Hi everyone,
I have the following dataset and I want to create a forest plot for some correlation coefficients. Each correlation coefficient has an assigned weight and confidence interval that I calculated as these correlation coefficents are for specific cities in one region, and weights are assigned by population size. I have been using the following commands to make the forest plots, however, I am not sure how to include my own weights in the code. I am new to STATA so I have selected the random effects model, and the weights are being used from there. Below is my inputted dataset, code and output.
In my code, I calculated the relative difference and used civartolerence() as the meta set command was not accepting my original confidence intervals.
Code:
generate double relative_diff = reldif(correlationcoefficient-lowercl, uppercl-correlationcoefficient)
summarize relative_diff
meta set correlationcoefficient lowercl uppercl, civartolerance(.18)
meta forestplot _plot _esci _weight _es lhin
Output:
. summarize relative_diff
Variable | Obs Mean Std. dev. Min Max
-------------+---------------------------------------------------------
relative_d~f | 14 .0969017 .0380708 .0107149 .1718803
. meta set correlationcoefficient lowercl uppercl, civartolerance(.18)
Meta-analysis setting information
Study information
No. of studies: 14
Study label: Generic
Study size: N/A
Effect size
Type: <generic>
Label: Effect size
Variable: correlationcoefficient
Precision
Std. err.: _meta_se
CI: [_meta_cil, _meta_ciu]
CI level: 95%, controlled by level()
User CI: [lowercl, uppercl]
User CI level: 95%, controlled by civarlevel()
Model and method
Model: Random effects
Method: REML
.
. meta forestplot _plot _esci _weight _es lhin
Effect-size label: Effect size
Effect size: correlationcoefficient
Std. err.: _meta_se
Note: I have not included full results here.
Any suggestions for how to include my own weights into the forest plot would be greatly appreciated.
Thanks,
Maliha
I have the following dataset and I want to create a forest plot for some correlation coefficients. Each correlation coefficient has an assigned weight and confidence interval that I calculated as these correlation coefficents are for specific cities in one region, and weights are assigned by population size. I have been using the following commands to make the forest plots, however, I am not sure how to include my own weights in the code. I am new to STATA so I have selected the random effects model, and the weights are being used from there. Below is my inputted dataset, code and output.
In my code, I calculated the relative difference and used civartolerence() as the meta set command was not accepting my original confidence intervals.
Code:
generate double relative_diff = reldif(correlationcoefficient-lowercl, uppercl-correlationcoefficient)
summarize relative_diff
meta set correlationcoefficient lowercl uppercl, civartolerance(.18)
meta forestplot _plot _esci _weight _es lhin
Output:
. summarize relative_diff
Variable | Obs Mean Std. dev. Min Max
-------------+---------------------------------------------------------
relative_d~f | 14 .0969017 .0380708 .0107149 .1718803
. meta set correlationcoefficient lowercl uppercl, civartolerance(.18)
Meta-analysis setting information
Study information
No. of studies: 14
Study label: Generic
Study size: N/A
Effect size
Type: <generic>
Label: Effect size
Variable: correlationcoefficient
Precision
Std. err.: _meta_se
CI: [_meta_cil, _meta_ciu]
CI level: 95%, controlled by level()
User CI: [lowercl, uppercl]
User CI level: 95%, controlled by civarlevel()
Model and method
Model: Random effects
Method: REML
.
. meta forestplot _plot _esci _weight _es lhin
Effect-size label: Effect size
Effect size: correlationcoefficient
Std. err.: _meta_se
Note: I have not included full results here.
Any suggestions for how to include my own weights into the forest plot would be greatly appreciated.
Thanks,
Maliha
