Date:         Fri, 2 Dec 2005 11:28:53 -0500
Reply-To:     Jerry Davis <jwd@GRIFFIN.UGA.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Jerry Davis <jwd@GRIFFIN.UGA.EDU>
Subject:      Re: Multiple Imputation in a Narrow Range
In-Reply-To:  <200512012200.jB1LQMgH010036@mailgw.cc.uga.edu>
Content-Type: text/plain; charset="US-ASCII"

Talbot Michael Katz wrote:

> finished.  These values are not true missing values; they are somewhere
> between 0 and the calibration threshhold, but they just can't be measured
> with the same accuracy as readings above the calibration threshhold, so
> there are no accurate values within that range.  How would you handle such
> values?  (Discarding is not an option.)

I recommend that they be excluded from the analysis.  Not missing, not
zero, not the lower detection limit but excluded.  If the instrument
can't accurately or precisely measure the response and you know
beforehand what the detection limits are, these values are meaningless.
I don't see any reason to guess or estimate these responses.

Sure, keep them around for informational purposes but don't model them.
Bad data doesn't add a lot of insight, IMHO.

Jerry

--
Jerry Davis
Experimental Statistics
UGA, CAES, Griffin Campus