Split Hamiltonian Monte Carlo

We show how the Hamiltonian Monte Carlo algorithm can sometimes be speeded up by “splitting” the Hamiltonian in a way that allows much of the movement around the state space to be done at low computational cost. One context where this is possible is when the log density of the distribution of interest (the potential energy function) can be written as the log of a Gaussian density, which is a quadratic function, plus a slowly-varying function. Hamiltonian dynamics for quadratic energy functions can be analytically solved. With the splitting technique, only the slowly-varying part of the energy needs to be handled numerically, and this can be done with a larger stepsize (and hence fewer steps) than would be necessary with a direct simulation of the dynamics. Another context where splitting helps is when the most important terms of the potential energy function and its gradient can be evaluated quickly, with only a slowly-varying part requiring costly computations. With splitting, the quick portion can be handled with a small stepsize, while the costly portion uses a larger stepsize. We show that both of these splitting approaches can reduce the computational cost of sampling from the posterior distribution for a logistic regression model, using either a Gaussian approximation centered on the posterior mode, or a Hamiltonian split into a term that depends on only a small number of critical cases, and another term that involves the larger number of cases whose influence on the posterior distribution is small.     

An Examination of the Communication Styles Associated with Psychopathy and Their Influence on Observer Impressions

Psychopathic individuals are characterized as “intra-species predators”—callous, impulsive, aggressive, and proficient at interpersonal manipulation. For example, despite their high risk for re-offending, psychopathic offenders often receive early release on parole. While reputed to be social chameleons, research suggests that even naive observers can accurately infer high levels of psychopathic traits in others with very brief exposures to behavior, but accuracy degrades with extended observation. We utilized a lens model approach to examine the communication styles (emotional facial expressions, body language, and verbal content) of offenders varying in levels of psychopathic traits using “thin slice” video clips of psychological assessment interviews and to reveal which cues observers use to inform their evaluations of psychopathy. Psychopathic traits were associated with more (a) Duchenne smiles, (b) negative (angry) emotional language, and (c) hand gestures (illustrators). Further, psychopathy was associated with a marked behavioral incongruence; when individuals scoring high in psychopathic traits engaged in Duchenne smiles they were also more likely to use angry language. Naïve observers relied on each of these valid behavioral signals to quickly and accurately detect psychopathic traits. These findings provide insight into psychopathic communication styles, opportunities for improving the detection of psychopathic personality traits, and may provide an avenue for understanding successful psychopathic manipulation.     

Recent Mortality Trend Reversal in Russia: Are Regions Following the Same Tempo?

After several decades of negative trends and short-term fluctuations, life expectancy has been increasing in Russia since 2004. Between 2003 and 2014, the length of life rose by 6.6 years among males and by 4.6 years among females. While positive trends in life expectancy are observed in all regions of Russia, these trends are unfolding differently in different regions. First, regions entered the phase of life expectancy growth at different points in time. Second, the age- and cause-specific components of the gains in life expectancy and the number of years added vary noticeably. In this paper, we apply decomposition techniques—specifically, the stepwise replacement algorithm—to examine the age- and cause-specific components of the changes in inter-regional disparities during the current period of health improvement. The absolute inter-regional disparities in length of life, measured by the population-weighted standard deviation, decreased slightly between 2003 and 2014, from 3.3 to 3.2 years for males, and from 2.0 to 1.8 years for females. The decomposition of these small changes by ages and causes of death shows that these shifts were the result of diverse effects of mortality convergence at young and middle ages, and of mortality divergence at older ages. With respect to causes of death, the convergence is mainly attributable to external causes, while the inter-regional divergence of trends is largely determined by cardiovascular diseases. The two major cities, Moscow and Saint Petersburg, are currently pioneering mortality improvements in Russia and are making the largest contributions to the inter-regional divergence.     

Modeling probability density through ultraspherical polynomial transformations

We present a method for fitting parametric probability density models using an integrated square error criterion on a continuum of weighted Lebesgue spaces formed by ultraspherical polynomials. This approach is inherently suitable for creating mixture model representations of complex distributions and allows fully autonomous cluster analysis of high-dimensional datasets. The method is also suitable for extremely large sets, allowing post facto model selection and analysis even in the absence of the original data. Furthermore, the fitting procedure only requires the parametric model to be pointwise evaluable, making it trivial to fit user-defined models through a generic algorithm.     

Myths about Russia

This article provides an interdisciplinary theoretical analysis of contemporary social mythology and summarizes the results of an empirical study. The main groups of mythologized images of Russia in the consciousness of modern youth include: mythologized images of Russian domain and the historical perspective of Russia; symbolic and metaphorical images of Russia; and the types of social world views that young people have. The aspects of the constructive and destructive impacts of myths about Russia described in this article may be used to analyze and interpret individual and group images of Russia among young people and to develop recommendations for preventing the destructive impact of the mythologization of Russia's image on the social processes and consciousness of modern youth.     

Female Perceptions of Organizational Justice

This study examines women's conceptualization of the pervasive construct of organizational justice. A comprehensive four factor model was used to represent organizational justice while outcome variables were the important employee attitudes of job satisfaction, organizational commitment and turnover intentions. Structural equation modelling was used to analyse responses from 301 male and 147 female respondents. Differences were found for procedural, interpersonal and informational justices. It would appear female and male employees have differing responses to perceptions of justice. Justice had a diffuse effect for males, but not for females. For men, interpersonal justice predicted an increase in organizational commitment while procedural justice predicted a decrease in turnover intentions. For female employees, informational justice was found to increase commitment and reduce turnover intentions. The study highlights the need for all justice factors to be considered and for sex differences to be considered in future justice research.     

Bayesian approach to epsilon-skew-normal family

The estimation problem of epsilon-skew-normal (ESN) distribution parameters is considered within Bayesian approaches. This family of distributions contains the normal distribution, can be used for analyzing the asymmetric and near-normal data. Bayesian estimates under informative and non informative Jeffreys prior distributions are obtained and performances of ESN family and these estimates are shown via a simulation study. A real data set is also used to illustrate the ideas.     

Assessing the Need for Higher Levels of Care Among Problem Gambling Outpatients

Most treatment for gambling disorder is provided on an outpatient basis. Only a small number of jurisdictions in North America provide higher levels of gambling treatment, such as residential or intensive outpatient (IOP) care, despite the potential need for these services. Further, there appear to be few guidelines for determining appropriate level of gambling treatment. The aim of the present study was to assess the appropriateness of higher levels of problem gambling care among clients receiving outpatient treatment. Problem gamblers and their therapists independently completed questionnaires that assessed the need and desire for residential and IOP treatment. About 42% of problem gambling outpatients noted that they would be “probably” or “definitely” willing to attend residential treatment, and about half indicated they would be equally likely to attend IOP. Therapists recommended about a third of their clients as appropriate for higher levels of care. For both client and therapist assessments, there was a significant association between desire or recommendation for level of treatment and severity of gambling and co-occurring problems. Further, therapist recommendations for level of care were significantly associated with client willingness to attend higher levels of treatment. Our data reveal the potential need for higher levels of care for problem gambling, as evaluated by clients and their therapists. Policy implications for the funding of residential and IOP treatment are discussed.     

Correlation and variable importance in random forests

This paper is about variable selection with the random forests algorithm in presence of correlated predictors. In high-dimensional regression or classification frameworks, variable selection is a difficult task, that becomes even more challenging in the presence of highly correlated predictors. Firstly we provide a theoretical study of the permutation importance measure for an additive regression model. This allows us to describe how the correlation between predictors impacts the permutation importance. Our results motivate the use of the recursive feature elimination (RFE) algorithm for variable selection in this context. This algorithm recursively eliminates the variables using permutation importance measure as a ranking criterion. Next various simulation experiments illustrate the efficiency of the RFE algorithm for selecting a small number of variables together with a good prediction error. Finally, this selection algorithm is tested on the Landsat Satellite data from the UCI Machine Learning Repository.