Using interviewer random effects to remove selection bias from HIV prevalence estimates

Background: Selection bias in HIV prevalence estimates occurs if non-participation in testing is correlated with HIV status. Longitudinal data suggests that individuals who know or suspect they are HIV positive are less likely to participate in testing in HIV surveys, in which case methods to correct for missing data which are based on imputation and observed characteristics will produce biased results. Methods: The identity of the HIV survey interviewer is typically associated with HIV testing participation, but is unlikely to be correlated with HIV status. Interviewer identity can thus be used as a selection variable allowing estimation of Heckman-type selection models. These models produce asymptotically unbiased HIV prevalence estimates, even when non-participation is correlated with unobserved characteristics, such as knowledge of HIV status. We introduce a new random effects method to these selection models which overcomes non-convergence caused by collinearity, small sample bias, and incorrect inference in existing approaches. Our method is easy to implement in standard statistical software, and allows the construction of bootstrapped standard errors which adjust for the fact that the relationship between testing and HIV status is uncertain and needs to be estimated. Results: Using nationally representative data from the Demographic and Health Surveys, we illustrate our approach with new point estimates and confidence intervals (CI) for HIV prevalence among men in Ghana (2003) and Zambia (2007). In Ghana, we find little evidence of selection bias as our selection model gives an HIV prevalence estimate of 1.4% (95% CI 1.2% – 1.6%), compared to 1.6% among those with a valid HIV test. In Zambia, our selection model gives an HIV prevalence estimate of 16.3% (95% CI 11.0% - 18.4%), compared to 12.1% among those with a valid HIV test. Therefore, those who decline to test in Zambia are found to be more likely to be HIV positive. Conclusions: Our approach corrects for selection bias in HIV prevalence estimates, is possible to implement even when HIV prevalence or non-participation is very high or very low, and provides a practical solution to account for both sampling and parameter uncertainty in the estimation of confidence intervals. The wide confidence intervals estimated in an example with high HIV prevalence indicate that it is difficult to correct statistically for the bias that may occur when a large proportion of people refuse to test.

Likelihood estimate of treatment effects under selection bias

We consider methods for estimating causal effects of treatment in the situation where the individuals in the treatment and the control group are self selected, i.e., the selection mechanism is not randomized. In this case, simple comparison of treated and control outcomes will not generally yield valid estimates of casual effects. The propensity score method is frequently used for the evaluation of treatment effect. However, this method is based onsome strong assumptions, which are not directly testable. In this paper, we present an alternative modeling approachto draw causal inference by using share random-effect model and the computational algorithm to draw likelihood based inference with such a model. With small numerical studies and a real data analysis, we show that our approach gives not only more efficient estimates but it is also less sensitive to model misspecifications, which we consider, than the existing methods.

Publication selection bias in minimum-wage research? A meta-regression analysis

Card and Krueger's meta-analysis of the employment effects of minimum wages challenged existing theory. Unfortunately, their meta-analysis confused publication selection with the absence of a genuine empirical effect. We apply recently developed meta-analysis methods to 64 US minimum-wage studies and corroborate that Card and Krueger's findings were nevertheless correct. The minimum-wage effects literature is contaminated by publication selection bias, which we estimate to be slightly larger than the average reported minimum-wage effect. Once this publication selection is corrected, little or no evidence of a negative association between minimum wages and employment remains.

A partial dependence factorial analysis to deal with selection bias in observational studis

This thesis presents a creative and practical approach to dealing with the problem of selection bias. Selection bias may be the most important vexing problem in program evaluation or in any line of research that attempts to assert causality. Some of the greatest minds in economics and statistics have scrutinized the problem of selection bias, with the resulting approaches – Rubin’s Potential Outcome Approach(Rosenbaum and Rubin,1983; Rubin, 1991,2001,2004) or Heckman’s Selection model (Heckman, 1979) – being widely accepted and used as the best fixes. These solutions to the bias that arises in particular from self selection are imperfect, and many researchers, when feasible, reserve their strongest causal inference for data from experimental rather than observational studies. The innovative aspect of this thesis is to propose a data transformation that allows measuring and testing in an automatic and multivariate way the presence of selection bias. The approach involves the construction of a multi-dimensional conditional space of the X matrix in which the bias associated with the treatment assignment has been eliminated. Specifically, we propose the use of a partial dependence analysis of the X-space as a tool for investigating the dependence relationship between a set of observable pre-treatment categorical covariates X and a treatment indicator variable T, in order to obtain a measure of bias according to their dependence structure. The measure of selection bias is then expressed in terms of inertia due to the dependence between X and T that has been eliminated. Given the measure of selection bias, we propose a multivariate test of imbalance in order to check if the detected bias is significant, by using the asymptotical distribution of inertia due to T (Estadella et al. 2005) , and by preserving the multivariate nature of data. Further, we propose the use of a clustering procedure as a tool to find groups of comparable units on which estimate local causal effects, and the use of the multivariate test of imbalance as a stopping rule in choosing the best cluster solution set. The method is non parametric, it does not call for modeling the data, based on some underlying theory or assumption about the selection process, but instead it calls for using the existing variability within the data and letting the data to speak. The idea of proposing this multivariate approach to measure selection bias and test balance comes from the consideration that in applied research all aspects of multivariate balance, not represented in the univariate variable- by-variable summaries, are ignored. The first part contains an introduction to evaluation methods as part of public and private decision process and a review of the literature of evaluation methods. The attention is focused on Rubin Potential Outcome Approach, matching methods, and briefly on Heckman’s Selection Model. The second part focuses on some resulting limitations of conventional methods, with particular attention to the problem of how testing in the correct way balancing. The third part contains the original contribution proposed , a simulation study that allows to check the performance of the method for a given dependence setting and an application to a real data set. Finally, we discuss, conclude and explain our future perspectives.

Impact of random and systematic recall errors and selection bias in case--control studies on mobile phone use and brain tumors in adolescents (CEFALO study)

Whether the use of mobile phones is a risk factor for brain tumors in adolescents is currently being studied. Case--control studies investigating this possible relationship are prone to recall error and selection bias. We assessed the potential impact of random and systematic recall error and selection bias on odds ratios (ORs) by performing simulations based on real data from an ongoing case--control study of mobile phones and brain tumor risk in children and adolescents (CEFALO study). Simulations were conducted for two mobile phone exposure categories: regular and heavy use. Our choice of levels of recall error was guided by a validation study that compared objective network operator data with the self-reported amount of mobile phone use in CEFALO. In our validation study, cases overestimated their number of calls by 9% on average and controls by 34%. Cases also overestimated their duration of calls by 52% on average and controls by 163%. The participation rates in CEFALO were 83% for cases and 71% for controls. In a variety of scenarios, the combined impact of recall error and selection bias on the estimated ORs was complex. These simulations are useful for the interpretation of previous case-control studies on brain tumor and mobile phone use in adults as well as for the interpretation of future studies on adolescents.

Selection bias and the cross-validation of regression models for prediction

Strategies are compared for the development of a linear regression model with stochastic (multivariate normal) regressor variables and the subsequent assessment of its predictive ability. Bias and mean squared error of four estimators of predictive performance are evaluated in simulated samples of 32 population correlation matrices. Models including all of the available predictors are compared with those obtained using selected subsets. The subset selection procedures investigated include two stopping rules, C$\sb{\rm p}$ and S$\sb{\rm p}$, each combined with an 'all possible subsets' or 'forward selection' of variables. The estimators of performance utilized include parametric (MSEP$\sb{\rm m}$) and non-parametric (PRESS) assessments in the entire sample, and two data splitting estimates restricted to a random or balanced (Snee's DUPLEX) 'validation' half sample. The simulations were performed as a designed experiment, with population correlation matrices representing a broad range of data structures.^ The techniques examined for subset selection do not generally result in improved predictions relative to the full model. Approaches using 'forward selection' result in slightly smaller prediction errors and less biased estimators of predictive accuracy than 'all possible subsets' approaches but no differences are detected between the performances of C$\sb{\rm p}$ and S$\sb{\rm p}$. In every case, prediction errors of models obtained by subset selection in either of the half splits exceed those obtained using all predictors and the entire sample.^ Only the random split estimator is conditionally (on $\\beta$) unbiased, however MSEP$\sb{\rm m}$ is unbiased on average and PRESS is nearly so in unselected (fixed form) models. When subset selection techniques are used, MSEP$\sb{\rm m}$ and PRESS always underestimate prediction errors, by as much as 27 percent (on average) in small samples. Despite their bias, the mean squared errors (MSE) of these estimators are at least 30 percent less than that of the unbiased random split estimator. The DUPLEX split estimator suffers from large MSE as well as bias, and seems of little value within the context of stochastic regressor variables.^ To maximize predictive accuracy while retaining a reliable estimate of that accuracy, it is recommended that the entire sample be used for model development, and a leave-one-out statistic (e.g. PRESS) be used for assessment. ^

Sources of selection bias in evaluating social programs: An interpretation of conventional measures and evidence on the effectiveness of matching as a program evaluation method

This paper decomposes the conventional measure of selection bias in observational studies into three components. The first two components are due to differences in the distributions of characteristics between participant and nonparticipant (comparison) group members: the first arises from differences in the supports, and the second from differences in densities over the region of common support. The third component arises from selection bias precisely defined. Using data from a recent social experiment, we find that the component due to selection bias, precisely defined, is smaller than the first two components. However, selection bias still represents a substantial fraction of the experimental impact estimate. The empirical performance of matching methods of program evaluation is also examined. We find that matching based on the propensity score eliminates some but not all of the measured selection bias, with the remaining bias still a substantial fraction of the estimated impact. We find that the support of the distribution of propensity scores for the comparison group is typically only a small portion of the support for the participant group. For values outside the common support, it is impossible to reliably estimate the effect of program participation using matching methods. If the impact of participation depends on the propensity score, as we find in our data, the failure of the common support condition severely limits matching compared with random assignment as an evaluation estimator.

Quantifying Selection Bias from Birth History Estimates of Child Mortality

Thesis (Master's)--University of Washington, 2016-08

Bias from farmer self-selection in genetically modified crop productivity estimates: Evidence from Indian data

In the continuing debate over the impact of genetically modified (GM) crops on farmers of developing countries, it is important to accurately measure magnitudes such as farm-level yield gains from GM crop adoption. Yet most farm-level studies in the literature do not control for farmer self-selection, a potentially important source of bias in such estimates. We use farm-level panel data from Indian cotton farmers to investigate the yield effect of GM insect-resistant cotton. We explicitly take into account the fact that the choice of crop variety is an endogenous variable which might lead to bias from self-selection. A production function is estimated using a fixed-effects model to control for selection bias. Our results show that efficient farmers adopt Bacillus thuringiensis (Bt) cotton at a higher rate than their less efficient peers. This suggests that cross-sectional estimates of the yield effect of Bt cotton, which do not control for self-selection effects, are likely to be biased upwards. However, after controlling for selection bias, we still find that there is a significant positive yield effect from adoption of Bt cotton that more than offsets the additional cost of Bt seed.

Bias from farmer self-selection in genetically modified crop productivity estimates: Evidence from Indian data

In the continuing debate over the impact of genetically modified (GM) crops on farmers of developing countries, it is important to accurately measure magnitudes such as farm-level yield gains from GM crop adoption. Yet most farm-level studies in the literature do not control for farmer self-selection, a potentially important source of bias in such estimates. We use farm-level panel data from Indian cotton farmers to investigate the yield effect of GM insect-resistant cotton. We explicitly take into account the fact that the choice of crop variety is an endogenous variable which might lead to bias from self-selection. A production function is estimated using a fixed-effects model to control for selection bias. Our results show that efficient farmers adopt Bacillus thuringiensis (Bt) cotton at a higher rate than their less efficient peers. This suggests that cross-sectional estimates of the yield effect of Bt cotton, which do not control for self-selection effects, are likely to be biased upwards. However, after controlling for selection bias, we still find that there is a significant positive yield effect from adoption of Bt cotton that more than offsets the additional cost of Bt seed.

Who selects the 'right' directors? An examination of the association between board selection, gender diversity and outcomes

Using a sample of companies from the top 500 listed firms in Australia, we investigate whether the presence of a designated nomination committee and female representation on the nomination committee affect board gender diversity. We also examine whether gender diversity on the board affects firm risk and financial performance. We find that board gender diversity is significantly and positively associated with the presence of a designated nomination committee and that female representation on the nomination committee is a significant explanatory factor of increasing board gender diversity following the release of the 2010 Australian Securities Exchange Corporate Governance Council (ASXCGC) recommendations. Further, our results support the business case for board gender diversity as we find greater gender diversity moderates excessive firm risk which in turn improves firms’ financial performance. Our results are robust after correcting for selection bias and controlling for other board, firm and industry characteristics.