Mahalanobis distance matching on test and control groups

[#1]

Hi there,

How do I use JMP to do Mahalanobis distance matching on a test and control group of data so as to better balance the covariates?

Thx,

Ed

[#2]

Hi @rocdoc ,

  The easiest way to do that is to go: Analyze > Multivariate (cast your columns into the selected roles). If you have a Data Set column that is split into test and control, you could use that as your "By" column.

  Then, the hot button next to Multivariate select Outlier Analysis > Mahalanobis Distance. You could also select T^2, which is the square of the M-D.

  You could also do something similar with Analyze > Principal Components, red hot button, and then select Outlier Analysis (this can be adjusted for the number of PCs that you chose to use). So, depending on what you're after there's a couple ways to do it.

Hope this helps!,

DS

[#3]

Thx DS.
Follow-up query. I did try the MD option as you suggest, but it seems to provide analyses separately for each group? I am trying to ensure that test and control groups are "similar" in terms of their covariates, using distance matching. How do I do that?
Regards,
Ed