A biclustering algorithm for binary matrices based on penalized Bernoulli likelihood

We propose a new biclustering method for binary data matrices using the maximum penalized Bernoulli likelihood estimation. Our method applies a multi-layer model defined on the logits of the success probabilities, where each layer represents a simple bicluster structure and the combination of multiple layers is able to reveal complicated, multiple biclusters. The method allows for non-pure biclusters, and can simultaneously identify the 1-prevalent blocks and 0-prevalent blocks. A computationally efficient algorithm is developed and guidelines are provided for specifying the tuning parameters, including initial values of model parameters, the number of layers, and the penalty parameters. Missing-data imputation can be handled in the EM framework. The method is tested using synthetic and real datasets and shows good performance.     

Graphical methods for comparing confounding in two or more designs

In many circumstances it is necessary to design an experiment with partial confounding of parameter estimates. In this situation, one would like to have tools to assess and compare candidate designs graphically. In this paper we discuss two graphical representations of design properties, and illustrate their application with two examples.     

Web-Enhanced Preservice Training for Prospective Resource Parents: A Randomized Trial of Effectiveness and User Satisfaction

ABSTRACT

Traditionally, prospective resource parents must attend all preservice training in person. Although live sessions are necessary for screening applicants, instructional portions of training could be enhanced by Web-based sessions. This pilot study compares the effectiveness of online and classroom versions of one session from a widely used preservice training program. Ninety-two individuals enrolled in the program in two states were randomly assigned to a treatment group that viewed an online version of the class on child abuse and neglect or to a comparison group that took the same class in person. Written questionnaires were completed before and after the class. Significant group differences on knowledge of child maltreatment and empathy toward birth parents plus high user satisfaction were hypothesized. Analysis of covariance results showed the online training was more effective than the live training at increasing knowledge. Multivariate analysis of covariance findings on empathy were not significant but trended toward greater empathy for the online group. Feedback indicated high satisfaction with the online course. The finding that online instruction is more effective than live instruction has positive implications for practice, because Web-based training offers advantages like standardizing instruction, cutting agency and trainee costs, and providing greater flexibility.     

Inside the blogosphere: A taxonomy and framing analysis of abortion weblogs

This study analyses weblog abortion-related content sampled via a search and snowball method. When weblogs located inside of established media Web sites and other weblogs are compared, they are equivalent in the use of advocacy versus objectivist frames. Media weblog posts contain a political frame more frequently than non-media weblog posts do. In terms of diversity of frame, media weblogs employ more frames in their stories. Various explanations for these findings are offered in the context of current abortion-related content patterns.     

Heterogeneity in Parent‐reported Social Skill Development in Early Elementary School Children

Despite known risks associated with aberrant social skill development, there has been a relative dearth of literature on typical developmental changes in social skills over time. In this study, we examine systematic changes in social skills from kindergarten (typical age of 5–6 years) to third grade (typical age of 8–9 years), and focus on systematic heterogeneity across these developmental trajectories. Data came from the National Head Start—Public School Early Childhood Transition Demonstration Project (N = 6964). Mixture models provide evidence for multiple classes of individuals representing heterogeneity in the development of social skills. Classes were defined as a majority class (whose trajectories remained relatively stable over time), an increasing class (whose trajectories increased at a faster rate than the majority class), and a decreasing class (whose trajectories decreased at a faster rate than the majority class). Developmental trends accounted for a substantial proportion of the variance in social skill components. Implications and limitations of the study are discussed.     

Minority comparison model: Effects of Whites’ multiracial evaluation on symbolic racism and racialized policy preferences

Most Racial Studies primarily focus on African Americans without paying attention to nonblack minorities, and it fails to capture recent increase in racial diversity. Based on previous theories and empirical findings, we propose a new model, minority comparison model, which accounts for theoretical shortcomings in Racial Studies. This model (1) captures psychological processes that compare blacks and nonblacks, and (2) explores the effects of whites’ multiracial evaluation on racialized policy preferences. Drawing on a 2008 national representative sample, this study finds that whites who have positive stereotype of nonblacks (e.g., Hispanics and/or Asians) but negative stereotype of blacks show substantially higher symbolic racism and stronger opposition to Affirmative Action, whereas whites who have positive stereotype of blacks but negative stereotypes of nonblacks have stronger opposition to expansive immigration policy. Our study offers new ways of understanding and accounting for symbolic racism in modern context, and shows how whites’ preferences in racialized policies are influenced by multiracial evaluation.     

The State of Sleep Among College Students at a Large Public University

Abstract

Objective: Data about college student sleep were collected and used to develop an education campaign to improve sleep. Participants: On-campus residents at a large state university were surveyed on 4 occasions, October 2005 to April 2007. Sample size was 675 to 1,823 students. Fall 2005 mean age = 18.5 years, SD = 1.03 (range 18–30) years. Initial survey included 935 males and 1,859 females (2005–2006). Matched pairs data (2006–2007) included 91 males and 107 females. Twenty-six males and 22 females participated in interviews. Methods: A survey administered online included the Pittsburgh Sleep Quality Index, along with an 8-question in-person interview. Results: Poor sleep interacted with academics and mental health, and an education campaign positively affected student sleep. Conclusions: Teaching students how to effectively manage sleep can improve their well-being. Sleep may also be a gateway topic for health care professionals to address sensitive health issues such as depression.     

South Korean College Students' Holland Types and Career Compromise Processes

This study explored South Korean college students' career compromise processes and examined whether preferences for sex type, prestige, or interests would be differentiated by Holland theme interest types, gender, or college major. Participants were South Korean undergraduate students from 2 universities in Seoul, South Korea. They were asked to choose 1 occupation from each of 168 pairs of occupations using a forced‐choice format. A total of 376 surveys were analyzed. There were significant main effects for Holland interest types and for gender but not for college major on their career compromise processes. Implications for career counselors and researchers are discussed.     

Terrestrial habitat and individual fitness increase survival of a freshwater turtle in an urban landscape

Urban Ecosystems - Urbanisation is causing rapid land-use change worldwide. Populations of freshwater turtles are vulnerable to impacts of urbanisation such as habitat loss, fragmentation and...