Model selection using information criteria under a new estimation method: least squares ratio

In this study, we evaluate several forms of both Akaike-type and Information Complexity (ICOMP)-type information criteria, in the context of selecting an optimal subset least squares ratio (LSR) regression model. Our simulation studies are designed to mimic many characteristics present in real data – heavy tails, multicollinearity, redundant variables, and completely unnecessary variables. Our findings are that LSR in conjunction with one of the ICOMP criteria is very good at selecting the true model. Finally, we apply these methods to the familiar body fat data set.     

Bayesian parametric accelerated failure time spatial model and its application to prostate cancer

Prostate cancer (PrCA) is the most common cancer diagnosed in American men and the second leading cause of death from malignancies. There are large geographical variation and racial disparities existing in the survival rate of PrCA. Much work on the spatial survival model is based on the proportional hazards (PH) model, but few focused on the accelerated failure time (AFT) model. In this paper, we investigate the PrCA data of Louisiana from the Surveillance, Epidemiology, and End Results program and the violation of the PH assumption suggests that the spatial survival model based on the AFT model is more appropriate for this data set. To account for the possible extra-variation, we consider spatially referenced independent or dependent spatial structures. The deviance information criterion is used to select a best-fitting model within the Bayesian frame work. The results from our study indicate that age, race, stage, and geographical distribution are significant in evaluating PrCA survival.     

More than just nickels and dimes: a cross-national analysis of working poverty in affluent democracies

Despite its centrality to contemporary inequality, working poverty is often popularly discussed but rarely studied by sociologists. Using the Luxembourg Income Study (2009), we analyze whether an individual is working poor across 18 affluent democracies circa 2000. We demonstrate that working poverty does not simply mirror overall poverty and that there is greater cross-national variation in working than overall poverty. We then examine four explanations for working poverty: demographic characteristics, economic performance, unified theory, and welfare generosity. We utilize Heckman probit models to jointly model the likelihood of employment and poverty among the employed. Our analyses provide the least support for the economic performance explanation. There is modest support for unified theory as unionization reduces working poverty in some models. However, most of these effects appear to be mediated by welfare generosity. More substantial evidence exists for the demographic characteristics and welfare generosity explanations. An individual's likelihood of being working poor can be explained by (a) a lack of multiple earners or other adults in one's household, low education, single motherhood, having children and youth; and (b) the generosity of the welfare state in which he or she resides. Also, welfare generosity does not undermine employment and reduces working poverty even among demographically vulnerable groups. Ultimately, we encourage a greater role for the welfare state in debates about working poverty.     

Varieties of Ignorance

This paper considers three types and levels of ignorance about the professions: expert, amateur, and professional. The empirical data comprise my own book about the professions, the Wikipedia article on that subject, and the 105 papers that cited my book in 2008. In these three I separate ignorance of facts, of literatures, and of skills, characterizing each type of ignorance by its mix of the three. Amateur ignorance is mostly of skills, and professional ignorance mostly ignorance of collateral literatures. Expert ignorance reflects the use of theory as a mnemonic device and so is particularly insidious.     

Discrimination Hurts: The Academic,Psychological, and Physical Well‐Being of Adolescents

Little is known about the frequency of ethnic or racial discrimination and its implications for Latin American and Asian youths' development. In this study, we examined if there were ethnic and generation differences among 601 12th graders from Latin American (36%), Asian (43%), and European (19%) backgrounds in the frequency of peer, adult, and daily discrimination, and whether discrimination predicted their well‐being. Adolescents from Latin American and Asian backgrounds reported more adult and peer discrimination than their peers from European backgrounds. Latin American youth reported more adult discrimination than their Asian peers. Discrimination predicted lower grade point averages and self‐esteem, and more depressive symptoms, distress, and physical complaints. Ethnic identity, ethnic socialization, and race rejection sensitivity did not moderate the associations between discrimination and well‐being.     

Understanding predictive information criteria for Bayesian models

We review the Akaike, deviance, and Watanabe-Akaike information criteria from a Bayesian perspective, where the goal is to estimate expected out-of-sample-prediction error using a bias-corrected adjustment of within-sample error. We focus on the choices involved in setting up these measures, and we compare them in three simple examples, one theoretical and two applied. The contribution of this paper is to put all these information criteria into a Bayesian predictive context and to better understand, through small examples, how these methods can apply in practice.     

Uniform Tail Asymptotics for the Sum of Two Correlated Classes with Stochastic Returns and Dependent Heavy Tails

Consider a continuous-time risk model with two correlated classes of insurance business and risky investments whose price processes are geometric Lévy processes. By assuming that the correlation comes from a common shock, and the claim sizes are heavy-tailed and pairwise quasi-asymptotically independent, we investigate the tail behavior of the sum of the stochastic present values of the two correlated classes, and a uniform asymptotic formula is obtained.     

Can a treatment be licenced on the basis of post‐treatment predictive biomarkers?

In recent years, there has been a great deal of literature published concerning the identification of predictive biomarkers and indeed, an increasing number of therapies have been licenced on this basis. However, this progress has been made almost exclusively on the basis of biomarkers measured prior to exposure to treatment. There are quite different challenges when the responding population can only be identified on the basis of outcomes observed following exposure to treatment, especially if it represents only a small proportion of patients. The purpose of this paper is to explore whether or when a treatment could be licenced on the basis of post‐treatment predictive biomarkers (PTPB), the focus is on oncology but the concepts should apply to all therapeutic areas. We review the potential pitfalls in hypothesising the presence of a PTPB. We also present challenges in trial design required to confirm and licence on the basis of a PTPB: what's the control population?, could there be a detriment to non‐responders by exposure to the new treatment?, can responders be identified rapidly?, could prior exposure to the new treatment adversely affect performance of the control in responders? Nevertheless, if the patients to be treated could be rapidly identified after prior exposure to treatment, and without harm to non‐responders, in appropriately designed and analysed trials, may be more targeted therapies could be made available to patients. Copyright © 2014 John Wiley & Sons, Ltd.