[Babase] Checking dates against statdate

Karl O. Pinc babase@www.eco.princeton.edu
Sun, 7 Nov 2004 16:47:58 -0600


On 2004.11.02 10:59 Daphne A. Onderdonk wrote:
> Susan Alberts wrote:
> 

>>> I know I can't check sexual cycle information against statdate
>>> until after the animal is dead, but can I check rankdate
>>> and interaction date to be sure they don't occur after the
>>> statdate or can I only do this after the individual is dead?
>>> In other words, do you always enter the census before the
>>> interactions and before you do the ranking?
>> 
> I suppose you could check interaction dates and rankdate (not sure if 
> you mean Ranked column in Biograph or Rnkdate in Ranks, but it 
> shouldn't matter) against the statdate - the interaction data and 
> rankings are always done after demography for a given time period.

Ok.  FYI.  I'm not allowing RANKDATES.Ranked to ever be before
BIOGRAPH.Statdate.

>>> 
>>> As you can see I'm still worried that waiting to check
>>> statdate until the individual is dead will miss catching
>>> bogus dates, like reversing month and day, etc., which
>>> may be 'valid' by the time the individual dies.  Regardless
>>> who wants to wait that long to catch errors?
>>> 
>>> While we're at it, how about the point sample dates?
>>> I don't know how much checking I'll really do on the
>>> point sample stuff, because it's perfect and all that
>>> having already been checked by the Psion, but I'm
>>> curious anyhow.
>> 
>> 
>> 
>> Sorry, I am not really sure what the question is on the Psion dates. 
>> Are you asking whether they should be included in integrity checks? 
>> I think it is possible that someone could manually edit the point 
>> samples and mess up dates. However, if the likelihood of this is so 
>> low as to be almost nonexistent, then it is really not an issue and 
>> we would catch it when we used the data.
>> 
> Psion data can be updated before demography, so I wouldn't check 
> those dates against the statdate.  I'm not sure how the date gets 
> entered with the Psion data, but from Susan's description, it sounds 
> like it's fairly automatic and there's very little chance for error 
> anyway.

I am checking point sample individuals and neighbors to be sure they 
are alive
when sampled.  Partly because it's mostly easy and partly to guard
against erro when birthdates or deathdates get changed.

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