Ladner says, essentially, that increased numbers of illegally excluded ESL students brought heroic DC chancellor Rhee to grief – that is, when NAEP no longer allowed them to be excluded. Making their scores count made his heroine look bad, he concludes.
His solution? Exclude them all and look what wonders appear! The poor little fourth-graders in DCPS made great strides — nay, unprecedented, unequalled strides — under the great Michelle Rhee and Kaya Henderson, after all, and despite the nay-sayers.
Despite all his statistical legerdemain – which I have not attempted to replicate, at least not yet — I still don’t see any miracle occurring in years 2009 and 2011, not even using his own graph and data. What I see is a continuation of past trends. No miracle.
For your edification, I attach both his graph and the original graph. Ladner has clearly learned his lesson from the great “How To Lie With Statistics” — he has stretched the vertical axis to make it appear that the gains were big. He uses about twice as much room going from 180 to 215 as NAEP does in going from 170 to 220.
Look for yourself. And wonder why he didn’t also do this for 4th grade math, 8th grade math, and 8th grade reading.
The top one is from Matthew Ladner at Jay P. Greene’s blog; I added the vertical and horizontal lines.
The bottom one is from the NAEP TUDA site.
According to the data produced by Ladner, there was a rise of 27 points in 13 years from 1998 to 2011, a shade more than 2 points per year, or 4 points every 2 years. Some years a bit more (like from 1998 to 2002, when they rose by 12 points in 4 years), some times a bit less. What happened in 2009 and 2011 fits a pure-and-simple straight-line test for all the data quite well.
And, by the way, the folks at NAEP, who are way better statisticians than you, me, or Ladner, I bet, don’t think much of his argument. Although the official NAEP TUDA 4th grade scores for 2011 and 2009 are different from each other by 2 points, they don’t believe that they are significantly different, statistically speaking.