[Babase] MCE gender mistakes, and proposed valinter changes
Karl O. Pinc
babase@www.eco.princeton.edu
Wed, 18 Feb 2004 13:37:39 -0600
On 2004.02.18 12:52 Daphne Onderdonk wrote:
> Hi everyone,
>
> The errors that Karen emailed about earlier (with CHE as an actor for
> MCEs) are corrected in Parts.
>
> I did some querying to see if there were any other cases for MCEs in
> Interact and Parts where the actor was a female, or the actee was a
> male. There were 9 altogether (in addition to Karen's 5), out of
> almost
> 23,000 MCEs, so we're talking about a very low percentage.
Thanks.
A couple
> more were entry mistakes like the ones that Karen found. One was
> recorded backwards (female M male) and thus was entered backwards. I
> have corrected these in Interacts and Parts.
Ah, so there's a 'polarity' at issue too, some can be 'backwards'.
>
> The others were cases where the data were recorded as female M female
> or
> male M male (one case involved a consort, but in that case I think the
> sname was recorded wrong). We don't know whether these mounts
> actually
> happened and were recorded (though these types of mounts are not
> supposed to be recorded), or if the wrong sname was recorded. Is
> everyone agreed that they should be deleted from Interact and Parts?
Well, if it were dogs or hyenas you would see samesex M samesex as
there's
dominance stuff going on. Dunno about baboons. I bring it up because
I _thought_ there was an actual observed case once. Would that be
noted in demography notes? I'd think you guys would be the ones who'd
really remember such a thing.
> Do we want to put a better check on this is the valinter program?
> Assuming that this is possible (Karl?), we could have it check that,
> for
> all MCEs, all actors are male and all actees are female. Further,
> since
> the monitoring guide says that MCEs should only be recorded between
> males that are at least 4 years old and females that have reached
> menarche, it could also check that these conditions are true for the
> actors and actees. This would be a back-up system to the field
> assistants recording MCEs by these rules. Interactions are not
> updated
> until the demography for that period is updated, so that information
> should be in Biograph by the time the MCEs are updated.
>
> Susan, do you think it's a good idea to put these checks in valinter?
> Karl, is it possible? Or does it not matter since this is so rare?
I like to setup the system so it's behavior is driven by data, which
codes exist and whatnot. These sorts of checks are likely to make
some codes 'special' so that programming is required to add new codes,
change code meanings, etc. It might make sense to have a separate
process that does regular checks of the database by running queries
and reporting any problems. Then you guys could re-write the queries.
Dunno. We can certainly do something, and probably should. Ultimately
the system needs to work, even if that makes the programming ugly.
(Although if the code is ugly enough, the system will eventually
degrade to the point where it doesn't work.)
I'm thinking of keeping a 'todo' list of things to do after getting
the basics up and running. I'll put these on the list as it
doesn't seem urgent.
Karl <kop@meme.com>
Free Software: "You don't pay back, you pay forward."
-- Robert A. Heinlein