data mining

Yesterday I got back from spending a week in Nakuru, the capital of Rift Valley Province. Rift Valley is the largest of Kenya’s 8 administrative provinces, and is home to some of the most beautiful terrain in Kenya, courtesy of the eponymous natural feature, a massive intercontinental gash that stretches from Israel all the way down to Mozambique. Nakuru itself is one of the gems of the Rift Valley, as it is home to the beautiful Lake Nakuru national park, which home to tons of great wildlife (in particular, hundreds upon hundreds of flamingos) and has probably become the second biggest tourist destination in Kenya (after Maasai Mara).

Unfortunately, the only glimpse I got of this beauty was from the roof of my hotel (the $5/night Mt. Sinai Boarding and Lodging) at sunrise. The rest of my daylight hours were spent in the beautiful 1970s post-colonial bureaucratic modernist interior of the Provincial Headquarters office.

Typically, data-mining is considered a bad thing in the social sciences, as the term refers to the practice of running a bunch of theoretically uninformed analyses on a mass of information and using the specification that works the best to support an argument – in reality I think this could be more accurately called “result mining”. Data mining could be much more accurately used to describe what I’ve been up to over the last few weeks, as my relentless pursuit of district-level government data on administrative presence and law enforcement outcomes made me feel very much like I was a prospector hacking away with a pickaxe deep beneath the earth.

If you will indulge me that simile, then, without a doubt, I hit paydirt this week. The brief history of my (retrospectively) not-so-quixotic quest is that I’ve been politely pushed from one bureaucratic office to another, with assistance being withheld either on account of allegations of insufficient authorization, or by claims that what I was looking for just didn’t exist. I kept on digging because I was confident that neither of these claims actually hold any water, and were just tossed off by bureaucrats who couldn’t be bothered to spend a little bit of time figuring out where to find what I was looking for.

The real breakthrough came when I went to see the Permanent Secretary of the Ministry of State for Provincial Administration and Internal Security (PAIS) in the Office of the President (OP), on the advice of the deputy commander of forces for the north-eastern province. Although this didn’t pay off immediately (and led to me spending a few days getting bounced back and forth between dead ends in the Police Administration and Office of the President in Nairobi), my contact in the OP gave me a letter authorizing Provincial Commissioners (and their underlings) to give me access to any and all files I required, including personnel statistics from their Human Resources department and district level crime statistics.

When I reported to the Provincial Administration office in Nakuru, I was shown into the deputy Provincial Comissioner’s office, just as I was at the Coast and North Eastern Provinces. As I mentioned in an earlier post, when I was at the Coast, I was completely stonewalled because my contact at the Ministry of Education had neglected to give me a letter of any sort. Although the situation improved a bit when I visited the North East provincial headquarters in Garissa, the deputy PC pushed me off to the officer who dealt with all of the ministries OTHER THAN provincial administration and internal security. However, with a letter directly from the head of PAIS, the Rift Valley deputy Provincial Commissioner promptly set me up with two Administrative Officers (the Provincial Government equivalent of middle managers), who let me set up camp in their office and proceeded to send clerks digging around in the archives until they found pretty much exactly what I was looking for.

More on what they actually found for me a bit later, as well as on some details my interactions with these guys, but at least in the short run, I hope this shows that sometimes you can learn as much from trying to get data as you hoped to get from the information itself.