[Babase] Morris (MOR) and interpolation rules in members table
+ incomplete censuses!
Niki Learn
nlearn at princeton.edu
Wed Aug 19 15:17:40 EDT 2009
On 08/19/2009 10:27:58 AM, Niki Learn wrote:
> So, for now at least, I will plan to go ahead and enter any recorded
> or implied (by lack of being on the census sheets) absences even if
> that means entering an entire month of just absences, particularly
> for
> males when they enter a group at the beginning of a month or leave at
> the end of a month.
There is no particular reason to enter an entire month, but that's
from the perspective of the database. I did check and it's ok
to census somebody absent after they are dead.
Niki writes:
No, but from an uploading perspective, it's either enter zeros for the whole
month or create a separate file just to upload one or more absences for part
of the month. Entering an entire month is simpler and more efficient from
this perspective (and occasionally the team does actually mark someone
absent for an entire month anyway).
Absences do come into play with the following error message:
'DISPERSEDATES.Dispersed that are after the Statdate must be
on or before the next recorded absence from the group
in which the individual was last censused'
It sounds like this is another case where it's good to
ensure the entry of at least one absence after an individual
leaves the group.
Niki writes:
That makes sense. There usually is some notation of a male being absent
after he seemingly disperses and this first absence date is used to
calculate his disperse date (for a male leaving his natal group for the
first time) so I would notice quickly if there was no record of an absence.
> Karl, in addition to Tabby not mentioning this issue to me, I just
> searched through the babase update protocol I was given for anything
> census related and found no references to this issue, even in the
> section talking about making sure you catch individuals that were
> present at the end of one month and absent the next month - there was
> no mention of entering additional zeros, just of noticing that the
> individuals were absent so that one could determine whether they died
> or dispersed and take the appropriate action for that.
You might ask Catherine if she recalls anything, if we really
care. It's probably a discussion that took place some
time ago when I was first re-doing the interpolation for
Babase 2.0.
Niki writes:
I may look into it when I get back to Morris.
> On a related question, Jeanne and Susan, we have in this update a lot
> of incomplete census data. The standing rule is that incomplete
> census information does not need to be entered unless it marks
> present
> an individual that was absent in the prior complete census
This makes sense.
Niki writes:
Indeed, since their presence in between census days is interpolated anyway.
- this is
> then entered as a demography note, which in turn marks the individual
> as present on that day.
Offhand, I can't think of a reason why the database wants a demography
note instead of a census, but on the other hand there's nothing wrong
with doing it this way either. You do get to make a demography
note that says the census was incomplete. :-)
Niki writes:
Yes, we mark it as being from an incomplete census. And, again, we would
need separate files (currently we use one per month for all individuals)
with just the individuals censused on the incomplete census days (either
that or an additional code to mark individuals not censused as not censused
instead of either present or absent, as it now stands).
> Usually with team incomplete censuses nobody
> is marked absent because an incomplete census in this case typically
> means the team did not have time to search for all individuals or
> there was bad visibility or something that prevented them from
> verifying either presence or absence that were not marked present.
> However, for other observers incomplete censuses are usually more the
> result of the observer either not knowing all of the individuals
> and/or focusing on specific individuals so they may actually have an
> incomplete census where they have verified that certain individuals
> they do recognize are not present.
>
> In the case of current data generated by Courtney (and probably older
> data generated by others), this means that sometimes males who move
> around a lot are verified as absent on an incomplete census day
> though
> they had been present on the previous full census day (and may be
> either present or absent on the next full census day). With the way
> uploads work I cannot enter such incomplete census data with the full
> census data because every individual listed in the sheet must be
> marked either present or absent and we don't have that data for every
> individual on those dates.
Well, that's a good reason for making a demography note. :)
> I could, if desired, either use a demog
> note to mark them as present with an unknown group (9.0) or create a
> separate file just for incomplete census days for those males that
> move around a lot to better capture their movements (and in some
> cases
> avoid interpolation of presence with a group when we know they were
> in
> fact absent). The latter method would help me track the males
> movements between groups and wo!
> uld probably be less prone to error than marking them as present
> with
> an unknown group. Sometimes we already know where they were because
> they showed up in another group on or before the incomplete census
> day
> on which they were marked absent but in other cases their whereabouts
> are unknown. Please let me know if this is an issue that is worth
> the
> extra trouble of creating and uploading additional data for these
> males.
I could also change the census upload program so that there's a
way to mark "make no census row for this individual on this day".
It'd just be another character to enter into the matrix instead
of "X". Sounds quick and easy.
Niki writes:
Hey, you were reading my thoughts ahead of time... Currently I don't have
to enter X either - blank = X and 0 = absent. Having a "not censused" code
would make entering incomplete census data much simpler. Jeanne and Susan -
any thoughts on this?
More information about the Babase
mailing list