In this paper I aim to disentangle the mechanisms underlying the well-documented association between race and the location of waste management facilities (WMFs). To do so, I first study whether WMFs were more likely to be located in neighborhoods with a higher share of black population after controlling for relevant covariates. Secondly, I analyze the socioeconomic consequences following the introduction of WMFs. Using historical data from the 1920-1940 U.S. Censuses and GIS data from the Urban Transition Project, I focus on the cities of Baltimore, Boston, Chicago, Cincinnati, Houston, and Nashville in the United States. The results suggest that, contrary to expectations, there was no targeting of predominantly black neighborhoods for WMF placement after accounting for other geographic and socioeconomic factors. Instead, a more flexible model reveals an inverted U-shaped relationship, indicating that neighborhoods with mixed racial compositions were most affected by the presence of WMFs. These findings are consistent with the potential existence of blocking coalitions based on race, implying that homogeneous racial neighborhoods could resist the establishment of WMFs. Furthermore, I explore the consequences of WMF introduction on the affected regions. I find a decrease in urban development, measured by a decline in population density, following the construction of WMFs. Additionally, I find evidence of an increase in the share of black population in the areas where WMFs were introduced. These findings are consistent with a white flight from the affected neighborhoods.

Over half of the parents who use childcare spend more than 20 percent of their household income on childcare expenses. Recent research has shown that childcare subsidies increase women's attachment to the labor force. However, most political focus has been on interventions for preschool-aged children, which may come after many women have already made their decision to rejoin the labor force or "stay-at-home". This paper investigates the long-run impact of childcare subsidies offered soon after first birth on a women's employment, earnings, and total hours worked. Using restricted data which links earnings records from the SSA/IRS to the SIPP 1984-2009 panels and the PSID, I implement a dynamic difference-indifference model that exploits variation in the timing of birth and the 2003 expansion to the Child and Dependent Care Credit. I find that women exposed to the expansion when their children are young are more likely to work and have higher earnings. Early exposed women are 1.2 to 2.5 percentage points more likely to be employed zero to four years after giving birth, suggesting they return to work sooner. This effect grows in the long run. Early exposed women are 3.1 to 5.2 percentage points more likely to be employed five to ten years after giving birth, suggesting they are more likely to remain employed. This translates to large earnings returns for early exposed women. Five to ten years after giving birth early exposed women earn between $5,600-$6,300 more per year.

Objective: The aim of this analysis is to understand why young adult students in the US South marry in college or shortly afterwards. Background: While the median age at first marriage in the United States has risen and marriage delay has become culturally normative, a substantial minority of young adults marry in their late teens or early 20's, particularly in the US South. Young adult college students who marry negotiate conflicting imperatives of marriage and educational attainment, and how these are negotiated or resolved sheds light on the social, cultural, and economic forces that inform family formation. Method: This study draws from in-depth interviews conducted with 45 18–23-year-old engaged or married college students in Mississippi. Results: Entering college, most participants had expected to marry close to median ages, with marriage perceived as a hindrance to educational achievement. This marriage timeline was negotiated downwards with the influence of four interrelated factors: personal orientation to marriage; a marriage-oriented culture, being in the “right” relationship; and social and financial supports for marriage. Conclusion: The norm of marriage delay that these young adults embraced became perceived as inapplicable given evidence that marriage would not compromise college completion or financial stability. There were also social and cultural pressures toward marriage. “Why wait?” captures participants' sense of few barriers but clear incentives to marriage. Incentives were sometimes financial but also social, as marriage enabled students to establish themselves as more mature than peers.

Since President Clinton’s 1994 Executive Order 12898, federal agencies have been required to conduct environmental justice (EJ) analyses of federal rules and regulations. More recently, the Biden Administration has instituted several major efforts to reform regulatory review and promote a more equitable distribution of environmental benefits and burdens. This paper seeks to understand how prior guidelines have been implemented in federal regulatory reviews related to the Clean Water Act and provide a baseline for future studies of the distributional effects of clean water regulations. We reviewed 18 regulatory impact assessments relating to the Clean Water Act conducted since 1992. Only five of these studies conducted a quantitative analysis of distributional impacts and none of the 18 rules were determined to have disproportionately adverse effects on low-income or minority communities. Anticipating that future regulatory review will require more comprehensive distributional analyses, we combine national data on the location of all regulated point sources of water pollution with demographic characteristics to develop a baseline assessment of the distribution of water pollution facilities. Overall, we find that discharge locations tend to be located in areas that are poorer, have a higher White population share, and have less education. We find that rurality partly explains this pattern. The top 40% of census block groups in terms of rural population share contain almost all water pollution discharge locations. We conclude with a discussion of the policy implications of these analyses and suggestions for future work.

Importance Educational attainment in the US is associated with life expectancy. As the opioid crisis worsens, it is critical to understand how overdose death rate trends evolve across education groups. Objective To investigate the association between educational attainment and overdose death rates, with emphasis on trends during the COVID-19 pandemic. Design, Setting, and Participants This cross-sectional study used National Vital Statistics System Mortality Multiple Cause-of-Death data describing overdose death rates in the US by educational attainment from January 1, 2000, to December 31, 2021, with a focus on 2018 to 2021. Overdose deaths were aggregated by year and educational level for decedents aged 25 years or older. Exposure Educational attainment, categorized as no high school (HS) diploma, HS diploma (or General Educational Development) but no college, some college but no bachelor’s degree, and bachelor’s degree or more. Main Outcomes and Measures The main outcomes were rates of all overdose deaths, overdose deaths involving opioids, and overdose deaths involving synthetic opioids. Results Of 912 057 overdose deaths with education information from 2000 to 2021 (mean [SD] age at death, 44.9 [12.3] years; 64.1% male), there were 625 400 deaths (68.6%) among individuals with no college education and 286 657 deaths (31.4%) among those with at least some college. The overdose death rate was 19.9 per 100 000 population. From 2018 to 2021, there were 301 557 overdose deaths, including 58 319 (19.3%) among individuals without an HS diploma, 153 603 (50.9%) among people with an HS diploma, 64 682 (21.4%) among individuals with some college, and 24 953 (8.3%) among individuals with a bachelor’s degree. There were 3324 overdose deaths (1.1%) among American Indian or Alaska Native individuals, 2968 (1.0%) among Asian American or Pacific Islander individuals, 49 152 (16.3%) among Black individuals, 31 703 (10.5%) among Hispanic individuals, 211 359 (70.1%) among White individuals, and 3051 (1.0%) among multiracial individuals. From 2018 to 2021, the overdose death rate was 33.4 per 100 000 population, the opioid-related overdose death rate was 24.2 per 100 000 population, and the synthetic opioid overdose death rate was 19.1 per 100 000 population. From 2018 to 2021, the overdose death rate for those without a HS diploma increased by 35.4 per 100 000 population compared with 1.5 per 100 000 population for those with a bachelor’s degree. This differential growth was primarily due to increased rates of death involving synthetic opioids. Conclusions and Relevance In this cross-sectional study, lower educational attainment was found to be associated with higher growth in overdose deaths. As the opioid crisis has transitioned to fentanyl and polysubstance use, overdose deaths have become more prevalent in groups with lower socioeconomic status, potentially exacerbating existing life-expectancy disparities.

Public discourse about in-kind transfers often appeals to "preferences for rights"-for instance, the "right to health care" or "right to counsel" for indigent legal defense. Preferences for rights are "non-welfarist" if the person values the right per se, holding fixed how the right instrumentally affects others' utilities. We test for non-welfarist preferences for rights, and their relationship to redistributive choices, with incentivized online experiments (N = 1,800). Participants face choices about allocating rights goods (lawyers, health care) and benchmark goods (bus passes, YMCA memberships) to tenants facing eviction. We implement a share of choices. In two of three experiments, more than half of participants allocate rights goods in ways that are consistent with preferences for rights and dominated if preferences were entirely welfarist. Dominated behaviors are more common with rights goods than benchmarks. In a fourth experiment, those with preferences for rights also exhibit "anti-targeting," where they redistribute lawyers and health care more universally than benchmark goods to recipients whose incomes differ. At least 26% of participants are non-welfarist, while at most 31% are welfarist.

Purpose: Climate change is expected to cause larger and more frequent precipitation events in key agricultural regions of the United States, damaging crops and soils. Subsurface tile drainage is an important technology for mitigating the risks of a wetter climate in crop production. In this study, the authors examine how quickly farmers adapt to increased precipitation by investing in drainage technology. Design/methodology/approach: Using farm-level data from the 2018 Agricultural Resource Management Survey (ARMS) of soybean producers, the authors construct a drainage adoption timeline based on when the operator began farming their land and when tile drainage was installed, if at all. The authors examine both the initial investment decision and the speed with which drainage is installed by adopters. A Heckman-style Poisson regression is used to model the count nature of adoption speed (measured in years taken to install tile drainage) and to correct for potential sample-selection bias. Findings: The authors find that local precipitation is not a significant determinant of the drainage investment decision but may be highly influential in the timing of adoption among drainage users. Farms exposed to crop-damaging levels of precipitation install tile drainage faster than those with low to moderate levels of rainfall. Estimates of farm adaptation speeds are heterogeneous across farm and operator characteristics, most notably land tenure status. Originality/value: Understanding how US farmers adapt to extreme weather through technology adoption is key to predicting the long-term impacts of climate change on America's food system. This study extends the existing climate adaptation literature by focusing on the speed of adoption of an important and increasingly common climate-mitigating technology – subsurface tile drainage.

To estimate social class inequities in US mortality using a relational measure based on power over productive property and workers’ labor. Methods. We used nationally representative 1986–2018 National Health Interview Survey data with mortality follow-up through December 31, 2019 (n 5 911 850). First, using business-ownership, occupational, and employment-status data, we classified respondents as incorporated business owners (IBOs), unincorporated business owners (UBOs), managers, workers, or not in the labor force (NLFs). Next, using inverse-probability-weighted survival curves, we estimated class mortality inequities overall, after subdividing workers by employment status and occupation, and by period, gender, race/ethnicity, and education. Results. UBOs, workers, and NLFs had, respectively, 6.3 (95% confidence interval [CI] 5 28.1, 24.6), 6.6 (95% CI 5 28.1, 25.0), and 19.4 (95% CI 5 221.0, 217.7) per 100 lower 34-year survival rates than IBOs. Mortality risk was especially high for unemployed, blue-collar, and service workers. Inequities increased over time and were greater among male, racially minoritized, and less-educated respondents. Conclusions. We estimated considerable mortality inequities by class, gender, and race/ethnicity. We also estimated that class mortality inequities are increasing, threatening population health. Public Health Implications. Addressing class inequities likely requires structural, worker-empowering nterventions.

Education and skills have always been crucial for improving individuals’ economic opportunities, and the fact that U.S. college enrollment has not fully rebounded after the COVID-19 pandemic is troubling to many. As with other advanced economies, the United States is heading towards a demographic cliff caused by low birth rates and an aging, barely growing population—all of which do not bode well for a robust labor force. Without the luxury of an expanding labor pipeline, current and incoming workers must be more productive to help sustain U.S. global competitiveness. This means that U.S. adults need to possess skills and credentials that allow them to respond to a work environment rapidly evolving due to the increased use of automation and artificial intelligence. In the face of these challenges, policymakers and practitioners in postsecondary education and workforce development must think harder about how they will deliver a skilled and productive workforce in the years ahead.

This paper proposes a new method for testing the spatial equilibrium model that is widely used to describe the spatial economy of the United States. To do this, I examine the relationship between workers' wages and the population density they face, namely the urban wage premium (UWP). Using U.S. data from 2000 to 2019, I find that unskilled workers in non-tradable service occupations consistently have a larger UWP compared with other unskilled workers, and the difference has been increasing over time, which I take as evidence against the spatial equilibrium model being a suitable approximation for unskilled workers' behavior. In contrast, skilled workers in both service and non-service occupations have a similar UWP that has remained fairly stable over the same period. My findings also suggest that the significant decrease in unskilled workers' UWP in recent decades was driven not only by the movement of unskilled workers from middle-wage tradable jobs to low-wage non-tradable service jobs, which was particularly pronounced in big cities, but also by the decrease in UWP within both tradable and non-tradable occupations.

Background: Contraception and abortion access are essential to reproductive justice and women’s equality. These services are critical for women to achieve their reproductive goals. Extensive studies have measured the association between individual-level characteristics with abortion access and contraceptive behavior, especially with different types of contraceptive methods. Limited studies were completed to investigate how living in a disadvantaged neighborhood can impact contraceptive behavior and abortion access. These limited studies mostly used income/poverty to measure neighborhood socioeconomic and structural conditions, which are single-level measurements and do not capture resources and opportunities available in each neighborhood. In addition, studies that measure contraceptive behavior often use type of contraceptive methods as the outcome; no study has been completed to investigate how living in a disadvantaged neighborhood is associated with whether women report using their preferred contraceptive method. Limited studies have been completed to investigate how the COVID-19 pandemic affected the change of contraceptive behavior. Thus, the goal of my dissertation is to fill the aforementioned gaps. Methods: To measure how living in a disadvantaged neighborhood affects abortion access and contraceptive behavior, this dissertation used two data sources. The first data source used to measure abortion access was collected by the Ohio Policy Evaluation Network (OPEN), where a 20% sample of 2014 to 2018 abortion records was drawn from three abortion facilities in the State of Ohio. The second data source used to measure contraceptive behavior was the Ohio Survey of Women, an Ohio population-representative survey collected by the National Opinion Research Center (NORC). To measure neighborhood socioeconomic and structural conditions, we used the Ohio Opportunity Index (OOI), which includes seven domain scores (transportation, education, employment, housing, health, environment, and crime) and a composite of all seven domain scores that captures that neighborhood’s opportunity and resources. In Aim 1, I used Generalized Estimating Equation (GEE) models to measure the association between neighborhood effects using the OOI and access to abortion. I assessed four outcomes: 1) whether the individual received an abortion after visiting an abortion facility, 2) the type of abortion method patients chose, 3) the number of days between the first abortion appointment visit to the surgical abortion, and 4) the gestational stage at the surgical abortion. In Aim 2, I used the weighted GEE to measure the association between neighborhood effects using the OOI and contraceptive behavior. I assessed four outcomes: 1) choosing female sterilization or long-acting reversible contraception (LARC) over other highly effective methods (e.g., injection, oral contraception, patch, and ring), 2) choosing highly effective methods (≤10% failure rate) over not highly effective methods (e.g., cervical cap, diaphragm, sponge, male condom, female condom, spermicides, and fertility awareness-based methods), 3) using any contraceptive method compare to not using any, and 4) using their preferred method compared to a non-preferred method. In Aim 3, I used a Sankey diagram to show how the non-response (loss to follow-up) and type of contraceptive method change over time. I performed descriptive analyses to explore individual and neighborhood characteristics in relation to non-response and contraceptive method change over time, and how COVID related to these changes. Results: Of the neighborhood-level characteristics examined, all variables except health resources were significantly associated with whether a patient received a surgical abortion at a later gestational stage (>64 days). Living in a neighborhood with more resources and opportunities (transportation, education, employment, and housing), as well as lower crime, lowers the odds of getting a surgical abortion at a later gestational stage; whereas, living in a neighborhood with more environmental resources increases the odds of getting a surgical abortion at a later gestational stage. In this study, neighborhoods’ environmental resources were highly correlated to rural/urban characteristics of the neighborhood, where rural areas have less pollution and more green space. Although most neighborhood-level characteristics were not significantly associated with contraceptive behavior, a consistent pattern was observed when modeling type of contraceptive method as the outcome. After controlling individual-level characteristics, women who lived in a disadvantaged neighborhood had increased odds of using contraception (compared to not using), increased odds of choosing LARC and female sterilization over other highly effective methods, and increased odds of choosing highly effective methods over not highly effective methods, compared to those who living in a neighborhood with more resources and opportunities. However, opposite relationships were observed when using the preferred contraception method as the outcome. When investigating the distribution of methods across different neighborhoods, among participants who reported using their preferred method, women who reported using LARC and living in neighborhoods with the lowest resources and opportunities were significantly more likely to report using their preferred method. However, they were less likely to report other non-LARC methods as their preferred method compared to women living in a neighborhood with more resources and opportunities. During the early and mid-pandemic phases (2019-2021), there was an increase in changes to less effective contraceptive methods compared to the late pandemic (2021-2022). Black women, women with less than high school degree, and women living in neighborhoods with less resources and opportunities were more likely to report changing to less effective methods, and more likely to be lost to follow-up. Discussion: This study shows that living in disadvantaged neighborhoods is associated with getting an abortion at a later gestational stage. Compared to neighborhoods with more resources, women who live in disadvantaged neighborhoods are more likely to report using highly effective contraception, especially LARC and female sterilization. However, they are more likely to report they are not using their preferred contraception, because dissatisfaction with their less effective and non-LARC contraceptive methods. The result indicates that neighborhood interventions can help women access abortion services in time, and more studies are needed to investigate why women living in disadvantaged neighborhoods are dissatisfied with non-LARC methods. Interventions that target women who use non-LARC methods in disadvantaged neighborhoods can improve access to their preferred methods. Disparities in access to contraception during the pandemic suggest that interventions should be tailored to specific subgroups, such as Black women, women with low education levels, and women living in disadvantaged neighborhoods.

Most Mobilians who make up Gen Z have struggled to live independently, according to a new study. Two-thirds of so-called Zoomers still live with their parents, according to RentCafe.

A wide range of sub-national immigration and immigrant-related policies have been adopted in the United States in the last two decades. Using data from the Urban Institute and IPUMS, the authors analyze the impact of state-level policy climates related to enforcement, integration, and public benefits on inter-state mobility for immigrants. The findings of the present analysis show that immigrants are responsive to changes in the state-level immigration policy climate. They are less likely to move out of states that are on balance more supportive of the foreign-born population and more likely to relocate from restrictive states. In addition, the policy impacts do not vary significantly by immigrants’ education, language fluency, or employment status, calling into question whether the policies miss their intended targets. These findings have implications for local governments and agencies in policy design and implementation as they continue to develop policies targeting the foreign-born population.

This study illustrates trends in self-employment among Americans age 50 years and older during the 23-year period from 2000 to 2022. More so, it distinguishes rates between those who have incorporated their businesses and those who have not, with the goal of identifying potential differences in rates by sociodemographic groups. Data are from the Annual Social and Economic Supplement (ASEC) within the Current Population Survey that were downloaded from IPUMS CPS. Results include weights to create a nationally representative sample of Americans. For this study, more than 1.25 million observations from respondents age 50 and older were analyzed. In general, we found that unincorporated self-employment is more common than incorporated self-employment among older adults, and rates of unincorporated self-employment have declined over the past two decades, whereas rates of incorporated self-employment have held steady. We also revealed differences over this 23-year period by age group, gender, race, Hispanic ethnicity, education, income, marital status, U.S. citizenship, workplace pension access, and health insurance. These findings illustrate the complicated nature of self-employment and its relationship with important sociodemographic factors among older Americans.

Prior literature analyzing gender differences in commuting has reported that men commute longer distance/time than do women, and one explanation for this gender gap is based on household responsibilities falling on women. But most of the literature examining gender differences in commuting has not considered the interdependence that exists between the members of couples. We analyze gender differences in commuting time, and we take into account the inter-relatedness of decisions, estimating Seemingly Unrelated Regressions (SUR) on commuting time and mode of transport (private, public, and active transport) for dual-earner couples in Spain, Italy, Korea, and the United Kingdom. Our results indicate that the number of children is signicantly related to shorter female commuting times in Spain, Italy, and the UK, with no associations found for men, while in Korea there is a reduction in both male and female commuting times. Additionally, the presence of children signicantly increases (decreases) the proportion of commuting time done by private (public) transport, but results differ signicantly by gender, number of children, and country. Our evidence indicates that couples' decisions on commuting are complementary, which may shed light on the relationship in commuting that theoretical models should address.

A large literature has focused on how capital – human, social, and economic – influences immigrant economic attainment. However, few studies have explored how other factors, namely racialization processes, contribute to these within group differences. We examine the role of skin tone as a source of intra-group difference that stratifies outcomes among Mexican immigrants in the early twentieth century. We create a new dataset of Mexican border crossing records between 1910 and 1940 that we then link to the 1940 census. We use at-entry characteristics to predict income in 1940 and find that—in line with dominant assimilation theories—standard measures of capital are associated with within-group differences in attainment. However, we also find that skin tone was a source of within-group stratification: being perceived as having darker skin is associated with lower subsequent economic attainment compared to those with lighter skin. Furthermore, we find that these patterns vary across contexts such that skin tone had a greater effect in Texas, while standard measures of skill had stronger effects in California. Taken together, we argue that while standard measures of assimilation typically predict later outcomes, the stratifying effects of skin tone has been a long-run feature of Mexican immigrant history.

It is often asserted that secure property rights and legal frameworks conducive to the functioning of markets are essential institutional foundations of a capitalist economy. It is sometimes even claimed that they are preconditions of economic growth. Efforts to implement those institutions have, however, produced heterogeneous outcomes for different groups of people. This dissertation considers the effects of two waves of institutional change in North America: the nineteenth-century privatization and subsequent alienation of communal property in the United States and Mexico and the late-twentieth-century neoliberal reforms in Mexico. Both episodes contributed to profound structural transformations in the North American economy. In the process of shaping important aspects of the present capitalist economies of Mexico and the United States, the above-mentioned institutional changes resulted in land loss, dispossession, the destruction of traditional livelihoods and, for many people, insertion into labor markets on the lowest rungs, with reduced autonomy, and with little or no job security. The dissertation examines three cases of communal property privatization. First, it considers the effect of the 1887 Dawes General Allotment Act on American Indian migration using data from the 1930 U.S. Census. The results suggest that individuals who were likely to have lost land due to allotment had a higher propensity to migrate to cities and to other states. Second, historical literature is reviewed to understand how the privatization of communal property under Mexico’s 1856 Lerdo Law exacerbated land loss and inequality. That episode iii inspired subsequent efforts to reverse the effects of privatization through the creation of a new form of communal property known as ejidos during and after the Mexican Revolution. Third, the consequences of 1992 constitutional reforms allowing the privatization of ejidos are considered. The main finding is that municipalities with larger relative declines in ejido and agrarian community membership (as a percentage of population) and more land sales to non- ejido-members experienced larger increases in income inequality. Mexico’s 1992 ejido reforms were part of a broader set of neoliberal reforms aimed at seamlessly integrating the country into the North American and global economies. Trade and investment regulations were liberalized, which contributed to the spread of cross-border production sharing or “offshoring” arrangements in the manufacturing sector. The last section of the dissertation considers the effects of those arrangements on employment volatility. The main finding is that reliance on offshoring-related revenues generally had a large, positive impact on manufacturing-sector employment volatility in Mexico over the 2007 to 2020 period. In contrast, trade that was not related to offshoring had, at most, a weak impact on volatility. The main policy implication is that attracting jobs in the labor-intensive stages of transnational manufacturing production processes may entail the risk of increasing employment volatility.

The Office of Management and Budget (OMB) seeks to update race and ethnicity statistics in the United States through a proposed revision to the existing 1997 standards.1 The most consequential update is the proposed creation of a new Middle Eastern or North African (MENA) category, while other updates would combine race and ethnicity into a single question and require additional detail in race reporting. The stated goal is to represent the diverse American population and simplify self‐​reporting on federal surveys and forms. These statistical standards are crucial as they govern civil rights enforcement, federal funding allocation, program eligibility, and statistical reporting. This brief approximates responses to the proposed MENA question, as well as the creation of a new Hispanic or Latino race category, using ancestry and country of origin responses from the American Community Survey (ACS). I find a total MENA population of 3.8 million in 2021—about 1 percent of the total U.S. population. The MENA population is most demographically similar to the Non‐​Hispanic White and Asian populations in terms of age, earnings, education, and occupation. Specifically, the MENA population is more highly educated and has higher earnings than Non‐​Hispanic Whites, but lower levels of education and earnings than the Asian population. (This brief uppercases races and ethnicities to be consistent with U.S. government stylistic guidelines.)

In a departure from the traditional narrative of young adults leaving the nest, 68% of Gen Z adults and 20% of Millennials are opting to stay within familial confines, according to a recent RentCafe analysis of Integrated Public Use Microdata Series (IPUMS) data. The analysis opens up a conversation about generational adaptation, economic hardships, and the "evolving familial structures that are reshaping the American Dream," writes Adina Dragos, creative writer, RentCafe.

Our research questions are often chosen based on the existence of suitable data for analysis or prior research in the area. For new interdisciplinary research areas, such as occupational health equity, suitable data might not yet exist. In this manuscript, we describe how we approached a research question in the absence of suitable data using the example of identifying inequities in adequate bathrooms in US workplaces. We created a conceptual model that hypothesises causal mechanisms for occupational health inequities, and from this model we identified a series of questions that could be answered using separate data sets to better understand inequities in adequate workplace bathrooms. Breaking up the analysis into multiple steps allowed us to use multiple data sources and analysis methods, which helped compensate for limitations in each data set.