- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Understanding Effect test in regression
Can anybody show how sum of squares is calculated in effect test of regression.
Seen the help manual but could not find it.
Accepted Solutions
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Understanding Effect test in regression
Hi,
Look at the Analysis of Variance for the model. You will see that the total sum of squares is broken down into Model and Error. Model sum of squares can be further broken down into the sum of squares for each effect. This is what you see in the Effect Test report.
Try removing an effect from the model. You should see that the Model sum of squares is reduced by the sum of squares for the effect in the effect test.
So the effect test sum of squares is the difference in the Model sum of sqaures between the model with and without that effect.
I have assumed that you are comfortable with the concept of sum of squares but let me know if you need to know more.
Regards,
Phil
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Understanding Effect test in regression
@umbreon :
You might find a lot of helpful documentation on your question here (http://www.jmp.com/support/help/Regression_Reports.shtml)
Uday
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Understanding Effect test in regression
Thanks for replying.
But I have already gone through the pages from jmp. Apart from the hypothesis about effect test could find the way we carry out the analysis.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Understanding Effect test in regression
Hi,
Look at the Analysis of Variance for the model. You will see that the total sum of squares is broken down into Model and Error. Model sum of squares can be further broken down into the sum of squares for each effect. This is what you see in the Effect Test report.
Try removing an effect from the model. You should see that the Model sum of squares is reduced by the sum of squares for the effect in the effect test.
So the effect test sum of squares is the difference in the Model sum of sqaures between the model with and without that effect.
I have assumed that you are comfortable with the concept of sum of squares but let me know if you need to know more.
Regards,
Phil
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content