turn on suggestions

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

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- Test to determine if I need to account for correla...

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
- Highlight
- Email to a Friend
- Report Inappropriate Content

10-19-2016 11:14 AM

I have a dataset with the following form:

year location outcome covariate1 covariate2 covariate3

2007 00001 1 29.7 15.4 1

2008 00001 1 22.5 23.7 3

2007 00002 0 15.5 33.8 2

2008 00002 1 20.9 19.3 2

Outcome is binary, and covariates are a mix of categorical and continuous. What is a good approach to determine if my logistic regression needs to account for the repeated measures?

Accepted Solutions

Solution

10-19-2016
02:58 PM

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

10-19-2016 11:46 AM

All Replies

Solution

10-19-2016
02:58 PM

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

10-19-2016 11:46 AM

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

11-02-2016 12:41 PM

And yet, you might still want to consider the two years as non-independent. To me, one of the really nice things about modeling the error structure is that you can accommodate even small correlations. With only two repeated observations, an unstructured covariance matrix is the most flexible statement of the situation. Yes, the two years may have only a small correlation, but if so, that will not greatly affect any standard errors of the difference of fixed effect means. And there may not be enough data to adequately estimate confidence bounds, so that 0 might be included, even if the covariance is substantially away from 0.

My mantra is: If you measure the same experimental unit for the same endpoint, you ought to assume that those points are likely to be more closely related than points from independent units.

Steve Denham