David C. Folch’s research while affiliated with Northern Arizona University and other places

Objectives. To examine whether a previously reported association between airborne lead exposure and children’s cognitive function replicates across a geographically diverse sample of the United States. Methods. Residential addresses of children (< 5 years) were spatially joined to the Risk-Screening Environmental Indicators model of relative airborne lead toxicity. Cognitive outcomes for children younger than 8 years were available for 1629 children with IQ data and 1476 with measures of executive function (EF; inhibitory control, cognitive flexibility). We used generalized linear models using generalized estimating equations to examine the associations of lead, scaled by interquartile range (IQR), accounting for individual- and area-level confounders. Results. An IQR increase in airborne lead was associated with a 0.74-point lower mean IQ score (b = −0.74; 95% confidence interval = −1.00, −0.48). The association between lead and EF was nonlinear and was modeled with a knot at the 97.5th percentile of lead in our sample. Lead was significantly associated with lower mean inhibitory control but not with cognitive flexibility. This effect was stronger among males for both IQ and inhibitory control. Conclusions. Early-life exposure to airborne lead is associated with lower cognitive functioning. ( Am J Public Health. 2024;114(3):309–318. https://doi.org/10.2105/AJPH.2023.307519 )

The American Community Survey (ACS) is an indispensable tool for studying the United States (US) population. Each year the US Census Bureau (BOC) publishes approximately 11 billion ACS estimates, each of which is accompanied by a margin of error (MOE) specific to that estimate. Researchers, policy makers, and government agencies combine these estimates in myriad ways, which requires an accurate measurement of the MOE on that combined estimate. We compare three options for computing this MOE: the analytic approach uses standard statistically derived formulas, the simulation approach builds an empirical distribution of the combined estimate based on simulated values of the inputs, and the replicate approach uses simulated values published by the BOC based on their internal model that statistically replicates the entire ACS 80 times. We find that since the replicate approach is the only one of the three to incorporate covariance between the input variables, it performs the best. We further find that the simulation and analytic approaches generally match one another and can both overestimate and underestimate the MOE; they have their places when the replicate approach is not feasible.

Background The growth of geolocated data has opened the door to a wealth of new research opportunities in the health fields. One avenue of particular interest is the relationship between the spaces where people spend time and their health outcomes. This research model typically intersects individual data collected on a specific cohort with publicly available socioeconomic or environmental aggregate data. In spatial terms: individuals are represented as points on map at a particular time, and context is represented as polygons containing aggregated or modeled data from sampled observations. Uncertainty abounds in these kinds of complex representations. Methods We present four sensitivity analysis approaches that interrogate the stability of spatial and temporal relationships between point and polygon data. Positional accuracy assesses the significance of assigning the point to the correct polygon. Neighborhood size investigates how the size of the context assumed to be relevant impacts observed results. Life course considers the impact of variation in contextual effects over time. Time of day recognizes that most people occupy different spaces throughout the day, and that exposure is not simply a function residential location. We use eight years of point data from a longitudinal study of children living in rural Pennsylvania and North Carolina and eight years of air pollution and population data presented at 0.5 mile (0.805 km) grid cells. We first identify the challenges faced for research attempting to match individual outcomes to contextual effects, then present methods for estimating the effect this uncertainty could introduce into an analysis and finally contextualize these measures as part of a larger framework on uncertainty analysis. Results Spatial and temporal uncertainty is highly variable across the children within our cohort and the population in general. For our test datasets, we find greater uncertainty over the life course than in positional accuracy and neighborhood size. Time of day uncertainty is relatively low for these children. Conclusions Spatial and temporal uncertainty should be considered for each individual in a study since the magnitude can vary considerably across observations. The underlying assumptions driving the source data play an important role in the level of measured uncertainty.

Although policies to remove lead from gasoline have resulted in a substantial reduction in airborne lead, multiple industries are known to generate lead that is released in the air. The present study examines the extent to which residential proximity to a documented source of airborne lead is associated with intellectual and executive function in children. Data were available for n = 849 children from the Family Life Project. Geolocation for children's residences between birth and 36 months were referenced against the Environmental Protection Agency's Risk Screening Environmental Indicators (RSEI) database, which estimates exposure for each ½ mile grid in the contiguous United States. Instrumental variable models were employed to estimate causal associations between exposure and cognitive outcomes measured at 36, 48, and 60 months, using census-documented density of manufacturing employment as the instrument. Models of continuous lead dosage indicated small negative effects for both child IQ and executive function (EF). These results indicate that RSEI estimates of airborne lead exposure are meaningfully associated with decrements in cognitive development.

Background: The growth of geolocated data has opened the door to a wealth of new research opportunities in the health fields. One avenue of particular interest is the relationship between the spaces where people spend time and their health outcomes. This research model typically intersects individual data collected on a specific cohort with publicly available socioeconomic or environmental aggregate data. In spatial terms: individuals are represented as points on map at a particular time, and context is represented as polygons containing aggregated or modeled data from sampled observations. Uncertainty abounds in these kinds of complex representations. Methods: We present four sensitivity analysis approaches that interrogate the stability of spatial and temporal relationships between point and polygon data. Positional accuracy assesses the significance of assigning the point to the correct polygon. Neighborhood size investigates how the size of the context assumed to be relevant impacts observed results. Life course considers the impact of variation in contextual effects over time. Time of day recognizes that most people occupy different spaces throughout the day, and that exposure is not simply a function residential location. We use eight years of point data from a longitudinal study of children living in rural Pennsylvania and North Carolina and eight years of air pollution and population data presented at 0.5 mile (0.805 km) grid cells. We first identify the challenges faced for research attempting to match individual outcomes to contextual effects, then present methods for estimating the effect this uncertainty could introduce into an analysis and finally contextualize these measures as part of a larger framework on uncertainty analysis. Results: Spatial and temporal uncertainty is highly variable across the children within our cohort and the population in general. For our test datasets, we find greater uncertainty over the life course than in positional accuracy and neighborhood size. Time of day uncertainty is relatively low for these children. Conclusions: Spatial and temporal uncertainty should be considered for each individual in a study since the magnitude can vary considerably across observations. The underlying assumptions driving the source data play an important role in the level of measured

As a concept, social vulnerability describes combinations of social, cultural, economic, political, and institutional processes that shape socioeconomic differentials in the experience of and recovery from hazards. Quantitative measures of social vulnerability are widely used in research and practice. In this paper, we establish criteria for the evaluation of social vulnerability indicators and apply those criteria to the most widely used measure of social vulnerability, the Social Vulnerability Index (SoVI). SoVI is a single quantitative indicator that purports to measure a place’s social vulnerability. We show that SoVI has some critical shortcomings regarding theoretical and internal consistency. Specifically, multiple SoVI-based measurements of the vulnerability of the same place, using the same data, can yield strikingly different results. We also show that the SoVI is often misaligned with theory; increases in variables that contribute to vulnerability, like the unemployment rate, often decrease vulnerability as measured by the SoVI. We caution against the use of the index in policy making or other risk-reduction efforts, and we suggest ways to more reliably assess social vulnerability in practice.

This paper compares the occupational structure of cities in Brazil and United States aiming to evaluate the extent to which the economic structure of these urban agglomerations is associated with the different stages of development, specifically when comparing a rich country with a developing one. Using a harmonized occupational database and microdata from the Brazilian 2010 Demographic Census and the U.S. American Community Survey (2008-2012), results show that Brazilian cities have a stronger connection between population size, both with occupational structure and human capital distribution, than the one found for cities in the United States. These findings suggest a stronger primacy of large cities in Brazil’s urban network and a more unequal distribution of economic activity across cities when compared to USA, indicating a strong correlation between development and occupational structure.

To accommodate network allocation, population polygons are frequently transformed into singular, weighted centroids which are then integrated into the network either by snapping each centroid to the nearest network segment or by generating an artificial connector that becomes part of the network. In this article, an investigation of the connection method of network allocation is undertaken with two primary foci: (1) how the density of centroid connectors effects travel cost along the network; and (2) how the algorithms utilized to determine the placement of connectors are affected by the density of connectors. We hypothesize that both issues have an effect on network travel cost and, therefore, on network‐based modeling. These hypotheses are tested on three nested spatial networks in the New England region of the United States. Two fundamental facility location models, the p‐median problem, and p‐center problem, are solved at each spatial extent while varying the density of connectors from one to four. When generating more than one connector thought must be given to the method of connection, the angle of dispersion, the acceptable tolerance of connector length, segment crossing, and saturated connectivity. A novel and thorough framework is proposed to address these concerns.

Scholars frequently use counts of populations aggregated into geographic units like census tracts to represent measures of neighborhood context. Decades of research confirm that variation in how individuals are aggregated into geographic units can dramatically alter analyses conducted with these units. While most researchers are aware of the problem, they have lacked the tools to determine its magnitude or its capacity to affect analytical results obtained using these contextual measures. Using confidential access to the complete 2010 U.S. Decennial Census, we can construct—for all persons in the U.S.—individual‐specific contexts, which we group according to Census‐assigned block, block group, and tract. We compare these individual‐specific measures to the published statistics at each scale, and we then determine the degree to which published measures could be affected by how boundaries are drawn using a simple statistic, the standard deviation of individual context (SDIC). For three key measures (percent Black, percent Hispanic, and Entropy—a measure of ethno‐racial diversity), we find that block‐level Census statistics frequently contain a high degree of uncertainty meaning that they may not capture the actual context of individuals within them. More problematic, we uncover systematic spatial patterns in the uncertainty associated with contextual variables at all three scales.