Turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- JMP Academic
- :
- JMP Academic Discussions Forum
- :
- Power calculations for categorical variables.

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Jul 8, 2015 7:24 AM
(12914 views)

Can someone please explain how JMP calculates statistical power associated with the individual levels and the whole effect of a categorical variable? I know the equations used for calculating power in case of a continuous variable but they don't seem to work in the case of categorical variables.

1 ACCEPTED SOLUTION

Accepted Solutions

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Hi agn3284,

It sounds like you're calculating power with DOE > Sample Size and Power > K Sample Means, is that correct? The power calculated there is for the overall, omnibus ANOVA only. Pairwise tests have to be considered separately. Here is a link to some of the technical details for that calculation: Technical Details. Scroll down to "Power for a Categorical Effect."

To calculate power for pairwise companions you would need to use "Two Sample Means" rather than K sample means. Be sure to consider any of the alpha control procedures you'll be using when calculating prospective power.

I hope this helps!

4 REPLIES 4

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Hi agn3284,

It sounds like you're calculating power with DOE > Sample Size and Power > K Sample Means, is that correct? The power calculated there is for the overall, omnibus ANOVA only. Pairwise tests have to be considered separately. Here is a link to some of the technical details for that calculation: Technical Details. Scroll down to "Power for a Categorical Effect."

To calculate power for pairwise companions you would need to use "Two Sample Means" rather than K sample means. Be sure to consider any of the alpha control procedures you'll be using when calculating prospective power.

I hope this helps!

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Power calculations for categorical variables.

Hi Julian,

Thanks for the answer. I created a design using the custom design option consisting of just 1 categorical variable with 3 levels and N = 6 runs. I used the DOE> Evaluate Design option to look at the power values. Will the calculations be the same for those power values?

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Power calculations for categorical variables.

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Power calculations for categorical variables.

Hi Julian,

The answer was very helpful but I am a little confused on how to get the L matrix. I took the anticipated betas as 1 and -1 for levels 1 and 2 respectively. Can you please explain how to get the L matrix?