topic group matching (case control matching?) in Discussions

group matching (case control matching?)

OldGugga — Fri, 09 Jun 2023 00:27:50 GMT

how do i use JMP to equalize or match 6 groups of children on age, for a 3 way ANOVA? the groups range in n from about 30 to 200. i would prefer not to use child age as a covariate in the analysis.

Re: group matching (case control matching?)

Kevin_Anderson — Fri, 22 Jan 2021 22:11:57 GMT

Hi, OldGugga!

There can be dangerous Rodents Of Unusual Size in the Fire Swamp of vaguely-posed statistical questions like yours, but I will boldly try to help you in spite of knowing nothing about what you're trying to do or how the experiment you're analyzing was designed or performed.

A titan, George Edward Pelham Box FRS has an attributed quote about variation in experiments that goes "Block what you can, randomize what you cannot." You may be able to block-align the groups of Age and proceed with your analysis. But, be aware that blocking can harm the sensitivity of the analysis in the presence of block-treatment interactions, which with Age you may indeed have. You can check for this in the analysis. And it would be better if your experiment had been designed as a Randomized Block Design in the first place...

You might be well served by seeking professional help and consulting with a statistician.

Good luck!

Re: group matching (case control matching?)

OldGugga — Sat, 23 Jan 2021 00:10:09 GMT

thank you sincerely for your bold reply.
a trusted statistician suggested covarying on S age. i see problems in that but there could be problems in any analysis approach
i have to think about this
again thank you