 in Appendix G and reveal that the pattern of results in Figure 2 is confined to late-adolescents high in empathy.
Discussion
This research makes a vital contribution to our growing understanding of the consequences of rampant economic inequality on the economic optimism of the American people. Using surveys of over 1.3 million late-adolescent Americans entering college, we demonstrate that routine exposure to economic inequality in these youths’ adolescent residential context is associated with decreased belief in a central feature of the nation’s cultural ethos upon entering adulthood. We find that Americans raised in high-income households seem the most affected by having resided in a high-inequality setting, and ancillary analyses (appearing in the supplementary material) suggest this heightened sensitivity among the economically privileged may be due to their empathy towards the ‘have-nots’. Our findings add real-world texture to experimental research on inequality and the American dream relying on brief exposure to information about inequality (Davidai Reference Davidai2018; McCall et al. Reference McCall, Burk, Laperrière and Richeson2017).
Given that our analysis focuses on young people, it raises the question of the persistence of the observed association between adolescent exposure to inequality and belief in the American dream beyond late-adolescence and into adulthood. While our data cannot answer this question, prior research suggests that location-based experiences occurring during adolescence can have persisting long-term effects on socioeconomic and political outcomes. For example, adolescent exposure to ethnically targeted violence (Antman and Duncan Reference Antman and Duncan2024) and incarceration (Komisarchik et al. Reference Komisarchik, Sen and Velez2022) have been found to exert negative long-term impacts on home ownership and political trust. Additionally, exposure to ethno-racial diversity during adolescence can have long-term impacts on adult party identification (Billings et al. Reference Billings, Chyn and Haggag2021; Brown et al. Reference Brown, Enos, Feigenbaum and Mazumder2021) and racial attitudes (Goldman and Hopkins Reference Goldman and Hopkins2020). Finally, teenage participation in the ‘Moving to Opportunity’ program, which involved relocating from a high- to low-poverty neighborhood, was found to depress voter registration and turnout in adulthood (Elder et al. Reference Elder, Enos and Mendelberg2024). Our findings bear weight in light of this evidence that Americans’ pre-adult residential context can have powerful long-term impacts throughout adulthood.
With respect to limitations, the data we use are not designed to be representative samples of incoming college students or late-adolescent Americans, which limits the generalizability of our findings. This said, the large-N nature of the data we use affords an unrivaled opportunity to precisely estimate the relationship between adolescent economic context and belief in meritocratic ideology. Indeed, few commonly used nationally representative samples (for example, American National Election Study or Cooperative Election Study) include questions about meritocracy, and among those that do (for example, Pew Research Center surveys), researchers would confront small subsamples of late-adolescent respondents aged 18–20 years, resulting in underpowered hypothesis tests (Arel-Bundock et al. Reference Arel-Bundock, Briggs, Doucouliagos, Mendoza Aviña and Stanley2022). Finally, while the 2005–2008 TFS data used in this analysis overlap in time with the data used in a recent exchange over local exposure to inequality and belief in meritocracy (Newman et al. Reference Newman, Johnston and Lown2015; Solt et al. Reference Solt, Hu, Hudson, Song and Yu2016), our data are now 17–20 years old and future research could explore whether our findings replicate using more recent data.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0007123425101270.
Data availability statement
Replication data for this article can be found in Harvard Dataverse at: https://doi.org/10.7910/DVN/W1X0JQ.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interests
The authors declare none.
Ethical standards
The research meets all ethical guidelines, including adherence to the legal requirements of the study country.
Open access
