We've written before (here and here and probably elsewhere) about bubble kids, those students whose past academic performance suggests they might be within striking distance of reaching the "proficiency" cutoff score for the NCLB/state tests.  The concern is that other students -- both high-performers and extremely low performers -- will be ignored because NCLB pressures the school to focus on the bubble kids. 

And a new (to me) Ed Week blog, written by David Hoff, has news of a bubble-kid finding in a just-released report on Chicago's schools.  Hoff writes, "This study puts (sic) lends credence to common critiques of that law (sic) encourages teachers to focus on the so-called bubble kids--the ones that are close to reaching proficiency."

It also calls to mind a critique* of a recent report from the Center on Education Policy, which found, based on data from 13 states, that achievement had risen faster since enactment of NCLB.  But CEP's finding of rising achievement was based in part on before-and-after data for states that reported only the percentage of students reaching proficiency: Massachusetts, New Hampshire, New York, Oregon, Virginia and Wyoming.  So, if those states took the bubble approach, redistributing resources to bubble kids while ignoring high- and low-performers, it's not really fair to see overall student achievement rose.  There's simply no data in the CEP report that shows overall achievement rising in these states.

There is average test score data for 7 other states.  Of those, just 2 have consistent increases since 2002, 3 have increases that would disappear if just one score for one grade had gone the other way, 1 has a decrease that would disappear if one went the other way, and 1 has consistent decreases.
 
So, editorials and op-eds and Bush administration officials are claiming a national trend based on reliable data from just 7 states, only two of which -- Kansas and Kentucky -- show consistent gains.

*Thanks to my sharp-eyed colleague Dan (not pictured above) for digging into Table 10 and letting me know that CEP relied on something other than average test scores for many of the 13 states.