Extraordinary!

March 5th, Superintendent Lewis presented preliminary results on 3rd and 5th grade reading scores on the Texas mandated TAKS tests.

The preliminary results reported that Monday on 3rd grade reading scores indicate an overall increase of roughly 10% over last years’ scores.

This is the second year in a row of extraordinary gains on 3rd grade reading scores for the Lancaster ISD.

In a sample of 11 districts around Dallas in the Texas Educational Region 10 over the past 12 years, the ordinary (both median and mean) change of scores on this TAKS (formerly TASS) indicator has fallen year to year by about 2%. For only those year-to-year comparisons where districts have shown a gain against all other Texas districts tested, the median gain is 2% and the mean is 4%. No other district sampled since 1996 has posted more than a 10% year-to-year gain in 3rd grade reading, and only one district in this sample, going back to 1994, showed a greater gain than Lancaster '06. That district was Crandall ISD in school year ending 1996.

Well, except for Wilmer-Hutchins.

In 2001, the Wilmer-Hutchins district posted 17% year-to-year-gains for 3rd grade reading. 2001 was a year when state scores for reading in that grade were down by 2%.

The Wilmer Hutchins results were later proven to be fraudulent. Too Good to Be True.


(WH 2001 stories, for the newcomers, can be found here: initial elation --- http://www.clipfile.org/2002/04/28/566/
followed by the sade aftermath WH 2004 http://www.clipfile.org/2004/08/22/701/

Good stuff from DMN's education reporter, Josh Benton.)

The district with dramatic gains that have never been challenged was Crandall. This comparatively small district, due East of Lancaster, ranged from 21.9% gains to 16.3% losses in the 12 years sampled. Small absolute changes in the numbers of 3rd graders passing the tests had outsized impact on their statistics. In 1997, following their extraordinary gains, the gain has scaled back to 3.3% and by 1998 Crandall actually saw comparative scores fall by 10.7% . See more about little Cradndall ISD at http://www.crandall-isd.net/

Anyhow, moving on from the historically fraudulent to the currently statistically extraordinary:

Lancaster ISD’s 2006 3rd grade reading passing percentage scores gained by an (extraordinary!) 12.7% over 2005. This, in a year where the state scores overall were down by 7.7%, indicating the 2006 TAKS test for 3rd graders was more difficult -- even unusually more difficult -- than in 2005.

Every district sampled between 1994 and 2006 saw 3rd Grade reading gains fall off sharply after any double-digit gain. 2001 gains of 17.9% in Wilmer-Hutchins were followed by losses of 36.4%. This fall is a common statistical pattern known, technically, as “regression to the mean.”. Random chance may produce dramatic changes in any measure; but when those changes are purely due to chance they are typically offset in subsequent measures. Lancaster will, if the results hold up, be the first district in the area to show back-to-back, double-digit, gains on the 3rd grade reading measure. The extraordinary statistics, in turn, indicate extraordinary changes in instructional, or testing, procedures.

Friday, the Texas Education Agency released overall 3rd and 5th AEIS reading 2007 scores for all districts. Statewide, reading scores among 3rd graders were unchanged. So it’s reasonable to infer the TAKS test this year is no harder or easier than last year. LISD's preliminary reports of gains on this test indicate an advance not only year to year, but in comparison to neighboring districts.

I calulate “Gains”, for the purposes of analysis, by comparing the local district’s passing percentage on an AEIS measure to the state’s passing percentage on that same measure. The difference of the ratio of current year’s comparison to the prior year, from the expected value of 1, is reported as a gain if greater than 1 or a fall if less than 1. Readers are invited to replicate this analysis by obtaining copies of the AEIS reports for area school districts at
http://www.tea.state.tx.us/perfreport/aeis/.

Or e-mail me for copies of the sampled data in MS-Excel format.