Conditional vs marginal estimation of the predictive loss of hierarchical models using WAIC and cross-validation

The predictive loss of Bayesian models can be estimated using a sample from the full-data posterior by evaluating the Watanabe-Akaike information criterion (WAIC) or using an importance sampling (ISCVL) approximation to leave-one-out cross-validation loss. With hierarchical models the loss can be specified at different levels of the hierarchy, and in the published literature, it is routine for these estimators to use the conditional likelihood provided by the lowest level of model hierarchy. However, the regularity conditions underlying these estimators may not hold at this level, and the behaviour of conditional-level WAIC as an estimator of conditional-level predictive loss must be determined on a case-by-case basis. Conditional-level ISCVL does not target conditional-level predictive loss and instead is an estimator of marginal-level predictive loss. Using examples for analysis of over-dispersed count data, it is shown that conditional-level WAIC does not provide a reliable estimator of its target loss, and simulations show that it can favour the incorrect model. Moreover, conditional-level ISCVL is numerically unstable compared to marginal-level ISCVL. It is recommended that WAIC and ISCVL be evaluated using the marginalized likelihood where practicable and that the reliability of these estimators always be checked using appropriate diagnostics.     

Dog-Women and She-Devils: The Queering Field of Monstrous Women

Contemporary British writers Angela Carter, Jeanette Winterson, and Fay Weldon all employ representations of excessive and powerful female characters in their fiction. Although these often monstrous characterizations may be read as posing an onslaught to the notion of the subject, such representations ultimately allow for greater complexity in exploring possibilities of sexual subjectivity and identifications. All three authors craft their characters through a strongly-invested materialism, one derived through Bakhtinian excesses, pornography, and S/M. This essay investigates how these monstrous bodies model both feminist and queer applications, and how the narratives open spaces for a queer configuration of female heterosexuality.     

Visualizing the landscape and evolution of leadership research

Complementing prior quantitative and qualitative reviews of the leadership literature, we conduct a bibliometric analysis of leadership articles. Our bibliometric review provides a different perspective by portraying the landscape and developmental trajectory of leadership research over time via co-citation and co-occurrence analyses. Using a scientific visualization tool CiteSpace and 6528 leadership works collected from the Web of Science database from 1990 to 2017, we detect and visualize the landscape of leadership research and track how this landscape has evolved. After mapping the landscape, we discuss the insights gleaned from our bibliometric review, with a focus on open questions, future research directions, and implications. In doing so, our review provides readers with a systematic understanding of the development of the leadership field and a roadmap to spark leadership research and move this literature forward.     

A defence of subjective fiducial inference

This paper defends the fiducial argument. In particular, an interpretation of the fiducial argument is defended in which fiducial probability is treated as being subjective and the role taken by pivots in a more standard interpretation is taken by what are called primary random variables, which in fact form a special class of pivots. The resulting methodology, which is referred to as subjective fiducial inference, is outlined in the first part of the paper. This is followed by a defence of this methodology arranged in a series of criticisms and responses. These criticisms reflect objections that are often raised against standard fiducial inference and incorporate more specific concerns that are likely to exist with respect to subjective fiducial inference. It is hoped that the responses to these criticisms clarify the contribution that a system of fiducial reasoning can make to statistical inference.     

Mental Health and Online,Land-Based and Mixed Gamblers

Although high rates of problem gambling have been identified among Internet gamblers, most studies have failed to identify the relative contribution of multiple forms of gambling as opposed to the exclusive participation in online forms. The aim of this study was to investigate differences in mental health status in exclusive online, exclusive land-based, and mixed Internet and land-based samples of gamblers drawn from the general population. A sample of 4594 respondents completing an online survey were categorised as exclusive online, land-based and mixed form gamblers. Participants completed a questionnaire eliciting demographic details, participation on all forms of gambling, use of alcohol, tobacco and drugs, help-seeking, and personal problems experienced due to gambling, as well as measures of problem gambling and psychological distress. Findings indicated that mixed gamblers exhibited higher problem gambling scores, level of gambling involvement, and consumption of alcohol during gambling than exclusive online gamblers. Land-based gamblers experienced higher levels of psychological distress, self-acknowledged need for treatment, and help-seeking behaviour. These findings suggest that exclusive online gamblers represent a different subpopulation at lower risk of harm compared to gamblers engaging in multiple forms. Understanding the characteristics of different problem gambling subpopulations may inform the development of more effective targeted interventions.     

The Public Stigma of Problem Gambling: Its Nature and Relative Intensity Compared to Other Health Conditions

Problem gambling attracts considerable public stigma, with deleterious effects on mental health and use of healthcare services amongst those affected. However, no research has examined the extent of stigma towards problem gambling within the general population. This study aimed to examine the stigma-related dimensions of problem gambling as perceived by the general public compared to other health conditions, and determine whether the publicly perceived dimensions of problem gambling predict its stigmatisation. A sample of 2000 Australian adults was surveyed, weighted to be representative of the state population by gender, age and location. Based on vignettes, the online survey measured perceived origin, peril, concealability, course and disruptiveness of problem gambling and four other health conditions, and desired social distance from each. Problem gambling was perceived as caused mainly by stressful life circumstances, and highly disruptive, recoverable and noticeable, but not particularly perilous. Respondents stigmatised problem gambling more than sub-clinical distress and recreational gambling, but less than alcohol use disorder and schizophrenia. Predictors of stronger stigma towards problem gambling were perceptions it is caused by bad character, is perilous, non-recoverable, disruptive and noticeable, but not due to stressful life circumstances, genetic/inherited problem, or chemical imbalance in the brain. This new foundational knowledge can advance understanding and reduction of problem gambling stigma through countering inaccurate perceptions that problem gambling is caused by bad character, that people with gambling problems are likely to be violent to other people, and that people cannot recover from problem gambling.     

Managing chronic conditions in college: Findings from prompted health incidents diaries

Objective: This pilot study assessed an electronic health diary method designed to collect data about critical health incidents experienced by college students who have chronic health conditions. Participants: Nine university students with chronic medical conditions were recruited to complete a series of e-mail–based surveys, sent once every 3 days across the fall 2014 semester. Methods: In each survey, participants described a health-related incident that occurred within the past day and cited resources that helped or could have helped in that situation. They completed follow-up interviews and ranked the importance of cited resources. Results: The diary completion rate was 78.3% (141/180). Most frequently affected management areas were activities (61.3%), monitoring (34.9%), and problem-solving (34.3%). Resources considered helpful included situational knowledge, campus health professionals, peer support, and relaxation opportunities. Conclusions: Prompted health incidents diary method achieved a high completion rate and provided data that could be useful for college health researchers and practitioners.     

The Intergenerational Effect of Cambodia's Genocide on Children's Education and Health

We investigate the intergenerational impact of conflict on the educational and health outcomes of children born years after the conflict ended by exploiting geographical variation in the intensity of the genocide that occurred during the Khmer Rouge (KR) regime in Cambodia. We find that children of individuals who were of prime marriage age during the genocide and experienced greater intensity of genocide have worse educational and health outcomes. In particular, for each standard deviation increase in the intensity of the genocide, average children's normal grade progression rate decreases by 0.03 standard deviations and average children's height‐for‐ age Z‐score decreases by 0.06 standard deviations. We examine several channels through which genocide could affect children born to survivors after the conflict and find suggestive evidence that the marriage market acts as a channel that transmits the adverse impact of conflict across generations. Our findings are robust to alternative measures of mortality rates and post‐KR internal migration.     

The Interaction Between Child Respiratory Sinus Arrhythmia and Early Sensitive Parenting in the Prediction of Children's Executive Functions

This study investigated the interaction between children's parasympathetic functioning and maternal sensitive parenting behaviors during infancy and toddlerhood in the prediction of children's executive functions (EF) at the age of 5 years. Participants included 137 children and their mothers who were followed from the age of 3 months to 5 years. Children's cardiac activity was recorded at rest at multiple times from ages 3 to 36 months, and estimates of respiratory sinus arrhythmia (RSA; a measure of parasympathetic functioning) were calculated. Sensitive parenting was assessed during a mother–child play task at ages 6, 12, 24, and 36 months, and 5 years. Children completed age appropriate EF tasks at the age of 5 years. The link between sensitive parenting during toddlerhood (ages 24 and 36 months) and children's later EF was moderated by children's RSA such that this positive link was evident only among children who had low levels of baseline RSA, and not among those who had high levels of baseline RSA. These findings were obtained while controlling for concurrent sensitive parenting and maternal and child verbal abilities. Results from this study provide evidence for the significant role of biopsychosocial processes in early childhood in the development of EF.