nach oben

2012 | Buch

Kapitel lesen Erstes Kapitel lesen

Advances in Meta-Analysis

verfasst von: Terri D. Pigott

Verlag: Springer US

Buchreihe : Statistics for Social and Behavioral Sciences

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Wirtschaft" , Springer Professional "Technik"

Einloggen, um Zugang zu erhalten

Über dieses Buch

The subject of the book is advanced statistical analyses for quantitative research synthesis (meta-analysis), and selected practical issues relating to research synthesis that are not covered in detail in the many existing introductory books on research synthesis (or meta-analysis). Complex statistical issues are arising more frequently as the primary research that is summarized in quantitative syntheses itself becomes more complex, and as researchers who are conducting meta-analyses become more ambitious in the questions they wish to address. Also as researchers have gained more experience in conducting research syntheses, several key issues have persisted and now appear fundamental to the enterprise of summarizing research.

Specifically the book describes multivariate analyses for several indices commonly used in meta-analysis (e.g., correlations, effect sizes, proportions and/or odds ratios), will outline how to do power analysis for meta-analysis (again for each of the different kinds of study outcome indices), and examines issues around research quality and research design and their roles in synthesis. For each of the statistical topics we will examine the different possible statistical models (i.e., fixed, random, and mixed models) that could be adopted by a researcher. In dealing with the issues of study quality and research design it covers a number of specific topics that are of broad concern to research synthesists. In many fields a current issue is how to make sense of results when studies using several different designs appear in a research literature (e.g., Morris & Deshon, 1997, 2002). In education and other social sciences a critical aspect of this issue is how one might incorporate qualitative (e.g., case study) research within a synthesis. In medicine, related issues concern whether and how to summarize observational studies, and whether they should be combined with randomized controlled trials (or even if they should be combined at all).

For each topic, included is a worked example (e.g., for the statistical analyses) and/or a detailed description of a published research synthesis that deals with the practical (non-statistical) issues covered.

Inhaltsverzeichnis

Frontmatter

Chapter 1. Introduction

Abstract

This chapter introduces the topics that are covered in this book. The goal of the book is to provide reviewers with advanced strategies for strengthening the planning, conduct and interpretations of meta-analyses. The topics covered include planning a meta-analysis, computing power for tests in meta-analysis, handling missing data in meta-analysis, including individual level data in a traditional meta-analysis, and generalizations from a meta-analysis. Readers of this text will need to understand the basics of meta-analysis, and have access to computer programs such as Excel and SPSS. Later chapters will require more advanced computer programs such as SAS and R, and some advanced statistical theory.

Terri D. Pigott

Chapter 2. Review of Effect Sizes

Abstract

This chapter provides an overview of the three major effect sizes that will be used in the book: the standardized mean difference, the correlation coefficient, and the log odds ratio. The notation that will be used throughout the book is also introduced.

Terri D. Pigott

Chapter 3. Planning a Meta-analysis in a Systematic Review

Abstract

This chapter provides guidance on planning a meta-analysis. The topics covered include choosing moderators for effect size models, considerations for choosing between fixed and random effects models, issues in conducting moderator models in meta-analysis such as confounding of predictors, and computing meta-regression. Examples are provided using data from a meta-analysis by Sirin (2005). The chapter’s appendix also provides SPSS and SAS program code for the analyses in the examples.

Terri D. Pigott

Chapter 4. Power Analysis for the Mean Effect Size

Abstract

This chapter provides methods for computing the a priori power of the test of the mean effect size. Both fixed and random effects models tests are discussed. In addition, examples are provided for computing the number of studies needed to detect a substantively important effect size, and the detectable effect size with a given number of studies.

Terri D. Pigott

Chapter 5. Power for the Test of Homogeneity in Fixed and Random Effects Models

Abstract

This chapter will illustrate methods for the power of the test of homogeneity in fixed and random effects models. In fixed effects models, the test of homogeneity provides evidence about whether the effect sizes in a meta-analysis are measuring a common effect size. The test of homogeneity in random effects models is a test of the statistical significance of the variance component, the between-studies variance. The chapter gives examples of how to compute the power for the test of homogeneity in both fixed and random effects models.

Terri D. Pigott

Chapter 6. Power Analysis for Categorical Moderator Models of Effect Size

Abstract

This chapter provides methods for computing power with moderator models of effect size. The models discussed are analogues to one-way ANOVA models for effect sizes. Examples are provided for both fixed and random effects categorical moderator models. The power for meta-regression models requires knowledge of the values of the predictors for each study in the model, and is not provided here.

Terri D. Pigott

Chapter 7. Missing Data in Meta-analysis: Strategies and Approaches

Abstract

This chapter provides an overview of missing data issues that can occur in a meta-analysis. Common approaches to missing data in meta-analysis are discussed. The chapter focuses on the problem of missing data in moderators of effect size. The examples demonstrate the use of maximum likelihood methods and multiple imputation, the only two methods that produce unbiased estimates under the assumption that data are missing at random. The methods discussed in this chapter are most useful in testing the sensitivity of results to missing data.

Terri D. Pigott

Chapter 8. Including Individual Participant Data in Meta-analysis

Abstract

This chapter introduces methods for including individual participant data in a traditional meta-analysis. Since meta-analyses use data aggregated to the study, there is potential for aggregation bias, finding relationships between the effect size and study characteristics that may hold only at the level of the study. The potential of aggregation bias may limit the application of meta-analysis results to practice and policy. This chapter provides an example of including publicly available data in a traditional meta-analysis.

Terri D. Pigott

Chapter 9. Generalizations from Meta-analysis

Abstract

This chapter discusses the kinds of inferences and generalizations we can make from a meta-analysis. The chapter reviews the framework outlined by Shadish et al. (2002) for meta-analysis, and provides examples from two recent syntheses that had an influence on policy.

Terri D. Pigott

Chapter 10. Recommendations for Producing a High Quality Meta-analysis

Abstract

This chapter provides a set of recommendations based on prior chapters in the book for improving the quality of meta-analyses.

Terri D. Pigott

Chapter 11. Data Appendix

Abstract

Sirin (2005) conducted a systematic review of studies reporting a correlation between socioeconomic status (SES) and academic achievement. A number of different measures have been used in the literature for both SES and achievement; the goal of the meta-analysis was to examine whether variation in the strength of the association between SES and achievement varies depending on the types of measures used, and characteristics of the studies and their samples. The data used to construct Table 3.1 through 3.6 are given in Table 11.1 below.

Terri D. Pigott

Backmatter

Titel: Advances in Meta-Analysis
verfasst von: Terri D. Pigott
Verlag: Springer US
Electronic ISBN: 978-1-4614-2278-5
Print ISBN: 978-1-4614-2277-8
DOI: https://doi.org/10.1007/978-1-4614-2278-5