More Meetings at the UVG

I met yesterday with Sandra de Urioste-Stone and Edwin Castellanos, two professors at the University del Valle, Guatemala, with whom I talked about choosing my second and third Guatemalan cases--a high outlier and a low outlier. These guys are colleagues of Krister, and have been very helpful to me so far, in providing general information about the forest situation in Gutemala (originally) and now, in providing some specific information about cases that I am interested in studying specifically.

They made up for me a map (using ArgGIS--I asked) of the potential cases that I was thinking about doing as outliers in Guatemala. I'm going to put that in here, so y'all can see how fantastic these guys are. Uploading images to Blogger, I've found, is a little questionable and slow at times, so this may work or not.

I chose a series of outliers and a typical case based on the Lieberman article (2006, I think) about case selection--since there have been three outcome variables in Krister's past publications using this type of data, I tried to find cases that were outliers in all three regression models, and all on the same side (in other words, municipalities that have low outcomes in all three variables, instead of high in one and low in another--there seem to be quite a few cases like this, where there the community is a low outlier in terms of "forestry personnel as a share of total municipal employees" and a high outlier in terms of "percent of municipal budget in forestry".

I think this is a story about clientilism, but there are other combinations of these variables, some of which may be due to measurement error (mayors may not, for example, know how much of their budget is in forestry).

But I wanted two outlier cases that were really high or really low. So I tried to find cases that are consistent.

And I came up with this list of about thirteen or so, plus the "typical case" that I'm doing, which is Santa Catarina Barahona.

Broadly, what Edwin and Sandra suggested to me would be to choose two cases which were geographically close--the result would be a high and a low case which have similar ethnic makeup, pretty similar ecological dynamics, and where I have controlled for a range of other variables through the statistical process. They then suggested that the next step would be to contact the people who actually conducted the surveys in these places, to see if there is any further information I could get out of them that would help me pick which pair of cases would be most appropriate.

They seemed to shy away from Totonicapan, a consistently high outlier which has been pretty heavily studied in the past--they seem to think that I wouldn't be adding as much if I went there. I haven't read any other academic work on this municipality, but I also wonder if prior work would make much of a difference in terms of the kind of research I'm trying to do: mixed-method research where the case studies reinforce or enrich the statistical analysis. But I'll probably take their advice.

They seemed big on Uspantan (it's the T-shaped municipality in the map, above), which was one place I wanted to go, anyways, although it is different in some ways from Sta. Catarina, most notably in the size of its municipal staff and budget. But I may go back to the data to double-check--they seemed to think that my concerns were exaggerated.

I guess I'll take a look and see...