topic Re: Comparison of two groups that have same mean, but have very different varia in Statistical Procedures

Comparison of two groups that have same mean, but have very different varia

deleted_user — Tue, 15 Jun 2010 20:02:06 GMT

Hello,

I am infant in using JMP, but Ia m very impressed to the capability of JMP.

I wonder there is a procedure available in JMP to compare two groups that have the same mean, but have very different different range of variation?

How can I address the difference in two group in a statistical way using JMP?

Thank you for your comment.

Myo

Re: Comparison of two groups that have same mean, but have very different varia

Cynthia_sas — Tue, 15 Jun 2010 20:13:28 GMT

Hi:
There is a special forum just for JMP users. You may find more assistance there.
http://support.sas.com/forums/forum.jspa?forumID=43

cynthia

Re: Comparison of two groups that have same mean, but have very different varia

Doc_Duke — Tue, 15 Jun 2010 21:23:22 GMT

Cynthia's right, the JMP forum will get you more directly to the "how to" part of your answer.

Statistically, "range of variation" can mean different things. If you are saying "variance" (e.g. assuming a normal distribution in each group), then the "Folded-F" test that usually is part of a t-test output will give it to you. This test is sensitive to non-normality.

If your data are not normally distributed, then you are likely to end up with some sort of bootstrap procedure to compare percentiles of the distributions.

Re: Comparison of two groups that have same mean, but have very different varia

deleted_user — Tue, 15 Jun 2010 21:26:54 GMT

Thank you very much for your comment. I will look for what I need there. Best regards, Myo

Re: Comparison of two groups that have same mean, but have very different varia

deleted_user — Tue, 15 Jun 2010 21:39:38 GMT

Thank you so much your kind explanation. The importance of normal distribution makes sense in this comparison. Would you please give me some comment on the situation that if one group shows normal distribution while the other group does not show normal distribution. What could be the consequence of the analysis either by the Folded F-test or bootstrap procedure? Is there any preference or other option available?

Thank you again for your comment.

Best regards,

Myo

Re: Comparison of two groups that have same mean, but have very different varia

Doc_Duke — Wed, 16 Jun 2010 00:01:34 GMT

If one is normally distributed and the other is not, then the Folded-F will likely be conservative. I would consider a bootstrap approach at that juncture.

Re: Comparison of two groups that have same mean, but have very different varia

deleted_user — Wed, 16 Jun 2010 15:47:12 GMT

Thank you very much! I will try it. Best regards, Myo