Why Meta-Mar?
Meta-Mar is a free online meta-analysis service developed as an adjunctive tool for running a full meta-analysis (including meta-regression and subgroup analysis) or can be used as a calculator/convertor of effect sizes!
- Possibility of choosing the Data entry methods between manual Data entry or .xlsx upload.
- Calculation of effect sizes based on SMD , Correlation and Ratios models for every single study.
- Calculation of the overall effect size of the analysis based on fixed and random effect models.
- Calculation of Fail-N Safe based on fixed and random effect models.
- Calculation of heterogeneity of the analysis (Q Cochrane, I2 and Tau2).
- Possibility of meta regression and subgroup analysis.
- Visualization of Forest Plot and Funnel Plot.
- Possibility of exporting the results of the analysis via a .xlsx file.
- Finally and regardless of your analysis, you may just want to use an Effect Size Calculator
Or take a look at a solved example by Meta-Mar:
Download the results as an Excel file
Table.1 summary of studies
Study | N1 | Mean1 | Sd1 | N2 | Mean2 | Sd2 | Moderator | subgroup | Cohen's d | CorrectionFactor | Hedges'g (SMD) | SEg | 95%CI-Lower | 95%CI-Upper | weight(%)-fixed model | weight(%)-random model % | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | A | 23 | 30 | 2.50 | 23 | 32 | 3.3 | 24 | subgroup1 | 0.683187 | 0.982857 | 0.671475 | 0.298164 | 0.087074 | 1.255876 | 3.063359 | 9.703777 |
2 | B | 47 | 33 | 3.34 | 47 | 35 | 3.1 | 22 | subgroup1 | 0.620687 | 0.991826 | 0.615613 | 0.209466 | 0.205059 | 1.026168 | 6.206951 | 9.982981 |
3 | C | 44 | 39 | 2.30 | 44 | 41 | 4.1 | 33 | subgroup1 | 0.601657 | 0.991254 | 0.596395 | 0.216064 | 0.172908 | 1.019881 | 5.833654 | 9.965102 |
4 | D | 78 | 47 | 4.10 | 78 | 55 | 2.5 | 40 | subgroup2 | 2.356000 | 0.995122 | 2.344508 | 0.207386 | 1.938030 | 2.750985 | 6.332079 | 9.988515 |
5 | E | 311 | 26 | 1.40 | 311 | 33 | 6.1 | 51 | subgroup2 | 1.581744 | 0.998790 | 1.579830 | 0.091769 | 1.399962 | 1.759698 | 32.337834 | 10.214405 |
6 | F | 144 | 75 | 2.60 | 144 | 80 | 4.2 | 41 | subgroup3 | 1.431496 | 0.997375 | 1.427739 | 0.131738 | 1.169531 | 1.685946 | 15.692129 | 10.155089 |
7 | G | 79 | 55 | 2.20 | 79 | 66 | 4.3 | 31 | subgroup1 | 3.220700 | 0.995185 | 3.205191 | 0.239966 | 2.734859 | 3.675524 | 4.729433 | 9.896295 |
8 | H | 59 | 26 | 6.60 | 59 | 34 | 2.1 | 29 | subgroup1 | 1.633504 | 0.993521 | 1.622919 | 0.211237 | 1.208896 | 2.036943 | 6.103360 | 9.978232 |
9 | I | 214 | 98 | 1.80 | 214 | 110 | 2.5 | 38 | subgroup1 | 5.508878 | 0.998238 | 5.499173 | 0.211284 | 5.085056 | 5.913291 | 6.100603 | 9.978104 |
10 | J | 110 | 72 | 5.10 | 110 | 77 | 5.6 | 42 | subgroup2 | 0.933561 | 0.996556 | 0.930345 | 0.141506 | 0.652994 | 1.207697 | 13.600598 | 10.137501 |
Table.2 Summary of results - fixed and random effect models
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Figure.1 Forestplot - fixed and random effect models
Figure.2 Funnel Plot - fixed and random effect models
Table.3 Fail-N Safe
Bias of the analysis regarding the file-drawer problem: Fail-N Safe, the number of studies (or samples) with a null effect (g = 0)needed to bring the calculated significance level of the pooled effect (p value < 0.0001) near the critical significance level (p value = 0.05),
is calculated as follows: References:
* Rosenberg, M. S. (2005). The file‐drawer problem revisited: a general weighted method for calculating fail‐safe numbers in meta‐analysis. Evolution, 59(2), 464-468.
* Rosenthal, R. (1979). The file drawer problem and tolerance for null results. Psychological bulletin, 86(3), 638.
|
tc(α = 0.05, df = 10) = 1.812 |
Zc(α = 0.05) = 1.645 |
|
|
|
Table.4 Results of Meta regression
Dep. Variable: | y | R-squared: | 0.050 |
---|---|---|---|
Model: | OLS | Adj. R-squared: | -0.069 |
Method: | Least Squares | F-statistic: | 0.4228 |
Date: | Thu, 19 May 2022 | Prob (F-statistic): | 0.534 |
Time: | 20:50:35 | Log-Likelihood: | -17.649 |
No. Observations: | 10 | AIC: | 39.30 |
Df Residuals: | 8 | BIC: | 39.90 |
Df Model: | 1 | ||
Covariance Type: | nonrobust |
coef | std err | t | P>|t| | [0.025 | 0.975] | |
---|---|---|---|---|---|---|
const | 0.5062 | 2.125 | 0.238 | 0.818 | -4.395 | 5.407 |
Moderator | 0.0383 | 0.059 | 0.650 | 0.534 | -0.097 | 0.174 |
Omnibus: | 9.889 | Durbin-Watson: | 2.280 |
---|---|---|---|
Prob(Omnibus): | 0.007 | Jarque-Bera (JB): | 4.420 |
Skew: | 1.515 | Prob(JB): | 0.110 |
Kurtosis: | 4.194 | Cond. No. | 154. |
Warnings:
[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.
Table.5 Results of Subgroup Analysis - fixed and random effect models
fixed model
k | Hedges's g | SEg | 95%CI lower | 95%CI upper | z score | p value | Heterogeneity % | df | |
---|---|---|---|---|---|---|---|---|---|
subgroup1 | 6 | 2.121569 | 0.092199 | 1.940860 | 2.302279 | 23.010814 | 3.632910e-117 | 98.770556 | 5 |
subgroup2 | 3 | 1.503470 | 0.072181 | 1.361994 | 1.644945 | 20.829051 | 2.360932e-96 | 94.037451 | 2 |
subgroup3 | 1 | 1.427739 | 0.131738 | 1.169531 | 1.685946 | 10.837672 | 2.282038e-27 | 0.000000 | 0 |
total | 10 | 1.689609 | 0.052186 | 1.587324 | 1.791893 | 32.376697 | 5.842358e-230 | 98.096375 | 9 |
random model
k | Hedges's g | SEg | 95%CI lower | 95%CI upper | z score | p value | Heterogeneity % | df | |
---|---|---|---|---|---|---|---|---|---|
subgroup1 | 6 | 2.036985 | 0.835576 | 0.399257 | 3.674714 | 2.437822 | 1.477604e-02 | 98.770556 | 5 |
subgroup2 | 3 | 1.601711 | 0.331891 | 0.951204 | 2.252218 | 4.826013 | 1.392938e-06 | 94.037451 | 2 |
subgroup3 | 1 | 1.427739 | 0.131738 | 1.169531 | 1.685946 | 10.837672 | 2.282038e-27 | 0.000000 | 0 |
total | 10 | 1.848747 | 0.395247 | 1.074064 | 2.623430 | 4.677453 | 2.904607e-06 | 98.096375 | 9 |
Results of ANOVA for subgroups
F value = 0.09313109449240511, p value = 0.9121839700696838