Tuesday, July 15, 2008
Coding and analyzing results of reading test
We did a simple reading test in the five villages where we support community libraries, as part of a small research project with Burkinabe colleagues Alain Sissao and Felix Compaore. We're just analyzing the results now (they are part of the "pre-intervention" baseline for our summer reading camps and reading discussion groups). In this kind of very low budget, locally run research there is always room for misunderstandings, so it's always a relief when the results come in normal. Lots of variation in the responses, and good correlation across the different test elements. The one result that immediately pops out was how girls were reading considerably worse than boys. On the order of 30% worse, which is pretty darn big difference! These are kids at the CM1 (4th grade) level. Felix is pretty depressed about one young girl in Sara, who couldn't read any of the letters, words, phrases and passages in the oral test. Either too traumatized or really just unable to read. it is good to do this analysis early, because we'll adjust some of the reading camp components to include more "tutoring" time, where we pair strong readers (plenty of them) with weak readers (plenty there also). And the animators will be much more careful about suggesting reading materials to the students linked to their levels. And we'll spend some more attention to girls reading than we might have had we assumed that girls and boys were about equal. I was a little surprised to see that all the schools except one had very good gender balance.