[START DATA MANIAC RANT]
is a bad graph.
[ I took it from http://www.ed.gov/images/title-one.jpg on July 31 ]
Here's why:
1. Most obviously, it's ignoring the number of kids in school. Which probably accounts for plenty of the spending changes. It should probably be per pupil spending.
2. The scale on the left goes from 0 to 22.5 billion, the scale on the left goes from ? to 500 to ?. This provides almost no information.
3. Design wise, they've made a number of poor choices - not the least of which is using a graph with two Y axes.
4. They've made a choice to exclude test scores after 2000 (though a cursory Google search will lead you to the scores)
5. Notice on the #4 link that the scores range from 213 to 219 with SDs that indicate significance in some of the differences. Going back to point #2 - the highest score possible is 500 (17 year olds score in the 300s or so). And a score of 500 is 1/4 the way up the scale. Could this be the worst choice they made?
Anyway, here's the D of Ed budgets since 1980 (which gives ESEA grant spending explicitly, though none of the numbers match what these people are selling) and you can see if there could possibly be the tiniest, slightest, smallest, most miniscule correlation between ESEA spending and population of the US (I'll save you the time - without inflation adjustments it's 0.96, with inflation adjustments it's 0.93). ESEA appropriation/population (a better number) has risen from a low of $20.93 in 1982 to $42.60 in 2004 (in 2003 dollars, assumes 1.6% CPI growth and 0.9% population growth before 2004). But appropriation/population doesn't account for economic growth/lack thereof - so you GDP deflate it and the relative appropriations are like this: kind of a quartic function with 2 valleys around 1986 and 2000, then the middle hill around 1992. Things to think about: When economic times are good (mid 80s, late 90s) the appropriation relative to GDP is smaller. When economic times are bad (early 80s, early 90s, 2001 forward) the appropriation relative to GDP is larger. The appropriation hurts a little more when the business cycle is taking nap and doesn't hurt as bad when it's done a pound of blow and is knocking at your door at 4am wondering why you are in bed.
The way the test scores are lumped together blurs a few things, too. Some things pop out when you look with a little finer granularity. Between 1971 and 1994 the overall increase was 3.4. The increase for all males was 6.1. All white/non-Hispanic kids = +4.0. All black kids = +15.3. All females = +0.8, but they are quite a bit higher to begin with and to end with.
And I bring this up because, in regards to Good Secretary Rod Paige's quote on ed.gov - "Every child can learn, and we man it... excuses are not good enough; we need results" - if a group of kids is already scoring higher (females), their marginal increases will be smaller and bring down the overall median increase, so this relatively small overall increase will appear disappointing in a newspaper, but it will make a fine soundbite and a nice thing for reactionaries to hang on to.
All I'm saying is this: the EDUCATION department should choose better statistics to use for their kneejerk graphs.
[END DATA MANIAC RANT]
