Subscribe Bookmark RSS Feed

Trying to understand the SNR used in JMP 10 power analysis

wakester

Community Trekker

Joined:

Jun 11, 2014

In JMP 10 power analysis, the SNR is specified by the user.  Is the unknown sigma in that SNR my pilot study's standard dev divided by sqrt of my design size, or is it just the pilot study standard dev?

Dave Wakefield

1 ACCEPTED SOLUTION

Accepted Solutions
statcon

Community Trekker

Joined:

Dec 21, 2011

Solution

Hi Dave,

it's all about the anticipated RMSE of your model. So if you have a pilot study you would fit an adequate regression model and take that RMSE as an estimator. Another solution would be to use the standard deviation of some replicates in your data.

Bradley Jones wrote a pice on the different components of power analysis in JMP10 (compared to JMP 9): Power Analysis in Custom Design for a designed experiment in JMP 10

Hope this helps.

2 REPLIES
statcon

Community Trekker

Joined:

Dec 21, 2011

Solution

Hi Dave,

it's all about the anticipated RMSE of your model. So if you have a pilot study you would fit an adequate regression model and take that RMSE as an estimator. Another solution would be to use the standard deviation of some replicates in your data.

Bradley Jones wrote a pice on the different components of power analysis in JMP10 (compared to JMP 9): Power Analysis in Custom Design for a designed experiment in JMP 10

Hope this helps.

wakester

Community Trekker

Joined:

Jun 11, 2014

Thanks Statcon.  I was thinking I had to figure out what JMP was doing with the non-centrality parameter to back out effect size.

Respectfully,

Dave