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
- /
- PROC MI FCS documentation appears to be incorrect

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

05-04-2017 05:47 AM

The documentation in (SAS/STAT 14.1) PROC MI 'FCS Methods for Data Sets with Arbitrary Missing Patterns' doesn't look correct. Specifically, in the two sets of equations for the filled in and imputation phases, for each equation corresponding to a draw of the missing data, the equation implies that the draw is from a marginal distribution, without covariates. For example, Y^(t+1)_{p*} ~ P(Y_p | theta^(t+1)_p), in which there are no covariates shown in the conditional distribution.

The accompanying text describes the use of covariates in the conditional distributions which are drawn from, but if the equations are shown, they ought to make clear the dependence on covariates, otherwise they do not faithfully describe the algorithm's implementation.