Questions tagged [meta-analysis]
Methods focused on contrasting and combining results from different studies, in the hope of increasing precision and external validity.
1,559 questions
0
votes
0
answers
35
views
Combining prior on effect size with new effect size estimate (Bayesian)
CONTEXT
I am a researcher planning experiments. My goal is to combine a prior on the effect size from prior literature with pilot data.
SETUP
Assume I have a prior on an effect size,
$d \sim N(d_{\...
- 219
2
votes
0
answers
45
views
How to deal with missing standard errors from primary studies in meta-analyses?
I'm conducting a meta-regression in the field of Economics where my primary studies are regression analyses. To ensure comparability between regression results, I'm only considering regressions that ...
- 21
1
vote
1
answer
51
views
chi-squared significant for dose-response function obtained through meta-analysis but confidence interval around plotted predicted values not
The question:
I obtained a dose-response function (predictor vs relative risk) through meta-analysis using the dosresmeta package for R. The package applies a random effects model (REML)and I fitted a ...
- 413
1
vote
0
answers
40
views
Compare two independent meta-analysis
I extracted the overall effect from an existing meta-analysis that I would like to compare with the overall effect of my meta-analysis by performing an equivalence test (These are two independent meta-...
- 83
2
votes
1
answer
81
views
Which model is used by default in subgroup analysis by meta package for meta-analysis in R: random effects or mixed effects?
I posted this question on Stack Overflow but it was deemed off-topic there, so now I'm trying here.
I have undertaken a random effects meta-analysis using the meta package for R as follows:
...
- 413
1
vote
1
answer
48
views
Judging heterogeneity when it depends on study precision
My understanding of heterogeneity is that as the precision of included studies increases (i.e., due to larger sample sizes), heterogeneity metrics (e.g. $I^2$, $\tau^2$) will also increase.
If that is ...
- 81
1
vote
0
answers
46
views
Deriving effective sample size and effective residual error from random-effects model in meta-analysis
I am performing a meta-analysis between two cohorts. I want to aggregate the estimates I obtained across a series of variables for each cohort.
I know that two main models are used in meta-analysis: ...
- 267
3
votes
1
answer
135
views
What is the exact method of calculating a number needed to treat (NNT) in a meta-analysis?
I found the following table in a meta-analysis (https://doi.org/10.1016/s1474-4422(25)00068-7)
I noticed that if I were to "naively" calculate the NNT by taking the reciprocal of the ...
- 115
1
vote
0
answers
49
views
Meta-analysis when one study uses robust standard errors
We circulated an analysis plan and received logistic regression results back to summarize in a meta-analysis. However, one study (out of 6) used robust standard errors (because it was a family study). ...
- 11
4
votes
2
answers
188
views
Correlation values falling outside the estimated range derived from the p-value
According to the publication “Extracting Pre-Post Correlations for Meta-Analyses of Repeated Measures Designs” (available here), correlation values can be estimated if the p-value from a paired t-test ...
- 241
3
votes
1
answer
109
views
Calculating a new risk ratio from two risk ratios that share a control group?
Often in meta-analysis, we have to work with the effect sizes reported in the papers (because raw data is not available). Consider a non-randomised study comparing rates of death under the ...
1
vote
0
answers
50
views
Checking assumptions for meta-analytic three-level models
I'm conudcting meta-analysis and trying to fit three-level random- and mixed-effects models using rma.mv function of metafor package. I have raw mean difference as an effect size metrics.
I have read ...
- 11
4
votes
1
answer
161
views
Is Kaplan-Meier Analysis valid with aggregated study data?
I recently came across a paper in Scientific Reports and started questioning whether the methodology used to prepare data for Kaplan-Meier analysis is appropriate. Since I’m working with similar data ...
- 158
3
votes
2
answers
214
views
Minimum reporting set for results tables?
In results tables we might include $\beta$, SE, $t$, $d$, 95% CI, $p$, sample size, and model fit.
Often, though, just $\beta$, 95% CI, and $p$ are reported—or even just $\beta$ and $p$. That may be ...
- 1,049
2
votes
1
answer
60
views
Struggling in what to do with meta analysis multilevel
Im woking on meta analysis in accuracy models of genomic prediction in maize using pearson correlation, i also converted to fisher Z using the funcion escalc, and in my case, i considered as ...
- 21
5
votes
1
answer
236
views
Meta analysis for one-sample proportion with 0 events in some studies?
I am doing meta analysis for a one-sample proportion where some of the studies have 0 events. My understanding of the statistical literature is that:
Traditional meta analysis methods that require a ...
- 925
1
vote
0
answers
74
views
Understanding "V appears to be not positive definite" warning in rma.mv() when using rcalc() for multilevel meta-analysis of correlations
I am conducting a meta-analysis of correlations between two constructs using the metafor package in R. I am using the rcalc() ...
0
votes
0
answers
42
views
Is it correct to use the stratification for subgroup analysis in a meta-analysis of proportions when having study(s) include both types of cases?
I am conducting a meta-analysis of proportions, specifically focusing on the prevalence of a certain infectious disease.
I am planning to conduct a subgroup analysis of the prevalence of infectious ...
- 13
5
votes
1
answer
109
views
Question about metafor::escalc calculations (SMN)
I'm attempting to use metafor for an atypical project and settled on using a "single-group standardized mean" (SMN) as an effect size and now have a question about how escalc is calculating ...
- 76
4
votes
2
answers
319
views
Definition of the random-effect model
I am studying the random-effects model as a preparation for a meta-analysis. Unfortunately I am confused because there seem to be two inequivalent descriptions of the model in the literature:
In the ...
- 283
0
votes
0
answers
58
views
Difference between fixed-effect model and random-effect model [duplicate]
I am studying the fixed-effect model and the random-effect model as a preparation for a meta-analysis. In the Handbook of Meta-Analysis, equations (4.2) and (4.6) and Applied Meta-Analysis with R and ...
- 283
1
vote
1
answer
71
views
Meta-analysis on incidence - how to manage missing follow-up duration in some studies
I'm interested in performing a meta-analysis estimating the incidence (event/person-year) of a certain adverse event of interest in those who have undergone a procedure. I have identified a list of ...
- 105
2
votes
2
answers
137
views
Meta-analysis using slopes from non-linear regressions as the effect sizes
As discussed here, using linear regression slopes as effect sizes in meta-analysis is legitimate, provided that all the regression models from which they are derived use the exact same variable across ...
1
vote
1
answer
72
views
P-values in pre post treatment design (meta analysis)
I have a question regarding the use of a pre-post design without a control group (i.e., treatment-only) in a meta-analysis. In our dataset, some studies used nonparametric tests such as the Wilcoxon ...
- 11
0
votes
0
answers
58
views
When calculating odds ratios (ORs) between subgroups in a meta-analysis of proportions, is it required that all subgroups include the same studies?
I am conducting a meta-analysis of proportions, specifically focusing on the prevalence of a infectious disease. I have attempted to calculate the odds ratios using the meta package in R.
...
- 13
1
vote
1
answer
75
views
Is it possible to calculate odds ratios when performing subgroup analyses for two or more groups in conducting a meta-analysis of proportions?
I am conducting a meta-analysis of proportions, specifically focusing on the prevalence of a certain infectious disease. When performing subgroup analyses, such as comparing the prevalence rates of ...
- 13
2
votes
1
answer
165
views
Meta analysis combining relative risks and hazard ratios?
Is it possible to do meta analysis for combining relative risks and hazard ratios? Some of the studies I have only report relative risks for a binary outcome, while other studies report hazard ratios ...
- 925
0
votes
1
answer
212
views
Egger's Regression test
I am performing an Egger's regression test on a meta-analysis of single case studies. I used SPSS to perform this test and received the following output (see picture). I am not entirely sure how to ...
- 1
1
vote
0
answers
55
views
Entering data into a Tau-U effect size calculator
When calculating Tau-U effect sizes using an online single-case research calculator (https://singlecaseresearch.org/calculators/tau-u/), what is the best approach. It compares up to 5 AB contrasts (...
- 11
1
vote
1
answer
82
views
Meta regression: standardized effect size for how each study feature affects between study variability?
I'm doing meta analysis for a one-sample proportion, where I've obtained the estimated proportions from multiple studies.
There is substantial heterogeneity in the proportion estimates between studies ...
- 925
2
votes
0
answers
66
views
How to calculate weighted average odds ratios?
I am doing a meta-analysis on five papers about wound complications. I have calculated the ORs from the separate type of wound complications, but would like to calculate a weighted average to get the ...
- 21
5
votes
1
answer
203
views
Random-effects modeling vs Stouffer's method in meta-analysis
Suppose both point estimates and their corresponding $Z$-statistic values are available for meta-analysis. A random-effects meta-analysis model is generally considered the appropriate approach, as it ...
- 2,831
4
votes
1
answer
101
views
Partitioning $I^2$ heterogeneity estimates by each random effect when using the Jackson et al. (2012) pseudo-$I^2$ method
As established elsewhere on Stack [1], both the Jackson et al. (2012) [2] and the Nakagawa & Santos et al. (2012) [3] methods are viable methods for calculating $I^2$-like heterogeneity estimates ...
4
votes
1
answer
176
views
Calculating total and per random effect I2 where one random effect is nested
Following on from some other stack discussions (I2 (I-squared) estimates for heterogeneity of independent meta-analysis versus a single meta-regression Computing I2 heteorogeneity in metafor when ...
0
votes
0
answers
44
views
meta-SEM and correlation matrix
I want to conduct a meta-SEM to test the predictors and outcomes of a variable.
My question is: I don’t have the correlations between each independent variable (IV) and other IVs, nor between each ...
- 1
2
votes
1
answer
102
views
Computing I2 heteorogeneity in metafor when using inner|outer specification of random effect
I am trying to compute I2 heterogeneity for a multilevel model in metafor sensu Nakagawa & Santos, 2012, as described here: https://www.metafor-project.org/doku.php/tips:i2_multilevel_multivariate....
0
votes
1
answer
95
views
RMD (Relative mean difference) effect size
In Animal Sciences some people use relative mean difference (RMD) effect size in meta-analysis,
RMD= [(treatment mean - placebo mean)/placebo mean)] × 100.
My question is if we calculate ES this way ...
0
votes
0
answers
42
views
How can I conduct a meta-analysis on a construct with five dimensions?
I want to conduct a meta-analysis on the antecedents, outcomes, and moderators of work engagement. However, when searching the database, I noticed that some studies examine the five dimensions of work ...
- 1
0
votes
0
answers
65
views
Number of correlation terms needed for multi-(four-level)-meta-analysis
I am conducting a multi-level meta-analysis of randomized controlled trials (RCTs) examining the effects of treatment X for a particular condition (A). My goal is to appropriately model variance ...
- 25
0
votes
1
answer
116
views
Is it possible to convert the meta-analytic SMD (e.g., Cohen's d) to the percentage change between the two conditions?
Is it possible to convert average summary effect of random-effects meta-analysis in the form of Cohen's d to percentage change between the two conditions.
For example: A Cohen's d of X corresponds to ...
- 136
3
votes
1
answer
177
views
Convert partial eta squared to Cohen's d and find variance
I am conducting a metanalysis in R. For most studies, I am computing Hedges's g using the means and SDs of groups. However, for some studies I only have the partial $\eta ^2$ and 95% confidence ...
- 688
0
votes
1
answer
46
views
Is Homogeneity of Measurement Tools Necessary When Pooling Odds Ratios for Disease Classification?
When pooling estimates in meta-analysis, it is well-known that maintaining homogeneity among the studies is crucial, particularly when dealing with continuous data. In such cases, it is important to ...
0
votes
0
answers
44
views
Model simplification for nested data in meta-analysis
I am conducting a meta-analysis with differing amounts of heterogeneity in subgroups. Since the data structure is nested (effect sizes nested in studies) I took this into account by adding random ...
- 158
0
votes
0
answers
22
views
How to do pooled meta analysis of single arm studies with continous data (pre-post intervention) in R? [duplicate]
I tried to do meta analysis of single arm studies (without control) using R studio, but I still can't find how to do the pooled analysis. I only have the mean and SD from pre treatment and mean with ...
- 1
0
votes
1
answer
108
views
What is the relationship between OR, se, and CI
Apologies if this is a duplicate question, but I have been unable to find a clear answer.
What is the relationship between Odds Ratio, standard error, and confidence intervals?
I try to do a meta-...
- 13
1
vote
0
answers
77
views
Wald test with permutation standard error?
Many statistical tests are based on a Wald-like z-statistic: $z = \frac{\hat{\theta} - \theta_0}{\text{SE}(\hat{\theta})}$ where $\theta$ is the parameter of interest, $\theta_0$ is the value under ...
- 925
0
votes
1
answer
80
views
How to obtain one study effect from several within-study effects?
I want to do a narrative review (as a meta-analysis is not possible), and I want to conclude per study whether the predictor was significant yes/no. However, studies have several effect sizes per ...
- 13
3
votes
1
answer
147
views
Stouffer's Method: Restriction on underlying hypothesis tests producing p values?
When using Stouffer's method to combine p values, they first have to be converted to z scores. Does this mean that p values that are derived for example from a permutation test are not valid inputs, ...
- 33
0
votes
2
answers
191
views
Pre-post design (no control) at baseline and multiple follow up times (Meta analysis)
I am working on a meta-analysis project and need advice on choosing the right statistical method. We have access to around 18 publications. The design is similar to pre-post treatment design (no ...
- 11
3
votes
1
answer
181
views
Different results meta vs metafor for three-level meta-analysis of proportions
The R package “meta” has a function “metaprop” for meta-analysis of proportions and the argument “cluster” allows for three-level meta-analyses to be fitted. The R documentation (version 8.0-1) ...
