Tuesday, September 7, 2010

Randomness in Rwanda

(Sorry for the long pause in blog updates. Our internet got “fixed” last Friday, so I haven’t been able to get online until today.)

One of my more favorite graduation presents was a book. Not only was it my more favorite because it was given to me by my favorite person (hi KTL!), or because it was called “The Drunkard’s Walk”…but it also so confused and angered me one night that I ripped it into three pieces. That’s right…a book incited such a reaction in me that I tore it apart. That’s got to be the sign of a good present.

I still brought a third of it with me to Rwanda—as I explained to KTL, so I could read finish reading it before writing my thank you note. (KTL, consider this blog your thank you note. Check.) Besides angering me, this mathematic-theory-paper-disguised-as-a-book also caused me to laugh at a few points. It delves into the theories of randomness and of probability, explaining at one point how Mac had to make its shuffle option LESS random because the general population thought it was broken if the shuffle randomly played the same song twice in a row. It also discussed the probability of a randomizer returning a list of zeros to whatever power, etc. etc. I liked it and I promise it is relevant.

The set up: Celestin (former mayor) and I have already designed and gotten underway with our second community survey.

[Side note: When planning our meeting to launch the survey, Celestin said he’d want to schedule it for the next day. I asked him if it would be possible…since over 50 people, including some local officials, would need to attend the meeting. After a slightly confused expression, Celestin said to me “oh yes yes, I see...I will have to check to make sure I’m not busy tomorrow…yes yes…I should make sure that this meeting will not interfere with my plans.” Not quite what I was asking…but I guess that his answer still answered my question.]

Our plan was to randomly select 116 households from the health center’s catchment area. In my morning oblivion, I did not realize that the plan for doing this was to assign every household a number, to write all those numbers on pieces of paper, and then to pick 116 of them out of an envelope. I did not realize until the thousands of pieces of paper were cut and we were on our way to the health center, my laptop in my bag.

“Wait one sec,” I thought to myself, “I bet Excel can do this.”

Sure enough, all it took was searching the help for “random” for me to find the randbetween command. In a matter of minutes I quickly made a worksheet that chose random households by cell (bigger than a town, smaller than a county) and looked pretty too. The local authorities all accepted this method and were eager to hear which households were selected.

Then they were confused why the same household was selected twice, twice. They thought my randomizer was broken. I just thought back to the iPod shuffle story, and refreshed the worksheet to get a new number.

Then they were confused how two people in the room of 50 were randomly selected to survey themselves.

My favorite, though, was when the carefully randomized process became carefully not random. After announcing the assigned households, a flurry of hands went up to explain to us that certain respondents would most definitely not answer the survey questions, were unknown to the surveyors, were out of town, etc. etc. And so continued the process of Celestin asking them to survey their neighbors instead…which resulted in even more people suddenly realizing that their assigned households would not be able to answer the survey…so they were re-assigned to question their neighbors…and then I was informed of the changes made.

And so goes randomness in Rwanda.

But at least it resulted in 116 completed surveys being returned to us by noon the next day. ..and we’re off to begin the analysis.

1 comment:

Kristine said...

You're very welcome. So happy I decided today was the day to catch up on these and find my shoutout. Miss you!