Turn on suggestions

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

Showing results for

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Estimating random effect when there exist none (simulation)

Options

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

🔒 This topic is **solved** and **locked**.
Need further help from the community? Please
sign in and ask a **new** question.

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

Posted 02-07-2018 04:27 PM
(1299 views)

Hi all,

I am trying to run simulation for a mixed model. For one of the variable (xij) I do not add random effect (non-varying) at the data generation process but try to estimate the random effect during the estimation process. The expectation is that estimated variance component should be zero but it is not. This is the case even group-mean centering the variable. Is there an obvious explanation for this? I am not familiar with the estimation procedure of the PROC MIXED. The data generating mechanism and the model is below

```
* Level-1 Model
* yij = beta0j + beta1j*xij + .50*vij + rij
* Level-2 Model
* beta0j = gamma00 + gamma01*tj + u0j
* beta1j = gamma10 + gamma11*tj
```

I am trying to estimate random term for xij with PROC MIXED as

```
proc mixed data=SimData2 covtest NOCLPRINT method = REML;
class SchID;
model yij = xij / ddfm = BW solution;
random INTERCEPT xij /subject = SchID type=UN g gcorr;
ods output CovParms=CovParms1 SolutionF=SolutionF1;
by SampleID;
run;quit;
```

Please help,

Thanks

1 ACCEPTED SOLUTION

Accepted Solutions

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

But the average doesn't have to be zero, no matter how many runs you make (unless you can do an infinite number of runs).

The real test is as I described, that the values are not statistically different than zero, 95% of the time.

--

Paige Miller

Paige Miller

4 REPLIES 4

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

Perhaps this is just a difference in wording, but I would not agree with this statement:

The expectation is that estimated variance component should be zero but it is not

I would say there that the estimated variance component should *not be statistically different* than zero, 95% of the time (assuming you do the test with alpha = 0.05 and the errors are iid normal)

But in an individual run a non-zero estimated variance component does not bother me.

--

Paige Miller

Paige Miller

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

But the average doesn't have to be zero, no matter how many runs you make (unless you can do an infinite number of runs).

The real test is as I described, that the values are not statistically different than zero, 95% of the time.

--

Paige Miller

Paige Miller

- Mark as New
- Bookmark
- Subscribe
- Mute
- RSS Feed
- Permalink
- Report Inappropriate Content

**Available on demand!**

Missed SAS Innovate Las Vegas? Watch all the action for free! View the keynotes, general sessions and 22 breakouts on demand.

What is ANOVA?

ANOVA, or Analysis Of Variance, is used to compare the averages or means of two or more populations to better understand how they differ. Watch this tutorial for more.

Find more tutorials on the SAS Users YouTube channel.