We are concerned with testing for differential expression and consider three different aspects of such testing procedures. First, we develop an exact ANOVA type model for discrete gene expression data, produced by technologies such as a Massively Parallel Signature Sequencing (MPSS), Serial Analysis of Gene Expression (SAGE) or other next generation sequencing technologies. We adopt two Bayesian hierarchical models—one parametric and the other semiparametric with a Dirichlet process prior that has the ability to borrow strength across related signatures, where a signature is a specific arrangement of the nucleotides. We utilize the discreteness of the Dirichlet process prior to cluster signatures that exhibit similar differential expression profiles. Tests for differential expression are carried out using non-parametric approaches, while controlling the false discovery rate. Next, we consider ways to combine expression data from different studies, possibly produced by different technologies resulting in mixed type responses, such as Microarrays and MPSS. Depending on the technology, the expression data can be continuous or discrete and can have different technology dependent noise characteristics. Adding to the difficulty, genes can have an arbitrary correlation structure both within and across studies. Performing several hypothesis tests for differential expression could also lead to false discoveries. We propose to address all the above challenges using a Hierarchical Dirichlet process with a spike-and-slab base prior on the random effects, while smoothing splines model the unknown link functions that map different technology dependent manifestations to latent processes upon which inference is based. Finally, we propose an algorithm for controlling different error measures in a Bayesian multiple testing under generic loss functions, including the widely used uniform loss function. We do not make any specific assumptions about the underlying probability model but require that indicator variables for the individual hypotheses are available as a component of the inference. Given this information, we recast multiple hypothesis testing as a combinatorial optimization problem and in particular, the 0-1 knapsack problem which can be solved efficiently using a variety of algorithms, both approximate and exact in nature.

Teams are pervasive in organizations and provide an important contribution to organizational productivity. Since Hackman's (1987) seminal work, the team research focus has shifted from describing teams to outlining how researchers might use points of leverage, such as team design, to increase team effectiveness. There has been a wealth of research on team design variables that relate to team effectiveness. However, more than 15 years later, the team design literature remains fragmented and is inconsistent, and conclusions regarding optimal team design are difficult to make. The present study sought to unify the team design research by proposing a conceptual model and testing hypothesized relationships between specified design variables and team effectiveness using meta-analytic techniques. Specifically, the objectives of this study were to: (a) identify team design variables over which researchers and practitioners have some degree of control, (b) summarize the literature related to each of these variables, (c) hypothesize how each of the design variables are related to team effectiveness, (d) assess the relationship between these variables and team effectiveness using meta-analysis, (e) assess the influence of specified moderator variables (e.g., study setting, team tenure) on the team design variable/team effectiveness relationships, (f) make theoretically- and empirically-based recommendations for the design of effective teams, and (g) highlight areas in need of additional research. Results indicated that several team design variables show promise as a means of increasing team effectiveness. The strength of the team composition variable/team performance relationships was dependent on the study setting (lab or field); however, the study setting had considerable overlap with the type of team assessed (intellectual or physical). For lab studies (intellectual teams), team general mental ability (GMA) and task-relevant expertise were strong predictors of team performance, while team personality variables were unrelated to team performance. In field studies (physical teams), team agreeableness and conscientiousness had stronger relationships with team performance than team GMA and team task-relevant expertise. Team task design variables (e.g., task significance) had consistent, positive relationships with team performance, and several team structure variables (e.g., degree of self- management) were also related to team performance.

Includes vita. / Title from PDF of title page. / Document formatted into pages; contains 134 pages. / Thesis (Ph.D.)--University of South Florida, 2003. / Includes bibliographical references. / Text (Electronic thesis) in PDF format. / ABSTRACT: Validity generalization work over the past 25 years has called into question the veracity of the assumption that validity is situationally specific. Recent theoretical and methodological work has suggested that validity coefficients may be transportable even if true validity is not a constant. Most transportability work is based on the assumption that the distribution of rho ( ) is normal, yet, no empirical evidence exists to support this assumption. The present study used a competing model approach in which a new procedure for assessing transportability was compared with two more commonly used methods. Empirical Bayes estimation (Brannick, 2001; Brannick & Hall, 2003) was evaluated alongside both the Schmidt-Hunter multiplicative model (Hunter & Schmidt, 1990) and a corrected Hedges-Vevea (see Hall & Brannick, 2002; Hedges & Vevea, 1998) model. The purpose of the present study was two-fold. The first part of the study compared the accuracy of estimates of the mean, standard deviation, and the lower bound of 90 and 99 percent credibility intervals computed from the three different methods across 32 simulated conditions. The mean, variance, and shape of the distribution varied across the simulated conditions. The second part of the study involved comparing results of analyses of the three methods based on previously published validity coefficients. The second part of the study was used to show whether choice of method for determining whether transportability is warranted matters in practice. Results of the simulation analyses suggest that the Schmidt-Hunter method is superior to the other methods even when the distribution of true validity parameters violates the assumption of normality. Results of analyses conducted on real data show trends consistent with those evident in the analyses of the simulated data. Conclusions regarding transportability, however, did not change as a function of method used for any of the real data sets. Limitations of the present study as well as recommendations for practice and future research are provided. / System requirements: World Wide Web browser and PDF reader. / Mode of access: World Wide Web.

The purpose of this meta-analysis was to examine the effect of situated learning on the academic performance of students with and without disabilities in inclusive general education classrooms. While previous research has reported the overall effectiveness of situated learning, relatively few studies have been conducted to investigate how situated learning influences students' academic performances in inclusive settings where students with and without disabilities work together. Moreover, although the main interest of situated learning is about how to apply basic knowledge and skills to an authentic context and, beyond this, how to transfer them into a similar but novel situation in everyday life, little has been known about its effectiveness on students' achievement in terms of knowledge transfer. In this study, a meta-analytical statistical method was employed to investigate the effect of situated learning, and its effectiveness was examined according to the three levels of knowledge transfer (knowledge acquisition, application, and transfer). A total of 19 situated-learning studies, both published and unpublished, were analyzed. Each primary study's effect sizes were calculated using Hedges' g with the bias correction and then combined into the three weighted average effect sizes regarding the levels of knowledge transfer. This meta-analytic study found that, on all of the levels of knowledge transfer, the situated learning is effective for the learning of students with and without disabilities in inclusive general education classrooms. In the random effects model, the situated instruction produced a weighted mean effect size estimate of 2.049 for knowledge acquisition, 1.836 for knowledge application, 1.185 for knowledge transfer. In addition, the percentage of students with special needs in general education classrooms had a negative influence on the effectiveness of situated learning. However, the pattern of results also showed that the proportion of students with special needs in general education classrooms does not influence as greatly the learning of knowledge transfer as it does knowledge acquisition or application. / text

Because both validity and reliability indices are a function of the scores on a given administration of a measure, their values can often vary across samples. It is a common mistake to say that a test is reliable when in fact it is not the test that is reliable but the scores on the test that are reliable. In 1998, vacha-haase proposed a fixed-effects meta-analytic method for evaluating reliability that is similar to validity generalization studies called reliability generalization (rg). This study was conducted to evaluate alternative analysis strategies for the meta-analysis method of reliability generalization when the reliability estimates are not statistically independent. Five approaches for handling the violation of independence were implemented: ignoring the violation and treating each observation as independent, calculating one mean or median from each study, randomly selecting only one observation per study, or using a mixed effects model. This Monte Carlo study included five factors in the method. These factors were (a) the coefficient alpha, (b) sample size in the primary studies, (c) number of primary studies in the rg study, (d) number of reliability estimates from each, and (e) the degree of violation of independence where the strength of the dependence is related to the number of reliability indices (i.e. coefficient alpha) derived from a simulated set of examines and the magnitude of the correlation between the journal studies (with intra-class correlation icc = 0, .0l , .30, and .90). These factors were used to simulate samples under known and controlled population conditions. In general, the results suggested that the type of treatment does not have a noticeable impact on the accuracy of the reliability results but that researchers should be cautious when the intra-class correlation is relatively large. In addition, the simulations in this study resulted in very poor confidence band coverage. This research suggested that RG meta-analysis methods are appropriate for describing the overall average reliability of a measure or construct but the RG researcher should be careful in regards to the construction of confidence intervals.

Since Bowlby (1958, cited in Bowlby, 1969) originally introduced the theory of attachment it has been written about extensively and a vast amount of research has contributed to the development of the theory. In more recent years research has focused on the possible link between attachment and psychopathology. The major aim of the present meta-analysis was to contribute to this research effort by establishing the magnitude of the effect size for the relationship between attachment security and internalizing psychopathology; and attachment security and externalizing psychopathology, in children and adolescents. Four separate meta-analyses were conducted investigating internalizing and externalizing problems in cross-sectional and prospective studies. A comprehensive literature search was conducted to identify relevant studies for inclusion in the analysis. Identified studies were assessed for eligibility according to stringent inclusion and exclusion criteria. A total of 23 studies contributing 45 effect size correlations, involving 3793 different participants were considered eligible for inclusion. Relevant information was extracted and coded from the studies before the analyses were conducted. For cross-sectional studies the mean effect size correlation for attachment security and internalizing psychopathology was r = -0.24 (k = 14; p <0.01; 95% CI = -0.31, -0.17). For attachment security and externalizing psychopathology the mean effect size was r = -0.28 (k = 16; p <0.01; 95% CI = -0.34, -0.21). In terms of prospective studies the mean effect size correlation for attachment security and internalizing psychopathology was r = -0.17 (k = 8; p = 0.01; 95% CI = -0.28, -0.04); and for externalizing psychopathology it was r = -0.09 (k = 7; p = 0.02; 95% CI = -0.16, -0.01). When attachment security and psychopathology were measured concurrently, there was evidence of a negative association for both internalizing and externalizing psychopathology. Although the magnitude of effect was smaller for prospective studies evidence was also found for the predictive validity of a lower level of attachment security in the development of both internalizing and externalizing psychopathology. Theoretical explanations for these findings are presented and the research and clinical implications are discussed in terms of the limitations of the study.

Central to the goals of public health are obtaining and interpreting timely and relevant information for the benefit of humanity. In this dissertation, we propose methods to monitor and assess the spread HIV in a more rapid manner, as well as to improve decisions regarding patient treatment options. In Chapter 1, we propose a method, extending the previously proposed dual-testing algorithm and augmented cross-sectional design, for estimating the HIV incidence rate in a particular community. Compared to existing methods, our proposed estimator allows for shorter follow-up time and does not require estimation of the mean window period, a crucial, but often unknown, parameter. The estimator performs well in a wide range of simulation settings. We discuss when this estimator would be expected to perform well and offer design considerations for the implementation of such a study. Chapters 2 and 3 are concerned with obtaining a more complete understanding of the impact of treatment in randomized clinical trials in which multiple patient outcomes are recorded. Chapter 2 provides an illustration of methods that may be used to address concerns of both risk-benefit analysis and personalized medicine simultaneously, with a goal of successfully identifying patients who will be ideal candidates for future treatment. Riskbenefit analysis is intended to address the multivariate nature of patient outcomes, while “personalized medicine” is concerned with patient heterogeneity, both of which complicate the determination of a treatment’s usefulness. A third complicating factor is the duration of treatment use. Chapter 3 features proposed methods for assessing the impact of treatment as a function of time, as well as methods for summarizing the impact of treatment across a range of follow-up times. Chapter 4 addresses the issue of meta-analysis, a commonly used tool for combining information for multiple independent studies, primarily for the purpose of answering a clinical question not suitably addressed by any one single study. This approach has proven highly useful and attractive in recent years, but often relies on parametric assumptions that cannot be verified. We propose a non-parametric approach to meta-analysis, valid in a wider range of scenarios, minimizing concerns over compromised validity.

Multilevel modeling represents a potentially viable method for meta-analyzing single-subject research, but questions remain concerning its methodological properties with regard to characteristics of single-subject data. For this dissertation, Monte Carlo methods were used to investigate the properties of a 3 level model (i.e., with a quadratic equation at level 1), and three different level 1 error specifications (i.e., different variance components and covariances of 0, lag-1 autoregressive covariance structures, and separate error terms for each phase, with different variance components and covariances of 0). Data for simulated subjects were generated to have characteristics typical of published single-subject data (e.g., typical variances and magnitudes of effect). Samples were simulated for conditions which varied in number of data points per phase, number of subjects per study, number of studies meta-analyzed, level of autocorrelation in residuals, and continuity of variance across phases. Outcome variables examined included rates of convergence of analyses, power for statistical tests of fixed effects, and relative parameter bias of estimates of fixed effects, random effects’ variance components, and autocorrelation estimates. Convergence rates were found to be 100% for all level 1 error specifications and data conditions. Power for statistical tests of fixed effects was observed to be adequate when 10 or more data points were generated per phase and 60 or more total subjects were included in meta-analyses. The relative biases of estimates of fixed effects were found to have limited associations with numbers of data points per phase, levels of autocorrelation, and the continuity/discontinuity of variance across phases. Random effects’ variance components were observed to be frequently biased. Associations between relative bias and data conditions were found to vary by random effect. Finally, autocorrelation estimates were found to be biased in all conditions for which autocorrelation was generated. Results are discussed with regard to study strengths and limitations, and their implications for the meta-analysis of single subject data and primary single subject research. / text

Incorporating the best findings from current, high-quality research into routine clinical practice is the basis of evidence-based care. Chapter 1: "Systematic Review and Meta-Analysis in Evidence-Based Care" is a review of the systematic review process, including meta-analysis, aimed at clinical professionals with limited statistical training. It advocates the use of the systematic review process, outlines some general techniques, and provides selected resources where individuals can acquire additional assistance. The typical steps involved include: formulating a clear research question, defining inclusion and exclusion criteria, extracting the data and assessing the study quality, summarizing and synthesizing the evidence, and then interpreting the findings. When effort is made to minimize bias and locate as many articles on a particular topic as possible, systematic reviews and meta-analyses can produce invaluable findings for evidence-based care. Chapter 2: "The Effect of Macronutrient Distribution on the Lipid Profile in Adults: A Systematic Review and Meta-Analysis" describes a systematic review and meta-analysis that examined the impact total macronutrients had on blood lipid levels. This chapter builds upon the concepts introduced in chapter one, and assesses the effect of manipulating macronutrient distribution on the lipid profile of adults, and compares these effects to recommendations regarding macronutrients, such as the Acceptable Macronutrient Distribution Ranges (AMDRs). Suggestions related to improving the quality of meta-analyses are also outlined, and supplemental analyses are provided at the end of the dissertation.

A variety of psychotherapies have been demonstrated to be efficacious and effective treatments for depression. The cost of psychotherapy, however, and its low availability in some contexts pose significant treatment barriers for many depressed individuals. Based on the idea that peers (i.e., individuals who have successfully recovered from similar problems) may be uniquely able to provide empathy and support to those currently receiving treatment, some community mental health centers have implemented peer treatment models that employ recovered former clients as cost-effective adjunct providers. The effectiveness of these and other peer-administered interventions (PAIs) has not been well-established. The current study is a meta-analysis of the existing outcome research on PAIs for depression. Twenty-six studies were identified as eligible for inclusion and yielded 30 between-groups effect sizes and 29 pre-post PAI effect sizes. Study characteristics and methodological quality were coded and random-effects models were used to calculate and compare mean effect sizes. PAIs produced significant pre-to-post treatment reductions in depression symptoms that were comparable to those found in well-established professionally-administered interventions (.4554). In direct comparisons, PAIs performed as well as professionally-administered treatments (.0848). but not significantly better than treatment-as-usual (e.g., periodic physician check-ins or availability of community mental health services) and wait-list control conditions (.0978). These findings did not change after adjusting for the moderate degree of publication bias in the data. Moderation models revealed that professionally-co-administered PAIs produced significantly worse outcomes than those that were purely peer-administered, and that educational/skills-based PAIs (but not supportive PAIs) produced better outcomes compared with professional treatments. Limitations of this analysis included the heterogeneity of the included interventions and the lack of data on mediators and moderators. Still, these findings suggest that PAIs have promise as effective depression treatments and are worthy of further study.