Is regression adjustment supported by the Neyman model for causal inference?

This paper examines both theoretically and empirically whether the common practice of using OLS multivariate regression models to estimate average treatment effects (ATEs) under experimental designs is justified by the Neyman model for causal inference. Using data from eight large U.S. social policy experiments, the paper finds that estimated standard errors and significance levels for ATE estimators are similar under the OLS and Neyman models when baseline covariates are included in the models, even though theory suggests that this may not have been the case. This occurs primarily because treatment effects do not appear to vary substantially across study subjects.     

Using quantile averages in matched observational studies

In two observational studies, one investigating the effects of minimum wage laws on employment and the other of the effects of exposures to lead, an estimated treatment effect's sensitivity to hidden bias is examined. The estimate uses the combined quantile averages that were introduced in 1981 by B. M. Brown as simple, efficient, robust estimates of location admitting both exact and approximate confidence intervals and significance tests. Closely related to Gastwirth's estimate and Tukey's trimean, the combined quantile average has asymptotic efficiency for normal data that is comparable with that of a 15% trimmed mean, and higher efficiency than the trimean, but it has resistance to extreme observations or breakdown comparable with that of the trimean and better than the 15% trimmed mean. Combined quantile averages provide consistent estimates of an additive treatment effect in a matched randomized experiment. Sensitivity analyses are discussed for combined quantile averages when used in a matched observational study in which treatments are not randomly assigned. In a sensitivity analysis in an observational study, subjects are assumed to differ with respect to an unobserved covariate that was not adequately controlled by the matching, so that treatments are assigned within pairs with probabilities that are unequal and unknown. The sensitivity analysis proposed here uses significance levels, point estimates and confidence intervals based on combined quantile averages and examines how these inferences change under a range of assumptions about biases due to an unobserved covariate. The procedures are applied in the studies of minimum wage laws and exposures to lead. The first example is also used to illustrate sensitivity analysis with an instrumental variable.     

Nonparametric Panel Estimation of Labor Supply

In this article, we estimate structural labor supply with piecewise-linear budgets and nonseparable endogenous unobserved heterogeneity. We propose a two-stage method to address the endogeneity issue that comes from the correlation between the covariates and unobserved heterogeneity. In the first stage, Evdokimov’s nonparametric de-convolution method serves to identify the conditional distribution of unobserved heterogeneity from the quasi-reduced model that uses panel data. In the second stage, the conditional distribution is plugged into the original structural model to estimate labor supply. We apply this methodology to estimate the labor supply of U.S. married men in 2004 and 2005. Our empirical work demonstrates that ignoring the correlation between the covariates and unobserved heterogeneity will bias the estimates of wage elasticities upward. The labor elasticity estimated from a fixed effects model is less than half of that obtained from a random effects model.     

An alternative parameterization of Bayesian logistic hierarchical models for mixed treatment comparisons

Mixed treatment comparison (MTC) models rely on estimates of relative effectiveness from randomized clinical trials so as to respect randomization across treatment arms. This approach could potentially be simplified by an alternative parameterization of the way effectiveness is modeled. We introduce a treatment‐based parameterization of the MTC model that estimates outcomes on both the study and treatment levels. We compare the proposed model to the commonly used MTC models using a simulation study as well as three randomized clinical trial datasets from published systematic reviews comparing (i) treatments on bleeding after cirrhosis, (ii) the impact of antihypertensive drugs in diabetes mellitus, and (iii) smoking cessation strategies. The simulation results suggest similar or sometimes better performance of the treatment‐based MTC model. Moreover, from the real data analyses, little differences were observed on the inference extracted from both models. Overall, our proposed MTC approach performed as good, or better, than the commonly applied indirect and MTC models and is simpler, fast, and easier to implement in standard statistical software. Copyright © 2015 John Wiley & Sons, Ltd.     

The challenges of evaluating dose response in flexible‐dose trials using marginal structural models

Assessing dose response from flexible‐dose clinical trials is problematic. The true dose effect may be obscured and even reversed in observed data because dose is related to both previous and subsequent outcomes. To remove selection bias, we propose marginal structural models, inverse probability of treatment‐weighting (IPTW) methodology. Potential clinical outcomes are compared across dose groups using a marginal structural model (MSM) based on a weighted pooled repeated measures analysis (generalized estimating equations with robust estimates of standard errors), with dose effect represented by current dose and recent dose history, and weights estimated from the data (via logistic regression) and determined as products of (i) inverse probability of receiving dose assignments that were actually received and (ii) inverse probability of remaining on treatment by this time. In simulations, this method led to almost unbiased estimates of true dose effect under various scenarios. Results were compared with those obtained by unweighted analyses and by weighted analyses under various model specifications. The simulation showed that the IPTW MSM methodology is highly sensitive to model misspecification even when weights are known. Practitioners applying MSM should be cautious about the challenges of implementing MSM with real clinical data. Clinical trial data are used to illustrate the methodology. Copyright © 2012 John Wiley & Sons, Ltd.     

Inferences from logistic regression models in the presence of small samples,rare events,nonlinearity, and multicollinearity with observational data

The logistic regression model has been widely used in the social and natural sciences and results from studies using this model can have significant policy impacts. Thus, confidence in the reliability of inferences drawn from these models is essential. The robustness of such inferences is dependent on sample size. The purpose of this article is to examine the impact of alternative data sets on the mean estimated bias and efficiency of parameter estimation and inference for the logistic regression model with observational data. A number of simulations are conducted examining the impact of sample size, nonlinear predictors, and multicollinearity on substantive inferences (e.g. odds ratios, marginal effects) when using logistic regression models. Findings suggest that small sample size can negatively affect the quality of parameter estimates and inferences in the presence of rare events, multicollinearity, and nonlinear predictor functions, but marginal effects estimates are relatively more robust to sample size.     

Errors-in-variables regression using estimated latent variables

This note considers a method for estimating regression parameters from the data containing measurement errors using some natural estimates of the unobserved explanatory variables. It is shown that the resulting estimator is consistent not only in the usual linear regression model but also in the probit model and regression models with censoship or truncation. However, it fails to be consistent in nonlinear regression models except for special cases.     

Errors-in-variables regression using estimated latent variables

This note considers a method for estimating regression parameters from the data containing measurement errors using some natural estimates of the unobserved explanatory variables. It is shown that the resulting estimator is consistent not only in the usual linear regression model but also in the probit model and regression models with censoship or truncation. However, it fails to be consistent in nonlinear regression models except for special cases.     

Testing Inference from Logistic Regression Models in Data with Unobserved Heterogeneity at Cluster Levels

Clustering due to unobserved heterogeneity may seriously impact on inference from binary regression models. We examined the performance of the logistic, and the logistic-normal models for data with such clustering. The total variance of unobserved heterogeneity rather than the level of clustering determines the size of bias of the maximum likelihood (ML) estimator, for the logistic model. Incorrect specification of clustering as level 2, using the logistic-normal model, provides biased estimates of the structural and random parameters, while specifying level 1, provides unbiased estimates for the former, and adequately estimates the latter. The proposed procedure appeals to many research areas.     

Mixed Correlated Bivariate Ordinal and Negative Binomial Longitudinal Responses with Nonignorable Missing Values

Regression models with random effects are proposed for joint analysis of negative binomial and ordinal longitudinal data with nonignorable missing values under fully parametric framework. The presented model simultaneously considers a multivariate probit regression model for the missing mechanisms, which provides the ability of examining the missing data assumptions and a multivariate mixed model for the responses. Random effects are used to take into account the correlation between longitudinal responses of the same individual. A full likelihood-based approach that allows yielding maximum likelihood estimates of the model parameters is used. The model is applied to a medical data, obtained from an observational study on women, where the correlated responses are the ordinal response of osteoporosis of the spine and negative binomial response is the number of joint damage. A sensitivity of the results to the assumptions is also investigated. The effect of some covariates on all responses are investigated simultaneously.     

Multilevel regression mixture analysis

Summary.  A two-level regression mixture model is discussed and contrasted with the conventional two-level regression model. Simulated and real data shed light on the modelling alternatives. The real data analyses investigate gender differences in mathematics achievement from the US National Education Longitudinal Survey. The two-level regression mixture analyses show that unobserved heterogeneity should not be presupposed to exist only at level 2 at the expense of level 1. Both the simulated and the real data analyses show that level 1 heterogeneity in the form of latent classes can be mistaken for level 2 heterogeneity in the form of the random effects that are used in conventional two-level regression analysis. Because of this, mixture models have an important role to play in multilevel regression analyses. Mixture models allow heterogeneity to be investigated more fully, more correctly attributing different portions of the heterogeneity to the different levels.     

Frailty models of manufacturing effects

The median service lifetime of respirator safety devices produced by different manufacturers is determined using frailty models to account for unobserved differences in manufacturing processes and raw materials. The gamma and positive stable frailty distributions are used to obtain survival distribution estimates when the baseline hazard is assumed to be Weibull. Frailty distributions are compared using laboratory test data of the failure times for 104 respirator cartridges produced by 10 different manufacturers tested with three different challenge agents. Likelihood ratio tests indicate that both frailty models provide a significant improvement over a Weibull model assuming independence. Results are compared to fixed effects approaches for analysis of this data.     

Smoothing a Time Series by Segments of the Data Range

We consider the problem of estimating a trend with different amounts of smoothness for segments of a time series subjected to different variability regimes. We propose using an unobserved components model to consider the existence of at least two data segments. We first fix some desired percentages of smoothness for the trend segments and deduce the corresponding smoothing parameters involved. Once the size of each segment is chosen, the smoothing formulas here derived produce trend estimates for all segments with the desired smoothness as well as their corresponding estimated variances. Empirical examples from demography and economics illustrate our proposal.     

A markov regression model for the analysis of the postpartum lactational amenorrhea

Failure time data represent a particular case of binary longitudinal data. The corresponding analysis of the effect of explanatory covariates repeatedly collected over time on the failure rate has been largely facilitated by the Cox semi-parametric regression model. However, neither the interpretation of the estimated parameters associated with time-dependent covariates is straight-forward, nor does this model fully account for the dynamics of the effect of a covariate over time. Markovian regression models appear as complementary tools to address these specific issues from the predictive point of view. We illustrate these aspects using data from the WHO multicenter study, which was designed to analyze the relation between the duration of postpartum lactational amenorrhea and the breastfeeding pattern. One of the main advantage of this approach applied to the field of reproductive epidemiology was to provide a flexible tool, easily and directly understood by clinicians and fieldworkers, for simulating situations, which were still unobserved, and to predict their effects on the duration of amenorrhea.     

Mixture Multi-state Markov Regression Model

Although heterogeneity across individuals may be reduced when a two-state process is extended into a multi-state process, the discrepancy between the observed and the predicted for some states may still exist owing to two possibilities, unobserved mixture distribution in the initial state and the effect of measured covariates on subsequent multi-state disease progression. In the present study, we developed a mixture Markov exponential regression model to take account of the above-mentioned heterogeneity across individuals (subject-to-subject variability) with a systematic model selection based on the likelihood ratio test. The model was successfully demonstrated by an empirical example on surveillance of patients with small hepatocellular carcinoma treated by non-surgical methods. The estimated results suggested that the model with the incorporation of unobserved mixture distribution behaves better than the one without. Complete and partial effects regarding risk factors on different subsequent multi-state transitions were identified using a homogeneous Markov model. The combination of both initial mixture distribution and homogeneous Markov exponential regression model makes a significant contribution to reducing heterogeneity across individuals and over time for disease progression.     

Propensity score applied to survival data analysis through proportional hazards models: a Monte Carlo study

Propensity score methods are increasingly used in medical literature to estimate treatment effect using data from observational studies. Despite many papers on propensity score analysis, few have focused on the analysis of survival data. Even within the framework of the popular proportional hazard model, the choice among marginal, stratified or adjusted models remains unclear. A Monte Carlo simulation study was used to compare the performance of several survival models to estimate both marginal and conditional treatment effects. The impact of accounting or not for pairing when analysing propensity‐score‐matched survival data was assessed. In addition, the influence of unmeasured confounders was investigated. After matching on the propensity score, both marginal and conditional treatment effects could be reliably estimated. Ignoring the paired structure of the data led to an increased test size due to an overestimated variance of the treatment effect. Among the various survival models considered, stratified models systematically showed poorer performance. Omitting a covariate in the propensity score model led to a biased estimation of treatment effect, but replacement of the unmeasured confounder by a correlated one allowed a marked decrease in this bias. Our study showed that propensity scores applied to survival data can lead to unbiased estimation of both marginal and conditional treatment effect, when marginal and adjusted Cox models are used. In all cases, it is necessary to account for pairing when analysing propensity‐score‐matched data, using a robust estimator of the variance. Copyright © 2012 John Wiley & Sons, Ltd.     

Simulation-based model comparison methodology with application to road accident models

This article assumes the goal of proposing a simulation-based theoretical model comparison methodology with application to two time series road accident models. The model comparison exercise helps to quantify the main differences and similarities between the two models and comprises of three main stages: (1) simulation of time series through a true model with predefined properties; (2) estimation of the alternative model using the simulated data; (3) sensitivity analysis to quantify the effect of changes in the true model parameters on alternative model parameter estimates through analysis of variance, ANOVA. The proposed methodology is applied to two time series road accident models: UCM (unobserved components model) and DRAG (Demand for Road Use, Accidents and their Severity). Assuming that the real data-generating process is the UCM, new datasets approximating the road accident data are generated, and DRAG models are estimated using the simulated data. Since these two methodologies are usually assumed to be equivalent, in a sense that both models accurately capture the true effects of the regressors, we are specifically addressing the modeling of the stochastic trend, through the alternative model. Stochastic trend is the time-varying component and is one of the crucial factors in time series road accident data. Theoretically, it can be easily modeled through UCM, given its modeling properties. However, properly capturing the effect of a non-stationary component such as stochastic trend in a stationary explanatory model such as DRAG is challenging. After obtaining the parameter estimates of the alternative model (DRAG), the estimates of both true and alternative models are compared and the differences are quantified through experimental design and ANOVA techniques. It is observed that the effects of the explanatory variables used in the UCM simulation are only partially captured by the respective DRAG coefficients. This a priori, could be due to multicollinearity but the results of both simulation of UCM data and estimating of DRAG models reveal that there is no significant static correlation among regressors. Moreover, in fact, using ANOVA, it is determined that this regression coefficient estimation bias is caused by the presence of the stochastic trend present in the simulated data. Thus, the results of the methodological development suggest that the stochastic component present in the data should be treated accordingly through a preliminary, exploratory data analysis.     

Mixtures of Linear Regression with Measurement Errors

Existing research on mixtures of regression models are limited to directly observed predictors. The estimation of mixtures of regression for measurement error data imposes challenges for statisticians. For linear regression models with measurement error data, the naive ordinary least squares method, which directly substitutes the observed surrogates for the unobserved error-prone variables, yields an inconsistent estimate for the regression coefficients. The same inconsistency also happens to the naive mixtures of regression estimate, which is based on the traditional maximum likelihood estimator and simply ignores the measurement error. To solve this inconsistency, we propose to use the deconvolution method to estimate the mixture likelihood of the observed surrogates. Then our proposed estimate is found by maximizing the estimated mixture likelihood. In addition, a generalized EM algorithm is also developed to find the estimate. The simulation results demonstrate that the proposed estimation procedures work well and perform much better than the naive estimates.     

Nonparametric Discrete Choice Models With Unobserved Heterogeneity

In this research, we provide a new method to estimate discrete choice models with unobserved heterogeneity that can be used with either cross-sectional or panel data. The method imposes nonparametric assumptions on the systematic subutility functions and on the distributions of the unobservable random vectors and the heterogeneity parameter. The estimators are computationally feasible and strongly consistent. We provide an empirical application of the estimator to a model of store format choice. The key insights from the empirical application are: (1) consumer response to cost and distance contains interactions and nonlinear effects, which implies that a model without these effects tends to bias the estimated elasticities and heterogeneity distribution, and (2) the increase in likelihood for adding nonlinearities is similar to the increase in likelihood for adding heterogeneity, and this increase persists as heterogeneity is included in the model.     

How Cox models react to a study-specific confounder in a patient-level pooled dataset: random effects better cope with an imbalanced covariate across trials unless baseline hazards differ

Combining patient-level data from clinical trials can connect rare phenomena with clinical endpoints, but statistical techniques applied to a single trial may become problematical when trials are pooled. Estimating the hazard of a binary variable unevenly distributed across trials showcases a common pooled database issue. We studied how an unevenly distributed binary variable can compromise the integrity of fixed and random effects Cox proportional hazards (cph) models. We compared fixed effect and random effects cph models on a set of simulated datasets inspired by a 17-trial pooled database of patients presenting with ST segment elevation myocardial infarction (STEMI) and non-STEMI undergoing percutaneous coronary intervention. An unevenly distributed covariate can bias hazard ratio estimates, inflate standard errors, raise type I error, and reduce power. While uneveness causes problems for all cph models, random effects suffer least. Compared to fixed effect models, random effects suffer lower bias and trade inflated type I errors for improved power. Contrasting hazard rates between trials prevent accurate estimates from both fixed and random effects models.