A house price index defined in the potential outcomes framework

Current methods for constructing house price indices are based on comparisons of sale prices of residential properties sold two or more times and on regression of the sale prices on the attributes of the properties and of their locations. The two methods have well recognised deficiencies, selection bias and model assumptions, respectively. We introduce a new method based on propensity score matching. The average house prices for two periods are compared by selecting pairs of properties, one sold in each period, that are as similar on a set of available attributes (covariates) as is feasible to arrange. The uncertainty associated with such matching is addressed by multiple imputation, framing the problem as involving missing values. The method is applied to aregister of transactions ofresidential properties in New Zealand and compared with the established alternatives.

Introducing a first AIDS vaccine in Switzerland: a Policy Delphi analysis.

This preliminary exploration was limited by a number of factors. The format of the study has necessarily induced some form of selection bias of the panelists, because of the complexity of some questions, and the time required to complete the questionnaires. Several issues have not been addressed. One example could be the response to HIV infection occurring in a vaccinee. The study also did not address the difficulties related to the licensing of the vaccine. Indeed, the proposed scenario assumed that the vaccine had been registered as a starting point for the analysis. Finally, it has not been possible to conduct a sensitivity analysis, in order to evaluate how the responses would have been modified if some important characteristics of the vaccine had been modified.Very diverse evaluations were given in response to questions related with attitudes and perception of AIDS and AIDS vaccine. The possibility that vaccine availability or usage can be associated with an increased frequency in risky behaviors was spontaneously mentioned by half of the panelists. The estimation of the proportion of persons at highest risk who would choose to use this vaccine also indicated a high degree of uncertainty. This study offers important lessons. According to a broad and diverse panel of individuals, an incompletely effective AIDS vaccine would result in an additional level of complexity for the AIDS prevention strategy, rather than a simplification. The use of such a vaccine would have to be coupled with counselling. This implies a sustained emphasis on the recommendations which have been central to the STOP AIDS campaigns until now. In addition, consensual issues, as well as other issues more likely to be controversial have been identified. This should greatly help focusing the work of any committee designated to develop and implement a vaccination policy if an AIDS vaccine became available. Finally, our experience with the Policy Delphi indicates that this mode of structured communication could be usefully applied to other public health issues presenting a high visibility as well as a complex relationship with public perception.

Estimating uncertainty of alcohol-attributable fractions for infectious and chronic diseases.

Background: Alcohol is a major risk factor for burden of disease and injuries globally. This paper presents a systematic method to compute the 95% confidence intervals of alcohol-attributable fractions (AAFs) with exposure and risk relations stemming from different sources.Methods: The computation was based on previous work done on modelling drinking prevalence using the gamma distribution and the inherent properties of this distribution. The Monte Carlo approach was applied to derive the variance for each AAF by generating random sets of all the parameters. A large number of random samples were thus created for each AAF to estimate variances. The derivation of the distributions of the different parameters is presented as well as sensitivity analyses which give an estimation of the number of samples required to determine the variance with predetermined precision, and to determine which parameter had the most impact on the variance of the AAFs.Results: The analysis of the five Asian regions showed that 150 000 samples gave a sufficiently accurate estimation of the 95% confidence intervals for each disease. The relative risk functions accounted for most of the variance in the majority of cases.Conclusions: Within reasonable computation time, the method yielded very accurate values for variances of AAFs.

The timing of foreign direct investment under uncertainty: Evidence from the Spanish banking sector

This paper investigates the timing of foreign direct investment (FDI) in the banking sector. The importance of this issue would arise from the existence of differential benefits associated to be the first entrant in a foreign location. Nevertheless, when uncertainty is considered, the existence of some Ownership-Location-Internalization (OLI) advantages can make FDI less reversible and/or more delayable and therefore it may be optimal for the firm to delay the investment until the uncertainty is resolved. In this paper, the nature of OLI advantages in the banking sector has been examined in order to propose a prognostic model of the timing of foreign direct investment. The model is then tested for the Spanish case using duration analysis.

Examining non-response bias in substance use research-Are late respondents proxies for non-respondents?

BACKGROUND: Non-response is a major concern among substance use epidemiologists. When differences exist between respondents and non-respondents, survey estimates may be biased. Therefore, researchers have developed time-consuming strategies to convert non-respondents to respondents. The present study examines whether late respondents (converted former non-participants) differ from early respondents, non-consenters or silent refusers (consent givers but non-participants) in a cohort study, and whether non-response bias can be reduced by converting former non-respondents. METHODS: 6099 French- and 5720 German-speaking Swiss 20-year-old males (more than 94% of the source population) completed a short questionnaire on substance use outcomes and socio-demographics, independent of any further participation in a cohort study. Early respondents were those participating in the cohort study after standard recruitment procedures. Late respondents were non-respondents that were converted through individual encouraging telephone contact. Early respondents, non-consenters and silent refusers were compared to late respondents using logistic regressions. Relative non-response biases for early respondents only, for respondents only (early and late) and for consenters (respondents and silent refusers) were also computed. RESULTS: Late respondents showed generally higher patterns of substance use than did early respondents, but lower patterns than did non-consenters and silent refusers. Converting initial non-respondents to respondents reduced the non-response bias, which might be further reduced if silent refusers were converted to respondents. CONCLUSION: Efforts to convert refusers are effective in reducing non-response bias. However, converted late respondents cannot be seen as proxies of non-respondents, and are at best only indicative of existing response bias due to persistent non-respondents.

Random measurement error and regression dilution bias.

Summary points: - The bias introduced by random measurement error will be different depending on whether the error is in an exposure variable (risk factor) or outcome variable (disease) - Random measurement error in an exposure variable will bias the estimates of regression slope coefficients towards the null - Random measurement error in an outcome variable will instead increase the standard error of the estimates and widen the corresponding confidence intervals, making results less likely to be statistically significant - Increasing sample size will help minimise the impact of measurement error in an outcome variable but will only make estimates more precisely wrong when the error is in an exposure variable

Bias related to body mass index in pediatric echocardiographic z scores.

In pediatric echocardiography, cardiac dimensions are often normalized for weight, height, or body surface area (BSA). The combined influence of height and weight on cardiac size is complex and likely varies with age. We hypothesized that increasing weight for height, as represented by body mass index (BMI) adjusted for age, is poorly accounted for in Z scores normalized for weight, height, or BSA. We aimed to evaluate whether a bias related to BMI was introduced when proximal aorta diameter Z scores are derived from bivariate models (only one normalizing variable), and whether such a bias was reduced when multivariable models are used. We analyzed 1,422 echocardiograms read as normal in children ≤18 years. We computed Z scores of the proximal aorta using allometric, polynomial, and multivariable models with four body size variables. We then assessed the level of residual association of Z scores and BMI adjusted for age and sex. In children ≥6 years, we found a significant residual linear association with BMI-for-age and Z scores for most regression models. Only a multivariable model including weight and height as independent predictors produced a Z score free of linear association with BMI. We concluded that a bias related to BMI was present in Z scores of proximal aorta diameter when normalization was done using bivariate models, regardless of the regression model or the normalizing variable. The use of multivariable models with weight and height as independent predictors should be explored to reduce this potential pitfall when pediatric echocardiography reference values are evaluated.

Batch effect confounding leads to strong bias in performance estimates obtained by cross-validation.

BACKGROUND: With the large amount of biological data that is currently publicly available, many investigators combine multiple data sets to increase the sample size and potentially also the power of their analyses. However, technical differences ("batch effects") as well as differences in sample composition between the data sets may significantly affect the ability to draw generalizable conclusions from such studies. FOCUS: The current study focuses on the construction of classifiers, and the use of cross-validation to estimate their performance. In particular, we investigate the impact of batch effects and differences in sample composition between batches on the accuracy of the classification performance estimate obtained via cross-validation. The focus on estimation bias is a main difference compared to previous studies, which have mostly focused on the predictive performance and how it relates to the presence of batch effects. DATA: We work on simulated data sets. To have realistic intensity distributions, we use real gene expression data as the basis for our simulation. Random samples from this expression matrix are selected and assigned to group 1 (e.g., 'control') or group 2 (e.g., 'treated'). We introduce batch effects and select some features to be differentially expressed between the two groups. We consider several scenarios for our study, most importantly different levels of confounding between groups and batch effects. METHODS: We focus on well-known classifiers: logistic regression, Support Vector Machines (SVM), k-nearest neighbors (kNN) and Random Forests (RF). Feature selection is performed with the Wilcoxon test or the lasso. Parameter tuning and feature selection, as well as the estimation of the prediction performance of each classifier, is performed within a nested cross-validation scheme. The estimated classification performance is then compared to what is obtained when applying the classifier to independent data.

Measuring risk aversion with lists: A new bias

Various experimental procedures aimed at measuring individual risk aversion involve alist of pairs of alternative prospects. We first study the widely used method by Holt andLaury (2002), for which we find that the removal of some items from the lists yields asystematic decrease in risk aversion. This bias is quite distinct from other confoundsthat have been previously observed in the use of the Holt and Laury method. It may berelated to empirical phenomena and theoretical developments where better prospectsincrease risk aversion. Nevertheless, we have also found that the more recent elicitationmethod due to Abdellaoui et al. (2011), also based on lists, does not display anystatistically significant bias when the corresponding items of the list are removed. Ourresults suggest that methods other than the popular Holt and Laury one may bepreferable for the measurement of risk aversion.

Uncertainty Quantification with Support Vector Regression Prediction Models

Uncertainty quantification of petroleum reservoir models is one of the present challenges, which is usually approached with a wide range of geostatistical tools linked with statistical optimisation or/and inference algorithms. The paper considers a data driven approach in modelling uncertainty in spatial predictions. Proposed semi-supervised Support Vector Regression (SVR) model has demonstrated its capability to represent realistic features and describe stochastic variability and non-uniqueness of spatial properties. It is able to capture and preserve key spatial dependencies such as connectivity, which is often difficult to achieve with two-point geostatistical models. Semi-supervised SVR is designed to integrate various kinds of conditioning data and learn dependences from them. A stochastic semi-supervised SVR model is integrated into a Bayesian framework to quantify uncertainty with multiple models fitted to dynamic observations. The developed approach is illustrated with a reservoir case study. The resulting probabilistic production forecasts are described by uncertainty envelopes.

Systematic review of the empirical evidence of study publication bias and outcome reporting bias - an updated review.

BACKGROUND: The increased use of meta-analysis in systematic reviews of healthcare interventions has highlighted several types of bias that can arise during the completion of a randomised controlled trial. Study publication bias and outcome reporting bias have been recognised as a potential threat to the validity of meta-analysis and can make the readily available evidence unreliable for decision making. METHODOLOGY/PRINCIPAL FINDINGS: In this update, we review and summarise the evidence from cohort studies that have assessed study publication bias or outcome reporting bias in randomised controlled trials. Twenty studies were eligible of which four were newly identified in this update. Only two followed the cohort all the way through from protocol approval to information regarding publication of outcomes. Fifteen of the studies investigated study publication bias and five investigated outcome reporting bias. Three studies have found that statistically significant outcomes had a higher odds of being fully reported compared to non-significant outcomes (range of odds ratios: 2.2 to 4.7). In comparing trial publications to protocols, we found that 40-62% of studies had at least one primary outcome that was changed, introduced, or omitted. We decided not to undertake meta-analysis due to the differences between studies. CONCLUSIONS: This update does not change the conclusions of the review in which 16 studies were included. Direct empirical evidence for the existence of study publication bias and outcome reporting bias is shown. There is strong evidence of an association between significant results and publication; studies that report positive or significant results are more likely to be published and outcomes that are statistically significant have higher odds of being fully reported. Publications have been found to be inconsistent with their protocols. Researchers need to be aware of the problems of both types of bias and efforts should be concentrated on improving the reporting of trials.

Using graphical probability analysis (Bayes Nets) to evaluate a conditional DNA inclusion

This paper discusses the analysis of cases in which the inclusion or exclusion of a particular suspect, as a possible contributor to a DNA mixture, depends on the value of a variable (the number of contributors) that cannot be determined with certainty. It offers alternative ways to deal with such cases, including sensitivity analysis and object-oriented Bayesian networks, that separate uncertainty about the inclusion of the suspect from uncertainty about other variables. The paper presents a case study in which the value of DNA evidence varies radically depending on the number of contributors to a DNA mixture: if there are two contributors, the suspect is excluded; if there are three or more, the suspect is included; but the number of contributors cannot be determined with certainty. It shows how an object-oriented Bayesian network can accommodate and integrate varying perspectives on the unknown variable and how it can reduce the potential for bias by directing attention to relevant considerations and distinguishing different sources of uncertainty. It also discusses the challenge of presenting such evidence to lay audiences.