Misclassification Model Research Articles

Does recall bias explain the association of mood disorders with workplace harassment?

PurposeTo determine the contribution of recall bias to the observed excess in mental ill-health in those reporting harassment at work. MethodsA prospective cohort of 1885 workers in welding and electrical trades was contacted every six months for up to 5 years, asking whether they were currently anxious or depressed and whether this was made worse by work. Only at the end of the study did we ask about any workplace harassment they had experienced at work. We elicited sensitivity and specificity of self-reported bullying from published reliability studies and formulated priors that reflect the possibility of over-reporting of workplace harassment (exposure) by those whose anxiety or depression was reported to be made worse by work (cases). We applied the resulting misclassification models to probabilistic bias analysis (PBA) of relative risks. ResultsWe observe that PBA implies that it is unlikely that biased misclassification due to the study subjects' states of mind could have caused the entire observed association. Indeed, the results demonstrated that doubling of risk of anxiety or depression following workplace harassment is plausible, with the unadjusted relative risk attenuated with understated uncertainty. ConclusionsIt seems unlikely that risk of anxiety or depression following workplace harassment can be explained by the form of recall bias that we proposed.

A Bayesian Approach to Estimate Maternal Mortality Globally Using National Civil Registration Vital Statistics Data Accounting for Reporting Errors

Estimation and monitoring of mortality, across multiple population-periods, is vital to improving national and global health outcomes. Reducing maternal mortality is a key part of the Sustainable Development Goals (SGDs), which commit countries and international agencies to monitor progress toward improvement of global maternal health outcomes. Country-specific estimates of maternal mortality are used for monitoring country progress. Civil registration vital statistics (CRVS) data provide valuable information on maternal mortality but are often subject to substantial reporting errors due to misclassification of maternal deaths or incompleteness of CRVS-reporting on deaths to women of reproductive age. Motivated by the challenge of estimating maternal mortality using error-prone CRVS data, this article introduces a Bayesian two-stage approach to obtain country-specific trends in maternal mortality ratios (MMRs), accounting for CRVS-related misclassification errors. In the first stage, we produce country-specific trends of CRVS-related data quality metrics, in terms sensitivity and specificity, through a Bayesian hierarchical misclassification model. In the second stage, we produce data-driven country trends in MMRs and proportion maternal deaths out of all-cause deaths, which are constructed using a Bayesian maternal mortality estimation model, accounting for CRVS-related data quality assessed in Stage 1. We present the approach to estimating maternal mortality and illustrate its use for several country case studies. This approach is used by the United National Maternal Mortality Interagency Group (UN-MMEIG) to produce maternal mortality estimates for global monitoring.

An Assessment of Racial Disparities in Pretrial Decision-Making Using Misclassification Models

An Assessment of Racial Disparities in Pretrial Decision-Making Using Misclassification Models

Adjusting for misclassification of an exposure in an individual participant data meta-analysis.

A common problem in the analysis of multiple data sources, including individual participant data meta-analysis (IPD-MA), is the misclassification of binary variables. Misclassification may lead to biased estimators of model parameters, even when the misclassification is entirely random. We aimed to develop statistical methods that facilitate unbiased estimation of adjusted and unadjusted exposure-outcome associations and between-study heterogeneity in IPD-MA, where the extent and nature of exposure misclassification may vary across studies. We present Bayesian methods that allow misclassification of binary exposure variables to depend on study- and participant-level characteristics. In an example of the differential diagnosis of dengue using two variables, where the gold standard measurement for the exposure variable was unavailable for some studies which only measured a surrogate prone to misclassification, our methods yielded more accurate estimates than analyses naive with regard to misclassification or based on gold standard measurements alone. In a simulation study, the evaluated misclassification model yielded valid estimates of the exposure-outcome association, and was more accurate than analyses restricted to gold standard measurements. Our proposed framework can appropriately account for the presence of binary exposure misclassification in IPD-MA. It requires that some studies supply IPD for the surrogate and gold standard exposure, and allows misclassification to follow a random effects distribution across studies conditional on observed covariates (and outcome). The proposed methods are most beneficial when few large studies that measured the gold standard are available, and when misclassification is frequent.

Rotation group bias and the persistence of misclassification errors in the Current Population Surveys

We develop a general misclassification model to explain the so-called “Rotation Group Bias (RGB)” problem in the Current Population Surveys, where different rotation groups report different labor force statistics. The key insight is that responses to repeated questions in surveys can depend not only on unobserved true values, but also on previous responses to the same questions. Our method provides a framework to understand why unemployment rates in rotation group one are higher than those in other rotation groups in the CPS, without imposing any a priori assumptions on the existence and direction of RGB. Using our method, we provide new estimates of the U.S. unemployment rates, which are much higher than the official series, but lower than previous estimates that ignored persistence in misclassification.

Estimating misclassification errors in the reporting of maternal mortality in national civil registration vital statistics systems: A Bayesian hierarchical bivariate random walk model to estimate sensitivity and specificity for multiple countries and years with missing data.

Civil registration vital statistics (CRVS) systems provide data on maternal mortality that can be used for monitoring trends and to inform policies and programs. However, CRVS maternal mortality data may be subject to substantial reporting errors due to misclassification of maternal deaths. Information on misclassification is available for selected countries and periods only. We developed a Bayesian hierarchical bivariate random walk model to estimate sensitivity and specificity for multiple populations and years and used the model to estimate misclassification errors in the reporting of maternal mortality in CRVS systems. The proposed Bayesian misclassification (BMis) model captures differences in sensitivity and specificity across populations and over time, allows for extrapolations to periods with missing data, and includes an exact likelihood function for data provided in aggregated form. Validation exercises using maternal mortality data suggest that BMis is reasonably well calibrated and improves upon the CRVS‐adjustment approach used until 2018 by the UN Maternal Mortality Inter‐Agency Group (UN‐MMEIG) to account for bias in CRVS data resulting from misclassification error. Since 2019, BMis is used by the UN‐MMEIG to account for misclassification errors when estimating maternal mortality using CRVS data.

Global estimation of finite mixture and misclassification models with an application to multiple equilibria

We show that the identification results of finite mixture and misclassification models are equivalent in a widely used scenario except for an extra ordering assumption. In the misclassification model, an ordering condition is imposed to pin down the precise values of the latent variable, which are also of interest to researchers and need to be identified. In contrast, finite mixture models are usually identified up to permutations of a latent index, which results in local identification. This local identification is satisfactory because the latent index does not convey any economic meaning. However, reaching global identification is important for estimation, especially when researchers use bootstrap to estimate standard errors. This is because standard errors approximated by bootstrap may be incorrect without a global estimator. We demonstrate that games with multiple equilibria fit in our framework and the global estimator with ordering conditions provides more reliable estimates.

Correlated and misclassified binary observations in complex surveys

Misclassifications in binary responses have long been a common problem in medical and health surveys. One way to handle misclassifications in clustered or longitudinal data is to incorporate the misclassification model through the generalized estimating equation (GEE) approach. However, existing methods are developed under a non‐survey setting and cannot be used directly for complex survey data. We propose a pseudo‐GEE method for the analysis of binary survey responses with misclassifications. We focus on cluster sampling and develop analysis strategies for analyzing binary survey responses with different forms of additional information for the misclassification process. The proposed methodology has several attractive features, including simultaneous inferences for both the response model and the association parameters. Finite sample performance of the proposed estimators is evaluated through simulation studies and an application using a real dataset from the Canadian Longitudinal Study on Aging.

Marginal Bayesian Semiparametric Modeling of Mismeasured Multivariate Interval-Censored Data

ABSTRACTMotivated by data gathered in an oral health study, we propose a Bayesian nonparametric approach for population-averaged modeling of correlated time-to-event data, when the responses can only be determined to lie in an interval obtained from a sequence of examination times and the determination of the occurrence of the event is subject to misclassification. The joint model for the true, unobserved time-to-event data is defined semiparametrically; proportional hazards, proportional odds, and accelerated failure time (proportional quantiles) are all fit and compared. The baseline distribution is modeled as a flexible tailfree prior. The joint model is completed by considering a parametric copula function. A general misclassification model is discussed in detail, considering the possibility that different examiners were involved in the assessment of the occurrence of the events for a given subject across time. We provide empirical evidence that the model can be used to estimate the underlying time-to-event distribution and the misclassification parameters without any external information about the latter parameters. We also illustrate the effect on the statistical inferences of neglecting the presence of misclassification. Supplementary materials for this article are available online.

Accounting for false mortality in telemetry tag applications

Deaths of animals in the wild are rarely observed directly, which often limits understanding of survival rates. Telemetry transmitters offer field ecologists the opportunity to observe mortality events in cases as the absence of animal movement. When observations of mortality are based on factors such as the absence of animal movement, live individuals can be mistaken for dead, resulting in biased estimates of survival. Additionally, tag failure or emigration might also influence estimates of survival in telemetry studies. Failing to account for mis-classification, tag failure, and emigration rates can result in overestimates of mortality rates by up two-fold, even when the data are corrected for obviously mistaken entries. We use a multi-state capture–recapture model with a misclassification parameter in estimating both the rate of permanent emigration and/or tag failure and the rate at which individuals are mistakenly identified as dead. We use this method on an annual telemetry survey of three species of native fish in the Murray river, Australia: Murray cod (Maccullochella peelii), trout cod (Maccullochella macquariensis) and golden perch (Macquaria ambigua). Evidence for higher mortality rates in the first year post-implantation occurred for Murray cod and golden perch, which is likely an effect of tagging and/or the transmitter, or transmitters shedding. Using simulations, we confirm that our model approach is robust to a broad range of misclassification and transmitter failure rates. With these simulations we also demonstrate that misclassification models that do not account for emigration will likely be erroneous if live and dead animals have different probabilities of detection. These findings will have a broad interest to ecologists wishing to account for multiple sources of misclassification error in capture-mark-recapture studies, with the caveat that the specifics of the approach are dependent on species, transmitter types and other aspects of experimental design which may or may not be amenable to the misclassification framework.

Two Wrongs Make a Right: Addressing Underreporting in Binary Data from Multiple Sources.

Media-based event data-i.e., data comprised from reporting by media outlets-are widely used in political science research. However, events of interest (e.g., strikes, protests, conflict) are often underreported by these primary and secondary sources, producing incomplete data that risks inconsistency and bias in subsequent analysis. While general strategies exist to help ameliorate this bias, these methods do not make full use of the information often available to researchers. Specifically, much of the event data used in the social sciences is drawn from multiple, overlapping news sources (e.g., Agence France-Presse, Reuters). Therefore, we propose a novel maximum likelihood estimator that corrects for misclassification in data arising from multiple sources. In the most general formulation of our estimator, researchers can specify separate sets of predictors for the true-event model and each of the misclassification models characterizing whether a source fails to report on an event. As such, researchers are able to accurately test theories on both the causes of and reporting on an event of interest. Simulations evidence that our technique regularly out performs current strategies that either neglect misclassification, the unique features of the data-generating process, or both. We also illustrate the utility of this method with a model of repression using the Social Conflict in Africa Database.

Generalized partially linear regression with misclassified data and an application to labour market transitions

Large data sets that originate from administrative or operational activity are increasingly used for statistical analysis as they often contain very precise information and a large number of observations. But there is evidence that some variables can be subject to severe misclassification or contain missing values. Given the size of the data, a flexible semiparametric misclassification model would be good choice but their use in practise is scarce. To close this gap a semiparametric model for the probability of observing labour market transitions is estimated using a sample of 20 m observations from Germany. It is shown that estimated marginal effects of a number of covariates are sizeably affected by misclassification and missing values in the analysis data. The proposed generalized partially linear regression extends existing models by allowing a misclassified discrete covariate to be interacted with a nonparametric function of a continuous covariate.

A Bayesian approach to misclassified binary response: female employment and intimate partner violence in urban India

ABSTRACTWe examine the effect of female employment on the odds of physical spousal violence using a Bayesian misclassification model combined with propensity score regression estimation. While a classical propensity score model finds a significant violence-provoking effect of female employment, our model finds no evidence of a significant effect. This suggests that misleading inferences are caused by falsely small standard errors in a model that does not account for uncertainties around propensity scores. Further, we confirm our misclassification model as a preferred specification using Deviance Information Criterion (DIC).

A flexible AFT model for misclassified clustered interval-censored data.

Motivated by a longitudinal oral health study, we propose a flexible modeling approach for clustered time-to-event data, when the response of interest can only be determined to lie in an interval obtained from asequence of examination times (interval-censored data) and on top of that, the determination of the occurrence of the event is subject to misclassification. The clustered time-to-event data are modeled using an accelerated failure time model with random effects and by assuming a penalized Gaussian mixture model for the random effects terms to avoid restrictive distributional assumptions concerning the event times. A general misclassification model is discussed in detail, considering the possibility that different examiners were involved in the assessment of the occurrence of the events for agiven subject across time. A Bayesian implementation of the proposed model is described in a detailed manner. We additionally provide empirical evidence showing that the model can be used to estimate the underlying time-to-event distribution and the misclassification parameters without any external information about the latter parameters. We also provide results of asimulation study to evaluate the effect of neglecting the presence of misclassification in the analysis of clustered time-to-event data.

An MPEC estimator for misclassification models

In this paper, we propose a constrained maximum likelihood estimator for misclassification models, by formulating the estimation as an MPEC (Mathematical Programming with Equilibrium Constraints) problem. Our approach improves the numerical accuracy and avoids the singularity problem. Monte Carlo simulations confirm that the proposed estimator reduces bias and standard deviation of the estimator, especially when the sample is small/medium and/or the dimension of latent variable is large.

Statistical analysis with missing exposure data measured by proxy respondents: a misclassification problem within a missing-data problem.

In studies of older adults, researchers often recruit proxy respondents, such as relatives or caregivers, when study participants cannot provide self-reports (e.g., because of illness). Proxies are usually only sought to report on behalf of participants with missing self-reports; thus, either a participant self-report or proxy report, but not both, is available for each participant. Furthermore, the missing-data mechanism for participant self-reports is not identifiable and may be nonignorable. When exposures are binary and participant self-reports are conceptualized as the gold standard, substituting error-prone proxy reports for missing participant self-reports may produce biased estimates of outcome means. Researchers can handle this data structure by treating the problem as one of misclassification within the stratum of participants with missing self-reports. Most methods for addressing exposure misclassification require validation data, replicate data, or an assumption of nondifferential misclassification; other methods may result in an exposure misclassification model that is incompatible with the analysis model. We propose a model that makes none of the aforementioned requirements and still preserves model compatibility. Two user-specified tuning parameters encode the exposure misclassification model. Two proposed approaches estimate outcome means standardized for (potentially) high-dimensional covariates using multiple imputation followed by propensity score methods. The first method is parametric and uses maximum likelihood to estimate the exposure misclassification model (i.e., the imputation model) and the propensity score model (i.e., the analysis model); the second method is nonparametric and uses boosted classification and regression trees to estimate both models. We apply both methods to a study of elderly hip fracture patients.

A MPEC Estimator for Misclassification Models

In this paper, we propose a constrained maximum likelihood estimator for misclassification models, by formulating the estimation as an MPEC (Mathematical Programming with Equilibrium Constraints) problem. Our approach improves the numerical accuracy and avoids the singularity problem. Monte Carlo simulations confirm that the proposed estimator reduces bias and standard deviation of the estimator, especially when the sample is small/medium and/or the dimension of matrix is large.

Reference Analysis in a Misclassification Model for the Meiosis I Nondisjunction Fraction in Trisomies

The determination of the meiosis I nondisjunction fraction plays an important role in identifying the characteristics of affected individuals and their mothers, which can generate aneuploidies. The number of individuals with one, two, and three peaks pattern is used to obtain the information; however, the data are susceptible to misclassification. We review the misclassification model previously introduced in the literature which considers a common misclassification error. This paper aims to introduce a joint prior distribution for the meiosis I nondisjunction fraction and the misclassification error. We prove that the reference prior is a proper distribution. We analyze a Brazilian Down syndrome dataset and compare the results with those obtained through Bayes-Laplace and beta prior distributions.

Modeling of Multivariate Monotone Disease Processes in the Presence of Misclassification

Motivated by a longitudinal oral health study, the Signal–Tandmobiel® study, we propose a multivariate binary inhomogeneous Markov model in which unobserved correlated response variables are subject to an unconstrained misclassification process and have a monotone behavior. The multivariate baseline distributions and Markov transition matrices of the unobserved processes are defined as a function of covariates through the specification of compatible full conditional distributions. Distinct misclassification models are discussed. In all cases, the possibility that different examiners were involved in the scoring of the responses of a given subject across time is taken into account. A full Bayesian implementation of the model is described and its performance is evaluated using simulated data. We provide theoretical and empirical evidence that the parameters can be estimated without any external information about the misclassification parameters. Finally, the analyses of the motivating study are presented. Appendices 1–7 are available in the online supplementary materials.

Test, estimation and model comparison for the meiosis I nondisjunction fraction in trisomies

Trisomies are numerical chromosomal anomalies (aneuploidies) which are common causes of mental retardation, pregnancy losses and fetal death. The determination of the meiosis I nondisjunction fraction plays an important role in the identification of possible factors which could generate such aneuploidies. In this article, more flexible misclassification models for the number of peaks in a polymorphic locus of trisomic individuals are considered. They are compared to some others proposed in the literature. Estimation and tests for the nondisjunction fraction in meiosis I and for the misclassification errors are introduced extending previous works. Using the Decision Theory approach, we also build a criterion for making decisions under Jeffreys and Pereira–Stern tests. We apply the results to Down Syndrome data that is the most prevalent trisomy in humans.