Hi
I have calculated effect sizes (Cohen's d) or SMD (for those who prefer it) and am having problems running a meta analysis in STATA. Does anyone know the coding for the same?
Thanks
Hi Devika,
The generic command for a continuous outcome variable is
metan tsample tmean tsd csample cmean csd, options
where tsample & csample = number of observations in test and control groups, respectively; mean and sd refer to mean and standard deviation.
You can also use the mean difference between the two groups along with the 95% CI.
Pl let me know if you still have any problems.
Best
Jeeva
Dear Devika,
I think there are better tools for meta-analysis than stata. For more advanced meta-analysis, including meta-regression, I would recommend the r package metafor by Wolfgang Viechtbauer. For more basic meta-analysis our tool Meta-Essentials might be of interest. It consists of a set of workbooks for Microsoft Excel that are free and easy to use. It calculates all the required statistics and provides the accompanying plots (e.g., forest plot, moderator analysis, subgroup analysis and a wide range of publication bias analyses). You can find the tool at www.meta-essentials.com
Best regards,
Henk
43
Despite what Henk says, Stata will do exactly what you want here. Learning R is time-consuming, and I'd only recommend it if you need to do something that is impossible otherwise (that's how I started!).
There are many Stata commands for various types of meta-analysis, and I'd recommend the book "Meta-Analysis in Stata" for its clearly-worked examples. http://www.stata.com/bookstore/meta-analysis-in-stata/
http://www.stata.com/bookstore/meta-analysis-in-stata/
Dear Prof. Conroy,
I just want to note that I didn't say it is not possible with Stata, just that I think there are better tools. I fully agree with you that Stata can indeed do what Devika is looking for, but in my opinion other, more user friendly, tools are available. The reason why I mentioned the metafor package for r is that I think it is the best tool for advanced meta-analyses. For more basic meta-analyses I would recommend to use our easy and free tool called Meta-Essentials (see www.meta-essentials.com), rather than Stata.
By the way, do you know MAVIS (see http://kylehamilton.net/shiny/MAVIS/)? It provides a relatively simple web-based user interface for meta-analysis, primarily based on metafor.
Best regards,
Henk
142
MAVIS is really neat! Have to add that to my list of resources – thanks Henk!
Hi dear
You can use :
metan n1 mean1 sd1 n2 mean2 sd2, options ..........
Dear Devika,
I am about to run a mata-analysis using effect sizes in Stata, as you mentionned in your question, and I have problems as well to find the appropriate command. Have you find a solution?
Many thanks,
Marion
Hi Marion Carayol,
If you are doing a meta-analysis using continuous data, here are some commands:
1) When effect sizes were already extracted: metan effect size standard error, options
2) When you would like to compute the effect sizes through the sample size, mean difference and SD: metan n1 mean1 SD1 n2 mean2 SD2, options
If you still have problems, I suggest updating the package.
Cheers,
Pedro Lopez
Jeeva Sankar how to change effect size in meta forest plot. and lets suppose in my meta forest plot I wanna add extra column after author/study year how can I do that. I would appreciate your help in that.
Yasar Sattar Thanks for the query.
You can use the term lcols(varlist) as an option (i.e., after the comma) in the metan command to add additional columns to the left of the forest plot.
Jeeva
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis