A clustering approach to interpretable principal components

A new method for constructing interpretable principal components is proposed. The method first clusters the variables, and then interpretable (sparse) components are constructed from the correlation matrices of the clustered variables. For the first step of the method, a new weighted-variances method for clustering variables is proposed. It reflects the nature of the problem that the interpretable components should maximize the explained variance and thus provide sparse dimension reduction. An important feature of the new clustering procedure is that the optimal number of clusters (and components) can be determined in a non-subjective manner. The new method is illustrated using well-known simulated and real data sets. It clearly outperforms many existing methods for sparse principal component analysis in terms of both explained variance and sparseness.     

Matrix mean square error comparisons based on a certain covariance structure

For two given estimators of a parameter vector the covariance structure of their difference is used to compare them in terms of their mean square error matrices. The results obtained are applied to the covariance adjustment technique and regression     

A Collaborative Student Affairs and Faculty Health Assessment and Intervention Initiative

Abstract

The authors describe a collaborative partnership forged between faculty and student affairs staff to improve student health at a large urban university. They examine skills and reward structures of each constituency and the stages of the collaboration in the context of 2 theoretical models. A comprehensive data collection and dissemination process in the campus community provided goals for the initial stage of the partnership, leading to implementation of campus initiatives that use the reciprocal skills of each stakeholder. Outcomes of the collaboration included (1) a working relationship between faculty and student affairs staff, (2) increased dialogue with high-level administrators, (3) more coordinated campus efforts to decrease high-risk drinking, (4) use of outcome measures for implementing and evaluating health programs, and (5) an opportunity for interdisciplinary research. The authors offer suggestions for implementing the process on other campuses.     

Pregaming: An Exploratory Study of Strategic Drinking by College Students in Pennsylvania

Abstract

Objectives: This exploratory study examined pre-event drinking, or pregaming, by US college students. Participants: 112 undergraduates from 10 Pennsylvania colleges participated. Method: A focus group, including a written questionnaire, was conducted at each institution. Results: Only 35.7% of the participants had not pregamed during the last 2 weeks. Pregamers consumed an average of 4.9 (SD = 3.1) drinks during their most recent session. Gender, class year, and other demographic variables did not predict pregaming. Heavier drinkers, and those stating that the average student pregamed 3+ times in the last 2 weeks, were more likely to report pregaming in the last 2 weeks. How much students drink when pregaming is influenced by how much they expect to drink later on. Conclusion: Pregaming presents a growing challenge for campus officials. Additional research is needed on the nature of the problem and which combination of prevention strategies might best address this behavior.     

Book reviews and books received

Abstract

In the field of stress research it is often assumed that social support as well as coping behaviour contributes to feelings of health and well-being. However, until now it has not been altogether clear whether a main-effect model or a buffer model was the more appropriate for explaining the role of these factors in response to stressful situations. In this paper, a research project is described in which both the direct and indirect effects of social support and coping were studied in the work situation of nurses. The results offer more support for a main-effect model than for a buffer or moderating-effect model, especially with regard to the coping strategies ‘active problem-solving’ and ‘palliative behaviour’. However, some moderating effects on nurses‘ reactions to their work situation were found for social support and for ‘comforting cognitions’ as a coping strategy.     

Influence diagnostics in a multivariate normal regression model with general parameterization

Patriota and Lemonte [24] introduced a quite general multivariate normal regression model. This model considers that the mean vector and the covariance matrix share the same vector of parameters. In this paper we present some influence assessment for this model, such as the local influence, total local influence of an individual and generalized leverage which are discussed. Additionally, the normal curvatures for local influence studies are derived under some perturbation schemes.     

Visual Dominance and Visual Egalitarianism: Individual and Group-Level Influences of Sex and Status in Group Interactions

This study investigated visual dominance and visual egalitarianism of men and women (N = 94; 17 teams) in team meetings at diverse workplaces. Two novel gaze-related measures were developed: (a) a group visual dominance ratio (group-VDR) assessing each member’s visual dominance vis-à-vis all other members, and (b) a gaze distribution index (GDI) assessing each member’s visual egalitarianism to all group members. Multilevel analyses were conducted to account for influences of the team members’ sex and status on the individual level and for influences of sex and status composition of the teams, and the team leaders’ sex on the group level. Results suggested that high-status individuals displayed more visual dominance than low-status individuals. The significant interaction of individuals’ sex and status indicated that the positive relationship of status and visual dominance applied particularly to women. The more women in a team, the more visual dominance was displayed. The team leader’s sex significantly influenced visual egalitarianism: Gaze distribution was less egalitarian when the team leader was male.     

On the number of separable partitions

Consider partitions of a given set A of n distinct points in general position in ℝ ^d into parts where each pair of parts can be separated by a hyperplane that contains a given set of points E. We consider the problem of counting and generating all such partitions (correcting a classic 1967 result of Harding about the number of such partitions into two parts). Applications of the result to partition problems are presented.     

Asymptotic tests and monte carlo studies associated with the multiplicative interaction model

Several authors have proposed approximations to percentage points required for testing certain hypotheses associated with the multiplicative interaction model. Alternative approximations based on the asymptotic joint distribution of the characteristic roots of a noncentral Wishart matrix are proposed in this paper. The type I error rates of the resulting tests and the existing procedures are then compared using Monte Carlo methods.