[Babase] Daphne Question - Procedures for Incomplete Censuses
Karl O. Pinc
babase@www.eco.princeton.edu
Thu, 18 Aug 2005 14:01:10 +0000
On 08/17/2005 12:10:50 PM, Daphne A Onderdonk (donderdo@Princeton.EDU)
wrote:
> Hi all,
>
> The way I dealt with incomplete censuses (as passed on by Jessica)
> was, for the individuals not marked on the day of the incomplete
> census, to do my own interpolating on the census sheets, and then
> enter the census file as if those were real presence/absences. (On
> the census sheet itself, I marked the Xs/Os in pencil and made a note
> that they were interpolated due to an incomplete census.)
It's not clear to me what this means, although I'm sure it will
be upon examination of the data. So it's hard to comment.
But it does not sound particularly good, not just for reasons
of confusing what's interpolated and what's not (and I imagine
there's "M" rows involved so this can be sorted out) but also
because it means we can never, ever, change how we do the
interpolation. And of course we've just changed how we
do the interpolation a bit.
So, it's one of those things that needs investigating on
the data end.
>
> As far as how this shows up in Census/Members, it doesn't change much
> - you're just doing the interpolating instead of the computer. But it
> obviously makes some presence/absences look confirmed when they were
> not, and artificially reduces the number of interpolated days in
> Members. But you would lose some information by not including the
> incomplete censuses at all. If you're concerned about making these
> manually interpolated presence/absences look confirmed when they're
> not, either using demography notes or creating a separate census file
> with just the confirmed individuals, as you suggested, would take care
> of that (it would incorporate all the known information without
> falsely adding information).
>
> I don't know how the way Jessica and I dealt with this issue was
> decided upon or when, but that's what we did. I like your ideas
> better, I think...
I don't know either. It smacks of work-around, maybe not having
a DEMOG table or something. I probably had my finger in it
somehow. :-(
Karl <kop@meme.com>
Free Software: "You don't pay back, you pay forward."
-- Robert A. Heinlein