Hello,
I am hoping to get some help on a multivariate analysis that adjusts for clustering. I have been trying to use the mvreg command, but realize now that this may not allow me to adjust for clustering. I am wondering if there are any alternatives - I had thought that using a generalized estimating equations using the xtgee command would be appropriate, but it doesn't seem that I am able to including multiple dependent variables. I am using Stata SE version 14.1
Here are the details in what I have tried so far:
I initially tried running a multivariate multiple linear regression analysis, where my dependent variables are both continuous (depvar1=balance time on right foot, depvar2= balance time on left foot). My independent variables are sport specialization level (categorical; low, medium, or high), sex (categorical; male or female) and hours of sport participation (continuous). I would also like to adjust for clustering by school (total clusters=4).
I tried using the code --mvreg-- as follows:
I have had added the cluster command at the end as I normally would for a single outcome multiple linear regression (non multivariate), but have been receiving an error message saying "option cluster() not allowed". After searching many stata manuals and online forums, I realized that there may not be the option to adjust for cluster with this type of analysis. I'm not sure if this is a limitation of Stata, or if this is just not a function of this type of analysis in any software.
It was suggested to me to try a GEE model. This allows me to adjust for clustering, but I cannot figure if how to include two dependent variables (or if it is even possible to do so). The code that I have been using for this is:
but as you can see I am only able to include one of my dependent variables in the code, otherwise my second dependent variable is recognized as an independent variable.
Any advice on this type of analysis or adjustments to the above codes would be greatly appreciated.
Kind Regards.
I am hoping to get some help on a multivariate analysis that adjusts for clustering. I have been trying to use the mvreg command, but realize now that this may not allow me to adjust for clustering. I am wondering if there are any alternatives - I had thought that using a generalized estimating equations using the xtgee command would be appropriate, but it doesn't seem that I am able to including multiple dependent variables. I am using Stata SE version 14.1
Here are the details in what I have tried so far:
I initially tried running a multivariate multiple linear regression analysis, where my dependent variables are both continuous (depvar1=balance time on right foot, depvar2= balance time on left foot). My independent variables are sport specialization level (categorical; low, medium, or high), sex (categorical; male or female) and hours of sport participation (continuous). I would also like to adjust for clustering by school (total clusters=4).
I tried using the code --mvreg-- as follows:
Code:
mvreg depvar1 depvar2 = i.spec_level i.sex c.hours, cluster(school)
I have had added the cluster command at the end as I normally would for a single outcome multiple linear regression (non multivariate), but have been receiving an error message saying "option cluster() not allowed". After searching many stata manuals and online forums, I realized that there may not be the option to adjust for cluster with this type of analysis. I'm not sure if this is a limitation of Stata, or if this is just not a function of this type of analysis in any software.
It was suggested to me to try a GEE model. This allows me to adjust for clustering, but I cannot figure if how to include two dependent variables (or if it is even possible to do so). The code that I have been using for this is:
Code:
xtgee depvar1 i.spec_level i.sex c.hours, family(gaussian) link(identity) corr(independent) i(school)
but as you can see I am only able to include one of my dependent variables in the code, otherwise my second dependent variable is recognized as an independent variable.
Any advice on this type of analysis or adjustments to the above codes would be greatly appreciated.
Kind Regards.
Comment