Wednesday, June 09, 2010

More on the new HSA study

What gets me is that these pro-charter studies are publicized before they are peer-reviewed, making their validity hard to judge. Not sure if they ever go through this process.

http://www.gse.upenn.edu/pdf/fac/SupovitzHSA.pdf

In addition to Luis’ observations below, see this:

Also, fewer students were identified as Special Education students at HAS (15%) than either those who were not admitted in the lottery to attend HSA (24%) or in the other schools (21%). This difference may be a function of HSA’s lower tendency to label students as

special education.

Or to the practice of HSA to counsel out sped students.

There were also far fewer Latino students, 16%, compared to 23% in the group that didn’t attend, because they supposedly lost the lottery! Wonder how random that lottery really is. Also there were very high attrition levels:

From HSA, we learned that 19 of the original 79 (24%) students chosen by lottery to attend HSA were no longer in the school. Additionally, 21 of the 97 (22%) in the original comparison group (i.e. those not chosen in the lottery) were reported by the NYC DOE to be not attending New

York City public schools in 2009.

Because of this attrition in both the original HSA population and the comparison group, the analyses below include only those students who persisted in HSA as compared to those who remained in New York City public schools. We compared the demographics of the stayers and leavers separately for both the HSA population and the group not selected by lottery and found no statistically significant differences.

They do not report on those differences; and do not compare the relative achievement levels of those who stayed and left. Other studies of charters like KIPP find that the students who left had much lower performance levels. All in all, 24% of students leaving the cohort is a huge difference, and could really undermine the strength of the conclusions..

Not surprisingly, Special Education (SPED) students performed significantly worse than did regular education students. As explained previously, there were not enough students of different ethnicities to include ethnicity in the models.

Unmentioned in the study is the fact that those HSA applicants who applied to the lottery but did not attend because they apparently lost the lottery had higher scores than the matched students in the district as a whole, revealing a definite self-selection bias of those who apply to charter schools.

This is also what the Hoxby study showed, revealing evidence of “creaming.”

72% of students who applied to the charter but who didn’t attend the school met standards in ELA compared to 54% in the “matched” population.

17% had level fours in math, compared to 11% in the matched population, and 4% in reading compared to 2% in the matched population.

Leonie Haimson

No comments: