And now, some work.

So, then, a status update on the project itself. I've been out here 1.5 weeks now, but actual working time is only a week, due to the initial few days being taken up with various Agarwal-related social activities. I also lose about 30 to 60 minutes daily in pursuit of various quality of life initiatives (e.g., the continual pursuit of safe drinking water). That said, I'm starting to get a handle on the issues involved in scaling this project up from 40 villages in an area with decent health statistics to 230 villages in an area with not-so-good statistics (or infrastructure).

First, the good news. I have good staff -- they're pretty smart, and 2 of the 3 speak English well enough that between their English and my Hindi, we can communicate well. Both of them are constantly searching for new jobs, as being posted to the desert is not fun for anyone, but they're mine for the next two months. I also have good resources -- an Internet connection that's working whenever the power is, a new server coming to handle the database load, and soon, a nice room of 10 data entry terminals to feed that survey. Reliable techs to do the data entry remain a work in progress.

Now, the troubles. They really come down to one thing: the Rajgarh Block is BIG. Blocks are set up by population, and so in a sparsely populated environment, you get a big geographic area to cover. This then gets exacerbated by the fact that roads are somewhat lacking out here, case in point:



So, to get from any one point to another is a minimum 30 minute trip, and usually more like an hour, even by car. This affects the three main parts of a database project, namely data collection, data entry, and data checking. As far as collection, we're dependent on women in the villages. Sometimes they fill out the phonebook-size registers we hand them. Sometimes they don't. Sometimes they fill them out incorrectly, in which case we have to send them back as a do-over.

Entry goes OK when the power is on, but is dependent on the hiring of computer-wallahs. When the transport infrastructure fails them, they don't come. When they do come, they're rather tempted to spend their time surfing the Internet for music. They do less of that (and make fewer mistakes) when supervised, but we only have three professional staff, and if they're out in the field, there's not much they can be doing.

It's quality control that I see as my real task while I'm out here, and that's the one that seems like a particular bear. The only way you know if the data are trustworthy is to check them against reality. Reality is a bunch of women and their families in various structures throughout a village. Checking it means going house-to-house. The first REACH project serves 40 villages, and they're small. You can just do exhaustive surveying. This new REACH project covers 230, and about 270 total community centers within them. I'm going to have to figure out a successful technique for sampling villages and households within those villages, and ideally I need to oversample the places where the problems are so that I can start rooting them out. How I think I'm going to do that is the topic of the next work-related entry.