Quadratic estimation in mixed linear models with two variance components

Explicit expressions for Bayes invariant quadratic estimates, biased and unbiased, are presented and proved to cover the entire class of admissible estimates in the considered classes. An unbalanced genetic model is studied for demonstration.     

Formative research methods to understand patient and provider responses to heart attack symptoms

Formative research is often required for program planning, and for reducing uncertainty about generalizability of program effects. This article describes and justifies methods of formative research conducted for the REACT study (Rapid Early Action for Coronary Treatment), a multi-center collaborative randomized community trial aimed at reducing patient delay in seeking care for acute myocardial infarction (AMI). Formative research cast light on patient and community members’ decision-making process in seeking help for AMI, as well as barriers and facilitators of this process. Investigators at all five REACT Field Centers participated in the formative research. The process consisted of: (1) developing a common theoretical framework for the study intervention; (2) conducting a literature review and qualitative research to identify and address gaps in knowledge; and (3) developing a common protocol for the REACT study that accommodated the diversity of the target communities in terms of services, resources, history, and ethnicity. Analysis employed triangulation, defined as an explicit search for heterogeneous data sources to reduce uncertainty about forces at work and opportunities for intervention across settings and populations. Because the collection and interpretation of data went in stages, staff of several REACT Field Centers had independent input to the overall synthesis, then shared and revised the results. Advantages and limitations of this approach are discussed.     

The Dickensian Occupational Structure

Twelve of the novels of Charles Dickens provide a sample of 349 characters whose occupational achievement and mobility are investigated. The mobility patterns indicate more inheritance than in the United States. This cannot be accounted for by education, material inheritance, or morality (being a good or bad person). A model of quasi-perfect mobility, taking all diagonal frequencies as given, more closely represents the patterns. Determinants of achievement considered are social origins, number of siblings, education, marital status, morality. The characters from large families do better occupationally than those from small. Education functions primarily to redistribute success. The married are superior to the unmarried, with marriage functioning not as a selective mechanism, but exerting causal impact of its own. The evil are rewarded in this life. The basic path model of the stratification system of Dickens is similar to that of the United States.     

Using Choquet integral in economics

This paper is devoted to application of the Choquet integral with respect to the monotone set functions in economics. We present the application in decision making, finance, insurance, social welfare and quality of life. The Choquet integral is used as the numerical representation of preference relation in decision making, as the “expected value” of future price in financial decision problems, as the insurance premium and as the social evaluation function. Received: March 2000; revised version: August 2001     

Individual control treatment in split-plot experiments

The paper deals with experiments laid out in a complete or an incomplete split-plot design in which one control (standard) treatment occurs in addition to the usual treatments. Usually the control (standard) treatment has been treated as one specific factor level. In this paper, in contrast to others in this area, the control (standard) may not be strictly connected with treatment combinations. The new incomplete split-plot designs with control satisfy all generally accepted methodological requirements, with special reference to the problems of randomisation. Moreover, tools are described which allow checking of the general balance or efficiency of the design, as well as performance of experiments with inference.     

The essentially complete class of rules in multinomial identification

The problem of identification within two groups is considered, when the space 𝔛 of the possible values of the observed random variable X is finite. Using an essentially complete class of tests hypothesis ξ=0 (ξis real) for the multivariate exponential family, an essentially complete class of sample-based identification rules has been found. Comparison of the rules to those derived from density estimators has shown that the latter constitute a subclass of the former.