Fw: Lobbying on NCLB, July 10-12

["GERALD BRACEY" <gbracey1@verizon.net>]

----- Original Message ----- 
From: "GERALD BRACEY" <gbracey1@verizon.net>


> Here's a little something that the Department slipped into the public 
> domain with any fanfare: a report on the impact of choice and tutoring. 
> The study was conducted by researchers at the RAND corporation and 
> Mathematica Policy Research, two of the more reputable organizations 
> around.  So far, only AP has picked up on it and spun it positively.
>
> On choice the impact is simply stated:  There isn't any.
>
> How odd that this finding didn't make it to the AP story.  The AP reporter 
> cites RAND researcher Ron Zimmerman as saying there are too few students 
> to draw conclusions.  But the study reports data on 3,140 students with 
> reading scores and 2,944 students with math scores.  As someone brought up 
> on group sizes of 25-40, that sure looks like plenty to me.
>
> It's true that few eligibile students avail themselves of the choice 
> option. More whites than any other group (only 1.0% of all eligible white 
> students). And while choice students go to higher achieving, more racially 
> balanced schools, they don't get anything out of it in either reading or 
> mathematics.
>
> Not too many students use the tutoring option either, 10.1% of whites, 
> 16.9% of blacks, 11.6% of Hispanics, 13.1% of LEP students, 14.6% of SPED. 
> Those that did had lower test scores to start with than those who didn't.
>
> The students show statistically significant gains in math and reading and 
> more gains if they were tutored two or more years.  But...the gains are 
> quite small, .09 standard deviations in math, .08 in reading, .18 with two 
> or more years of math, .15 with two or more years of reading.  The 
> outcomes are statistically significant because enormous numbers of 
> students, almost 50,000 are involved.  The chances of finding 
> statistically significant results increase with increasing sample size. 
> For several districts, the samples are so large that effect sizes of 0.03 
> are significant.  As we used to say when we were impudent graduate 
> students, significant but meaningless.
>
> The effect in the two largest districts are tiny.  District A (the RAND 
> people don't have permission of all districts to use their names), with 
> 16,127 students, the effect in math is .04 standard deviations and .06 in 
> reading.  There is no two-year data for A.  In District C, the first year 
> impact is .03 in math and.02 in reading and these are statistically 
> significant!!! N = 22,757.  The two year impact in math drops to .02 and 
> the reading rises to .05.
>
> My reading: a couple of billion dollars a year essentially down the tubes.
>
> Except for District F.  Someone should take a special look at F.  Its 
> first year math is .36 and reading .57 and second year is .63 and .70. 
> There are 933 in the first year group, and 68 in the second year.
>
> The Districts are named, but not linked to data.  For what it's worth, my 
> guess is that District A is Chicago--large numbers, lots of blacks and 
> District C is Los Angeles, large numbers, mostly Hispanic.  I'd guess 
> district F is Long Beach, CA.  Other districts are Baltimore, 
> Philadelphia, Washington, DC, Denver, Palm Beach, and San Diego.
>
> Jerry Bracey