skip to main content


Title: Bayesian finite-population inference with spatially correlated measurements
Abstract

Community-based public health interventions often rely on representative, spatially referenced outcome data to draw conclusions about a finite population. To estimate finite-population parameters, we are posed with two challenges: to correctly account for spatial association among the sampled and nonsampled participants and to correctly model missingness in key covariates, which may be also spatially associated. To accomplish this, we take inspiration from the preferential sampling literature and develop a general Bayesian framework that can specifically account for preferential non-response. This framework is first applied to three missing data scenarios in a simulation study. It is then used to account for missing data patterns seen in reported annual household income in a corner-store intervention project. Through this, we are able to construct finite-population estimates of the percent of income spent on fruits and vegetables. Such a framework provides a flexible way to account for spatial association and complex missing data structures in finite populations.

 
more » « less
NSF-PAR ID:
10372992
Author(s) / Creator(s):
; ; ; ;
Publisher / Repository:
Springer Science + Business Media
Date Published:
Journal Name:
Japanese Journal of Statistics and Data Science
Volume:
5
Issue:
2
ISSN:
2520-8756
Page Range / eLocation ID:
p. 407-430
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract

    We develop a Bayesian model–based approach to finite population estimation accounting for spatial dependence. Our innovation here is a framework that achieves inference for finite population quantities in spatial process settings. A key distinction from the small area estimation setting is that we analyze finite populations referenced by their geographic coordinates. Specifically, we consider a two‐stage sampling design in which the primary units are geographic regions, the secondary units are point‐referenced locations, and the measured values are assumed to be a partial realization of a spatial process. Estimation of finite population quantities from geostatistical models does not account for sampling designs, which can impair inferential performance, whereas design‐based estimates ignore the spatial dependence in the finite population. We demonstrate by using simulation experiments that process‐based finite population sampling models improve model fit and inference over models that fail to account for spatial correlation. Furthermore, the process‐based models offer richer inference with spatially interpolated maps over the entire region. We reinforce these improvements and demonstrate scalable inference for groundwater nitrate levels in the population of California Central Valley wells by offering estimates of mean nitrate levels and their spatially interpolated maps.

     
    more » « less
  2. null (Ed.)
    We examine the uneven social and spatial distributions of COVID-19 and their relationships with indicators of social vulnerability in the U.S. epicenter, New York City (NYC). As of July 17th, 2020, NYC, despite having only 2.5% of the U.S. population, has [Formula: see text]6% of all confirmed cases, and [Formula: see text]16% of all deaths, making it a key learning ground for the social dynamics of the disease. Our analysis focuses on the multiple potential social, economic, and demographic drivers of disproportionate impacts in COVID-19 cases and deaths, as well as population rates of testing. Findings show that immediate impacts of COVID-19 largely fall along lines of race and class. Indicators of poverty, race, disability, language isolation, rent burden, unemployment, lack of health insurance, and housing crowding all significantly drive spatial patterns in prevalence of COVID-19 testing, confirmed cases, death rates, and severity. Income in particular has a consistent negative relationship with rates of death and disease severity. The largest differences in social vulnerability indicators are also driven by populations of people of color, poverty, housing crowding, and rates of disability. Results highlight the need for targeted responses to address injustice of COVID-19 cases and deaths, importance of recovery strategies that account for differential vulnerability, and provide an analytical approach for advancing research to examine potential similar injustice of COVID-19 in other U.S. cities. Significance Statement Communities around the world have variable success in mitigating the social impacts of COVID-19, with many urban areas being hit particularly hard. Analysis of social vulnerability to COVID-19 in the NYC, the U.S. national epicenter, shows strongly disproportionate impacts of the pandemic on low income populations and communities of color. Results highlight the class and racial inequities of the coronavirus pandemic in NYC, and the need to unpack the drivers of social vulnerability. To that aim, we provide a replicable framework for examining patterns of uneven social vulnerability to COVID-19- using publicly available data which can be readily applied in other study regions, especially within the U.S.A. This study is important to inform public and policy debate over strategies for short- and long-term responses that address the injustice of disproportionate impacts of COVID-19. Although similar studies examining social vulnerability and equity dimensions of the COVID-19 outbreak in cities across the U.S. have been conducted (Cordes and Castro 2020, Kim and Bostwick 2002, Gaynor and Wilson 2020; Wang et al. 2020; Choi and Unwin 2020), this study provides a more comprehensive analysis in NYC that extends previous contributions to use the highest resolution spatial units for data aggregation (ZCTAs). We also include mortality and severity rates as key indicators and provide a replicable framework that draws from the Centers for Disease Control and Prevention’s Social Vulnerability indicators for communities in NYC. 
    more » « less
  3. Abstract

    Young planets provide a window into the early stages and evolution of planetary systems. Ideal planets for such research are in coeval associations, where the parent population can precisely determine their ages. We describe a young association (MELANGE-3) in the Kepler field, which harbors two transiting planetary systems (KOI-3876 and Kepler-970). We identify MELANGE-3 by searching for kinematic and spatial overdensities around Kepler planet hosts with high levels of lithium. To determine the age and membership of MELANGE-3, we combine new high-resolution spectra with archival light curves, velocities, and astrometry of stars near KOI-3876 spatially and kinematically. We use the resulting rotation sequence, lithium levels, and color–magnitude diagram of candidate members to confirm the presence of a coeval 105 ± 10 Myr population. MELANGE-3 may be part of the recently identified Theia 316 stream. For the two exoplanet systems, we revise the stellar and planetary parameters, taking into account the newly determined age. Fitting the 4.5 yr Kepler light curves, we find that KOI-3876b is a 2.0 ± 0.1Rplanet on a 19.58 day orbit, while Kepler-970 b is a 2.8 ± 0.2Rplanet on a 16.73 day orbit. KOI-3876 was previously flagged as an eclipsing binary, which we rule out using radial velocities from APOGEE and statistically validate the signal as planetary in origin. Given its overlap with the Kepler field, MELANGE-3 is valuable for studies of spot evolution on year timescales, and both planets contribute to the growing work on transiting planets in young stellar associations.

     
    more » « less
  4. Abstract

    Multi‐view data, which is matched sets of measurements on the same subjects, have become increasingly common with advances in multi‐omics technology. Often, it is of interest to find associations between the views that are related to the intrinsic class memberships. Existing association methods cannot directly incorporate class information, while existing classification methods do not take into account between‐views associations. In this work, we propose a framework for Joint Association and Classification Analysis of multi‐view data (JACA). Our goal is not to merely improve the misclassification rates, but to provide a latent representation of high‐dimensional data that is both relevant for the subtype discrimination and coherent across the views. We motivate the methodology by establishing a connection between canonical correlation analysis and discriminant analysis. We also establish the estimation consistency of JACA in high‐dimensional settings. A distinct advantage of JACA is that it can be applied to the multi‐view data with block‐missing structure, that is to cases where a subset of views or class labels is missing for some subjects. The application of JACA to quantify the associations between RNAseq and miRNA views with respect to consensus molecular subtypes in colorectal cancer data from The Cancer Genome Atlas project leads to improved misclassification rates and stronger found associations compared to existing methods.

     
    more » « less
  5. ABSTRACT

    Longitudinal studies are often subject to missing data. The recent guidance from regulatory agencies, such as the ICH E9(R1) addendum addresses the importance of defining a treatment effect estimand with the consideration of intercurrent events. Jump-to-reference (J2R) is one classical control-based scenario for the treatment effect evaluation, where the participants in the treatment group after intercurrent events are assumed to have the same disease progress as those with identical covariates in the control group. We establish new estimators to assess the average treatment effect based on a proposed potential outcomes framework under J2R. Various identification formulas are constructed, motivating estimators that rely on different parts of the observed data distribution. Moreover, we obtain a novel estimator inspired by the efficient influence function, with multiple robustness in the sense that it achieves n1/2-consistency if any pairs of multiple nuisance functions are correctly specified, or if the nuisance functions converge at a rate not slower than n−1/4 when using flexible modeling approaches. The finite-sample performance of the proposed estimators is validated in simulation studies and an antidepressant clinical trial.

     
    more » « less