Racially motivated hate crime exposure and birth outcomes: Analysis of US vital statistics data 1995-2020 Authors: Yingyue Jiang, Emily Rauscher Published: 2026 ABSTRACT AbstractExposure to racially motivated hate crime represents a salient form of structural and interpersonal racism that is associated with detrimental mental and physical health. Although prior research shows that stress during pregnancy can influence birth outcomes, empirical evidence linking hate crime exposure to adverse birth outcomes remains limited. Using National Vital Statistics Birth Data from 1995 to 2020 linked to county-level hate crime data, we use linear probability models and leverage variation in the timing of hate crimes in the county to examine the association between maternal exposure to a hate crime targeting her own race during pregnancy and the likelihood of preterm birth or low birth weight among Black, Asian, and Hispanic mothers, compared with White mothers as the reference group. Hate crime exposure is associated with a higher likelihood of preterm birth, though not with low birth weight; this association is stronger among Black and Asian than White mothers. When stratified by maternal education, the relationship between hate crime exposure and the likelihood of preterm birth is significantly larger among Black and Hispanic mothers with at least some college education than similar White mothers. Hate crime, as a marker of the racialized social environment, may contribute to racial disparities in birth outcomes. Policies reducing hate crime and its health effects may improve maternal and infant health and narrow inequities at birth.Keywords: Racially motivated hate crime, Infant health, Preterm birth, Birth weight, Racially marginalized, Prenatal stress Introduction Racial disparities in health outcomes remain a persistent challenge in the United States. There is a great deal of evidence linking these gaps to the chronic stress of experiencing racism (Brondolo et al., 2009), as perceived micro aggressive discrimination increases negative emotion, and these disparities would be exacerbated with rising racial animosity (Cheadle et al., 2020). Beyond interpersonal experiences of discrimination, exposure to violence and crime also represents an important source of stress and can generate chronic psychological strain (Baranyi et al., 2021). This study focuses on racially motivated hate crimes as an indicator of such contextual stressors and as a potential threat to infant health outcomes. The birth outcomes examined include preterm birth (gestation<37 weeks) and low birth weight (<2500 grams), both well-established indicators of early-life health that reflect fetal nutritional status and growth pattern. Beyond their immediate clinical significance, these outcomes have lasting consequences for neurological development and human capital accumulation (Jańczewska et al., 2023; Saigal and Doyle, 2008). Although hate crimes can be motivated by various characteristics, such as religion or sexual orientation, racially motivated hate crimes constitute the most common form of hate crimes in the US (Federal Bureau of Investigation, 2023). This pattern reflects not only race as a visible marker of group membership, but fundamentally the deeply entrenched history of structural racism in US society. As such, focusing on racially motivated hate crimes provides a unique opportunity to investigate how structural and interpersonal racism, manifested through violent acts, impacts health outcomes at birth and produces intergenerational consequences. Previous studies have supported the intergenerational transmission and inheritance of risk factors via the germline, with epigenetic mechanisms strongly implicated (Perez and Lehner, 2019; Yeshurun and Hannan, 2019). Our study contributes an empirical test of whether racially motivated hate crimes, as a form of targeted traumatic stress, are associated with birth outcomes consistent with this intergenerational biological pathway. Existing studies on the relationship between crime and birth outcomes have largely examined overall crime exposure, with little attention to the characteristics of crime victims or the motivations behind crimes. Prior studies have shown that exposure to crime is associated with worse birth outcomes (Mark and Torrats-Espinosa, 2022, 2024). However, racially motivated hate crimes are distinct from general crime in both their nature and consequences. Unlike general crime, which selects victims based on opportunity and circumstance rather than racial identity, racially motivated hate crimes explicitly target victims because of their membership in a racial group, sending a message of threat not only to direct victims but to all members of that group. Compared to offenses with other motivations, hate crimes are reported to be more likely to result in severe injury and hospitalization, to involve multiple offenders and serial or repeated attacks, and to cause emotional and psychological distress (Messner et al., 2004; Brunton-Smith et al., 2026). This group-based threat mechanism makes racially motivated hate crime a more theoretically precise exposure for studying racial disparities in birth outcomes than general crime exposure. We conceptualize racially motivated hate crime as both a discrete event and a signal of a broader hostile racial environment. Some individuals become aware of specific incidents targeting their group through social networks, media, or word of mouth, while the prevalence of such incidents also marks an ambient climate of racial threat that can be felt even without knowledge of any particular event. Both pathways may operate through perceptions of racial threat and diminished safety that trigger stress responses associated with adverse health outcomes (Dunkel Schetter, 2011). Because our county-level measure captures recorded hate crime, it reflects both channels and cannot distinguish between them. A key contribution of this study is its analysis of the association of racially motivated hate crimes with birth outcomes across multiple racial groups. First, this study moves beyond prior research that focuses on a single racial comparison or examines outcomes within one racial group (Braveman et al., 2017; Jaskiewicz, 2026) by comparing White mothers with mothers of three racially marginalized groups: Black, Hispanic, and Asian. Second, this study brings attention to Asian mothers, a group that has been largely overlooked in research on the health impacts of stressors such as violent crimes (Mark and Torrats-Espinosa, 2022, 2024). Although there is emerging focus on the impact of hate crime on Asians given the surging violence towards them (Harii et al., 2025), research on whether and how hate crime exposure affects Asian mothers’ birth outcomes remains limited. Evidence suggests that Asians have birth outcomes similar to Whites, and generally healthier than other racially marginalized groups (Dongarwar et al., 2022; Zheng et al., 2024). However, this apparent health advantage may obscure the vulnerabilities that Asian mothers face. This study investigates the relationship between exposure to county-level hate crime and birth outcomes, examining variation by the race of hate crime victims and by maternal race. Literature review Association between hate crime and infant health outcomes Hate crimes can be conceptualized as a form of social trauma that collectivizes separation, rejection, isolation, and exclusion of a social group, becoming embedded in collective identity (Abrutyn, 2024). Beyond acute events, hate crimes function as ambient strains, permeating daily life and exerting disproportionate physical and psychological pressure on targeted racial groups, whose members may fear becoming the next victims. Birth outcomes are particularly sensitive to external stressors. Studies have documented adverse birth outcomes among Latina women following the 2016 US election (Gemmill et al., 2019), Arabic-named women after September 11 (Lauderdale, 2006), and women from countries targeted by the 2017 travel ban (Samari et al., 2020). These findings collectively suggest that racialized sociopolitical stressors have measurable intergenerational health consequences. According to Federal Bureau of Investigation statistics, incidents motivated by race, ethnicity, or ancestry remain the most common type of hate crime, which can be viewed as triggers for racism-related stress. Exposure to such stressors is associated with negative emotions such as fear and anxiety (Cheadle et al., 2020). Despite this evidence, we know relatively little about how hate crimes affect mothers across different racial groups from an intergenerational perspective. Studying racially motivated hate crime rather than general crime is a more theoretically precise exposure because it foregrounds the racial threat mechanism that differentially burdens racially marginalized mothers, enabling a more direct empirical test of the racism-health link. Based on existing literature, we hypothesize that exposure to racially motivated hate crimes is associated with adverse infant health outcomes, particularly among racially marginalized mothers (Hypothesis 1). Racial stereotypes and health outcomes Racially marginalized groups are sometimes hypothesized to have different coping strategies to deal with stress or racism, which may protect them from worse health outcomes. Asian Americans are often perceived as “model minorities:” the stereotype that the hardiness of Asian people contributes to their socioeconomic success and fewer struggles than other racial groups (Kim et al., 2021). The “model minority myth” for Asian people can extend to health, which finds partial support in birth outcomes. Asian American women exhibited lower odds of preterm birth than non-Hispanic White women (Dongarwar et al., 2022). While Asian infants had a higher prevalence of low birth weight than White infants, the rate was lower than Black infants (Zheng et al., 2024). However, the “model minority myth” deprioritizes Asian communities from interventions aimed at health equity and obscures the necessity to pay attention to health risks. Most studies on racial disparities in birth outcomes focus on Black and Hispanic mothers (Mark and Torrats-Espinosa, 2024; Braveman et al., 2017), leaving open the questions of whether Asian mothers are resilient to the adverse impact of hate crimes. Hispanic women present a similarly complex picture. Despite significant socioeconomic barriers, Hispanic women have exhibited favorable birth outcomes, a phenomenon known as the “Hispanic paradox” (Montoya-Williams et al., 2021). Early documentation of this pattern showed that Mexican-born women in California had lower rates of low birth weight than non-Hispanic White women (Williams et al., 1986), with proposed explanations including cultural buffering through strong marital and community ties (McGlade et al., 2004). However, more recent work using New York City data found that while Hispanic women have better birthweight and small-for-gestational-age outcomes than White women, they face worse preterm birth and infant mortality outcomes, suggesting the “Hispanic paradox” is not uniformly protective across all birth outcomes (Borrell et al., 2022). For Black mothers, the relevant coping framework is John Henryism, a high-effort, sustained behavioral response to external stressors (James et al., 1983). Early research linked John Henryism to better mental and physical health outcomes (Bronder et al., 2014; Bonham et al., 2004), suggesting it may buffer the health effects of racial discrimination. More recent evidence complicates this picture: high-effort coping appears harmful when directed at uncontrollable stressors such as structural racism (Nguyen et al., 2022), suggesting that John Henryism's protective capacity is context dependent and may not buffer the effects of ambient racial threat. Our study provides an opportunity to empirically assess whether this coping strategy buffers or amplifies the effects of hate crime exposure on birth outcomes among Black mothers. The “model minority,” “John Henryism,” and “Hispanic paradox” assumptions suggest that racially marginalized mothers possess coping strategies to reduce the potential impact of hate crimes. In other words, these stereotypes suggest that Black, Asian, and Hispanic mothers may be less negatively impacted by hate crime exposure than White mothers, who have weaker coping mechanisms. These assumptions can downplay the harms of racially motivated hate crimes and reduce the perceived urgency of addressing them. Building on this perspective, an alternative hypothesis posits that exposure to racially motivated hate crimes does not lead to significant differences in birth outcomes between White mothers and Black, Asian, or Hispanic mothers (Hypothesis 2). Variation by timing of exposure and maternal education Exposure to environmental shocks has negative implications for infant health through multiple pathways and the effects can vary depending on the timing of exposure during pregnancy. For example, exposure to an earthquake during gestation reduced infant health through maternal stress, but only among those exposed during the first trimester of pregnancy (Torche, 2011). Exposure to iron mining residues due to a collapsed dam during the third trimester reduced infant birth weight and increased infant mortality (Carrillo et al., 2020). A study also shows that consequences of exposure to rainfall shocks are concentrated in the second trimester of pregnancy (Le and Nguyen, 2022). Although findings are not consistent across studies, they all indicate the importance of timing and pattern of stress during pregnancy. There are potential explanations for why the effect of stress exposure may vary by gestational timing. One theory is that the first trimester is most sensitive, with physiological stress responsiveness declining as pregnancy progresses (Glynn et al., 2004). A second perspective emphasizes stress duration, suggesting earlier exposure may trigger prolonged psychological responses through rumination and anticipatory anxiety. In contrast, a third perspective points to late pregnancy as a critical endpoint where protective mechanisms may break down, particularly for mothers that face consistent stress (Glynn et al., 2008). While these theories agree that gestational timing shapes the health implications of stress exposure, previous studies have largely overlooked how this sensitivity varies across racial groups. We focus on how mothers of different racial groups respond to hate crimes targeting their own racial group, with particular attention to gestational timing. The health effects of hate crime exposure may vary by trimester depending on whether exposure functions as an acute or chronic stressor, a distinction that differs systematically across racial groups. Fig. A1 shows the proportion of births exposed during gestation to a racial hate crime in the mother's county of residence by maternal race/ethnicity and by birth year from 1995 to 2020. Prenatal exposure to anti-Asian hate crimes declined for much of the period since 1995 but increased sharply after 2018 in the context of the 2019 Coronavirus (COVID-19) pandemic. This episodic pattern suggests that for Asian mothers, hate crime exposure is more likely to function as an acute and unexpected stressor. Such exposure may induce rumination and heightened vigilance during sensitive developmental windows (Glynn et al., 2004), amplifying its impact on birth outcomes. Black mothers face a persistently high likelihood of exposure to anti-Black hate crimes, with prevalence remaining relatively stable at around 50 percent over time. For Black mothers, hate crime exposure is therefore less likely to represent an isolated shock and more likely to compound existing chronic racial stress. This pattern suggests that mid-to late-pregnancy, when cumulative physiological burden is greatest (Glynn et al., 2008), may be the most sensitive period for Black mothers. Prenatal exposure rates among Hispanic mothers fall in between those of Asian and Black mothers, suggesting effects of hate crime exposure could result in acute or chronic stress patterns or a pattern in between those expected among Asian and Black mothers. Taken together, these patterns suggest that the effects of hate crime exposure on birth outcomes differ by gestational timing across racial groups (Hypothesis 3). Although higher educational attainment is generally associated with better health outcomes, this protective effect is weaker among racially marginalized groups. For racially marginalized mothers, higher education may amplify rather than buffer vulnerability to racialized stressors such as hate crimes. Rates of chronic worry about racial discrimination are highest among Black women with higher income and education levels (Braveman et al., 2017), and higher educational attainment among Black women has been associated with greater discrimination and hypertension risk relative to less-educated Black counterparts (Gaston et al., 2023). This pattern reflects a paradox whereby socioeconomic advancement exposes racially marginalized individuals to new forms of discrimination, serving as a stark reminder that racial group membership supersedes individual achievement. Based on this literature, we hypothesize that among racially marginalized mothers, those with higher educational attainment are more adversely affected by hate crime exposure than those with lower educational attainment (Hypothesis 4). Methods Exposure to hate crime varies by location and the local economic, demographic, or environmental context could explain any variation in maternal or infant health. Mothers with means may also select where they live based on crime and other contextual characteristics. To address these challenges, we use within-county variation in exposure to a hate crime targeting one's own race, using the timing of infant conception to estimate the association of such exposure with infant health among racially marginalized mothers, and compare these estimates with those for White mothers. This approach addresses stable differences in infant health between counties and addresses variation in infant health over time unrelated to hate crime exposure. Identification is based on the trimester of gestation in which a hate crime occurs in the mother's county of residence. Mothers may time their birth seasonally, with variation by socioeconomic status (Buckles and Hungerman, 2013). However, it is unlikely that mothers can anticipate the timing of a local racially motivated hate crime or time their pregnancy based on potential future hate crimes. We estimate the relationship between exposure to a racially motivated hate crime in the county and infant health by trimester of exposure and by maternal race, using models with fixed effects for county and for year and month of birth. Data We use data from two main sources: National Vital Statistics System (NVSS) birth data and Federal Bureau of Investigation Uniform Crime Reports (UCR). We obtained restricted natality data from NVSS, which includes comprehensive birth certificate information from all live births in the US, including maternal county of residence. The natality data also documents the demographic characteristics of mothers. To allow meaningful statistical inference and a workable number of observations (Gibbs et al., 2017; Berk et al., 1995), we use a 10 percent random sample of all singleton births in each year from 1995 to 2020. We exclude multiple births (e.g., twins) because they often have poorer infant health measures, and the likelihood of multiple births is not randomly distributed (Saavedra, 2020; Matthews et al., 2015; Kulkarni et al., 2013; Luke and Martin, 2004; Russell et al., 2003). The hate crime data is publicly available from the Federal Bureau of Investigation website and documents the date, place, and type of each hate crime. Each hate crime has an Originating Agency Identifier code, which can be linked to county Federal Information Processing System codes for maternal counties of residence in the natality data. We use county as the geographic level for analysis for both practical and theoretical reasons. County is the finest geographic level at which UCR hate crime data can be reliably linked to birth records, and prior studies of birth outcomes adopt the same unit (Jaskiewicz, 2026; Rauscher and Cao, 2024). Theoretically, county captures local racial climate, law enforcement, and community context relevant to stress accumulation. To avoid conflating exposure across racial groups, we restrict our sample to hate crimes that target one single racial group. We identify hate crime types using the bias motivation recorded in UCR reports. According to the methodology of hate crime data, law enforcement agencies report hate crimes brought to their attention to the FBI where agencies can indicate bias motivations of each case based on investigation. Following the American College of Obstetricians and Gynecologists and the Society for Maternal-Fetal Medicine et al. (2016), we exclude births with non-viable gestational ages (<24 weeks gestation) and extremely low birth weight (<1000 grams), as birth below these thresholds are etiologically distinct from the outcomes under study and survival is exceptionally rare. In our sample of singleton births 1995-2020, there are a total of 9,020,440 live births in the sample, including births to 1,301,392 Black mothers, 483,255 Asian mothers, 2, 119,468 Hispanic mothers, and 5,116,325 White mothers. Measures The dependent variables are preterm birth (<37 weeks) and low birth weight (<2500 grams), which are binary variables. These measures indicate a high risk of health problems from infancy through later adulthood and have been related to educational and economic outcomes (Thomson et al., 2021). The independent variables are racially motivated hate crime exposure indicators, which are dummy variables indicating whether a hate crime targeting the same racial group as the mother occurred in her county of residence during a specific trimester of pregnancy. Preterm birth and trimester of gestation are determined based on estimated date of conception, calculated as birth date minus gestational length. First trimester exposure is coded 1 if a hate crime targeting the mother's race occurred in the mother's county of residence during the first 3 months of gestation and 0 otherwise. Second trimester exposure indicates whether a racially motivated hate crime occurred during months 4-6 of gestation, and third trimester exposure indicates whether a hate crime occurred after 6 months of gestation and before birth. Conception date is an estimate and is measured with error, but the alternative approach – measuring hate crime exposure by counting backward from birth date – is biased by variation in gestational length (Currie and Rossin-Slater, 2013). Gestational length and likelihood of premature birth are related to maternal race and other characteristics. The alternative approach based on birth date rather than estimated conception date would inaccurately code hate crime exposure for some premature births. Control variables include mother's education level (less than high school, high school, some college, or at least a bachelor's degree), mother's age, marital status (married or unmarried), infant's sex (male or female), and father's race (White, Black, Native American, Asian, Hispanic, multiracial, non-Hispanic unknown, or unknown). Observations missing father's race are recoded as unknown rather than dropped, retaining the full sample and avoiding selection bias. All observations included in the analytic sample have non-missing values on all covariates. Statistical analyses We estimate linear probability models separately by maternal race. Model 1 is the crude (unadjusted) model. The equation is written as: where Yijt is the health outcome for an individual birth i in county j at birth month-year t, α0 is the intercept, k is pregnancy trimester, ranging from 1 to 3. Tt=1 if a hate crime occurred in trimester k for birth month-yeart in county j. This coefficient estimates the association between the timing of exposure to a racially motivated hate crime and the probability of preterm birth or low birth weight. ConceptDateijt is a continuous measure of conception date for an individual birth i. δj is the county fixed effect capturing the time-invariant differences in the outcome between counties. μmonth(t) is the birth month fixed effect that absorbs monthly seasonal patterns, λyear(t) is the birth year fixed effect which captures differences in outcomes across years. εijt is the error term. In Model 2, we include maternal sociodemographic factors, which can be written as: where Xijt is a series of control variables, including mother's education, age, marital status, and infant's sex. We further control for the father's race in Model 3, which we refer as the full model throughout the paper. This sequential modeling approach enhances transparency by showing how the estimated associations between hate crime exposure and birth outcomes evolve as conceptually distinct sets of covariates are added. When predicting indicators for preterm birth and low birth weight, positive coefficients for hate crime exposure would be consistent with Hypothesis 1, suggesting more negative health outcomes associated with hate crime exposure. We test for significant differences in hate crime exposure coefficients by mother's race using z tests. For example, to test for different effects of first trimester hate crime exposure on birth outcomes between Asian and White mothers, we calculate z statistics (z=βAsian−βWhite∕SEAsian2+SEWhite2), where βAsian indicates the estimated effect of first trimester exposure to anti-Asian hate crime among births to Asian mothers and βWhite indicates the estimated effect of first trimester exposure to anti-Asian hate crime among births to White mothers. If these z statistics are not statistically significant, they would support Hypothesis 2 that there is not a significant difference in the response to anti-Asian hate crime in the first trimester between Asian and White mothers. If the hate crime exposure coefficient differs by trimester timing across racial groups, Hypothesis 3 is supported. Because anti-White hate crimes are relatively rare and not theoretically central to our mechanism of racially marginalized groups–targeted hostility, our main analysis focuses on hate crimes targeting racially marginalized groups. We compare birth outcomes across racial groups exposed to the same local incidents to assess differential effects. We compare effects of hate crime exposure targeting mother's own race in sensitivity analyses. We then stratify the sample by maternal education and examine birth outcomes separately for two groups: mothers with at least some college education and those without (high school or less). If a stronger association between hate crime exposure and adverse birth outcomes is observed among mothers with at least some college education versus mothers without, then Hypothesis 4 is supported. In all models, we use robust standard errors adjusted for clustering at the county level to account for correlated observations within each county. Sensitivity analyses We perform seven sensitivity analyses to assess robustness. First, to avoid the potential impact of COVID-19 pandemic, we reanalyze the data restricting the sample to births from 1995 to 2018. Second, we operationalize hate crime severity and magnitude using the number of victims, allowing us to check whether the results are consistent using continuous measures of hate crime exposure versus indicators for any exposure. Third, we repeat the main analyses by county-level poverty rate, categorizing counties into three groups: low, medium, and high poverty. Fourth, we estimate models using exposure to anti-White hate crimes and compare effects of anti-White hate crime and hate crimes towards racially marginalized mothers. Fifth, we run a falsification test using two post-birth windows (2-4 month and 5-7 month). These windows are selected because they are sufficiently close to birth to serve as a meaningful placebo test, while excluding the first month to avoid potential misclassification of gestational age. Sixth, we assess sensitivity to trimester boundary misclassification by shifting trimester assignments by ±1 month. Seventh, we estimate models using aggregate racially motivated hate crime exposure as an alternative independent variable. All analyses are conducted using STATA/SE 18.0. Results Table 1 provides descriptive statistics for the 10 percent random sample of singleton births from 1995 to 2020 stratified by mother's race. According to the US Centers for Disease Control and Prevention (CDC), rates of low birth weight and preterm birth were 7 percent and 10 percent for singleton births in 2020 (Osterman et al., 2022). In our 10 percent random sample spanning 1995 to 2020, low birth weight and preterm birth rates were comparable, at 6 and 10 percent, respectively. Consistent with the CDC, Black infants experience the highest rates of low birth weight and preterm birth. There is a clear difference in exposure to racially motivated hate crime by maternal race: about 41 percent of Black mothers were exposed to anti-Black hate crime within their county during each trimester of pregnancy, compared with about 24 percent of Asian mothers exposed to anti-Asian hate crime, and 33 percent of Hispanic mothers exposed to anti-Hispanic hate crime. White mothers have the lowest rate of exposure, with 18 percent of births to White mothers exposed to anti-White hate crime. Asian and White mothers are more likely to be married than mothers in other racial groups, and most mothers are partnered with someone of the same race. Asian mothers have the highest educational attainment, with more than half holding at least a bachelor's degree, followed by White mothers. Table 1Descriptive statistics.Black (N = 1,301,392)Asian (N = 483,255)Hispanic (N = 2,119,468)White (N = 5,116,325)Trimester 1 (%)40.5123.5832.3617.96Trimester 2 (%)40.8023.6932.6517.99Trimester 3 (%)40.9923.8432.8218.05Preterm birth (%)14.378.6310.078.35Low birth weight (%)10.146.115.214.68Education (%)Lower than high school21.379.7340.1810.52High school36.8018.5830.0026.72Some college28.2019.5419.1727.40Bachelor and above13.6352.1510.6535.37Age25.96 (6.24)30.41 (5.36)26.64 (6.16)28.27 (5.86)Male child (%)50.8851.5251.0251.36Married (%)29.8385.6251.9174.07Father's race (%)White2.6915.5710.1782.13Black59.232.213.152.43Native American0.120.180.240.45Asian0.2273.330.720.80Hispanic1.803.4469.844.47Unknown race, non-Hispanic0.970.500.390.53Multiracial0.280.480.620.35Unknown34.684.2914.878.84Average birth month6.606.606.636.54Average birth year2007200820082007Notes: Standard deviation in parentheses. Descriptive statistics. Notes: Standard deviation in parentheses. Table 2 presents estimates of the relationship between hate crime exposure and infant health among Black, Asian, and Hispanic mothers using the full model that includes all control variables. Fig. 1 displays preterm birth estimates with 95 percent confidence intervals. Linear probability models show consistent associations between hate crime exposure and preterm birth, even after adjusting for maternal and paternal characteristics (results of Models 1 and 2 are shown in Table A1-A3). However, no statistically significant association is observed between hate crime exposure and low birth weight for any group. Table 2Relationship between racially motivated hate crime exposure and birth outcomes, 1995-2020.Preterm birthLow birth weightAnti-BlackWhiteBlackWhiteBlackTri 1−0.00006 (0.00031)0.00064 (0.00067)0.00000 (0.00027)−0.00031 (0.00073)Tri 2 −0.00021 (0.00029) 0.00148* (0.00067) 0.00028 (0.00026)−0.00028 (0.00079)Tri 30.00036 (0.00030)−0.00067 (0.00074)0.00028 (0.00026)−0.00058 (0.00076)Anti-AsianWhiteAsianWhiteAsianTri 1 0.00029 (0.00045) 0.00228** (0.00086) −0.00064 (0.00041)0.00053 (0.00094)Tri 20.00055 (0.00043)0.00138 (0.00096)−0.00035 (0.00040)−0.00038 (0.00106)Tri 30.00043 (0.00044)0.00197* (0.00098)0.00027 (0.00038)0.00131 (0.00108)Anti-HispanicWhiteHispanicWhiteHispanicTri 1−0.00016 (0.00040)0.00078 (0.00064)−0.00014 (0.00032)−0.00065 (0.00049)Tri 20.00031 (0.00042)0.00063 (0.00065)0.00015 (0.00034)0.00060 (0.00044)Tri 30.00082 (0.00043)0.00134* (0.00062)0.00022 (0.00036)−0.00008 (0.00044)Notes: Robust standard errors adjusted for county clustering in parentheses.Estimates are from the full model which adjusts for mother's and father's characteristics including mother's age, mother's education, infant's sex, marital status, and father's race. Conception date and fixed effects for year and month of birth and county are controlled in all models.Bolded cells indicate statistically significant differences in coefficients for exposure to racially motivated hate crime between White mothers and racially marginalized mothers using z-tests with a 0.05 significance level.*p < .05**p < .01***p < 0.001 (two-tailed tests). Relationship between racially motivated hate crime exposure and birth outcomes, 1995-2020. Notes: Robust standard errors adjusted for county clustering in parentheses. Estimates are from the full model which adjusts for mother's and father's characteristics including mother's age, mother's education, infant's sex, marital status, and father's race. Conception date and fixed effects for year and month of birth and county are controlled in all models. Bolded cells indicate statistically significant differences in coefficients for exposure to racially motivated hate crime between White mothers and racially marginalized mothers using z-tests with a 0.05 significance level. p < .05 p < .01 p < 0.001 (two-tailed tests). Coefficient plots of racially motivated hate crime exposure on preterm birth by racial groups. Notes: results based on the full model from Table 2; tri1, tri2, and tri3 indicate trimester 1, trimester 2, and trimester 3. Error bars indicate 95 percent confidence intervals. *p < .05; **p < .01; ***p < 0.001 (two-tailed tests). Exposure to anti-Black hate crime during the second trimester is associated with a 0.15 percentage point increase in the probability of preterm birth among Black mothers, and this association is significantly larger than that observed among White mothers. Among Asian mothers, exposure to anti-Asian hate crime during the first and third trimesters is associated with a higher probability of preterm birth. Specifically, exposure during the first trimester is associated with a 0.23 percentage point increase in the probability of preterm birth, while exposure during the third trimester is associated with a 0.20 percentage point increase in probability. The first trimester estimate is significantly larger than that observed among White mothers, whereas the third trimester difference is not significant. Among Hispanic mothers, exposure to anti-Hispanic hate crime during the third trimester is associated with a 0.13 percentage point increase in the probability of preterm birth. However, the difference between coefficients for White and Hispanic mothers is not statistically significant. Variation by maternal education Results from models stratified by mothers’ education are shown in Table 3 and the corresponding coefficient plot for preterm birth is shown in Fig. 2. The Black-White difference in the response to hate crime exposure during the second trimester is significant among mothers with at least some college education. This difference is not significant among mothers with no college education. This pattern suggests a more negative association between hate crime exposure and birth outcomes among Black women with higher education. The Hispanic-White difference in the association between hate crime exposure and preterm birth is statistically significant during the third trimester, with a 0.30 percentage point higher probability of preterm birth among Hispanic mothers with at least some college compared to a 0.05 percentage point increase among White mothers with at least some college. Similar to estimates comparing Black and White mothers, there are no significant Hispanic-White differences among mothers with no college education. This suggests that Hispanic mothers with higher education may also show a stronger association between hate crime exposure and preterm birth. For Asian mothers, there is no statistically significant difference in the association between hate crime exposure and preterm birth compared with White mothers when the sample is stratified by education. Table 3Relationship between hate crime exposure and birth outcomes by maternal education, 1995-2020.Lower than some collegeAt least some collegePreterm birthWhiteBlackWhiteBlackTri1−0.00058(0.00051)0.00035(0.00087)0.00023(0.00035)0.00093(0.00084)Tri20.00011(0.00053)0.00132(0.00089) −0.00036 (0.00034) 0.00171 (0.00089) Tri30.00027(0.00048)−0.00119(0.00099)0.00045(0.00037)0.00002(0.00094)WhiteAsianWhiteAsianTri10.00086(0.00081)0.00443(0.00295)0.00022(0.00051)0.00156(0.00106)Tri20.00027(0.00071)0.00338(0.00224)0.00088(0.00050)0.00071(0.00105)Tri30.00150(0.00082)0.00288(0.00191)0.00022(0.00047)0.00176(0.00102)WhiteHispanicWhiteHispanicTri1−0.00007(0.00070)0.00072(0.00072)−0.00018(0.00043)0.00075(0.00087)Tri20.00131(0.00071)0.00057(0.00076)−0.00012(0.00047)0.00068(0.00092)Tri30.00168**(0.00065)0.00064(0.00066) 0.00049 (0.00050) 0.00296 *** (0.00089) Low birth weightWhiteBlackWhiteBlackTri10.00025(0.00054)−0.00045(0.00097)−0.00008(0.00029)−0.00008(0.00102)Tri2−0.00010(0.00054)−0.00052(0.00102)0.00049(0.00029)0.00012(0.00103)Tri3−0.00002(0.00053)−0.00082(0.00112)0.00047(0.00029)−0.00009(0.00090)WhiteAsianWhiteAsianTri1−0.00077(0.00087)0.00072(0.00234)−0.00045(0.00043)0.00046(0.00116)Tri2−0.00047(0.00082)−0.00227(0.00199)−0.00009(0.00042)0.00020(0.00124)Tri30.00111(0.00072)0.00196(0.00208)0.00014(0.00041)0.00108(0.00125)WhiteHispanicWhiteHispanicTri1−0.00025(0.00060)−0.00052(0.00054)−0.00010(0.00037)−0.00099(0.00091)Tri20.00079(0.00071)0.00032(0.00049)−0.00012(0.00040)0.00123(0.00073)Tri3−0.00044(0.00063)−0.00029(0.00048)0.00059(0.00044)0.00049(0.00083)Notes: Robust standard errors adjusted for county clustering in parentheses. Estimates are from the model which adjusts for mother's and father's characteristics including mother's age, infant's sex, marital status, and father's race. Conception date, year and month of childbirth, and county are controlled in all models.Bolded cells indicate statistically significant differences in coefficients for exposure to racially motivated hate crime between White mothers and racially marginalized mothers using z-tests with a 0.05 significance level.*p < .05**p < .01***p < 0.001 (two-tailed tests). Relationship between hate crime exposure and birth outcomes by maternal education, 1995-2020. Notes: Robust standard errors adjusted for county clustering in parentheses. Estimates are from the model which adjusts for mother's and father's characteristics including mother's age, infant's sex, marital status, and father's race. Conception date, year and month of childbirth, and county are controlled in all models. Bolded cells indicate statistically significant differences in coefficients for exposure to racially motivated hate crime between White mothers and racially marginalized mothers using z-tests with a 0.05 significance level. p < .05 p < .01 p < 0.001 (two-tailed tests). Coefficient plots of racially motivated hate crime exposure on preterm birth by racial groups among mothers who have at least some college education. *p < .05; **p < .01; ***p < 0.001 (two-tailed tests). Notes: results based on the model from Table 3 which adjusts for mother's and father's characteristics including mother's age, infant's sex, marital status, and father's race; tri1, tri2, tri3 indicate trimester 1, trimester 2, and trimester 3. Error bars indicate 95 percent confidence intervals Sensitivity analyses Results are broadly robust across sensitivity analyses. Analyses limited to births before 2019 largely confirm the main findings, though the Asian-White difference becomes not significant once post-2019 births are excluded, potentially reflecting lower rates and reduced salience of anti-Asian hate crimes prior to 2019 (Table A4-A6). Models using the number of hate crime victims as a continuous exposure measure show consistent patterns, with Black and Asian mothers exhibiting stronger responses during third and second trimester in terms of preterm birth (Table A7-A9). Stratification by county poverty level shows that mothers in medium-poverty counties experience the highest exposure to racially motivated hate crimes, with consistent associations between anti-Black and anti-Asian hate crime exposure and preterm birth (Table A10-A12). When comparing estimates of exposure to hate crimes targeting one's own race (i.e. White mothers' exposure to anti-White hate crime), we find no significant Black-White or Hispanic-White differences, though the Asian-White difference during the first trimester is significant (Table A13). These findings suggest that exposure to hate crimes directed at one's own racial group can also have varying impacts across racial groups. In the falsification test using post-birth hate crime exposure, none of the differences between White and racially marginalized mothers reach statistical significant at the 95 percent level (Table A14), though for Asian and Hispanic mothers the post-birth estimates are similar in magnitude to the prenatal estimates. This suggests that estimates for Asian and Hispanic mothers could reflect county-level factors related to both hate crime exposure and infant health and should be interpreted with caution. Sensitivity analyses shifting trimester boundaries by ±1 month show consistent results, with coefficients stable in sign and magnitude (Table A15). Models using aggregate racially motivated hate crime show no significant difference between White and racially marginalized mothers (Table A16), whereas target-specific models do. This pattern suggests that the type of hate crime matters and motivates our disaggregated approach. Conclusion and discussion Based on a random sample of over 9 million administrative birth records from NVSS, we estimate the relationship between maternal exposure to racially motivated hate crime during gestation and infant health. We find that Black, Asian, and Hispanic mothers exposed to a hate crime targeting their own group had a higher likelihood of preterm birth than those not exposed in the same county, though we find no significant association with low birth weight. These preterm birth associations vary across racial groups and trimesters. Estimates range from 0.13 to 0.30 percentage point increases in the likelihood of preterm birth from exposure to a hate crime during pregnancy (p < 0.05). Relative to White mothers, the difference in the association between hate crime and preterm birth is statistically significant for Black and Asian mothers but not among Hispanic mothers. When stratifying by maternal education, the relationship between hate crime exposure and birth outcomes varies by maternal race. For Black mothers, a larger association relative to White mothers is observed consistently in the pooled sample and for mothers with at least some college. These findings align with higher exposure to anti-Black hate crime and with existing evidence of more negative patterns of racism among highly educated Black people (Stepanikova and Oates, 2017). Among Hispanic mothers, a stronger association between hate crime exposure and preterm birth relative to White mothers emerges only after stratifying by maternal education, concentrated among those with at least some college, suggesting this group may occupy a precarious position. For Asian mothers, a higher association of hate crime exposure with preterm birth observed in the full sample disappears once analyses are stratified by maternal education. We interpret the patterns among Asian and Hispanic mothers cautiously, though the post-birth racial difference is not statistically significant, some post-birth estimates are similar in magnitude to the prenatal estimates, a pattern we cannot fully distinguish from residual confounding. In summary, the association is strongest among Black and Hispanic mothers with higher education, while for Asian mothers it does not persist once stratified by education. The estimates point toward more negative associations with hate crime exposure among racially marginalized mothers than among White mothers, which supports Hypothesis 1 and Hypothesis 4, most clearly for Black mothers. Comparing estimates for hate crime exposure to those among births to White women, we find significantly larger estimates among births to Black, Asian, and Hispanic women. Results contradict the stereotypes that racially marginalized women may be less affected by hate crime exposure due to better coping strategies. With respect to preterm birth, our evidence does not indicate that Black, Asian, and Hispanic women are unharmed by exposure to racially motivated hate crimes: they have poorer infant health outcomes than White women when exposed to racially motivated hate crimes, though we find no corresponding association for low birth weight. In this case, Hypothesis 2 - that there is no significant difference in the response to hate crime exposure between White and racially marginalized mothers - is not supported for preterm birth. We also find some evidence supporting Hypothesis 3, indicating that the timing of exposure to hate crime matters and varies across racial groups. Moreover, these timing patterns depend on contextual factors such as county poverty levels. Asian mothers appear to be more affected by exposure during the early-to mid-stages of pregnancy, whereas Black mothers are more likely to be affected during the mid-to late-stages, though these patterns vary by context. The association between hate crime exposure and preterm birth rates is relatively weak, but even small changes in infant health can have costly and long-term consequences for individuals and society. Preterm birth has long-term consequences for health and economic outcomes throughout life, including higher risk of death, neurodevelopmental impairment and cognitive dysfunction as well as lower educational and labor market outcomes (Saigal and Doyle, 2008; O'Nions et al., 2021). To illustrate the potential magnitude of these associations, we translate them into approximate cases and hospital costs. We emphasize that these figures are illustrative and rest on the strong assumption that the estimated associations are causal. Each late preterm birth and very preterm birth represents approximately $19,669 and $221,498 more than a full-term birth in hospital costs respectively (Phibbs et al., 2019). If the observed association reflected a causal effect, our sample sizes, hate crime exposure rate in the second trimester, and estimated increase in preterm birth rate would correspond to about 290 additional preterm births to Black mothers in 2018 (478,000 births in 2018 x 0.41 exposure rate x 0.00148) and about 92 additional preterm births to Asian mothers based on first trimester hate crime exposure (168,470 births in 2018 x 0.24 exposure rate x 0.00228). This would represent about $7.5 million more in hospital costs for one year alone. These magnitudes are consistent with those reported in studies of other county-level stressors on birth outcomes using comparable data and designs, such as Rauscher and Cao (2024), who found similarly modest effects of wildfire exposure on infant health. This study has several limitations. First, our fixed effects cannot absorb unmeasured time-varying county conditions that could co-occur with hate crimes. Second, our county-level hate crime measure is ecological and may not accurately reflect individual mothers’ actual exposure, preventing us from distinguishing those directly versus indirectly affected. Counties vary substantially in size and demographic composition, which affects how hate crimes are experienced and how knowledge of a hate crime is transmitted within a community. Because reporting and awareness likely vary systematically across counties and racial groups, this misclassification is could vary and may bias our estimates in an indeterminate direction. Third, we cannot directly examine mechanisms linking hate crime exposure to birth outcomes; although maternal stress is a likely pathway (Dunkel Schetter, 2011), it remains unmeasured in our data. Fourth, hate crime data drawn from UCR reports may understate the true incidence of hate crimes (Mills et al., 2024). Evidence collection poses inherent challenges, victims may decline to report, and many agencies fail to submit annual reports to the UCR (Piatkowska et al., 2019). With an average of 3032 counties from 1995 to 2020, about 8 percent of counties in our sample reported zero hate crimes from 1995 to 2020, and it remains uncertain whether these figures truly reflect an absence of hate crimes or deficiencies in reporting practices. Our within-county fixed-effects approach addresses stable between-county differences in reporting practices but cannot account for within-county temporal changes. Finally, while our use of county-level hate crime exposure is both empirically grounded and theoretically justified, future research employing finer-grained geographic units, alternative hate crime measures, and richer county-level characteristics would help address these limitations and provide more precise estimates of the impact of hate crime exposure on birth outcomes. Despite limitations, our findings have meaningful implications for understanding the intergenerational health consequences of racism. While the observational design precludes causal claims, the direction of the findings, most consistently for Black mothers, is compatible with a link between hate crime exposure and preterm birth. The Asian and Hispanic results are more suggestive and call for further research on these groups. Addressing the health consequences of hate crime exposure may benefit from coordinated action across institutions. Public health agencies could implement trauma-informed prenatal screening in high-exposure communities, integrating hate crime exposure into maternal risk assessments, and track birth outcomes in affected areas. Law enforcement agencies could prioritize hate crime reporting compliance, timely response, and community liaison programs that reduce fear and increase safety among vulnerable populations. Community organizations may complement these efforts through social support and advocacy for policies that reduce structural and interpersonal discrimination. Together, these directions suggest that treating racialized violence as a structural determinant of maternal and infant health, rather than solely a public safety issue, represents a worthwhile reorientation for future policy. Supplementary Material Acknowledgements This research benefited from support from the Population Studies and Training Center at Brown University, which receives funding from the NIH (P2C HD041020). Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.socscimed.2026.119817. Footnotes Ethics approval This study analyzed secondary administrative data from National Vital Statistics (https://www.cdc.gov/nchs/nvss/nvss-restricted-data.htm) and public available data from FBI (https://cde.ucr.cjis.gov/LATEST/webapp/#/pages/downloads). Because the dataset was completely de-identified prior to analysis, this study did not require Institutional Review Board (IRB) review. Declaration of competing interest The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Data availability The data that has been used is confidential. References Associated Data Supplementary Materials Data Availability Statement The data that has been used is confidential.