BAYESIAN SUBSET SELECTION AND MODEL AVERAGING USING A CENTRED AND DISPERSED PRIOR FOR THE ERROR VARIANCE

This article proposes a new data‐based prior distribution for the error variance in a Gaussian linear regression model, when the model is used for Bayesian variable selection and model averaging. For a given subset of variables in the model, this prior has a mode that is an unbiased estimator of the error variance but is suitably dispersed to make it uninformative relative to the marginal likelihood. The advantage of this empirical Bayes prior for the error variance is that it is centred and dispersed sensibly and avoids the arbitrary specification of hyperparameters. The performance of the new prior is compared to that of a prior proposed previously in the literature using several simulated examples and two loss functions. For each example our paper also reports results for the model that orthogonalizes the predictor variables before performing subset selection. A real example is also investigated. The empirical results suggest that for both the simulated and real data, the performance of the estimators based on the prior proposed in our article compares favourably with that of a prior used previously in the literature.     

International guidelines and the inclusion of disabled people. The Ugandan story

Uganda has made much progress towards including disabled people in its mainstream development, particularly in the political agenda. The exact process by which this has been achieved and the relationship between this and international guidelines and legislation are not known. This study undertakes to examine this from two specific perspectives: (1) How do international documents relevant to disabled people relate to national legislation in Uganda? (2) What can this comparison, together with the perceptions of stakeholders, tell us about how the legislation and services could improve? International legislation and Ugandan legislation is reviewed and compared. Data from 5 semi‐structured key informant interviews and 6 focus group discussions involving a total of 38 people are collected and analysed. The themes arising from the data are related to the documents, legislation, policies and other relevant literature. The results examine the barriers to service provision; the role of change; the importance of representation; policy and legislation issues; and the effects of devolution. Five specific findings relate to how legislation and services can improve: more resources to increase access for disabled people; strong leadership and collaboration between Disabled People’s Organisations (DPOs) funding bodies and governments; awareness raising and training; representation from all impairment groups; and raising the profile of disabled people through further legislation.     

Statistics for weighing benefits and harms in a proposed genetic substudy of a randomized cancer prevention trial

Summary.  When evaluating potential interventions for cancer prevention, it is necessary to compare benefits and harms. With new study designs, new statistical approaches may be needed to facilitate this comparison. A case in point arose in a proposed genetic substudy of a randomized trial of tamoxifen versus placebo in asymptomatic women who were at high risk for breast cancer. Although the randomized trial showed that tamoxifen substantially reduced the risk of breast cancer, the harms from tamoxifen were serious and some were life threaten-ing. In hopes of finding a subset of women with inherited risk genes who derive greater bene-fits from tamoxifen, we proposed a nested case–control study to test some trial subjects for various genes and new statistical methods to extrapolate benefits and harms to the general population. An important design question is whether or not the study should target common low penetrance genes. Our calculations show that useful results are only likely with rare high penetrance genes.     

Factor analysis for ranked data with application to a job selection attitude survey

Summary.  Factor analysis is a powerful tool to identify the common characteristics among a set of variables that are measured on a continuous scale. In the context of factor analysis for non-continuous-type data, most applications are restricted to item response data only. We extend the factor model to accommodate ranked data. The Monte Carlo expectation–maximization algorithm is used for parameter estimation at which the E-step is implemented via the Gibbs sampler. An analysis based on both complete and incomplete ranked data (e.g. rank the top q out of k items) is considered. Estimation of the factor scores is also discussed. The method proposed is applied to analyse a set of incomplete ranked data that were obtained from a survey that was carried out in GuangZhou, a major city in mainland China, to investigate the factors affecting people's attitude towards choosing jobs.     

HEAVY-TAILED-DISTRIBUTED THRESHOLD STOCHASTIC VOLATILITY MODELS IN FINANCIAL TIME SERIES

To capture mean and variance asymmetries and time‐varying volatility in financial time series, we generalize the threshold stochastic volatility (THSV) model and incorporate a heavy‐tailed error distribution. Unlike existing stochastic volatility models, this model simultaneously accounts for uncertainty in the unobserved threshold value and in the time‐delay parameter. Self‐exciting and exogenous threshold variables are considered to investigate the impact of a number of market news variables on volatility changes. Adopting a Bayesian approach, we use Markov chain Monte Carlo methods to estimate all unknown parameters and latent variables. A simulation experiment demonstrates good estimation performance for reasonable sample sizes. In a study of two international financial market indices, we consider two variants of the generalized THSV model, with US market news as the threshold variable. Finally, we compare models using Bayesian forecasting in a value‐at‐risk (VaR) study. The results show that our proposed model can generate more accurate VaR forecasts than can standard models.     

Population constraints on pooled surveys in demographic hazard modeling

In non-experimental research, data on the same population process may be collected simultaneously by more than one instrument. For example, in the present application, two sample surveys and a population birth registration system all collect observations on first births by age and year, while the two surveys additionally collect information on women’s education. To make maximum use of the three data sources, the survey data are pooled and the population data introduced as constraints in a logistic regression equation. Reductions in standard errors about the age and birth-cohort parameters of the regression equation in the order of three-quarters are obtained by introducing the population data as constraints. A halving of the standard errors about the education parameters is achieved by pooling observations from the larger survey dataset with those from the smaller survey. The percentage reduction in the standard errors through imposing population constraints is independent of the total survey sample size.     

Causal inference with generalized structural mean models

Summary.  We estimate cause–effect relationships in empirical research where exposures are not completely controlled, as in observational studies or with patient non-compliance and self-selected treatment switches in randomized clinical trials. Additive and multiplicative structural mean models have proved useful for this but suffer from the classical limitations of linear and log-linear models when accommodating binary data. We propose the generalized structural mean model to overcome these limitations. This is a semiparametric two-stage model which extends the structural mean model to handle non-linear average exposure effects. The first-stage structural model describes the causal effect of received exposure by contrasting the means of observed and potential exposure-free outcomes in exposed subsets of the population. For identification of the structural parameters, a second stage 'nuisance' model is introduced. This takes the form of a classical association model for expected outcomes given observed exposure. Under the model, we derive estimating equations which yield consistent, asymptotically normal and efficient estimators of the structural effects. We examine their robustness to model misspecification and construct robust estimators in the absence of any exposure effect. The double-logistic structural mean model is developed in more detail to estimate the effect of observed exposure on the success of treatment in a randomized controlled blood pressure reduction trial with self-selected non-compliance.     

How Welfare Policies Affect Adolescents' School Outcomes: A Synthesis of Evidence From Experimental Studies

Using data from 8 random assignment studies and employing meta‐analytic techniques, this article provides systematic evidence that welfare and work policies targeted at low‐income parents have small adverse effects on some school outcomes among adolescents ages 12 to 18 years at follow‐up. These adverse effects were observed mostly for school performance outcomes and occurred in programs that required mothers to work or participate in employment‐related activities and those that encouraged mothers to work voluntarily. The most pronounced negative effects on school outcomes occurred for the group of adolescents who had a younger sibling, possibly because of the increased home and sibling care responsibilities they assumed as their mothers increased their employment.     

Foster parenting: A developmental model

This paper proposes a developmental framework for foster parents and outlines four distinct growth stages. Such a framework can be of value to program administrators who are required to assess foster parent development during the crucial matching process. To draw a distinction between each developmental stage, specific instrumental tasks and indicators are outlined.