Integrating Results Through Meta-Analytic Review Using Sas Software

by Wang, Morgan C.; Bushman, Brad J.

• ISBN13: 9781580252935
• ISBN10: 1580252931
• Format: Paperback
• Copyright: 1998-11-01
• Publisher: Sas Inst

Summary

Finally ... a book that addresses the various needs, concepts, and approaches for SAS users who work with meta-analytic procedures! Wang and Bushman introduce you to the important concepts in meta-analysis and how to use SAS software for this specific type of analysis. They describe the meta-analytic (or quantitative) approach to reviewing results from a collection of studies that all investigate the same phenomenon. The authors thoroughly describe how you can use meta-analysis in "data-mining" projects to discover meaningful relations among variables in a collection of studies. Practicing meta-analysts or anyone interested in combining the results from several related studies, surveys, and experiments will benefit from this comprehensive book. It is assumed that the reader has an understanding of meta-analytic procedures and SAS software. Book jacket.

Author Biography

Morgan C. Wang, Ph. D., is a statistician with a solid theoretical and applied background in meta-analysis Brad J. Bushman, Ph. D., is a practicing meta-analyst and Associate Professor in the Department of Psychology at Iowa State University

Table of Contents

Using This Book	xi
Introduction	1	(24)
Narrative (Qualitative) and Meta-Analytic (Quantitative) Literature Reviews	2	(4)
Increasing Use of Meta-Analysis	6	(2)
Two Approaches to Conducting a Meta-Analysis	8	(3)
Operationally Defining Abstract Concepts in Research	11	(1)
Categorical (Qualitative) and Continuous (Quantitative) Variables	12	(5)
Types of Variables in Research	12	(3)
Effect-Size Measures for Categorical Variables	15	(1)
Effect-Size Measures for Continuous Variables	16	(1)
Some Issues to Consider When Conducting a Meta-Analysis	17	(4)
Publication Bias and Study Quality	17	(1)
Missing Effect-Size Estimates	18	(1)
Fixed- and Random-Effects Models	19	(1)
Correlated Effect-Size Estimates	20	(1)
Using the SAS System to Conduct a Meta-Analysis	21	(1)
References	22	(3)
Using the SAS® System to Conduct a Meta-Analysis	25	(36)
Introduction	26	(1)
Help for New Users of SAS Software	26	(2)
SAS/ASSIST Software	26	(2)
Creating a Meta-Analytic Data Set Using SAS Software	28	(7)
Creating a Meta-Analytic Data Set from the SAS Data Step	29	(4)
Creating a Meta-Analytic Data Set from an OBDC Data	33	(2)
Manipulating a SAS Data Set Using the SAS Data Step	35	(2)
Renaming Variables	35	(1)
Keeping and Dropping Variables in Output SAS Data Sets	35	(2)
Creating a Permanent SAS Data Library	37	(1)
Using a SAS Macro	38	(3)
Creating and Manipulating a SAS Graph	41	(3)
SAS/Graph Displays on the Computer Monitor	41	(2)
Creating a SAS/Graph CGM File	43	(1)
SAS Procedures Used in This Book	44	(14)
PROC Sort	45	(1)
PROC Print	46	(1)
PROC Means	47	(2)
PROC Univariate	49	(1)
PROC GLM	50	(1)
PROC Format	51	(1)
PROC Timeplot	52	(3)
PROC Shewhart	55	(3)
Conclusions	58	(1)
References	59	(2)
Graphical Presentation of Meta-Analytic Results	61	(46)
Introduction	62	(1)
Dot Plots	62	(2)
Funnel Plots	64	(16)
Using a Funnel Plot to Investigate Whether All Studies Come from a Single Population	64	(8)
Using a Funnel Plot to Search for Publication Bias	72	(5)
Problems with Funnel Plots	77	(3)
Normal Quantile Plots	80	(8)
Using a Normal Quantile Plot to Check the Normality Assumption	82	(2)
Using a Normal Quantile Plot to Investigate Whether All Studies Come from a Single Population	84	(1)
Using a Normal Quantile Plot to Search for Publication Bias	85	(3)
Problems with Normal Quantile Plots	88	(1)
Stem-and-Leaf Plots	88	(2)
Box Plots	90	(6)
Conclusions	96	(1)
References	97	(1)
Appendices	98	(9)
SAS Code for Output 3.1	98	(1)
SAS Macro for Finding the Minimum and Maximum Values of the Variables on the X and Y Axes	98	(1)
SAS Macro for Entering Parameters for a Funnel Plot	99	(2)
SAS Macro for Creating a Funnel Plot	101	(1)
SAS Code for Figure 3.2	102	(3)
SAS Code Used to Create Normal Quantile Plots	105	(2)
Combining Effect-Size Estimates Based on Categorical Data	107	(36)
Introduction	108	(1)
Two-Way Contingency Tables	108	(3)
The Odds Ratio ω	111	(4)
Combining Odds Ratios Using the Weighted Average Method	115	(1)
Heterogeneity Test for Odds Ratios	116	(4)
Combining Odds Ratios Using the Mantel-Haenszel Method	120	(5)
Controlling for the Effects of Covariates	125	(7)
Control by Logistic Regression	125	(4)
Control by Stratification	129	(3)
Conclusions	132	(1)
References	133	(2)
Appendices	135	(8)
SAS Macro for Computing the Odds Ratio	135	(2)
SAS Macro for Computing the Common Odds Ratio Based on the Weighted Average Method	137	(1)
SAS Macro for Heterogeneity Test of Odds Ratios	138	(2)
SAS Macro for Computing the Common Odds Ratio Based on the Mantel-Haenszel Method	140	(2)
SAS Code for Example 4.5	142	(1)
Combining Effect-Size Estimates Based on Continuous Data	143	(34)
Introduction	144	(1)
Two Families of Effect Sizes	144	(8)
The Standardized Mean Difference Family	145	(3)
The Correlation Family	148	(1)
Relationship between the Two Families of Effect Sizes	149	(3)
Converting Test Statistics to Effect-Size Estimates and Converting Effect-Size Estimators from One Type to Another	152	(3)
Combining Sample Standardized Mean Differences	155	(4)
Combining Sample Correlation Coefficients	159	(4)
Conclusions	163	(1)
References	164	(1)
Appendices	165	(12)
Formulas for Converting Cohen's d, Hedges' g, and the Point-Biseral Correlation to Hedges' gU	165	(1)
Formulas for Converting Cohen's d, Hedges' g, and Hedges' gU to Point-Biseral Correlations	166	(1)
SAS Macro for Computing Effect-Size Estimates	167	(2)
Formulas for Obtaining Hedges' g, Hedges' gU, and the Point-Biseral Correlation from a t Test Statistic	169	(2)
SAS Macro for Converting Test Statistics to Effect-Size Estimates and for Converting Effect-Size Estimators from One to Another	171	(4)
SAS Macro for Computing a Weighted Average of Effect-Size Estimates	175	(2)
Vote-Counting Procedures in Meta-Analysis	177	(28)
Introduction	178	(1)
The Conventional Vote-Counting Procedure	179	(1)
Level of Significance	180	(1)
Vote-Counting Situations	180	(14)
Vote-Counting Procedures for Estimating the Population Standardized Mean Difference	183	(6)
Vote-Counting Procedures for Estimating the Population Correlation Coefficient	189	(5)
Conclusions	194	(2)
References	196	(1)
Appendices	197	(8)
SAS/IML Module for Obtaining the Probability of the Vote-Counting Estimate Using the Large Sample Approximation Method	197	(1)
SAS/IML Module for Obtaining a Confidence Interval for the Population Standardized Mean Difference Using Vote-Counting Procedures	198	(3)
SAS Macro for Obtaining a Confidence Interval for the Population Correlation Coefficient Using Vote-Counting Procedures	201	(3)
Sas Macro for Estimating Population Effect Sizes Using Vote-Counting Procedures	204	(1)
Combining Effect-Size Estimates and Vote Counts	205	(18)
Introduction	206	(1)
Using the Combined Procedure to Estimate the Population Standardized Mean Difference	207	(3)
Using the Combined Procedure to Estimate the Population Correlation Coefficient	210	(5)
Conclusions	215	(1)
References	216	(1)
Appendices	217	(6)
SAS Code for Example 7.1	217	(2)
SAS Macro for Obtaining the Pearson Product-Moment Correlation Coefficient Based on the Method of Maximum Likelihood	219	(2)
SAS Code for Example 7.2	221	(2)
Fixed-Effects Models in Meta-Analysis	223	(50)
Introduction	224	(2)
Fixed- and Random-Effects Models in Individual Experiments	226	(2)
Fixed- and Random-Effects Models in Meta-Analysis	228	(1)
Testing the Moderating Effects of Categorical Study Characteristics in ANOVA Models	229	(20)
Fixed-Effects ANOVA Models with One Categorical Factor	229	(13)
Fixed-Effects ANOVA Models with Two Categorical Factors	242	(7)
Testing the Moderating Effects of Continuous Study Characteristics in Regression Models	249	(13)
Confidence Intervals for Individual Regression Coefficients	250	(4)
Omnibus Tests for Blocks of Regression Coefficients and Tests for Homogeneity of Effects	254	(6)
Multicollinearity Among Study Characteristics	260	(2)
Quantifying Variation Explained by Study Characteristics in ANOVA and Regression Models	262	(1)
Conclusions	263	(2)
References	265	(1)
Appendices	266	(7)
SAS Macro for Computing Q-Statistics in Fixed-Effects ANOVA Models	266	(2)
SAS Macro for Computing Confidence Intervals for Group Mean Effects in Fixed-Effects ANOVA Models	268	(1)
SAS Macro for Comparing Group Mean Effects in Fixed-Effects ANOVA Models	269	(2)
SAS Macro for Computing Confidence Intervals for Regression Coefficients in Fixed-Effects Models	271	(2)
Random-Effects Models in Meta-Analysis	273	(30)
Introduction	274	(3)
Testing the Moderating Effects of Categorical Study Characteristics in ANOVA Models	277	(14)
Random-Effects Models with One Categorical Factor	277	(9)
Random-Effects Models with Two Categorical Factors	286	(5)
Testing the Moderating Effects of Continuous Study Characteristics in Regression Models	291	(10)
Testing Whether the Random-Effects Variance is Zero	291	(2)
Estimating the Random-Effects Variance	293	(2)
Confidence Intervals for Individual Regression Coefficients	295	(1)
Omnibus Tests for Blocks of Regression Coefficients	296	(3)
Multicollinearity among Study Characteristics	299	(2)
Conclusions	301	(1)
References	302	(1)
Combining Correlated Effect-Size Estimates	303	(32)
Introduction	304	(2)
Combining the Results from Multiple-Treatment Studies	306	(9)
Combining the Results from Multiple-Endpoint Studies	315	(6)
Conclusions	321	(2)
References	323	(1)
Appendices	324	(11)
SAS Macro for Computing FMAX Statistics for Multiple-Treatment Studies	324	(2)
SAS Macro for Computing Combined Effect-Size Estimates and 95% Confidence Intervals in Multiple-Treatment Studies	326	(5)
SAS Macro for Computing Combined Effect-Size Estimates and 95% Confidence Intervals in Multiple End-Point Studies	331	(4)
Conducting and Reporting the Results of a Meta-Analysis	335	(28)
Introduction	336	(1)
Reporting the Results of the Literature Search	336	(2)
Reporting the Results of the Data Collection	338	(2)
Reporting the Results of the Data Analysis	340	(6)
Checking Statistical Assumptions	341	(1)
Reporting the Results of Subgroup Analyses	341	(2)
Reporting the Results of Sensitivity Analyses	343	(3)
Example of a Meta-Analysis	346	(14)
Reporting the Results of the Literature Search	347	(1)
Reporting the Results of the Data Collection	347	(2)
Reporting the Results of the Data Analysis	349	(11)
Conclusions about Example Meta-Analysis	360	(1)
Conclusions	360	(2)
References	362	(1)
Index	363