Feed on
Posts
Comments


Photo credit: Eye poking

I am sorely missing Eduwonkette. Really, no one combines her wit and irreverence, with statistical analysis. It was a killer combo. I felt reminiscent when I read this killer piece from Tom Hoffman at Tuttle SVC. He takes on a report from McKinsey & Company on The Economic Impact of the Achievement Gap on America’s Schools.

This apparently comes out of a larger kerfuffle that had it’s nexus in comments from Thomas Friedman in the NY Times citing the reports findings and a NY Times article on the study. Clay Burrell deals with it in three parts on Education – Change.org Basically, everyone is running around in circles screaming about the achievement gap, and using questionable analysis of achievement data to make their case. They compare the “huge” gaps in achievement in the U.S. to other OECD countries and, surprise, ours is larger. Tom points out succinctly and relentlessly that they also have MUCH smaller income level gaps. Clay takes it a step further, pointing out the gap was smaller in the U.S. when income gaps were smaller. SURPRISE!

This is a classic error in statistical analysis, the apples to oranges comparison, and  has more to do with logic than numbers. There is also the Correlation = Causation error, that Mark Pullen picks up on in this (along with some other recent examples).

My new grand unification theory is this; we have all, in the social sciences, been polluted by Chicago School economics, which is based on elaborate mathematical models that have little basis in reality, and are compromised by some really brain dead assumptions (hey, I don’t just think that, economist do to). Let’s take a look at a short list of fallacies offered up as “givens”:

  • People, in aggregate, will generally act to maximize their economic self interest (they’re rational actors).  I’ll just point to this post for a contrary view of how that works in real life, Economist’s View: Is Poverty Caused by Irrational Behavior? and I’ll leave it to you to think about its implications for school reform that has any basis in choice, changing parent behavior based on a putative model, etc.
  • Markets will maximize benefits for all. Look at this, for how market choice is working in NYC. You say, well, that’s not like a “real” market which is open and transparent, to which I reply, hedge funds, and derivatives.
  • There are some base assumptions in most models. In the recent housing bubble, risk models either assumed an ever expanding market, or did not factor in systemic risk. The big one I see in a lot of the studies like McKenzie is a version of “all things being equal” that underlies basic supply/demand curves. They are assuming all kids are equal and coming to school equal. Economist know all things are never equal, we know all kids are not coming to us with equal assets, so why do we treat them like they are when we look at these studies?

Links

Some posts that talk more realistically about poverty and it’s not always pretty intersection with education:

2 Responses to “Could you please stop trying to blind me with bad science?”

  1. Ira Socol says:

    Chicago School Economics ignores human reality and, of course, Nash’s Equilibrium, which “proves” that people are not solely selfish actors. But yes, any attempt to apply these statistical analyses to schools automatically buys into the industrial model of education – which implies that children are either “equal” or “defective” and that teachers are either machines or fatally flawed.

  2. Alice,

    Well said.

    As Mark Twain (and maybe Disraeli) said, “”There are three kinds of lies: lies, damned lies, and statistics.”

    Larry

Leave a Reply