11 September 2009

Friday Afternoon Chart: 'Throwing Money at Problems' Edition

From Andrew Coulson at Cato@Liberty, by way of Coyote Blog:


I would prefer if both vertical axes on this chart ran from 0, rather $3500 and 250 points. The math NAEP is scored from 0 to 500, I believe the reading is the same. That would be slightly more honest, though it would still show an order of magnitude increase in spending vs a flat trend in achievement. It's also not made explicit from that chart, but the spending figures do appear to be per pupil, not aggregate.

Coyote makes the following observation: "It is interesting that among the left, this chart is proof that we need to spend more money while the exact same chart in health care (say with scores replaced by life expectancy) is proof we need to spend less money. In fact, the health care chart would look better, because at least there the key metric of quality has increased over time."

2 comments:

  1. I chose the given axis start points to show the difference in trends in spending and achievement since 1970. I wanted to start them all at the same point in the bottom left corner of the chart and see where they went from there.

    If you start from zero on both axes, much of the chart is empty, spending starts at a level below achievement (thus losing the nice analogue I was going for), and the variation in test scores is all but lost so it's harder to see how scores and spending are related (if at all).

    If I were trying to be mean to the public schools, I would have included the science scores, which suffered a statistically significant drop over this period.

    ReplyDelete
  2. I understand completely, the chart looks much better as is, and that's a better way to make your point.

    I'm just overly fastidious about charts. I bitched about the NYTimes playing fast and loose with their charts a couple of days ago (in that overdraft fee story) so I wanted to be as critical as possible towards a chart I actually agree with in the interest of fairness.

    ReplyDelete