[Babase] possible interpolation problem

Karl O. Pinc kop at meme.com
Wed Jul 26 16:01:07 EDT 2006


On 07/26/2006 12:55:35 PM, Leah Gerber wrote:
> Hello Karl, Jeanne, Susan, Catherine (or anyone who could help)
> 
> Russ was looking at some immigrant males in babase and found the  
> following strange occurrences. The following males are listed as  
> immigrants (matgrp of 9) in biograph but are in study groups in  
> members and census from their birthdates on. I think that the matgrps  
> are correct and that the members and census rows are not (otherwise  
> some of these guys stayed in their matgrps for 18 years). I am  
> wondering if anyone has any idea why the data is like this and how we  
> would go about fixing it. Before I start blindly flipping through  
> binders to find out when they really did show up in study groups I  
> thought somebody might have some insight. Could it just be an  
> interpolation issue (I hope)? I checked and the data is the same in  
> Foxpro and in the new system.

My suspiction is that this is the result of the coding of the data
before it was first loaded into babase.  I only checked one row,
BJX.

Biograph:
> sname birth      	matgrp  members.grp   census.grp     
> census.status census.cen
> BJX   1960-12-29      	9.00   	1.00   		1.00 		  
> B        f

Census:
cenid  |    date    | sname |  grp  | status | cen
--------+------------+-------+-------+--------+-----
    1351 | 1960-12-29 | BJX   |  1.00 | B      | f
    1361 | 1960-12-30 | BJX   |  1.00 | B      | f

Look what the CENSUS doc has to say about the "B" code:

  B

     (birth date) The data comes from the old DISPERSE database where  
the record had a Dateout but not a Datein.


(IIRC, DISPERSE is Susan's old data.)

So, the old data had the indivdual leaving a group, in the case
of BJX that would be group 1.0, but never had any record of the
individual entering the group.  So, even though the Matgrp is
9.0, the individual was put in 1.0 from the time of birth
until they left group 1.0.

There's no good way out of this, it's why all those "old census
codes" give me willies.  So many of them are the result of
a compromise that's a bit spooky.  Probably every "B" code is
census is "wrong" this way, but nobody's ever bothered to get
rid of them.  It's always been one of those "Oh, we know the
data so we know to ignore that" issues.

Karl <kop at meme.com>
Free Software:  "You don't pay back, you pay forward."
                  -- Robert A. Heinlein



More information about the Babase mailing list