Population-level Inference Research Articles

Effects of connectivity hyperalignment (CHA) on estimated brain network properties: from coarse-scale to fine-scale.

Recent gains in functional magnetic resonance imaging (fMRI) studies have been driven by increasingly sophisticated statistical and computational techniques and the ability to capture brain data at finer spatial and temporal resolution. These advances allow researchers to develop population-level models of the functional brain representations underlying behavior, performance, clinical status, and prognosis. However, even following conventional preprocessing pipelines, considerable inter-individual disparities in functional localization persist, posing a hurdle to performing compelling population-level inference. Persistent misalignment in functional topography after registration and spatial normalization will reduce power in developing predictive models and biomarkers, reduce the specificity of estimated brain responses and patterns, and provide misleading results on local neural representations and individual differences. This study aims to determine how connectivity hyperalignment (CHA)-an analytic approach for handling functional misalignment-can change estimated functional brain network topologies at various spatial scales from the coarsest set of parcels down to the vertex-level scale. The findings highlight the role of CHA in improving inter-subject similarities, while retaining individual-specific information and idiosyncrasies at finer spatial granularities. This highlights the potential for fine-grained connectivity analysis using this approach to reveal previously unexplored facets of brain structure and function.

Fleeting but Not Forgotten: The Imprint of Escaping Hydrogen Atmospheres on Super-Earth Interiors

Small, close-in exoplanets are divided into two subpopulations: super-Earths and sub-Neptunes. Most super-Earths are thought to have lost their primordially accreted hydrogen-dominated atmospheres via thermally driven winds. We consider the global chemical equilibrium of super-Earths and the lasting impacts of their fleeting hydrogen atmospheres. We find that hydrogen is efficiently sequestered into the interior, oxidizing iron and endogenously producing ∼0.5%–1.0% water by mass. As the atmospheres of super-Earths are continuously sculpted by mass loss and chemical equilibration, they remain hydrogen-dominated by mole (number) fraction but become steam-dominated by mass, which may be observable with JWST for planets transitioning across the radius valley. One of the main effects of efficient sequestration of hydrogen into the interior is to produce an underdense bulk interior compared to that of Earth. We predict bulk densities of super-Earths to be ∼5.0 g cm−3 for a 1M ⊕ planet, which is consistent with high-precision mass measurements and also population-level inference analyses from atmospheric escape models.

PopSED: Population-level Inference for Galaxy Properties from Broadband Photometry with Neural Density Estimation

We present PopSED , a framework for the population-level inference of galaxy properties from photometric data. Unlike the traditional approach of first analyzing individual galaxies and then combining the results to determine the physical properties of the entire galaxy population, we directly make the population distribution the inference objective. We train normalizing flows to approximate the population distribution by minimizing the Wasserstein distance between the synthetic photometry of the galaxy population and the observed data. We validate our method using mock observations and apply it to galaxies from the GAMA survey. PopSED reliably recovers the redshift and stellar mass distribution of 105 galaxies using broadband photometry within <1 GPU hr, being 105–6 times faster than the traditional spectral energy distribution modeling method. From the population posterior, we also recover the star-forming main sequence for GAMA galaxies at z < 0.1. With the unprecedented number of galaxies in upcoming surveys, our method offers an efficient tool for studying galaxy evolution and deriving redshift distributions for cosmological analyses.

Addressing positivity violations in causal effect estimation using Gaussian process priors.

In observational studies, causal inference relies on several key identifying assumptions. One identifiability condition is the positivity assumption, which requires the probability of treatment be bounded away from 0 and 1. That is, for every covariate combination, it should be possible to observe both treated and control subjects the covariate distributions should overlap between treatment arms. If the positivity assumption is violated, population-level causal inference necessarily involves some extrapolation. Ideally, a greater amount of uncertainty about the causal effect estimate should be reflected in such situations. With that goal in mind, we construct a Gaussian process model for estimating treatment effects in the presence of practical violations of positivity. Advantages of our method include minimal distributional assumptions, a cohesive model for estimating treatment effects, and more uncertainty associated with areas in the covariate space where there is less overlap. We assess the performance of our approach with respect to bias and efficiency using simulation studies. The method is then applied to a study of critically ill female patients to examine the effect of undergoing right heart catheterization.

Maternal major depression disorder misclassification errors: Remedies for valid individual- and population-level inference.

Individual and population level inference about risk and burden of MDD, particularly maternal MDD, is often made using case‐finding tools that are imperfect and prone to misclassification error (i.e. false positives and negatives). These errors or biases are rarely accounted for and lead to inappropriate clinical decisions, inefficient allocation of scarce resources, and poor planning of maternal MDD prevention and treatment interventions. The argument that the use of existing maternal MDD case‐finding instruments results in misclassification errors is not new; in fact, it has been argued for decades, but by and large its implications and particularly how to correct for these errors for valid inference is unexplored. Correction of the estimates of maternal MDD prevalence, case‐finding tool sensitivity and specificity is possible and should be done to inform valid individual and population‐level inferences.

The US population-level burden of cow's milk allergy

BackgroundCow's milk is a food allergen of public health importance both in the United States and globally. Its natural history and epidemiology have been most studied among infants and young children, but the public health burden it places on adults and older pediatric populations remains unclear. This study comprehensively characterizes the US population-level burden of cow's milk allergy (CMA), including its prevalence, severity, health care utilization, psychosocial impact, natural history, and other associated factors—including ability to tolerate extensively heated milk products—within a large, nationally-representative survey sample of US households. MethodsA cross-sectional survey was administered via web and telephone. Population level inference was based on data collected from participants recruited from National Opinion Research Center (NORC) at the University of Chicago's probability-based AmeriSpeak panel. Established dual-sample complex survey calibration methods were used to augment this sample with additional participants recruited through Survey Sampling International to increase the precision of the resulting estimates. In total, surveys were administered to a nationally representative sample of 51 819 US households from October 1, 2015, through September 31, 2016. Self-report responses from 40 453 adults and parent-proxy report for 38 408 children were analyzed. ResultsAnalyzing survey responses from 78 851 individuals, an estimated 4.7% (95% CI, 4.4%–4.9%) of the US population reported current CMA, whereas 1.9% (95% CI, 1.8%–2.1%) met symptom-report criteria for convincing IgE-mediated allergy. An estimated 0.9% (95% CI, 0.8%–1.0%) had CMA that met convincing symptom-report criteria and was physician diagnosed. Female and White respondents were more likely to report outgrowing CMA relative to males and those reporting other races and ethnicities. Individuals with CMA who reported tolerance to baked milk products were less likely to report severe reaction histories (33.5% vs 42.7%; p = .03), a lifetime history of food allergy-related emergency department visits (43.4% vs. 55.8%; p = .005), and treating a milk-allergic reaction with epinephrine (10.6 vs. 18.9%; p = .003). These individuals also exhibited less psychosocial burden on the validated Food Allergy Independent Measure (FAIM) than their counterparts with CMA who were allergic to baked forms of milk. ConclusionThese data indicate a discrepancy in reported rates of allergy to cow's milk among the US general population ranging from approximately 1 in 20 with reported CMA to fewer than 1 in 50 with physician-confirmed CMA. However, they suggest a substantial population-level burden of CMA, including substantial healthcare utilization, psychosocial burden and nutritional impacts—particularly among the approximately 30% of individuals with CMA who cannot tolerate baked forms of milk.

Knowledge, attitudes, and behaviors regarding tick-borne disease prevention in Lyme disease-endemic areas of the Upper Midwest, United States

Lyme disease and other tick-borne diseases are a major public health threat in the Upper Midwestern United States, including Michigan, Minnesota, and Wisconsin. To prevent tick bites and tick-borne diseases, public health officials commonly recommend personal protective measures and property management techniques. Adoption of tick-borne disease prevention behaviors and practices by individuals are, however, highly variable. We aimed to characterize current tick-borne disease knowledge, attitudes, and prevention behaviors (KAB) practiced by the public in these states, as well as their willingness to use specific tick control methods. We conducted a population-based survey in summer 2019 in 48 high-risk counties (those having a five-year average (2013–2017) Lyme disease incidence of ≥ 10 cases per 100,000 persons per year), in Michigan, Minnesota, and Wisconsin. A total of 2713 surveys were analyzed; survey weights were used to account for household selection probability and post-stratified to match county-level joint age and sex population distributions in population-level inference. An estimated 98% of the population had heard of Lyme disease, with most perceiving it as very or extremely serious (91%); however, only an estimated 25% perceived tick-borne diseases as very or extremely common in their community. Among those who spent time in places with ticks from April through October, an estimated 68% check themselves thoroughly for ticks most of the time or always and 43% use bug repellent on skin or clothing most of the time or always. An estimated 13% of the population had ever treated their property with a pesticide to kill ticks, and 3% had ever used devices that apply pesticide to rodents to kill ticks on their property. Willingness to practice tick bite prevention behaviors, however, was estimated to be much higher; with 82% being willing to perform tick checks at least once a day, and more than 60% willing to use bug repellent, tick control products on pets, or to bathe within two hours of being outdoors. We found that residents would likely be willing to support a county-wide tick control program to reduce the risk of tick-borne disease in their community (81%) or to apply tick control products to their property to reduce the risk of tick-borne disease in their household (79%). Tick checks were more likely to be practiced among participants who perceived tick-borne diseases to be highly prevalent in their community, if they or a household member had been previously diagnosed with a tick-borne disease?, or if they perceived tick exposure to be likely around their home, cabin, or vacation home. In addition, property-based tick control methods were associated with perceived risk of encountering ticks around the home, cabin, or vacation home. Participants who had seen information from state health departments were also more likely to practice preventive measures. The most common reported barriers to using any of these methods were forgetfulness, safety concerns, and lack of awareness. Our survey findings shed light on how residents from these Upper Midwest states may adopt tick control and tick bite prevention measures and how public health outreach may be most effective for this population.

Modeling abundance, distribution, movement and space use with camera and telemetry data.

Studies of animal abundance and distribution are often conducted independently of research on movement, despite the important links between processes. Movement can cause rapid changes in spatial variation in density, and movement influences detection probability and therefore estimates of abundance from inferential methods such as spatial capture-recapture (SCR). Technological developments including camera traps and GPS telemetry have opened new opportunities for studying animal demography and movement, yet statistical models for these two data types have largely developed along parallel tracks. We present a hierarchical model in which both datasets are conditioned on a movement process for a clearly defined population. We fitted the model to data from 60 camera traps and 23,572 GPS telemetry locations collected on 17 male white-tailed deer in the Big Cypress National Preserve, Florida, USA during July 2015. Telemetry data were collected on a 3-4 h acquisition schedule, and we modeled the movement paths of all individuals in the region with a Ornstein-Uhlenbeck process that included individual-specific random effects. Two of the 17 deer with GPS collars were detected on cameras. An additional 20 male deer without collars were detected on cameras and individually identified based on their unique antler characteristics. Abundance was 126 (95% CI: 88-177) in the 228 km2 region, only slightly higher than estimated using a standard SCR model: 119 (84-168). The standard SCR model, however, was unable to describe individual heterogeneity in movement rates and space use as revealed by the joint model. Joint modeling allowed the telemetry data to inform the movement model and the SCR encounter model, while leveraging information in the camera data to inform abundance, distribution and movement. Unlike most existing methods for population-level inference on movement, the joint SCR-movement model can yield unbiased inferences even if non-uniform sampling is used to deploy transmitters. Potential extensions of the model include the addition of resource selection parameters, and relaxation of the closure assumption when interest lies in survival and recruitment. These developments would contribute to the emerging holistic framework for the study of animal ecology, one that uses modern technology and spatio-temporal statistics to learn about interactions between behavior and demography.

Unbiased likelihood-free inference of the Hubble constant from light standard sirens

Multi-messenger observations of binary neutron star mergers offer a promising path towards resolution of the Hubble constant ($H_0$) tension, provided their constraints are shown to be free from systematics such as the Malmquist bias. In the traditional Bayesian framework, accounting for selection effects in the likelihood requires calculation of the expected number (or fraction) of detections as a function of the parameters describing the population and cosmology; a potentially costly and/or inaccurate process. This calculation can, however, be bypassed completely by performing the inference in a framework in which the likelihood is never explicitly calculated, but instead fit using forward simulations of the data, which naturally include the selection. This is Likelihood-Free Inference (LFI). Here, we use density-estimation LFI, coupled to neural-network-based data compression, to infer $H_0$ from mock catalogues of binary neutron star mergers, given noisy redshift, distance and peculiar velocity estimates for each object. We demonstrate that LFI yields statistically unbiased estimates of $H_0$ in the presence of selection effects, with precision matching that of sampling the full Bayesian hierarchical model. Marginalizing over the bias increases the $H_0$ uncertainty by only $6\%$ for training sets consisting of $O(10^4)$ populations. The resulting LFI framework is applicable to population-level inference problems with selection effects across astrophysics.

Bayesian inference of population prevalence.

Within neuroscience, psychology, and neuroimaging, the most frequently used statistical approach is null hypothesis significance testing (NHST) of the population mean. An alternative approach is to perform NHST within individual participants and then infer, from the proportion of participants showing an effect, the prevalence of that effect in the population. We propose a novel Bayesian method to estimate such population prevalence that offers several advantages over population mean NHST. This method provides a population-level inference that is currently missing from study designs with small participant numbers, such as in traditional psychophysics and in precision imaging. Bayesian prevalence delivers a quantitative population estimate with associated uncertainty instead of reducing an experiment to a binary inference. Bayesian prevalence is widely applicable to a broad range of studies in neuroscience, psychology, and neuroimaging. Its emphasis on detecting effects within individual participants can also help address replicability issues in these fields.

Reclassifying stroke lesion anatomy

Cognitive and behavioural outcomes in stroke reflect the interaction between two complex anatomically-distributed patterns: the functional organization of the brain and the structural distribution of ischaemic injury. Conventional outcome models—for individual prediction or population-level inference—commonly ignore this complexity, discarding anatomical variation beyond simple characteristics such as lesion volume. This sets a hard limit on the maximum fidelity such models can achieve. High-dimensional methods can overcome this problem, but only at prohibitively large data scales. Drawing on one of the largest published collections of anatomically-registered imaging of acute stroke—N = 1333—here we use non-linear dimensionality reduction to derive a succinct latent representation of the anatomical patterns of ischaemic injury, agglomerated into 21 distinct intuitive categories. We compare the maximal predictive performance it enables against both simpler low-dimensional and more complex high-dimensional representations, employing multiple empirically-informed ground truth models of distributed structure–outcome relationships. We show our representation sets a substantially higher ceiling on predictive fidelity than conventional low-dimensional approaches, but lower than that achievable within a high-dimensional framework. Where descriptive simplicity is a necessity, such as within clinical care or research trials of modest size, the representation we propose arguably offers a favourable compromise of compactness and fidelity.

Hierarchically modelling Kepler dwarfs and subgiants to improve inference of stellar properties with asteroseismology

ABSTRACT With recent advances in modelling stars using high-precision asteroseismology, the systematic effects associated with our assumptions of stellar helium abundance (Y) and the mixing-length theory parameter (αMLT) are becoming more important. We apply a new method to improve the inference of stellar parameters for a sample of Kepler dwarfs and subgiants across a narrow mass range ($0.8 \lt M \lt 1.2\, \rm M_\odot$). In this method, we include a statistical treatment of Y and the αMLT. We develop a hierarchical Bayesian model to encode information about the distribution of Y and αMLT in the population, fitting a linear helium enrichment law including an intrinsic spread around this relation and normal distribution in αMLT. We test various levels of pooling parameters, with and without solar data as a calibrator. When including the Sun as a star, we find the gradient for the enrichment law, $\Delta Y / \Delta Z = 1.05\substack{+0.28\\-0.25}$ and the mean αMLT in the population, $\mu _\alpha = 1.90\substack{+0.10\\-0.09}$. While accounting for the uncertainty in Y and αMLT, we are still able to report statistical uncertainties of 2.5 per cent in mass, 1.2 per cent in radius, and 12 per cent in age. Our method can also be applied to larger samples that will lead to improved constraints on both the population level inference and the star-by-star fundamental parameters.

Context-dependent variability in the predicted daily energetic costs of disturbance for blue whales.

Assessing the long-term consequences of sub-lethal anthropogenic disturbance on wildlife populations requires integrating data on fine-scale individual behavior and physiology into spatially and temporally broader, population-level inference. A typical behavioral response to disturbance is the cessation of foraging, which can be translated into a common metric of energetic cost. However, this necessitates detailed empirical information on baseline movements, activity budgets, feeding rates and energy intake, as well as the probability of an individual responding to the disturbance-inducing stressor within different exposure contexts. Here, we integrated data from blue whales (Balaenoptera musculus) experimentally exposed to military active sonar signals with fine-scale measurements of baseline behavior over multiple days or weeks obtained from accelerometry loggers, telemetry tracking and prey sampling. Specifically, we developed daily simulations of movement, feeding behavior and exposure to localized sonar events of increasing duration and intensity and predicted the effects of this disturbance source on the daily energy intake of an individual. Activity budgets and movements were highly variable in space and time and among individuals, resulting in large variability in predicted energetic intake and costs. In half of our simulations, an individual's energy intake was unaffected by the simulated source. However, some individuals lost their entire daily energy intake under brief or weak exposure scenarios. Given this large variation, population-level models will have to assess the consequences of the entire distribution of energetic costs, rather than only consider single summary statistics. The shape of the exposure-response functions also strongly influenced predictions, reinforcing the need for contextually explicit experiments and improved mechanistic understanding of the processes driving behavioral and physiological responses to disturbance. This study presents a robust approach for integrating different types of empirical information to assess the effects of disturbance at spatio-temporal and ecological scales that are relevant to management and conservation.

Census data aggregation decisions can affect population-level inference in heterogeneous populations.

Conservation and population management decisions often rely on population models parameterized using census data. However, the sampling regime, precision, sample size, and methods used to collect census data are usually heterogeneous in time and space. Decisions about how to derive population‐wide estimates from this patchwork of data are complicated and may bias estimated population dynamics, with important implications for subsequent management decisions.Here, we explore the impact of site selection and data aggregation decisions on pup survival estimates, and downstream estimates derived from parameterized matrix population models (MPMs), using a long‐term dataset on grey seal (Halichoerus grypus) pup survival from southwestern Wales. The spatiotemporal and methodological heterogeneity of the data are fairly typical for ecological census data and it is, therefore, a good model to address this topic.Data were collected from 46 sampling locations (sites) over 25 years, and we explore the impact of data handling decisions by varying how years and sampling locations are combined to parameterize pup survival in population‐level MPMs. We focus on pup survival because abundant high‐quality data are available on this developmental stage.We found that survival probability was highly variable with most variation being at the site level, and poorly correlated among sampling sites. This variation could generate marked differences in predicted population dynamics depending on sampling strategy. The sample size required for a confident survival estimate also varied markedly geographically.We conclude that for populations with highly variable vital rates among sub‐populations, site selection and data aggregation methods are important. In particular, including peripheral or less frequently used areas can introduce substantial variation into population estimates. This is likely to be context‐dependent, but these choices, including the use of appropriate weights when summarizing across sampling areas, should be explored to ensure that management actions are successful.

Population and Stand-Level Inference in Forest Inventory with Penalized Splines

AbstractPenalized splines have potential to decrease estimates of variance in forest inventories with a design-based population-level inference, and a model-based domain-level inference by decreasing the likelihood of a model misspecification. We provide examples with second-order (B2) B-splines and radial basis (RB) functions as extensions to a linear working model (WM). Bias was not prominent, yet greater with B2 and in particular with RB than with WM, and decreased with sample size. Important reductions in the variance of a population mean were achieved with both B2 and RB, but at the domain-level only with RB. The proposed regression estimator of variance generated estimates of variance being slightly smaller than the observed variance. A consistent and larger underestimation was seen with the popular difference estimator of variance.Study Implications: Forest inventories supported by light detection and range (LiDAR) data require—in the estimation phase—a model for linking LiDAR metrics to attributes of interest. Formulating a parametric model can be a challenge and unsatisfactory if the goodness of fit varies across the range of the attribute of interest. A semiparametric model provides more flexibility and lessens the chance of a model misspecification, albeit with the potential of overfitting. A penalty directed at reducing overfitting is required. A flexible semiparametric model is potentially also better suited for applications to small areas like stands than a parametric model. We demonstrate that important reductions in variance are indeed possible, but also that they depend on the form of the nonparametric part of the chosen model and the level of inference (population versus domains). With regard to practical application, reliable estimates of forest attributes at stand-level are of special interest within the scope of forest-management planning, as silvicultural treatments are always stand-oriented, at least with small-scale forestry under Central European conditions, and stand-related volume (basal area, tree density) belongs to the set of relevant parameters for management decisions regarding harvest and regeneration measures.

Mechanical Turk data collection in addiction research: utility, concerns and best practices.

Amazon Mechanical Turk (MTurk) provides a crowdsourcing platform for the engagement of potential research participants with data collection instruments. This review (1) provides an introduction to the mechanics and validity of MTurk research; (2) gives examples of MTurk research; and (3) discusses current limitations and best practices in MTurk research. We review four use cases of MTurk for research relevant to addictions: (1) the development of novel measures, (2) testing interventions, (3) the collection of longitudinal use data to determine the feasibility of longer-term studies of substance use and (4) the completion of large batteries of assessments to characterize the relationships between measured constructs. We review concerns with the platform, ways of mitigating these and important information to include when presenting findings. MTurk has proved to be a useful source of data for behavioral science more broadly, with specific applications to addiction science. However, it is still not appropriate for all use cases, such as population-level inference. To live up to the potential of highly transparent, reproducible science from MTurk, researchers should clearly report inclusion/exclusion criteria, data quality checks and reasons for excluding collected data, how and when data were collected and both targeted and actual participant compensation. Although on-line survey research is not a substitute for random sampling or clinical recruitment, the Mechanical Turk community of both participants and researchers has developed multiple tools to promote data quality, fairness and rigor. Overall, Mechanical Turk has provided a useful source of convenience samples despite its limitations and has demonstrated utility in the engagement of relevant groups for addiction science.

Accuracy of medical billing data against the electronic health record in the measurement of colorectal cancer screening rates

ObjectiveMedical billing data are an attractive source of secondary analysis because of their ease of use and potential to answer population-health questions with statistical power. Although these datasets have known...

Bayesian Network Modeling Applied to Feline Calicivirus Infection Among Cats in Switzerland.

Bayesian network (BN) modeling is a rich and flexible analytical framework capable of elucidating complex veterinary epidemiological data. It is a graphical modeling technique that enables the visual presentation of multi-dimensional results while retaining statistical rigor in population-level inference. Using previously published case study data about feline calicivirus (FCV) and other respiratory pathogens in cats in Switzerland, a full BN modeling analysis is presented. The analysis shows that reducing the group size and vaccinating animals are the two actionable factors directly associated with FCV status and are primary targets to control FCV infection. The presence of gingivostomatitis and Mycoplasma felis is also associated with FCV status, but signs of upper respiratory tract disease (URTD) are not. FCV data is particularly well-suited to a network modeling approach, as both multiple pathogens and multiple clinical signs per pathogen are involved, along with multiple potentially interrelated risk factors. BN modeling is a holistic approach—all variables of interest may be mutually interdependent—which may help to address issues, such as confounding and collinear factors, as well as to disentangle directly vs. indirectly related variables. We introduce the BN methodology as an alternative to the classical uni- and multivariable regression approaches commonly used for risk factor analyses. We advise and guide researchers about how to use BNs as an exploratory data tool and demonstrate the limitations and practical issues. We present a step-by-step case study using FCV data along with all code necessary to reproduce our analyses in the open-source R environment. We compare and contrast the findings of the current case study using BN modeling with previous results that used classical regression techniques, and we highlight new potential insights. Finally, we discuss advanced methods, such as Bayesian model averaging, a common way of accounting for model uncertainty in a Bayesian network context.

ESTIMATING POPULATION AVERAGE CAUSAL EFFECTS IN THE PRESENCE OF NON-OVERLAP: THE EFFECT OF NATURAL GAS COMPRESSOR STATION EXPOSURE ON CANCER MORTALITY.

Most causal inference studies rely on the assumption of overlap to estimate population or sample average causal effects. When data suffer from non-overlap, estimation of these estimands requires reliance on model specifications, due to poor data support. All existing methods to address non-overlap, such as trimming or down-weighting data in regions of poor data support, change the estimand so that inference cannot be made on the sample or the underlying population. In environmental health research settings, where study results are often intended to influence policy, population-level inference may be critical, and changes in the estimand can diminish the impact of the study results, because estimates may not be representative of effects in the population of interest to policymakers. Researchers may be willing to make additional, minimal modeling assumptions in order to preserve the ability to estimate population average causal effects. We seek to make two contributions on this topic. First, we propose a flexible, data-driven definition of propensity score overlap and non-overlap regions. Second, we develop a novel Bayesian framework to estimate population average causal effects with minor model dependence and appropriately large uncertainties in the presence of non-overlap and causal effect heterogeneity. In this approach, the tasks of estimating causal effects in the overlap and non-overlap regions are delegated to two distinct models, suited to the degree of data support in each region. Tree ensembles are used to non-parametrically estimate individual causal effects in the overlap region, where the data can speak for themselves. In the non-overlap region, where insufficient data support means reliance on model specification is necessary, individual causal effects are estimated by extrapolating trends from the overlap region via a spline model. The promising performance of our method is demonstrated in simulations. Finally, we utilize our method to perform a novel investigation of the causal effect of natural gas compressor station exposure on cancer outcomes. Code and data to implement the method and reproduce all simulations and analyses is available on Github (https://github.com/rachelnethery/overlap).

Using an individual-based model to assess common biases in lek-based count data to estimate population trajectories of lesser prairie-chickens.

Researchers and managers are often interested in monitoring the underlying state of a population (e.g., abundance), yet error in the observation process might mask underlying changes due to imperfect detection and availability for sampling. Additional heterogeneity can be introduced into a monitoring program when male-based surveys are used as an index for the total population. Often, male-based surveys are used for avian species, as males are conspicuous and more easily monitored than females. To determine if male-based lek surveys capture changes or trends in population abundance based on female survival and reproduction, we developed a virtual ecologist approach using the lesser prairie-chicken (Tympanuchus pallidicinctus) as an example. Our approach used an individual-based model to simulate lek counts based on female vital rate data, included models where detection and lek attendance probabilities were <1, and was analyzed using both unadjusted counts and an N-mixture model to compare estimates of population abundance and growth rates. Using lek counts to estimate population growth rates without accounting for detection probability or density-based lek attendance consistently biased population growth rates and abundance estimates. Our results therefore suggest that lek-based surveys used without accounting for lek attendance and detection probability may miss important trends in population changes. Rather than population-level inference, lek-based surveys not accounting for lek attendance and detection probability may instead be better for inferring broad-scale range shifts of lesser prairie-chicken populations in a presence/absence framework.