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
- :
- Discussions
- :
- REML mixed model

Topic Options

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

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

Mar 6, 2017 9:19 AM
(864 views)

Hello,

I have few very general questions related to running a mixed model using REML.

1. Why is REML always recommneded ? what's the most obvious advantages of it over EMS method?

2. How do I make meaningful inferences for the given **residual by predicted plot**?

3. How do I generat %CV from REML?

Thank you.

5 REPLIES

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

Mar 6, 2017 9:40 AM
(860 views)

Also How do you interpret a __negative value__ **Variance component** of a random factor in REML?

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

Mar 10, 2017 1:22 AM
(782 views)

I am not sure that there is a meaningful interpretation of a negative variance component.

I would first check that it is significantly non-zero: do CIs overlap with zero? The most sensible interpretation is that it is a neglible component of total variation.

If you have a large significant negative variance component that might give you cause to question the model.

I believe you can specify to bound all variance components to be positive if you would prefer.

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

Mar 10, 2017 1:18 AM
(783 views)

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

Mar 10, 2017 6:18 AM
(772 views)

- Select
**Help**>**Books**>**Fitting Linear Models**.**Chapter 3**is devoted to fitting models with random effects using REML estimation procedure. In general, EMS only works in balanced designs. REML gives the same result in such cases so you lose nothing by always using it. The only disadvantage is that REML is more computationally expensive but that is a moot point today. - The Residual by Predicted Plot is not intended to be used for inference (hypothesis tests). It is a visual assessment of the estimated errors (residuals) to identify violations of the assumptions of the linear regression such as lack of fit, heteroscedasticity, and influential observations.
- You compute the %CV from the variance components as 100*Sqrt(VC)/Mean as usual.

The negative variance is actually a negative covariance, if that interpretation helps you to accept such a result. Use the confidence interval for the estimate to decide if it is not zero. REML is more flexible than EMS. By allowing negative estimates around zero and adjusting the degrees of freedom, you get the desired coverage from the inference about the fixed effects.

Learn it once, use it forever!

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

Mar 10, 2017 10:37 AM
(766 views)