The New York Times Plays Fast and Loose With School Statistics

Gary North busts the New York Times on their less than subtle promotion of public education:

when the look-say method of teaching reading, or its semi-literate cousin, whole-language method, produces poor readers, decade after decade, it is wise to compare math scores when you are being paid by the government to compare public school performance with private-school performance. (Just for the record, I have 12 semester credits in young child development, which I took in 1995. I know what the education system recommends for teaching reading.)

The survey involved ten times as many students as any previous survey. This seems to make the results authoritative. Furthermore, this study used "advanced statistical techniques." This implies that previous studies were conducted by people who did not understand statistics very well, and so used "obsolescent statistical techniques." Then again, perhaps "advanced statistical techniques" is a code phrase for "cooked-data techniques."

What were these techniques? They adjusted "for the effects of income, school and home circumstances." I see. They adjusted, among other factors, for school "circumstances." Not being trained in "advanced statistical techniques," I should have thought that adjusting for "school circumstances" was precisely what the techniques should not have done. The goal of the study, officially, was to discover differences or lack thereof with respect to schools and their circumstances.

You'll want to read the whole article.No word yet on whether the New York Times reporters that have kids want to send them to New York city non-magnet public schools.