Some Speculations

I did another couple of interviews today--three, in all. Not a bad
day's work. This brings me up to a total of five for Uspantán, which is
not a bad number for only having been in town for about two full days
(and a little bit more).

I would say, at this point, that I have some speculations about
Uspantán, and the reasons why it is such a low outlier.

First, there are some questions of measurement error and accuracy. The
data which is gathered here, at the level of the municipality, in the
forestry surveys I'm using to pick these cases masks a tremendous amount
of intra-municipality diversity. This would not have been a problem in
Santa Catarina Barahona, where the municipality was quite small, and the
aggregate-level dynamics weren't as noticeable. This is really a
notable problem, for one thing, because this level of analysis--the
municipality--is pretty small, compared the the national-level dynamics
which we often use in Comparative Politics. Presumably, in those cases,
these problems are even worse.

So forestry-level outcomes are likely not reflected accurately in terms
of the municipality-level outcomes we measure here, and this is likely
an even bigger problem for a number of independent variables. For
example, if poverty is a driver of these forestry outcomes, the per
capita income won't reflect the level (and concentration) of poverty in
the municipality, and therefore, the results will be biased for this
case, because poverty will be concentrated in rural regions of the
municipality, away from the main road, and therefore, deforestation will
be concentrated in those places.

I might get some of this effect by incorporating my geographic surface
area variable into the models, but that probably won't get all of it. A
better way to do it would be to use a multi-level model (with
measurement at the level of the Aldea--hamlet--or at the level of the
individual as well as the level of the municipality.) But, of course,
this would require going out and gathering all of this data again.

Maybe that's my next big project, after (a) the dissertation, (b) my
paper on the drug war, (c) the paper on education, (d) the paper on
populism I will eventually get around to talking to Zane about...

Second, there may be something of a problem of concept validity in terms
of some of the independent variables we're looking at. For example,
there is one person employed by the municipal government in forestry,
however, the measure of "one person employed in forestry," translated
into a percentage of total municipal employees in forestry fails to take
into account if that individual is especially effective or ineffective,
well-trained or not, respected by the community or not, etc...

Again, this is another issue which I may not be able to incorporate into
the current quantitative analysis, unless there's something I haven't
thought of yet.

Bloody cold here right now! I'm wearing two thermal shirts, and I'm
still shivering!

Take that, Boulder at 105 degrees!