Doing Meta-Analysis in R
I Introduction & R Basics
1
About this Guide
2
RStudio & Basics
2.1
Getting RStudio to run on your computer
2.1.1
Running R Code
2.1.2
Getting Help
2.2
The
dmetar
package
2.2.1
Installation of the
dmetar
package
3
Getting your data into R
3.1
Data preparation in Excel
3.1.1
Setting the columns of the Excel spreadsheet (raw effect size data)
3.1.2
Setting the columns of the Excel spreadsheet (pre-calculated effect size data)
3.2
Importing the Spreadsheet into Rstudio
3.2.1
Saving the data in the right format
3.2.2
Saving the data in your working directory
3.2.3
Loading the data
3.3
Data manipulation
3.3.1
Converting to factors
3.3.2
Converting to logicals
3.3.3
Selecting specific studies
3.3.4
Changing cell values
3.4
Exercises
(META-ANALYSIS) Meta-Analysis in R
4
Pooling Effect Sizes
4.1
Fixed-Effects-Model
4.1.1
Pre-calculated effect size data
4.1.2
Raw effect size data
4.2
Random-Effects-Model
4.2.1
Estimators for
tau
2
in the random-effects-model
4.2.2
Pre-calculated effect size data
4.2.3
Raw effect size data
4.3
Binary outcomes
4.3.1
Event rate data
4.3.2
Incidence rates
4.3.3
Pre-calculated effect sizes
4.4
Correlations
5
Forest Plots
5.1
Generating a Forest Plot
5.2
Layout types
5.3
Saving the forest plots
6
Between-study Heterogeneity
6.1
Assessing the heterogeneity of your pooled effect size
6.2
Detecting outliers & influential cases
6.2.1
Searching for extreme effect sizes (outliers)
6.3
Influence Analyses
7
Subgroup Analyses
7.1
Subgroup Analyses using the Mixed-Effects-Model
7.2
Subgroup Analyses using the Random-Effects-Model
8
Meta-Regression
8.1
Calculating meta-regressions in R
8.2
Plotting regressions
8.3
Multiple Meta-Regression
8.3.1
Common pitfalls of multiple meta-regression models
8.3.2
Model building methods
8.3.3
Using
metafor
to compute Multiple Meta-Regressions
9
Publication Bias
9.1
Small-study effect methods
9.1.1
Funnel plots
9.1.2
Testing for funnel plot asymmetry using Egger’s test
9.1.3
Duval & Tweedie’s trim-and-fill procedure
9.2
P
-Curve
9.2.1
Performing a
p
-curve analysis
9.2.2
Estimating the “true” effect
10
Risk of Bias summary
10.1
Preparing your Risk of Bias data
10.2
Plotting the summary
10.3
Saving the Summary Plot
II Advanced Topics
11
Network Meta-Analysis
11.1
Frequentist Network Meta-Analysis
11.1.1
The Network Meta-Analysis Model
11.1.2
Performing a Network Meta-Analysis using the
netmeta
package
11.1.3
Evaluating the validity of our results
11.1.4
Summary
11.2
Bayesian Network Meta-Analysis
11.2.1
The Network Meta-Analysis Model
11.2.2
Performing a Network Meta-Analysis using the
gemtc
package
11.2.3
Assessing inconsistency: the nodesplit method
11.2.4
Generating the network meta-analysis results
11.2.5
Summary
12
“Multilevel” Meta-Analysis
12.1
Fitting a three-level model
12.1.1
Data preparation
12.1.2
Model fitting
12.1.3
Distribution of total variance
12.1.4
Comparing the fit
12.2
Subgroup Analyses in Three-Level Models
13
Structural Equation Modeling Meta-Analysis
13.1
The idea behind Meta-Analytic SEM
13.1.1
Model Specification
13.1.2
Meta-Analysis from a SEM perspective
13.1.3
The Two-Stage Meta-Analytical SEM approach
13.2
Multivariate Meta-Analysis
13.2.1
Specifying the Model
III Helpful Tools
14
Effect size calculators
14.1
Hedges’ g from the Mean and SD
14.2
Hedges’
g
from a regression coefficient
14.2.1
Unstandardized regression coefficients
14.2.2
Standardized regression coefficents
14.3
Odds Ratio from
Chi-square
14.4
Hedges’
g
from a one-way ANOVA
14.5
Hedges’
g
from the Mean and SE
14.6
Hedges’
g
from a correlation
14.7
Hedges’
g
from an independent t-test
14.8
Hedges’
g
from Cohen’s
d
14.9
Multiple comparisons
14.10
Number Needed to Treat (
NNT
)
14.10.1
The
NNT
function
14.11
Standard Error from
p
-value
15
Power Analysis
15.1
Fixed-Effect Model
15.2
Random-Effects Model
15.3
Subgroup Analyses
15.4
Power Calculator Tool
References
How to cite this guide
Get the
dmetar
package
Doing Meta-Analysis in R
(META-ANALYSIS) Meta-Analysis in R
