Search

Well-chosen covariates boost the design sensitivity of individually and cluster-randomized trials. We provide guidance on covariate selection generating an extensive compilation of single- and multilevel design parameters on student achievement. Embedded in psychometric heuristics, we analyzed (a) covariate types of varying bandwidth-fidelity, namely domain-identical (IP), cross-domain (CP), and fluid intelligence (Gf) pretests, as well as sociodemographic characteristics (SC); (b) covariate combinations quantifying incremental validities of CP, Gf, and/or SC beyond IP; and (c) covariate time lags of 1–7 years, testing validity degradation in IP, CP, and Gf. Estimates from six German samples (1868 ≤ N ≤ 10,543) covering various outcome domains across grades 1–12 were meta-analyzed and included in precision simulations. Results varied widely by grade level, domain, and hierarchical level. In general, IP outperformed CP, which slightly outperformed Gf and SC. Benefits from coupling IP with CP, Gf, and/or SC were small. IP appeared most affected by temporal validity decay. Findings are applied in illustrative scenarios of study planning and enriched by comprehensive Online Supplemental Material (OSM) accessible via the Open Science Framework (OSF; ). [ABSTRACT FROM AUTHOR]

Regression calibration as developed by Rosner, Spiegelman, and Willett is used to adjust the bias in effect estimates due to measurement error in continuous exposures. The method involves two models: a measurement error model relating the mismeasured exposure to the true (or gold-standard) exposure and an outcome model relating the mismeasured exposure to the outcome. However, no comprehensive guidance exists for determining which covariates should be included in each model. In this article, we investigate the selection of the minimal and most efficient covariate adjustment sets under a causal inference framework. We show that to address the measurement error, researchers must adjust for, in both measurement error and outcome models, any common causes (1) of true exposure and the outcome and (2) of measurement error and the outcome. We also show that adjusting for so-called prognostic variables that are independent of true exposure and measurement error in the outcome model, may increase efficiency, while adjusting for any covariates that are associated only with true exposure generally results in efficiency loss in realistic settings. We apply the proposed covariate selection approach to the Health Professional Follow-up Study dataset to study the effect of fiber intake on cardiovascular disease. Finally, we extend the originally proposed estimators to a nonparametric setting where effect modification by covariates is allowed. [ABSTRACT FROM AUTHOR]

Understanding causal mechanisms is a central goal in the behavioral, developmental, and social sciences. When estimating and probing causal effects using observational data, covariate adjustment is a crucial element to remove dependencies between focal predictors and the error term. Covariate selection, however, constitutes a challenging task because availability alone is not an adequate criterion to decide whether a covariate should be included in the statistical model. The present study introduces a non-Gaussian method for covariate selection and provides a forward selection algorithm for linear models (i.e., non-Gaussian forward selection; nGFS) to select appropriate covariates from a set of potential control variables to avoid inconsistent and biased estimators of the causal effect of interest. Further, we demonstrate that the forward selection algorithm has properties compatible with principles of direction of dependence, i.e., probing whether the causal target model is correctly specified with respect to the causal direction of effects. Results of a Monte Carlo simulation study suggest that the selection algorithm performs well, in particular when sample sizes are large (i.e., n ≥ 250) and data strongly deviate from Gaussianity (e.g., distributions with skewness beyond 1.5). An empirical example is given for illustrative purposes. [ABSTRACT FROM AUTHOR]

Covariate identification is an important step in the development of a population pharmacokinetic/pharmacodynamic model. Among the different available approaches, the stepwise covariate model (SCM) is the most used. However, SCM is based on a local search strategy, in which the model-building process iteratively tests the addition or elimination of a single covariate at a time given all the others. This introduces a heuristic to limit the searching space and then the computational complexity, but, at the same time, can lead to a suboptimal solution. The application of genetic algorithms (GAs) for covariate selection has been proposed as a possible solution to overcome these limitations. However, their actual use during model building is limited by the extremely high computational costs and convergence issues, both related to the number of models being tested. In this paper, we proposed a new GA for covariate selection to address these challenges. The GA was first developed on a simulated case study where the heuristics introduced to overcome the limitations affecting currently available GA approaches resulted able to limit the selection of redundant covariates, increase replicability of results and reduce convergence times. Then, we tested the proposed GA on a real-world problem related to remifentanil. It obtained good results both in terms of selected covariates and fitness optimization, outperforming the SCM. [ABSTRACT FROM AUTHOR]

In clusterwise regression analysis, the goal is to predict a response variable based on a set of explanatory variables, each with cluster-specific effects. In many real–life problems, the number of candidate predictors is typically large, with perhaps only a few of them meaningfully contributing to the prediction. A well–known method to perform variable selection is the LASSO, with calibration done by minimizing the Bayesian Information Criterion (BIC). However, existing LASSO-penalized estimators are problematic for several reasons. First, only certain types of penalties are considered. Second, the computations may sometimes involve approximate schemes. Third, variable selection is usually time consuming, due to a complex calibration of the penalty term, possibly requiring several multiple evaluations of an estimator for each plausible value of the tuning parameter(s). We introduce a two–step approach to fill these gaps. In step 1, we fit LASSO clusterwise linear regressions with some pre–specified level of penalization (Fit step). In step 2 (Selection step), we perform covariate selection locally, i.e. on the weighted data, with weights corresponding to the posterior probabilities from the previous step. This is done by using a generalization of the Least Angle Regression (LARS) algorithm, which permits covariate selection with a single evaluation of the estimator. In addition, both Fit and Selection steps leverage on an Expectation Maximization (EM) algorithm, fully in closed forms, designed with a very general version of the LASSO penalty. The advantages of our proposal, in terms of computation time reduction, and accuracy of model estimation and selection, are shown by means of a simulation study, and illustrated with a real data application. [ABSTRACT FROM AUTHOR]

Inference for structural and treatment parameters while having high-dimensional covariates in the model is increasingly common. The Neyman-orthogonal (NO) estimators in Belloni, Chernozhukov, and Wei (2016, Journal of Business and Economic Statistics 34: 606–619) produce valid inferences for the parameters of interest while using generalized linear model lasso methods to select the covariates. Drukker and Liu (2022, Econometric Reviews 41: 1047–1076) extended the estimators in Belloni, Chernozhukov, and Wei (2016) by using a Bayesian information criterion stepwise method and a testing-stepwise method as the covariate selector. Drukker and Liu (2022) found a family of data-generating processes for which the NO estimator based on Bayesian information criterion stepwise produces much more reliable inferences than the lasso-based NO estimator. In this article, we describe the implementation of posw, a command for the stepwise-based NO estimator for the high-dimensional linear, logit, and Poisson models. [ABSTRACT FROM AUTHOR]

The performance and transferability of species distribution models (SDMs) depends on a number of ecological, biological, and methodological factors. There is a growing body of literature that explores how the choice of climate covariate combinations and model parameters can affect predictive performance, but relatively few that delve into covariate reduction methods and the optimisation of model parameters, and the resulting spatial and temporal transferability of those models. The present work used the citrus pest, Diaphorina citri Kuwayama (Hemiptera: Psyllidae), to illustrate how MaxEnt models trained on the insect's native range in Asia varied in their predictions of climatic suitability across the introduced range when eight different covariate reduction methods were applied during model building. Additionally, it showed how model sensitivity varied across these different covariate combinations using three sets of independently validated occurrence points in the invaded range of the psyllid. Climatically suitable areas for D. citri differed by as much as two-fold between the best and worst-performing models in selected areas. Great care should be taken in the selection of the highest-performing predictor combinations and model parameter settings for SDMs, particularly in the case of invasive species where the assumption of environmental equilibrium is likely violated in the introduced range. Understanding how the predictive ability of SDMs can be influenced by the methodological choices made during the model building phase is vital to ensuring that ecological and invasion management programmes do not over- or underestimate climatic suitability and subsequent invasion risk. • Climatic suitability and model sensitivity varied across covariate combinations. • Up to a two-fold underestimate of suitable climate reported in worse performing models. • Low model transferability found in the African invaded range of Diaphorina citri. • Tuned and default MaxEnt model parameters were not significantly different. • Pest monitoring and management could be affected by model accuracy. [Display omitted] [ABSTRACT FROM AUTHOR]

Summary: Various statistical methodologies embed a probability distribution in a more flexible family of distributions. The latter is called elaboration model, which is constructed by choice or a formal procedure and evaluated by asymmetric measures such as the likelihood ratio and Kullback–Leibler information. The use of asymmetric measures can be problematic for this purpose. This paper introduces two formal procedures, referred to as link functions, that embed any baseline distribution with a continuous density on the real line into model elaborations. Conditions are given for the link functions to render symmetric Kullback–Leibler divergence, Rényi divergence and phi‐divergence family. The first link function elaborates quantiles of the baseline probability distribution. This approach produces continuous counterparts of the binary probability models. Examples include the Cauchy, probit, logit, Laplace and Student's t links. The second link function elaborates the baseline survival function. Examples include the proportional odds and change point links. The logistic distribution is characterised as the one that satisfies the conditions for both links. An application demonstrates advantages of symmetric divergence measures for assessing the efficacy of covariates. [ABSTRACT FROM AUTHOR]

High-dimensional models that include many covariates which might potentially affect an outcome are increasingly common. This paper begins by introducing a lasso-based approach and a stepwise-based approach to valid inference for a high-dimensional model. It then discusses several essential extensions to the literature that make the estimators more usable in practice. Finally, it presents Monte Carlo evidence to help applied researchers choose which of several available estimators should be used in practice. The Monte Carlo evidence shows that our extensions to the literature perform well. It also shows that a BIC-stepwise approach performs well for a data-generating process for which the lasso-based approaches and a testing-stepwise approach fail. The Monte Carlo evidence also indicates the BIC-based lasso and plugin-based lasso can produce better inferential results than the ubiquitous CV-based lasso. Easy-to-use Stata commands are available for all the methods that we discuss. [ABSTRACT FROM AUTHOR]

Summary: Variable selection for interval‐censored failure time data has recently attracted a great deal of attention along with the analysis of interval‐censored data in both method developments and practical applications. Interval‐censored data are a general type of time‐to‐event or failure time data where the failure time of interest is known or observed only to lie in an interval instead of being observed exactly. They often occur in different forms and in many fields, including demographical studies, epidemiological studies, medical or public health researches and social sciences. This paper will discuss and review the existing methods for variable selection based on interval‐censored data, a relatively new but important topic, along with some directions or issues that need more research. [ABSTRACT FROM AUTHOR]

Background: With many disease-modifying therapies currently approved for the management of multiple sclerosis, there is a growing need to evaluate the comparative effectiveness and safety of those therapies from real-world data sources. Propensity score methods have recently gained popularity in multiple sclerosis research to generate real-world evidence. Recent evidence suggests, however, that the conduct and reporting of propensity score analyses are often suboptimal in multiple sclerosis studies. Objectives: To provide practical guidance to clinicians and researchers on the use of propensity score methods within the context of multiple sclerosis research. Methods: We summarize recommendations on the use of propensity score matching and weighting based on the current methodological literature, and provide examples of good practice. Results: Step-by-step recommendations are presented, starting with covariate selection and propensity score estimation, followed by guidance on the assessment of covariate balance and implementation of propensity score matching and weighting. Finally, we focus on treatment effect estimation and sensitivity analyses. Conclusion: This comprehensive set of recommendations highlights key elements that require careful attention when using propensity score methods. [ABSTRACT FROM AUTHOR]

It is essential for any epidemiologic and clinical investigation to determine the appropriate covariates for which to ascertain measures and subsequently model. A number of recent articles have sought to elucidate covariate selection in the context of data analysis. Unfortunately, few articles characterize covariate selection in the context of data collection and discuss their principles under the assumption that data are measured and available for analyses. Additionally, many articles delineating the appropriate principles use jargon that may be inaccessible to the audiences that need to understand them most. Considering these gaps, this paper first seeks to put forth a simple foundational guide to primary data collection by explaining four sets of covariates for which to ascertain measures: 1) all covariates that cause both the exposure and outcome; 2) selected covariates that cause the exposure; 3) selected covariates that cause the outcome; and 4) relevant sociodemographic and baseline covariates. To the extent possible, this paper attempts to communicate these principles clearly and in the absence of advanced causal inference terminology. Finally, this paper provides a conceptual framework for covariate inclusion and exclusion with respect to data analysis and regression modeling. Specifically, this framework suggests that regression models 1) include all known common cause covariates; 2) include all sociodemographic covariates; 3) exclude any covariate that is known to be both a consequence of the exposure and cause of the outcome; and 4) generally, for every term included in the statistical model, there should be at least 10 observations in the data set. [ABSTRACT FROM AUTHOR]

In randomized trials, adjustment for measured covariates during the analysis can reduce variance and increase power. To avoid misleading inference, the analysis plan must be pre-specified. However, it is often unclear a priori which baseline covariates (if any) should be adjusted for in the analysis. Consider, for example, the Sustainable East Africa Research in Community Health (SEARCH) trial for HIV prevention and treatment. There are 16 matched pairs of communities and many potential adjustment variables, including region, HIV prevalence, male circumcision coverage, and measures of community-level viral load. In this paper, we propose a rigorous procedure to data-adaptively select the adjustment set, which maximizes the efficiency of the analysis. Specifically, we use cross-validation to select from a pre-specified library the candidate targeted maximum likelihood estimator (TMLE) that minimizes the estimated variance. For further gains in precision, we also propose a collaborative procedure for estimating the known exposure mechanism. Our small sample simulations demonstrate the promise of the methodology to maximize study power, while maintaining nominal confidence interval coverage. We show how our procedure can be tailored to the scientific question (intervention effect for the study sample vs. for the target population) and study design (pair-matched or not). Copyright © 2016 John Wiley & Sons, Ltd.

Climate‐induced coral bleaching events are a leading threat to coral reef ecosystems and can result in coral–macroalgal regime shifts that are difficult to reverse. It is unclear how different factors causally influence regime shift or recovery trajectories after a bleaching event. Here, we use structural causal modeling (SCM) and its application of directed acyclic graphs (DAGs) to determine how key factors affect regime shift versus recovery potential across coral reefs in Seychelles, which were severely impacted by bleaching events in 1998 and 2016. Our causal models reveal additional causal drivers of regime shifts, including initial macroalgal cover, wave exposure, and branching coral cover. We also find that reduced depth and structural complexity and increased nutrients increase the likelihood of regime shifting. Further, we use a DAG‐informed predictive model to show how recovering reefs are expected to change after a recent 2016 bleaching event, suggesting that three out of 12 recovering reefs are expected to regime shift given their predisturbance conditions. Collectively, our results provide the first causally grounded analysis of how different factors influence postbleaching regime shift versus recovery potential on coral reefs. More broadly, SCM stands apart from previous observational analysis and provides a strong framework for causal inference across other observational ecological studies. [ABSTRACT FROM AUTHOR]

Abstract Climate‐induced coral bleaching events are a leading threat to coral reef ecosystems and can result in coral–macroalgal regime shifts that are difficult to reverse. It is unclear how different factors causally influence regime shift or recovery trajectories after a bleaching event. Here, we use structural causal modeling (SCM) and its application of directed acyclic graphs (DAGs) to determine how key factors affect regime shift versus recovery potential across coral reefs in Seychelles, which were severely impacted by bleaching events in 1998 and 2016. Our causal models reveal additional causal drivers of regime shifts, including initial macroalgal cover, wave exposure, and branching coral cover. We also find that reduced depth and structural complexity and increased nutrients increase the likelihood of regime shifting. Further, we use a DAG‐informed predictive model to show how recovering reefs are expected to change after a recent 2016 bleaching event, suggesting that three out of 12 recovering reefs are expected to regime shift given their predisturbance conditions. Collectively, our results provide the first causally grounded analysis of how different factors influence postbleaching regime shift versus recovery potential on coral reefs. More broadly, SCM stands apart from previous observational analysis and provides a strong framework for causal inference across other observational ecological studies.

Inferring the causal effect of a treatment on an outcome in an observational study requires adjusting for observed baseline confounders to avoid bias. However, adjusting for all observed baseline covariates, when only a subset are confounders of the effect of interest, is known to yield potentially inefficient and unstable estimators of the treatment effect. Furthermore, it raises the risk of finite‐sample bias and bias due to model misspecification. For these stated reasons, confounder (or covariate) selection is commonly used to determine a subset of the available covariates that is sufficient for confounding adjustment. In this article, we propose a confounder selection strategy that focuses on stable estimation of the treatment effect. In particular, when the propensity score (PS) model already includes covariates that are sufficient to adjust for confounding, then the addition of covariates that are associated with either treatment or outcome alone, but not both, should not systematically change the effect estimator. The proposal, therefore, entails first prioritizing covariates for inclusion in the PS model, then using a change‐in‐estimate approach to select the smallest adjustment set that yields a stable effect estimate. The ability of the proposal to correctly select confounders, and to ensure valid inference of the treatment effect following data‐driven covariate selection, is assessed empirically and compared with existing methods using simulation studies. We demonstrate the procedure using three different publicly available datasets commonly used for causal inference. [ABSTRACT FROM AUTHOR]

Aim: Our aim was to develop predictive statistical models for mapping the abundance of 18 waterfowl species at a pan‐Canadian level. We refined the previous generation of national waterfowl models by (a) developing new, more interpretable statistical models that (b) explicitly account for spatiotemporal variations in waterfowl abundance, while (c) testing for associations with an updated suite of habitat covariates. Location: All of Canada, excluding the Northern Arctic ecozone. Methods: Our response variables were annual species counts on 2,227 aerial‐survey segments over a period of 25 years (1990–2015). Combining machine‐learning and hierarchical regression modelling, we devised an innovative covariate selection strategy to select for each species the best subset of a panel of 232 candidate habitat covariates. With the selected covariates, we implemented hierarchical generalized linear models in a Bayesian framework, using the integrated nested Laplace approximation and stochastic partial differential equation approaches. Results: On average, our models explained 47% of the observed variance for spatiotemporal predictions and 74% for temporally averaged spatial predictions. The 18 species models included 94 significant waterfowl‐habitat associations involving 42 distinct habitat covariates, with an average of 5.3 covariates per model. Covariates for forest attributes were the most represented in our models. The proportional biomass of Populus tremuloides was the most frequently selected covariate (10/94 associations in 10/18 species). Model predictions generated spatial and spatiotemporal maps of species abundances over almost all of Canada. Main conclusions: We showed that it is possible to efficiently combine machine‐learning, variable selection and hierarchical Bayesian methods that exploit high‐dimensional covariate spaces. Our approach yielded powerful and easily interpretable species distribution models with very few covariates, while accounting for residual autocorrelation. Possible applications of the resulting models and maps include the development of biodiversity indicators, the evaluation and execution of conservation planning strategies, and ecosystem services monitoring. [ABSTRACT FROM AUTHOR]

AIC and BIC based on either empirical likelihood (EAIC and EBIC) or Gaussian pseudo-likelihood (GAIC and GBIC) are proposed to select variables in longitudinal data analysis. Their performances are evaluated in the framework of the generalized estimating equations via intensive simulation studies. Our findings are: (i) GAIC and GBIC outperform other existing methods in selecting variables; (ii) EAIC and EBIC are effective in selecting covariates only when the working correlation structure is correctly specified; (iii) GAIC and GBIC perform well regardless the working correlation structure is correctly specified or not. A real dataset is also provided to illustrate the findings. [ABSTRACT FROM AUTHOR]

Summary: To unbiasedly estimate a causal effect on an outcome unconfoundedness is often assumed. If there is sufficient knowledge on the underlying causal structure then existing confounder selection criteria can be used to select subsets of the observed pretreatment covariates, X, sufficient for unconfoundedness, if such subsets exist. Here, estimation of these target subsets is considered when the underlying causal structure is unknown. The proposed method is to model the causal structure by a probabilistic graphical model, for example, a Markov or Bayesian network, estimate this graph from observed data and select the target subsets given the estimated graph. The approach is evaluated by simulation both in a high‐dimensional setting where unconfoundedness holds given X and in a setting where unconfoundedness only holds given subsets of X. Several common target subsets are investigated and the selected subsets are compared with respect to accuracy in estimating the average causal effect. The proposed method is implemented with existing software that can easily handle high‐dimensional data, in terms of large samples and large number of covariates. The results from the simulation study show that, if unconfoundedness holds given X, this approach is very successful in selecting the target subsets, outperforming alternative approaches based on random forests and LASSO, and that the subset estimating the target subset containing all causes of outcome yields smallest MSE in the average causal effect estimation. [ABSTRACT FROM AUTHOR]

Selection of covariates is crucial in the estimation of average treatment effects given observational data with high or even ultra-high dimensional pretreatment variables. Existing methods for this problem typically assume sparse linear models for both outcome and univariate treatment, and cannot handle situations with ultra-high dimensional covariates. In this paper, we propose a new covariate selection strategy called double screening prior adaptive lasso (DSPAL) to select confounders and predictors of the outcome for multivariate treatments, which combines the adaptive lasso method with the marginal conditional (in)dependence prior information to select target covariates, in order to eliminate confounding bias and improve statistical efficiency. The distinctive features of our proposal are that it can be applied to high-dimensional or even ultra-high dimensional covariates for multivariate treatments, and can deal with the cases of both parametric and nonparametric outcome models, which makes it more robust compared to other methods. Our theoretical analyses show that the proposed procedure enjoys the sure screening property, the ranking consistency property and the variable selection consistency. Through a simulation study, we demonstrate that the proposed approach selects all confounders and predictors consistently and estimates the multivariate treatment effects with smaller bias and mean squared error compared to several alternatives under various scenarios. In real data analysis, the method is applied to estimate the causal effect of a three-dimensional continuous environmental treatment on cholesterol level and enlightening results are obtained. [ABSTRACT FROM AUTHOR]

The paper provides a simple test for deciding, from a given causal diagram, whether two sets of variables have the same bias-reducing potential under adjustment. The test requires that one of the following two conditions holds: either (1) both sets are admissible (i.e. satisfy the back-door criterion) or (2) the Markov boundaries surrounding the treatment variable are identical in both sets. We further extend the test to include treatment-dependent covariates by broadening the back-door criterion and establishing equivalence of adjustment under selection bias conditions. Applications to covariate selection and model testing are discussed.

Covariate selection when the number of available variables is large relative to the number of observations is problematic in epidemiology and remains the focus of continued research. Whilst a variety of statistical methods have been developed to attempt to overcome this issue, at present very few methods are available for wide data that include a clustered outcome. The purpose of this research was to make an empirical evaluation of a new method for covariate selection in wide data settings when the dependent variable is clustered. We used 3300 simulated datasets with a variety of defined structures and known sets of true predictor variables to conduct an empirical evaluation of a mixed model stability selection procedure. Comparison was made with an alternative method based on regularisation using the least absolute shrinkage and selection operator (Lasso) penalty. Model performance was assessed using several metrics including the true positive rate (proportion of true covariates selected in a final model) and false discovery rate (proportion of variables selected in a final model that were non-true (false) variables). For stability selection, the false discovery rate was consistently low, generally remaining ≤ 0.02 indicating that on average fewer than 1 in 50 of the variables selected in a final model were false variables. This was in contrast to the Lasso-based method in which the false discovery rate was between 0.59 and 0.72, indicating that generally more than 60% of variables selected in a final model were false variables. In contrast however, the Lasso method attained higher true positive rates than stability selection, although both methods achieved good results. For the Lasso method, true positive rates remained ≥ 0.93 whereas for stability selection the true positive rate was 0.73–0.97. Our results suggest both methods may be of value for covariate selection with high dimensional data with a clustered outcome. When high specificity is needed for identification of true covariates, stability selection appeared to offer the better solution, although with a slight loss of sensitivity. Conversely when high sensitivity is needed, the Lasso approach may be useful, even if accompanied by a substantial loss of specificity. Overall, the results indicated the loss of sensitivity when employing stability selection is relatively small compared to the loss of specificity when using the Lasso and therefore stability selection may provide the better option for the analyst when evaluating data of this type. [ABSTRACT FROM AUTHOR]

Nonparametric additive models are powerful techniques for multivariate data analysis. Although many procedures have been developed for estimating additive components both in mean regression and quantile regression, the problem of selecting relevant components has not been addressed much especially in quantile regression. We present a doubly-penalized estimation procedure for component selection in additive quantile regression models that combines basis function approximation with a ridge-type penalty and a variant of the smoothly clipped absolute deviation penalty. We show that the proposed estimator identifies relevant and irrelevant components consistently and achieves the nonparametric optimal rate of convergence for the relevant components. We also provide an accurate and efficient computation algorithm to implement the estimator and demonstrate its performance through simulation studies. Finally, we illustrate our method via a real data example to identify important body measurements to predict percentage of body fat of an individual. [ABSTRACT FROM AUTHOR]

Adjustment for covariates and specification of the correct covariate set are important issues in the analysis of clinical trials. Edwards (1999) proposes a model selection approach where the model is chosen on the final data set, which remains blinded for treatment group allocation. We investigate this method for time-to-event endpoints and compare its performance to variable selection within an adaptive design. This adaptive design integrates the methods of Schäfer and Müller (2001) and Keiding et al. (1987) and allows variable selection on the unblinded data during an interim analysis. Monte Carlo simulation shows that Edwards’ method—though blinded—outperforms the adaptive method in terms of ability to select the survival relevant covariates and power. The application of the methods is illustrated by a clinical trial example. [ABSTRACT FROM AUTHOR]

ABSTRACT This paper addresses strategies for selecting variables for adjustment in non-experimental comparative effectiveness research and uses causal graphs to illustrate the causal network that relates treatment to outcome. Variables in the causal network take on multiple structural forms. Adjustment for a common cause pathway between treatment and outcome can remove confounding, whereas adjustment for other structural types may increase bias. For this reason, variable selection would ideally be based on an understanding of the causal network; however, the true causal network is rarely known. Therefore, we describe more practical variable selection approaches based on background knowledge when the causal structure is only partially known. These approaches include adjustment for all observed pretreatment variables thought to have some connection to the outcome, all known risk factors for the outcome, and all direct causes of the treatment or the outcome. Empirical approaches, such as forward and backward selection and automatic high-dimensional proxy adjustment, are also discussed. As there is a continuum between knowing and not knowing the causal, structural relations of variables, we recommend addressing variable selection in a practical way that involves a combination of background knowledge and empirical selection and that uses high-dimensional approaches. This empirical approach can be used to select from a set of a priori variables based on the researcher's knowledge to be included in the final analysis or to identify additional variables for consideration. This more limited use of empirically derived variables may reduce confounding while simultaneously reducing the risk of including variables that may increase bias. Copyright © 2013 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

Propensity score matching is a widely-used method to measure the effect of a treatment in social as well as medicine sciences. An important issue in propensity score matching is how to select conditioning variables in estimation of the propensity scores. It is commonly mentioned that variables which affect both program participation and outcomes are selected. Using Monte Carlo simulation, this paper shows that efficiency in estimation of the Average Treatment Effect on the Treated can be gained if all the available observed variables in the outcome equation are included in the estimation of propensity scores. This result still holds in the presence of non-sampling errors in the observed control variables. [ABSTRACT FROM AUTHOR]

We provide necessary and sufficient conditions for effect identification, thereby characterizing the limits to identification. Our results link the nonstructural potential outcome framework for identifying and estimating treatment effects to structural approaches in economics. This permits economic theory to be built into treatment effect methods. We elucidate the sources and consequences of identification failure by examining the biases arising when the necessary conditions fail, and we clarify the relations between unconfoundedness, conditional exogeneity, and the necessary and sufficient identification conditions. A new quantity, the exogeneity score, plays a central role in this analysis, permitting an omitted variable representation for effect biases. This analysis also provides practical guidance for selecting covariates and insight into the price paid for making various identifying assumptions and the benefits gained. [ABSTRACT FROM PUBLISHER]

Abstract: Lindane [γ-hexachlorocyclohexane (γ-HCH)] is an organochlorine pesticide with toxic effects on humans. It is bioaccumulative and can remain in soils for long periods, and although its use for crop spraying was banned in France in 1998, it is possible that residues from before this time remain in the soil. The RMQS soil monitoring network consists of soil samples from 2200 sites on a 16km regular grid across France, collected between 2002 and 2009. We use 726 measurements of the Lindane concentration in these samples to (i) investigate the main explanatory factors for its spatial distribution across France, and (ii) map this distribution. Geostatistics provides an appropriate framework to analyze our spatial dataset, though two issues regarding the data are worth special consideration: first, the harmonization of two subsets of the data (which were analyzed using different measurement processes), and second, the large proportion of data from one of these subsets that fell below a limit of quantification. We deal with these issues using recent methodological developments in geostatistics. Results demonstrate the importance of land use and rainfall for explaining part of the variability of Lindane across France: land use due to the past direct input of Lindane on cropland and its subsequent persistence in the soil, and rainfall due to the re-deposition of volatilized Lindane. Maps show the concentrations to be generally largest in the north and northwest of France, areas of more intensive agricultural land. We also compare levels to some contamination thresholds taken from the literature, and present maps showing the probability of Lindane concentrations exceeding these thresholds across France. These maps could be used as guidelines for deciding which areas require further sampling before some possible remediation strategy could be applied. [Copyright &y& Elsevier]

Abstract: Partially linear regression models with fixed effects are useful tools for making econometric analyses and normalizing microarray data. Baltagi and Li (2002) proposed a computation friendly difference-based series estimation (DSE) for them. We show that the DSE is not asymptotically efficient in most cases and further propose a weighted difference-based series estimation (WDSE). The weights in it do not involve any unknown parameters. The asymptotic properties of the resulting estimators are established for both balanced and unbalanced cases, and it is shown that they achieve a semiparametric efficient boundary. Additionally, we propose a variable selection procedure for identifying significant covariates in the parametric part of the semiparametric fixed-effects regression model. The method is based on a combination of the nonconcave penalization (Fan and Li, 2001 ) and weighted difference-based series estimation techniques. The resulting estimators have the oracle property; that is, they can correctly identify the true model as if the true model (the subset of variables with nonvanishing coefficients) were known in advance. Simulation studies are conducted and an application is given to demonstrate the finite sample performance of the proposed procedures. [ABSTRACT FROM AUTHOR]

Evaluations of agri-environmental schemes (AES) based on observational farm data generally use a matching algorithm for comparing participating and non-participating farms. To mitigate the potential post-matching covariate imbalances between groups resulting from the use of large covariate sets, this paper proposes a method mix that reduces the covariate set and maximises the utilised number of observations. We test the approach on an evaluation of the European Union's AES in the programming period of 2000–2006, estimating the impacts of AES participation on typical measures of land management, i.e. fertiliser and plant protection expenditures and grassland share. We use Mahalanobis distance matching with exact matching on the entry year of the participating farms and kernel matching with automated bandwidth selection to maximise the utilised sample and increase the estimator's efficiency. Combining cause-and-effect path analysis with statistical covariate selection algorithms reduces the covariate set and improves balance on the characteristics that describe the production environment, farming intensity, productivity, and farmers' preferences. We find that AES generate moderate decreases in plant protection expenditure and moderate increases in grassland shares. We conclude that our proposed method mix ensures an efficient use of information and improves the reliability of AES impact evaluation. • Causal relationship between participation in agri-environmental schemes (AES) and farm land management measures. • Method mix for theory informed automated covariate selection to improve post matching covariate balance. • Focus on the role of farmers' environmental and land management prior to AES participation. • Matching on the selected covariate set improves covariate balance and yields robust estimates across matching algorithms. • AES participation generated moderate decreases in plant protection expenditure and moderate increases in grassland shares. [ABSTRACT FROM AUTHOR]

A newly established method for optimizing logistic models via a minorization–majorization procedure is applied to the problem of diagnosing acute coronary syndromes (ACS). The method provides a principled approach to the selection of covariates which would otherwise require the use of a suboptimal method owing to the size of the covariate set. A strategy for building models is proposed and two models optimized for performance and for simplicity are derived via 10-fold cross-validation. These models confirm that a relatively small set of covariates including clinical and electrocardiographic features can be used successfully in this task. The performance of the models is comparable with previously published models using less principled selection methods. The models prove to be portable when tested on data gathered from three other sites. Whilst diagnostic accuracy and calibration diminishes slightly for these new settings, it remains satisfactory overall. The prospect of building predictive models that are as simple as possible for a required level of performance is valuable if data-driven decision aids are to gain wide acceptance in the clinical situation owing to the need to minimize the time taken to gather and enter data at the bedside. [Copyright &y& Elsevier]

This article is concerned with variable selection methods for the Cox proportional hazards regression model. Including excessive covariates causes extra variability and inflated confidence intervals for regression parameters; thus regimes for discarding the less informative ones are needed. Our framework has p covariates designated as "protected," while variables from a further set of q covariates are examined for possible inclusion or exclusion. We develop a focused information criterion (FIC) that for given interest parameter finds the best subset of covariates. Thus the FIC might find that the best model for predicting median survival time is different than the best model for estimating survival probabilities, and the best overall model for analyzing men's survival might not be the same as the best overall model for analyzing women's survival. Methodology is also developed for model averaging, wherein the final estimate of a quantity is a weighted average of estimates computed for a range of submodels. Our methods are illustrated in simulations and for a survival study of Danish skin cancer patients. [ABSTRACT FROM AUTHOR]

Purpose. One of the main objectives of the nonlinear mixed effects modeling is to provide rational individualized dosing strategies by explaining the interindividual variability using intrinsic and/or extrinsic factors (covariates). The aim of the current study was to evaluate, using computer simulations and real data, methods for estimating the exact significance level for including or excluding a covariate during model building. Methods. Original data were simulated using a simple one-compartment pharmacokinetic model with (full model) or without (null model) covariates (one or two). The covariate values in the original data were resampled (using either permutations or parametric bootstrap methods) to generate data under the null hypothesis that there is no covariate effect. The original and permuted data were fitted to null and full models, using first-order and first-order condition estimation (with or without interaction) methods in NONMEM, to compare the asymptotic and conditional p-value. Target log-likelihood ratio cutoffs for assessing covariate effects were derived. Results. The simulations showed that for sparse as well as dense data, the first-order condition estimation methods yielded the best results while the first-order method performs somewhat better for sparse data. Depending on the modeling objective, the appropriate asymptotic p-value can be substituted for the conditional significance level. Target log-likelihood ratio cutoffs should be determined separately for each covariate when exact p-values are important. Conclusions. Resampling methods can be employed to estimate the exact significance level for including a covariate during nonlinear mixed effects model building. Some reasonable inferences can be drawn for potential application to design future population analyses. [ABSTRACT FROM AUTHOR]

Pharmacokinetics is the study of the time course of a drug and its metabolites following its introduction into the body. Population pharmacokinetic studies are becoming increasingly important as an aid to drug development. The data from such studies typically consist of dose histories, drug concentrations with associated sampling times, and often covariate measurements such as the age and weight of each subject. These studies aim to provide an understanding of the pharmacokinetics of the drug in question and so lead to an informed choice of dosage regimen. Such an understanding includes determining those covariates that are important predictors of fundamental pharmacokinetic parameters, such as clearance, defined as the volume of plasma cleared of drug in a unit of time. Determining those subpopulations (e.g., the elderly) with altered kinetics has implications for the choice of an appropriate dosage regimens, because predictive concentration profiles arising from a particular regimen in different populations may be very different. In this article a general Bayesian hierarchical model is described. Pharmacokinetic models relating concentration to time are generally nonlinear, and the data are often sparse and/or noisy. The number of individuals on whom data have been collected is often large, and so the dimensionality of the parameter space is large. Consequently, estimation, from a Bayesian or a classical perspective, is not straightforward. In this article the Hastings-Metropolis algorithm is used for learning about the posterior distribution. An analysis of concentration data collected after the administration of the antiarrhythmic drug quinidine is presented. The data consist of 361 measurements on a total of 136 patients. Nine covariates are also available for each individual. These covariates are a mixture of discrete and continuous measurements. Some of the covariates are constant within an individual during the course of the study, whereas others change. A covariate model is constructed, and the sensitivity of the inferences to distributional assumptions is examined. The importance of assessing the appropriateness of modeling assumptions is emphasized and extensive model checking is carried out for the quinidine data using graphical diagnostics. [ABSTRACT FROM AUTHOR]

Objective This work presents a review of the literature on reporting, practice and misuse of knowledge-based and data-driven variable selection methods, in five highly cited medical journals, considering recoding and interaction unlike previous reviews. Study Design and Setting Original observational studies with a predictive or explicative research question with multivariable analyses published in N. Engl. J. Med., Lancet, JAMA, Br. Med. J. and Ann. Intern. Med. between 2017 and 2019 were searched. Article screening was performed by a single reader, data extraction was performed by two readers and a third reader participated in case of disagreement. The use of data-driven variable selection methods in causal explicative questions was considered as misuse. Results 488 articles were included. The variable selection method was unclear in 234 (48%) articles, data-driven in 78 (16%) articles and knowledge-based in 176 (36%) articles. The most common data-driven methods were: Univariate selection (n = 22, 4.5%) and model comparisons or testing for interaction (n = 17, 3.5%). Data-driven methods were misused in 51 (10.5%) of articles. Conclusion Overall reporting of variable selection methods is insufficient. Data-driven methods seem to be used only in a minority of articles of the big five medical journals., (Copyright © 2021. Published by Elsevier Inc.)