The Great Migration and Implicit Bias in the Northern United States

Abstract

The spatial patterning of present-day racial bias in Southern states is predicted by the prevalence of slavery in 1860 and the structural inequalities that followed. Here we extend the investigation of the historical roots of implicit bias to areas outside the South by tracing the Great Migration of Black southerners to Northern and Western states. We found that the proportion of Black residents in each county (N = 1,981 counties) during the years of the Great Migration (1900–1950) was significantly associated with greater implicit bias among White residents today. The association was statistically explained by measures of structural inequalities. Results parallel the pattern seen in Southern states but reflect population changes that occurred decades later as cities reacted to larger Black populations. These findings suggest that implicit biases reflect structural inequalities and the historical conditions that produced them.

Keywords

implicit bias the Great Migration structural inequality IAT regional differences

During the First World War, Black Americans began leaving Southern farms for Northern and Western cities. States north and west offered manufacturing jobs that paid higher wages than farming and offered an escape from the Jim Crow system of racial subjugation. Between 1910 and 1970, nearly six million Black Americans left. This mass exodus, which became known as the Great Migration, was the largest internal relocation in U.S. history, dwarfing the 19th-century gold rush and the 20th-century dust bowl migration combined (Tolnay, 2003; Wilkerson, 2010).

The Great Migration transformed Northern and Western cities. Historians argue that the concentration of Black Americans in these cities contributed, for example, to the Harlem Renaissance and the development of the Civil Rights Movement (Arnesen, 2003). Simultaneously, White backlash to larger Black populations led to new structural barriers for Black Americans in the North and West. White authorities responded to the growing Black populations by segregating neighborhoods and schools (Massey & Denton, 1993; Rothstein, 2017). For example, restrictive covenants kept Black families in Black-only areas of cities. Federal housing policies known as redlining prevented Black residents from obtaining mortgages, limiting their ability to build wealth. Black workers were barred in many places from joining labor unions and were relegated to low-wage work. Black Americans seeking to escape Jim Crow segregation in the South found the system partially rebuilt around them in Northern and Western cities.

In this article, we examined the legacy of the Great Migration for present-day racial biases. We hypothesized that counties with larger populations of Black residents during the first half of the 20th century would show greater implicit bias among White residents today. We also predicted that this association would be partly explained by the level of systemic racism in those counties. Our aim was to show that, unlike the Southern United States, where regional implicit bias can be traced to slavery and Jim Crow systems, in the North and West, the roots of implicit bias for White Americans can be traced to the Great Migration.

Theoretical Framework: How Historical Racism Relates to Inequality and Implicit Bias

Implicit bias describes mental links between social groups and evaluations or stereotypes (Banaji, 2001; Fazio & Olson, 2003; Petty et al., 2008). It is measured using performance on cognitive tests, rather than self-report, making it more difficult to conceal or manipulate than explicit attitudes. Explicit attitudes, in contrast, are self-reported using traditional survey methods. Although both explicit and implicit bias have been declining over time, race bias on implicit measures remains more prevalent than in explicit reports (Charlesworth & Banaji, 2019; Schuman et al., 1997). The literature on implicit bias does not treat either type of bias as the “true” attitude; rather, implicit and explicit bias are often independently associated with discriminatory behavior (Amodio & Devine, 2006; Dovidio et al., 2002; Kurdi et al., 2019). Because of these independent effects, implicit bias has often been invoked to explain continued discrimination even among people who do not explicitly endorse prejudiced attitudes.

Research suggests that features of social environments can cue implicit bias. Although the biases activated in specific individuals can be transient, aggregated measures of bias in a city, county, or state reflect the collective biases in that context and are more enduring. As many aspects of environments remain stable over time, aggregate implicit bias can be highly stable (Payne et al., 2019; Vuletich & Payne, 2019). Several kinds of environmental features have been found to correlate with average levels of implicit bias. Racial biases among White Americans are higher, for example, in metro areas with poorer health and fewer health care providers (Hehman et al., 2021), and in states with lower Medicaid spending (Leitner et al., 2018). Implicit bias is higher on college campuses with fewer Black faculty, lower upward mobility among students, and more confederate monuments (Vuletich & Payne, 2019). Aggregate scores are also associated, in some cases quite strongly, with racial disparities in important outcomes. For example, county-, city-, or state-level biases are associated with racial disparities in health (Leitner et al., 2016; Orchard & Price, 2017), education (Chin et al., 2020), school discipline (Riddle & Sinclair, 2019), and police use of force (Hehman et al., 2018).

These associations raise questions about the origins of context-based implicit biases, and why some places are more racially biased than others. Some research suggests that aggregate implicit biases may track long-term structural features. For example, variability in implicit gender biases across cultures tracks stereotypes embedded in languages (Charlesworth et al., 2021; Lewis & Lupyan, 2020). Most relevant to the present research, the proportion of the population enslaved in each county in 1860 was associated with greater implicit racial bias among White residents today (Payne et al., 2019). That association was partially statistically explained by markers of systemic racism, including modern-day racial segregation and racial disparities in poverty and upward mobility. Areas more dependent on slavery in the antebellum South erected more structural barriers for Black Americans following the Civil War, which can still be detected today in patterns of segregation and economic disparities. Those present-day racial disparities, in turn, may cue biased associations in the minds of people who inhabit those spaces. Payne and colleagues (2019) suggested that large racial inequalities cue racial stereotypes in the minds of White residents, leading to higher levels of implicit preferences for White over Black individuals. These findings suggest that structural inequalities rooted in slavery and its aftermath may contribute to implicit bias and maintain it more strongly in some places than others.

Those findings primarily reflected variability in Southern states, where enslavement was widespread in 1860. However, implicit and explicit bias obviously exist outside of the South. In the present research, we focused on Northern and Western states following the Civil War. We examined U.S. Census estimates of the Black populations in each county in each decade to test whether—and when—the differential proportions of Black populations throughout the Great Migration became predictive of higher levels of implicit race bias today.

We hypothesized that, just as in the South, areas with larger Black populations would have higher levels of systemic inequalities (indexed by the proportion of Black residents among the poor, lack of economic mobility for Black residents, and residential segregation), and these systemic inequalities would in turn be associated with modern racial bias for White residents. However, we expected the relevant time frames to be later in the North and West, following the path of the Great Migration. We did not have a basis to predict which years during the Great Migration would be most relevant for contemporary implicit bias. Hence, we examined the link between implicit bias and Black populations for each census decade during that period (1910–1970). For comparison, we also examined the link between implicit bias and Black populations before and after the Great Migration (1870–1900 and 1980–2010). We did not expect population characteristics during those periods to be as strongly linked to implicit bias as the intervening years. In addition, we examined associations with explicit bias as a point of comparison. As we mentioned previously, although explicit and implicit bias tend to be positively correlated, they also show independent associations with other variables. Because explicit bias can be modulated to conform to local norms, self-presentation concerns, and other factors, we did not have an a priori hypothesis for this variable about the pattern of associations to population characteristics before, during, and after the Great Migration.

In exploratory analyses, we also examined implicit biases among Black residents and their relation to Black populations across the decades. Previous work has shown that Black residents show stronger implicit biases in favor of Black over White people in places where there is more structural inequality and a history of greater dependence on slavery (Payne et al., 2019). We expected our findings to be consistent with this previous work and reveal that larger Black populations (and greater structural inequality) are associated with a stronger bias among Black respondents favoring Black over White people. However, we did not have an a priori hypothesis about the pattern of associations across census decades.

Method

Sample

The implicit and explicit racial bias data used in this study came from the Project Implicit (https://implicit.harvard.edu/implicit/), an organization that collects data through a demonstration site visited by millions of users every year. We used county-level racial bias averages and only included counties: (1) in states¹ outside the confederacy (54% of the Implicit Association Test [IAT] data set), and (2) with nonmissing values (90% of the data set). These inclusion criteria yielded 37 states with 1,981 counties for White respondents (representing 1,636,473 IAT respondents) and 1,361 counties for Black respondents (representing 214,875 IAT respondents).

Variables

Table 1 presents the descriptive statistics and zero-order correlations for White respondents (full correlation tables for White and Black respondents are in the Supplementary Information).

Table 1

Descriptive Statistics and Zero-Order Correlation Matrix for the Focal Variables

	Year of measurement	Descriptive statistics		Zero-order correlations
Variable	Year of measurement	M	SD	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
(1) Implicit bias (White respondents)	2002–2018	0.37	0.03	1.00
(2) Explicit bias (White respondents)	2002–2018	0.58	0.21	.76***	1.00
(3) Past proportion of Black residents	1930	0.03	0.05	.16***	.05*	1.00
(4) Total population count	2010	1,022,027	1,680,172	−.12***	−.24***	.05*	1.00
(5) County area	2010	1,254	2,147	−.10***	−.14***	−.15***	.30***	1.00
(6) Present proportion of Black residents	2010	0.09	0.10	.07**	−.05*	.60***	.22***	−.17***	1.00
(7) Proportion of people living in poverty who are Black	2010	0.16	0.15	.13***	.05*	.58***	.18***	−.22***	.93***	1.00
(8) Intergenerational mobility among Black residents	1980–2012	34.01	3.19	−.23***	−.26***	−.14***	−.09***	.03	−.20***	−.30***	1.00
(9) Racial residential segregation	2010	29.02	21.38	.38***	.32***	.48***	.14***	−.30***	.66***	.69***	−.40***

Note. We used weights to account variations in the number of participants from one county to another.

***

p < .001. **p < .01. *p < .05.

Implicit and Explicit Bias (County-Level Averages)

Implicit Bias From 2002 to 2018 (Main Outcome)

Visitors to the Project Implicit demonstration website take an IAT that measures people’s associations between the racial categories “Black” and “White” and evaluations “Good” and “Bad” (Xu et al., 2014). Higher scores mean greater pro-White/anti-Black bias. Scores correspond to an effect size and range from negative to positive, with values of 0 reflecting no difference in the speed of categorizing racial categories with good/bad ratings. We averaged responses for all years from 2002 to 2018 at the county level for White and Black respondents separately.

Explicit Bias From 2002 to 2018 (Additional Outcome)

Visitors to the Project Implicit website also respond to questions about their racial attitudes. We used two “thermometer” scales to construct our measure of explicit bias. These items asked respondents to rate how warmly they feel toward White and Black people on a scale of 0 (very cold) to 10 (very warm). We subtracted warm feelings toward Black people from warm feelings toward White people to calculate an explicit bias score and averaged responses for all years from 2002 to 2018 at the county level for White and Black respondents separately.

Structural Variables (County-Level Sociodemographics)

Proportion of Black Residents From 1870 to 2010 (Focal Predictor)

Population data for each census year from 1870 to 2010, came from IPUMS-NHGIS (Manson et al., 2022). Higher values indicate greater proportions of Black residents among the total population in the county for each year. We also collected three potential population confounders: county area, total population count, and Black population in census year 2010. We used them as control variables when testing the effects of the past proportion of Black residents in the regression analysis. Because the distributions of the population variables were right-skewed, we log-transformed all values.

Proportion of People Living in Poverty Who Are Black, 2008–2012 (Intervening Variable 1)

Poverty data came from the American Community Survey “Table S1701, 5-year estimates (2008–2012),” disaggregated by county and race (U.S. Census Bureau, 2012). For our analyses, we used the proportion of individuals below the poverty line who are Black.

Intergenerational Economic Mobility Among Black Residents from 1980 to 2012 (Intervening Variable 2)

Economic mobility data came from Opportunity Insights and included commuting zone income rank statistics for Black children born between 1980 and 1991 of parents in the bottom 25th percentile of income level (Chetty et al., 2014).² Higher values on this measure indicated more economic mobility (children who have moved up the income ranks in 1996–2012 in reference to their parents in 1980–1991).

Racial Residential Segregation in 2010 (Intervening Variable 3)

Residential segregation data came from the American Communities Project (n.d.) and were based on the 2010 decennial census. We used an isolation index that measures, at the level of metropolitan areas, Black residents’ exposure to other Black residents. Values ranged from 0 to 100, with higher values indicating greater residential segregation. For instance, a value of 56.3 means that the average Black resident in a given metropolitan area lives in a neighborhood that is 56.3% Black.

Data Availability

Although the data used here are publicly available through the source sites, the merged data file, a codebook, and the scripts to reproduce the analysis are available on the OSF page of the project: https://osf.io/b8m9e/.

Results

Analysis Plan

Our first question was whether past proportions of Black residents in Northern and Western counties (using populations estimates from 1870 to 2010) are associated with modern-day implicit bias (using aggregated implicit bias data from 2002 to 2018). We used a bivariate correlation analysis and applied analytical weights to account for the fact that some counties had fewer participants and thus less accurate estimates of the aggregated bias, while other counties had more participants and thus more accurate estimates of the aggregated bias (Dupraz, 2013). We first report the correlations among White respondents, and then among Black respondents.

Next, we built a regression model with standard errors (SEs) adjusted for state clustering (accounting for the nonindependence of residuals within states). In this model, we used past population characteristics (from the period of the Great Migration) to predict implicit bias while statistically controlling for the proportion of Black residents in 2010. We also controlled for county area and total county population to address the possibility of selective migration to larger, more urban, and densely populated counties, which might have different levels of implicit bias compared with smaller, more rural, and less densely populated areas. As in the correlational analysis, we used weights to account for variations in the number of participants from one county to another.

Next, we performed a specification curve analysis as a robustness check, an analytical tool that enabled us to test all reasonable regression models testing the association between past proportion of Black residents and current implicit bias (Simonsohn et al., 2020). This analysis accounts for the fact that we made methodological and analytical decisions to construct what we think is the most reasonable model. However, some of these decisions can be arbitrary in nature (e.g., operationalizing the past proportion of Black residents using data from one year rather than another), and different researchers might have chosen to build a different model (for an empirical illustration, see Silberzahn et al., 2018). The following features were allowed to vary across models: the specific census year used to predict implicit bias, variable transformations, inclusion of control variables, and the method for accounting for unequal county number of responses (more details in the sections below).

Finally, we aimed to test the hypothesis that communities responded to how large the Black population was during the Great Migration by creating enduring systemic inequalities that continue to cue biases in modern-day county residents. Specifically, we tested the mediational role of systemic inequalities using the same three variables that were found, in the South, to mediate the association between enslaved populations and implicit bias: (1) proportion of people living in poverty who are Black, (2) intergenerational economic mobility among Black residents, (3) racial residential segregation (Payne et al., 2019). Each intervening variable was matched, as closely as possible, to the period covered by the implicit bias data. We used a structural equation model (SEM) with SEs adjusted for state clustering. Our model tested whether the relation between past proportion of Black residents and current implicit bias operated through a latent variable labeled “structural inequality,” which was based on these three components. We included the same control variables used in the regression analysis and we again used weights to account for the variations in the number of participants from one county to another.

Correlational Analyses

Implicit and Explicit Bias Among White Respondents

First, we tested the bivariate correlations between the proportion of Black residents in each county for each census year and measures of implicit bias among White residents today (with analytic weights). As seen in Figure 1 (left panel), the county-level proportion of Black residents across the years was positively associated with implicit bias favoring White over Black individuals today. Specifically, the association peaked in 1930, r = .16, 95% confidence interval (CI) = [0.12, 0.20], p < .001, and it was not significant in the years 1960 through 2000 (after the Great Migration). These results are consistent with our hypothesis that population characteristics during the years of the Great Migration relate more strongly to implicit bias today compared with years outside this period. Note that the Black population size in 2010 was correlated with implicit bias, which is accounted for by the fact that the implicit bias data span from 2002 to 2018, capturing the societal context of 2010.

We used the same procedure to test the correlations between the proportion of Black residents in each county for each census year and measures of explicit bias among White residents today. As seen in Figure 1 (right panel), explicit bias followed a very different pattern: Correlations were positive or null for years 1860 through 1940, and after 1940, counties with larger proportions of Black residents showed less explicit bias among White residents. An asymptotic chi-square test for the equality of two correlation matrices revealed that the matrix pertaining to implicit bias was significantly different from the matrix pertaining to explicit bias, χ²(120) = 5,088, p < .001. The dissociation between implicit and explicit bias as a function of Black populations is interesting because implicit and explicit bias are positively correlated across counties and tend to correlate with other variables in the same direction (Hehman et al., 2019).

Implicit and Explicit Bias Among Black Respondents

We used the same procedure described above to test the bivariate correlations between the proportion of Black residents in each county for each census year and measures of implicit bias among Black residents today. The pattern of findings was different from the pattern observed with White residents. As seen in Figure 2 (left panel), counties with larger proportions of Black residents displayed greater pro-Black implicit bias among Black residents, and this association was stronger when considering more recent population counts. Unlike the implicit biases of White respondents, the implicit biases of Black respondents appeared to be more strongly tied to contemporaneous demographic characteristics rather than past population characteristics. It is worth noting that, even though the link to past population characteristics was not as strong as the link to contemporaneous characteristics, the implicit biases of Black residents were still tied to structural inequalities, r = .20. (Black mobility), r = −.31 (proportion of the poor who are Black), and r = −.29 (residential segregation) (full correlation table in the Supplemental Information). The correlations suggest that higher levels of structural inequality in an area are associated with greater pro-Black (or anti-White) implicit bias among Black residents.

We again used the same procedure to test the bivariate correlations between the proportion of Black residents in each county for each census year and measures of explicit bias among Black residents today. As seen in Figure 2, this pattern of larger modern Black populations being associated with more pro-Black evaluations was consistent across implicit and explicit measures for Black respondents (the two correlation matrices were not different from one another, χ²(120) = 50.88, p = 1.00), and it mirrored the pattern observed for White residents’ explicit bias. The implicit biases of White residents were unique in both the direction of the association (positive) and the timing of the populations that mattered (decades within the Great Migration).

Regression Analyses

Next, we accounted for the nesting of counties within states and separated the effect of past population characteristics from the effect of current population characteristics. All variables were standardized for ease of interpretation. Below is the cluster-adjusted weighted regression equation (which accounts for nonindependent residuals within states):

\begin{matrix} \sqrt{n_{j}} \cdot {\bar{IAT}}_{j} = β_{0} + \sqrt{n_{j}} \cdot β_{1} \cdot \log ({\bar{Black Prop}}_{j}) \\ + \sqrt{n_{j}} \cdot β_{k} \cdot \log ({\bar{Control}}_{k_{j}}) + \sqrt{n_{j}} \cdot {\bar{μ}}_{j} \end{matrix}

. . . where j = 1, 2, . . . N [counties], $\sqrt{n_{j}}$ represents analytic weights (with n being the number of respondents per county), ${\bar{IAT}}_{j}$ represents the standardized county-aggregated implicit bias score, ${\bar{BlackProp}}_{j}$ represents the standardized proportion of Black residents in 1930, and ${\bar{Control}}_{k_{j}}$ represents a vector of the three standardized population control variables in 2010, and µ is the error term.

Table 2 presents the full findings. The association between the county-level proportion of Black residents in 1930 and implicit bias observed in the correlational analyses remained significant, β = 0.11, 95% CI = [0.02, 0.20], p = .016. Importantly, the proportion of Black population in 1930 explained variance over and above the proportion of Black population in 2010, as well as county area and total county population in 2010.

Table 2

White Respondents’ Implicit Bias Predicted by Black Populations in 1930

Predictor	β	SE ^a	p	95% CI
Proportion of Black population in 1930	0.11	0.04	.016	[0.02, 0.20]
Proportion of Black population in 2010	0.07	0.03	.051	[0, 0.13]
County area in 2010	0.02	0.03	.508	[−0.03, 0.07]
Total county population 2010	−0.15	0.03	<.001	[−0.20, −0.09]

Note. We used weights to account for variations in the number of participants from one county to another. Because the distribution of population variables is right-skewed, all predictors were log-transformed. All variables were standardized. CI = confidence interval.

Cluster-adjusted (number of state clusters = 37).

Robustness Check

As a robustness check, we conducted a specification curve analysis using the following four specification dimensions: (1) census year of measurement of past proportion of Black residents (ranging from 1900 to 1950, that is, the significant correlations for the Great Migration–related census years observed in the preliminary correlational analysis), (2) procedure accounting for the between-county variation in sample size (analytical weights or cutoffs ranging from n > 100 to >300 with a +50 increment; see Kurdi & Banaji, 2017), (3) set of population control variables (i.e., k = 0, 1, 2, or 3), and (4) data transformation method (untransformed or log-transformed variables). This resulted in 6 × 6 × $(\begin{matrix} 3 \\ k \end{matrix})$ × 2 = 576 possible model specifications.

We ran these 576 possible cluster-adjusted regression models and gathered the 576 standardized estimates associated with the effects of past proportion of Black residents. The median standardized estimate was different from zero $\bar{β} = . 047$ , 95% CI = [0.045, 0.049], p < .001; the direction of the effect was positive in nearly all cases, 98.26% [98.6%, 99.1%]; and the critical p value was below .05 in the clear majority of the models, 63.7% [59.6%, 67.6%]. As can be seen in Figure 3, this suggests that—despite variations from one model to another—the association was robust to model specifications.

Mediation Analysis

Next, we tested the mediational role of systemic inequalities using the same three variables that were found, in the South, to mediate the association between enslaved populations and implicit bias: (1) proportion of people living in poverty who are Black, (2) intergenerational economic mobility among Black residents, (3) racial residential segregation (Payne et al., 2019). Figure 4 offers a graphical representation of the finding. Proportion of Black residents in 1930 was a positive predictor of structural inequality circa 2010, β = 0.75, 95% CI = [0.66, 0.85], p < .001 (a path), which itself was a positive predictor of implicit bias, β = 0.61, 95% CI = [0.43, 0.79], p < .001 (b path). The association between proportion of Black residents in 1930 and implicit bias became nonsignificant when including structural inequality, β = 0.07, 95% CI = [−0.13, 0.28], p = .479 (c’ path). The indirect effect—calculated using the percentile bootstrap method with 10⁴ resamples (Yzerbyt et al., 2018)—was positive, indirect = 0.39, 95% CI = [0.10, 0.76], p = .003. Although these analyses cannot establish causality, they are consistent with the hypothesis that demographic population characteristics in the North and West during the Great Migration led to structural inequalities that in turn trigger higher levels of implicit bias in those counties today.

Discussion

We examined how the Great Migration set the stage for regional differences in implicit and explicit bias. For White respondents, the average implicit preference for White over Black individuals today was higher in counties that experienced larger-sized Black populations mid-century. This effect was statistically mediated by regional differences in structural inequalities. Counties with larger proportions of Black residents during this historical period became more residentially segregated, had larger proportions of Black individuals among the poor, and had less economic mobility among Black residents. These indirect effects are consistent with theories of implicit bias as a reflection of systemic racism in the environment (Payne & Hannay, 2021).

Explicit bias, in contrast, was lower in counties with larger Black populations. This negative correlation contrasts with the positive correlation between Black populations and explicit bias that has been documented in the South (Payne et al., 2019). Lower explicit bias in areas with larger Black populations is consistent with the contact hypothesis, which suggests that greater inter-racial interaction can reduce prejudice, at least under certain conditions (Paluck et al., 2019; Pettigrew & Tropp, 2006). The fact that the associations were stronger in more recent decades is consistent with the idea that it is present-day experiences rather than historical patterns that are responsible. Recent research examining regional variation in populations suggests that the effect of Black populations on White prejudice depends on the degree of positive inter-group contact (Rae et al., 2015). It is possible that present-day inter-racial contact in the Northern and Western states examined here is more positive than in Southern states. If so, this difference might explain why larger modern-day Black populations are associated with lower explicit bias in the North and West but greater explicit bias in the South. Future research should examine this question further.

For Black respondents, larger Black populations were associated with a larger preference for Black over White individuals on both implicit and explicit measures. One interpretation of this correlation is consistent with the contact hypothesis. Areas with larger Black populations may lead to less inter-racial contact with White residents, and hence less positive out-group attitudes. Another interpretation is that larger Black populations lead to more positive in-group contact and positive in-group regard among Black residents (Welch et al., 2001). Because the IAT is a relative measure, we are not able to distinguish between negative out-group attitudes and positive in-group attitudes in these data.

A third interpretation is that inequalities—which are more pronounced in places with larger Black populations—may cue racial discrimination in the minds of Black residents, leading to implicit preferences for Black over White people. Consistent with previous work (Payne et al., 2019), we found that greater structural inequality was associated with stronger anti-White (or pro-Black) biases among Black respondents. In this and previous work, however, historical events were less relevant to bias compared with more contemporary population characteristics. Overall, the findings suggest that White individuals are more susceptible to historical legacies, whereas Black individuals are more attuned to the contemporary context. Further research is needed to understand why this is the case. One possibility is that a racist historical context generates more individualistic explanations for inequality, which White people readily use when interpreting a situation. In contrast, Black people may more readily use structural explanations because of personal experiences, community experiences, or contemporary antiracist movements. Research has found that White people indeed tend to endorse more individualistic explanations for racial inequality compared with Black people, who tend to endorse more structural explanations (Hunt, 2007; Pew Research Center, 2016).

The correlational nature of these data prevents us from drawing causal conclusions, although the time series nature of the Census data provides some information about temporal precedence. Our emphasis on historical processes led us to focus on one direction of the theoretical causal arrow, but implicit biases can also contribute to the maintenance of structural inequalities insofar as they result in increased discrimination and reduced readiness to dismantle racist structures. Our measures of structural inequality were based on data that are representative of the general population, but implicit and explicit attitude variables from Project Implicit are based on an opt-in sample. Project Implicit is, however, the only source of data available for measuring biases at the level of counties and states and it represents millions of U.S. residents. Our mediation findings suggest that present-day structural inequalities may cue implicit biases. However, we cannot identify what aspects of modern social and physical environments cue biases. Future research should examine the environmental cues that connect inequalities in the social environment to implicit associations in the minds of residents.

The most novel and distinctive finding in the present research is the positive association between Black populations in the early and mid-20th century and implicit bias among White residents. Explicit measures among White residents, and both explicit and implicit measures among Black residents revealed negative associations that were driven by recent decades. Implicit bias among White residents, however, followed a pattern similar to that observed with enslaved populations in the South. This pattern emerged decades later in the North and West, however, as the Great Migration brought Black Americans to Northern and Western cities in unprecedented numbers. These results suggest that the foundation for implicit bias among White residents outside the south was laid decades ago and is embedded in the structural inequalities that characterize modern spaces.

Footnotes

Handling Editor: Monteith Margo

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: The work for this manuscript was supported in part by NSF grant 2041192 awarded to B. Keith Payne.

ORCID iD

Heidi A. Vuletich

Supplemental Material

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

Notes

Author Biographies

Heidi A. Vuletich is an assistant professor at the University of Denver and studies the psychological effects of inequality.

Nicolas Sommet, PhD, is an SNSF Ambizione Lecturer in social psychology at the LIVES Center, University of Lausanne, Switzerland. He works on the psychological consequences of economic inequalities and on achievement motivation.

B. Keith Payne, PhD, is a professor at the University of North Carolina at Chapel Hill and studies how inequality shapes the human mind.

References

American Communities Project. (n.d.). Diversity and disparities. https://s4.ad.brown.edu/projects/diversity/Data/data.htm

Amodio

D. M.

Devine

P. G.

(2006). Stereotyping and evaluation in implicit race bias: Evidence for independent constructs and unique effects on behavior. Journal of Personality and Social Psychology, 91(4), 652–661. https://doi.org/10.1037/0022-3514.91.4.652

Arnesen

(2003). Black protest and the great migration: A brief history with documents. Bedford/St. Martin’s.

Banaji

M. R.

(2001). Implicit attitudes can be measured. In Roediger

H. L.

III Nairne

III, J. S.

Neath

Surprenant

(Eds.), The nature of remembering: Essays in honor of Robert G. Crowder (pp. 117–150). American Psychological Association.

Charlesworth

T. E. S.

Banaji

M. R.

(2019). Patterns of implicit and explicit attitudes: I. long-term change and stability from 2007 to 2016. Psychological Science, 30(2), 174–192. https://doi.org/10.1177/0956797618813087

Charlesworth

T. E. S.

Yang

Mann

T. C.

Kurdi

Banaji

M. R.

(2021). Gender stereotypes in natural language: Word embeddings show robust consistency across child and adult language corpora of more than 65 million words. Psychological Science, 32(2), 218–240. https://doi.org/10.1177/0956797620963619

Chetty

Hendren

Kline

Saez

(2014). Where is the land of opportunity? The geography of intergenerational mobility in the United States. The Quarterly Journal of Economics, 129(4), 1553–1623. https://doi.org/10.1093/qje/qju022

Chin

M. J.

Quinn

D. M.

Dhaliwal

T. K.

Lovison

V. S.

(2020). Bias in the air: A nationwide exploration of teachers’ implicit racial attitudes, aggregate bias, and student outcomes. Educational Researcher, 49(8), 566–578. https://doi.org/10.3102/0013189X20937240

Dovidio

J. F.

Kawakami

Gaertner

S. L.

(2002). Implicit and explicit prejudice and interracial interaction. Journal of Personality and Social Psychology, 82(1), 62–68. https://doi.org/10.1037/0022-3514.82.1.62

10.

Dupraz

(2013). Using weights in Stata. https://www.parisschoolofeconomics.eu/docs/dupraz-yannick/using-weights-in-stata(1).pdf

11.

Fazio

R. H.

Olson

M. A.

(2003). Implicit measures in social cognition research: Their meaning and use. Annual Review of Psychology, 54, 297–327. https://doi.org/10.1146/annurev.psych.54.101601.145225

12.

Hehman

Calanchini

Flake

J. K.

Leitner

J. B.

(2019). Establishing construct validity evidence for regional measures of explicit and implicit racial bias. Journal of Experimental Psychology: General, 148(6), 1022–1040. https://doi.org/10.1037/xge0000623

13.

Hehman

Flake

J. K.

Calanchini

(2018). Disproportionate use of lethal force in policing associated with regional racial biases of residents. Social Psychological and Personality Science, 9(4), 393–401. https://doi.org/10.1177/1948550617711229

14.

Hehman

Ofosu

E. K.

Calanchini

(2021). Using environmental features to maximize prediction of regional intergroup bias. Social Psychological and Personality Science, 12(2), 156–164. https://doi.org/10.1177/1948550620909775

15.

Hunt

M. O.

(2007). African American, Hispanic, and White Beliefs about Black/White Inequality, 1977–2004. American Sociological Review, 72, 390–415. https://journals-sagepub-com.du.idm.oclc.org/doi/pdf/10.1177/000312240707200304

16.

Kurdi

Banaji

M. R.

(2017). Reports of the death of the individual difference approach to implicit social cognition may be greatly exaggerated: A commentary on Payne, Vuletich, and Lundberg. Psychological Inquiry, 28(4), 281–287. https://doi.org/10.1080/1047840X.2017.1373555

17.

Kurdi

Seitchik

A. E.

Axt

J. R.

Carroll

T. J.

Karapetyan

Kaushik

Tomezsko

Greenwald

A. G.

Banaji

M. R.

(2019). Relationship between the Implicit Association Test and intergroup behavior: A meta-analysis. American Psychologist, 74(5), 569–586. https://doi.org/10.1037/amp0000364

18.

Leitner

J. B.

Hehman

Ayduk

Mendoza-Denton

(2016). Racial bias is associated with ingroup death rate for Blacks and Whites: Insights from Project Implicit. Social Science & Medicine, 170, 220–227. https://doi.org/10.1016/j.socscimed.2016.10.007

19.

Leitner

J. B.

Hehman

Snowden

L. R.

(2018). States higher in racial bias spend less on disabled Medicaid enrollees. Social Science and Medicine, 208, 150–157. https://doi.org/10.1016/j.socscimed.2018.01.013

20.

Lewis

Lupyan

(2020). Gender stereotypes are reflected in the distributional structure of 25 languages. Nature Human Behaviour, 4, 1021–1028. https://doi.org/10.1038/s41562-020-0918-6

21.

Manson

Schroeder

Van Riper

Kugler

Ruggles

(2022). IPUMS National Historical Geographic Information System: Version 17.0 [Data set]. https://doi.org/10.18128/D050.V17.0

22.

Massey

Denton

N. A.

(1993). American apartheid: Segregation and the making of the underclass. Harvard University Press.

23.

Orchard

Price

(2017). County-level racial prejudice and the black-white gap in infant health outcomes. Social Science & Medicine, 181, 191–198. https://doi.org/10.1016/j.socscimed.2017.03.036

24.

Paluck

E. L.

Green

S. A.

Green

D. P.

(2019). The contact hypothesis re-evaluated. Behavioural Public Policy, 3(02), 129–158. https://doi.org/10.1017/bpp.2018.25

25.

Payne

B. K.

Hannay

J. W.

(2021). Implicit bias reflects systemic racism. Trends in Cognitive Sciences, 25(11), 927–936. https://doi.org/10.1016/J.TICS.2021.08.001

26.

Payne

B. K.

Vuletich

H. A.

Brown-Iannuzzi

J. L.

(2019). Historical roots of implicit bias in slavery. Proceedings of the National Academy of Sciences of the United States of America, 116(24), 11693–11698. https://doi.org/10.1073/pnas.1818816116

27.

Pettigrew

T. F.

Tropp

L. R.

(2006). A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology, 90(5), 751–783. https://doi.org/10.1037/0022-3514.90.5.751

28.

Petty

R. E.

Fazio

R. H.

Brinol

(2008). The new implicit measures: An overview. In Petty

R. E.

Fazio

R. H.

Brinol

(Eds.), Attitudes: Insights from the new implicit measures (pp. 3–19). Psychology Press.

29.

Pew Research Center. (2016). On views of race and inequality, blacks and whites are worlds apart. https://www.pewresearch.org/social-trends/2016/06/27/on-views-of-race-and-inequality-blacks-and-whites-are-worlds-apart/

30.

Rae

J. R.

Newheiser

A.-K.

Olson

K. R.

(2015). Exposure to racial out-groups and implicit race bias in the United States. Social Psychological and Personality Science, 6(5), 535–543. https://doi.org/10.1177/1948550614567357

31.

Riddle

Sinclair

(2019). Racial disparities in school-based disciplinary actions are associated with county-level rates of racial bias. Proceedings of the National Academy of Sciences of the United States of America, 116(17), 8255–8260. https://doi.org/10.1073/pnas.1808307116

32.

Rothstein

(2017). The color of law: A forgotten history of how our government segregated America. Liveright Publishing Corporation.

33.

Schuman

Steeh

Bobo

Krysan

(1997). Racial attitudes in America: Trends and interpretations. Harvard University Press.

34.

Silberzahn

Uhlmann

E. L.

Martin

D. P.

Anselmi

Aust

Awtrey

Bahník

Š.

Bai

Bannard

Bonnier

Carlsson

Cheung

Christensen

Clay

Craig

M. A.

Dalla Rosa

Dam

Evans

M. H.

Flores Cervantes

. . .Nosek

B. A.

(2018). Many analysts, one data set: Making transparent how variations in analytic choices affect results. Advances in Methods and Practices in Psychological Science, 1(3), 337–356. https://doi.org/10.1177/2515245917747646

35.

Simonsohn

Simmons

J. P.

Nelson

L. D.

(2020). Specification curve analysis. Nature Human Behaviour, 4(11), 1208–1214. https://doi.org/10.1038/s41562-020-0912-z

36.

Tolnay

S. E.

(2003). The African American “Great Migration” and beyond. Annual Review of Sociology, 29, 209–232. https://doi.org/10.1146/ANNUREV.SOC.29.010202.100009

37.

U.S. Census Bureau. (2012). Poverty status in the past 12 months; American Community Survey 5-year estimates [Data file]. https://data.census.gov/table?t=Income+and+Poverty&g=010XX00US$0500000&tid=ACSST1Y2021.S1701

38.

Vuletich

H. A.

Payne

B. K.

(2019). Stability and change in implicit bias. Psychological Science, 30(6), 854–862. https://doi.org/10.1177/0956797619844270

39.

Welch

Sigelman

Bledsoe

Combs

M. W.

(2001). Race and place: Race relations in an American city. Cambridge University Press.

40.

Wilkerson

(2010). The warmth of other suns: The epic story of America’s Great Migration. Random House.

41.

Nosek

Greenwald

(2014). Psychology data from the Race Implicit Association Test on the Project Implicit Demo website. Journal of Open Psychology Data, 2(1), Article e3. https://doi.org/10.5334/jopd.ac

42.

Yzerbyt

Muller

Batailler

Judd

C. M.

(2018). New recommendations for testing indirect effects in mediational models: The need to report and test component paths. Journal of Personality and Social Psychology, 115(6), 929–943. https://doi.org/10.1037/PSPA0000132