topic Re: Calculate correlation coefficients from a mixed models dataset in SAS Procedures

Calculate correlation coefficients from a mixed models dataset

AEP — Thu, 03 Aug 2017 17:50:41 GMT

Hello,

I was wondering if there was a way to calculate correlation coefficients from a dataset that was formated for repeated measure mixed models. I would like to do a correlation between variable A and B, during a pre-determined time. Before I reformat my dataset in a GLMs analysis format, I would like to know if it is possible. Thanks.

example of dataset

subno A B Time

001 4 7 1

001 2 3 1

001 2 4 1

001 9 7 2

001 5 6 2

001 4 0 2

002 4 7 1

002 2 3 1

002 2 4 1

002 9 7 2

002 5 6 2

002 4 0 2

Re: Calculate correlation coefficients from a mixed models dataset

PaigeMiller — Thu, 03 Aug 2017 17:55:03 GMT

Does PROC CORR do what you want? If not, why not?

Re: Calculate correlation coefficients from a mixed models dataset

Rick_SAS — Fri, 04 Aug 2017 14:23:04 GMT

I am tempted to suggest

proc corr data=have;
   by subno time;
   var a b;
run;

How is it that you have three measurements at each timepoint for each subject? Usually there is one measurement at each timepoint.

Re: Calculate correlation coefficients from a mixed models dataset

AEP — Sat, 05 Aug 2017 13:22:14 GMT

I found another way. But thank you for your suggestions.

Re: Calculate correlation coefficients from a mixed models dataset

PaigeMiller — Sat, 05 Aug 2017 13:39:40 GMT

@AEP wrote:

I found another way. But thank you for your suggestions.

@AEP Can you share it with the rest of the community?