On predictive distributions and Bayesian networks

In this paper we are interested in discrete prediction problems for a decision-theoretic setting, where the task is to compute the predictive distribution for a finite set of possible alternatives. This question is first addressed in a general Bayesian framework, where we consider a set of probability distributions defined by some parametric model class. Given a prior distribution on the model parameters and a set of sample data, one possible approach for determining a predictive distribution is to fix the parameters to the instantiation with the maximum a posteriori probability. A more accurate predictive distribution can be obtained by computing the evidence (marginal likelihood), i.e., the integral over all the individual parameter instantiations. As an alternative to these two approaches, we demonstrate how to use Rissanen's new definition of stochastic complexity for determining predictive distributions, and show how the evidence predictive distribution with Jeffrey's prior approaches the new stochastic complexity predictive distribution in the limit with increasing amount of sample data. To compare the alternative approaches in practice, each of the predictive distributions discussed is instantiated in the Bayesian network model family case. In particular, to determine Jeffrey's prior for this model family, we show how to compute the (expected) Fisher information matrix for a fixed but arbitrary Bayesian network structure. In the empirical part of the paper the predictive distributions are compared by using the simple tree-structured Naive Bayes model, which is used in the experiments for computational reasons. The experimentation with several public domain classification datasets suggest that the evidence approach produces the most accurate predictions in the log-score sense. The evidence-based methods are also quite robust in the sense that they predict surprisingly well even when only a small fraction of the full training set is used.     

Reconstructing the size of the African American population by age and sex, 1930–1990

We estimate the size of the African American population in five-year age groups at census dates from 1930 to 1990 using a three-part strategy. For cohorts born after 1935, we follow the U.S. Census Bureau in using classical demographic analysis. To estimate the size of cohorts born before 1895, we use extinct-generation estimates. For remaining cohorts, we implement an age/period/cohort model of census counts. All approaches are applied to a data set in which the age distribution of deaths has been corrected for age misreporting. Results provide strong confirmation of the basic validity of Census Bureau estimates of census undercounts for African Americans while extending estimates to new cohorts and periods. Our estimates are less consistent with an historical series prepared by Coale and Rives (1973).     

Collinearity, Bias, and Effect Size: Modeling “the” Effect of Track on Achievement

The estimated effect of any factor can be highly dependent on both the model and the data used for the analyses. This article presents an example of the estimated effect of one factor in two different data sets under three different forms of the standard linear model using the effect of track placement on achievement as an example. Some relative advantages and disadvantages of each model are considered. The analyses demonstrate that, given collinearity among the predictor variables, a model with a poorer statistical fit may be useful for some interpretive purposes.     

Gender roles,HIV risk behaviors,and perceptions of using female condoms among college students

The purpose of this study was to examine relationships between gender roles and HIV risk behavior, and perceptions and acceptance of the female condom among college students (n = 410). It was hypothesized that high hyperfeminine females and high hypermasculine males – those adhering to traditional gender roles – would engage in more HIV risk behaviors, including alcohol and drug use and various sexual practices, than those with lower hyperfemininity and hypermasculinity. It was also hypothesized that higher hyperfeminine females as well as higher hypermasculine males would perceive the female condom more negatively and would be less likely to view the female condom as a viable form of protection in the future. It was also hypothesized that high hyperfeminine females and high hypermasculine males would not accept the female condom as an alternative form of protection. Implications for prevention interventions are discussed.     

Preference in flower visits and partitioning in pollen diets of stingless bees in an Asian tropical rain forest

Floral resource partitioning among stingless bees (Trigona, Meliponini, Apidae) in a lowland rain forest in Sarawak, Malaysia, was investigated using tree towers and walkways in a 4-year study that included a general flowering period. We obtained 100 collections of insect visitors to flowers of varying floral location and shape representing 81 plant species. The tendency of 11 species of stingless bees to visit specific flowers with a particular floral location and shape was analyzed by logistic regression analysis. This analysis showed that the proportion of flower visitor collections containing Trigona fuscobalteata and T. melanocephala differed according to floral location. The former was frequently collected at canopy and gap flowers, whereas the latter was most often collected at understory flowers. The analysis also suggested that T. erythrogastra was more rarely collected at shallow flowers than at deep flowers. Analysis of the pollen diets of T. collina, T. fuscobalteata, T. melanocephala, and T. melina revealed that similarity of pollen sources differed among the six permutated pairs of the four species. The lowest mean rank of similarity found was between T. fuscobalteata and T. melanocephala. This result supports the hypothesis that preference in visiting flowers in different locations leads to pollen resource partitioning. Received: May 14, 1997 / Accepted: April 23, 1999     

Retrospective information on health status and its application for population health measures

Healthy life expectancies are almost always calculated by using health data from cross-sectional surveys. This type of calculation is done partly because data from longitudinal surveys are not always available, and when they are available, they are collected at intervals that are longer than one year. In such cases, collecting health information retrospectively for the years skipped by the survey is useful. The main purpose of this paper is to show how retrospective health information can be used to estimate life expectancies in different health states. Healthy life expectancies are estimated with and without using data on retrospective health information, and the corresponding estimates are compared. The two sets of estimates are similar. We conclude that retrospectively assessed health information based on a one-year recall period can be used to estimate years of life in various health states and that estimates based on such information will closely approximate estimates based on concurrent health information.     

MANAGING ACADEMIC INTERDISCIPLINARY RESEARCH PROJECTS*

This field study provides variables that can be used by practitioners as action levers and by future researchers as the basis for theoretical development. Conclusions from relevant literature and findings from interviews with interdisciplinary research management identified forty variables that were viewed as important to interdisciplinary research project success. After adjusting the data for reliability attenuation, these variables were further analyzed to identify the best prediction equation. The findings suggested that project age or the longevity of the project and open discussion of disagreements were the best predictors of performance.     

When Adaptations Go Awry: Functional and Dysfunctional Aspects of Social Anxiety

Adaptations are psychological and behavioral mechanisms designed through evolution to serve specific purposes ultimately related to reproductive success. Although adaptations are inherently functional, in some cases their operation can nevertheless cause personal and social dysfunction. We describe a theoretical framework for understanding, predicting, and reducing the dysfunctional consequences of psychological adaptations. We discuss three general sources of dysfunction: a) the existence of adaptive tradeoffs, b) mismatches between current environments and ancestral environments, and c) individual differences. The paper applies this framework primarily to the topic of social anxiety, a psychological phenomenon marked by concerns pertaining to social rejection and embarrassment. Although social anxiety can serve useful functions, it can also involve excessive worry, negative affect, and avoidance of social situations, leading to significant distress and social impairment. We consider sources of dysfunction in social anxiety and discuss implications for policy, including recommendations for psychological, situational, and biological interventions. We also discuss broader applications of this theoretical framework to other areas of social life.