Published Online: September 22, 2009
Published in Print: September 23, 2009, as Charter Quality's the Issue, Not Research Methods

Letter

Charter Quality's the Issue, Not Research Methods

Article Tools
  • PrintPrinter-Friendly
  • EmailEmail Article
  • ReprintReprints
  • CommentsComments

To the Editor:

Your recent online Commentary by Jeanne Allen of the Center for Education Reform ("Charter Laws and Flawed Research," Sept. 8, 2009) perpetuates a misconception she has about the compositions of “virtual twins” that were used in a report by Stanford University’s Center for Research on Education Outcomes, “Multiple Choice: Charter School Performance in 16 States.” Requires Adobe Acrobat Reader

While Ms. Allen has a distinguished record of advocacy for rigorous and fair study of charter school effectiveness, her continued insistence that the control populations used in the credo study are somehow mythic fabrications does a disservice to the quality of the work and detracts from the important findings that emerge. I write to set the record straight.

Our technique for creating control records collects all the students who attend the traditional public schools from which a given charter school typically gets its students. These students are real students, with real socioeconomic and demographic attributes, and with known past test scores. Then, for each charter school student in our sample, we identify all the traditional public school students who match that student on personal characteristics and on past test scores.

So for one real charter school student with a given past test score, we have multiple real students from traditional schools who look just like the student, including prior performance.

We then gather the subsequent test scores for all the students. Since the traditional public school students are now likely to have a range of subsequent scores, rather than select a single student to serve as the match, we average all the subsequent test scores. It is at this point that “virtual” takes over. We create a single record from the matched traditional schools’ students that includes the set of personal attributes (identical across all the records), prior test performance (identical across all the records), and the average of the real subsequent test experience. This “virtual twin” is very real.

I hope this explanation ends the confusion about the methods used in the study and opens the way to a clearer focus on the important issue of charter school quality.

Margaret E. Raymond
Director
Center for Research on Education Outcomes
Stanford University
Stanford, Calif.

Vol. 29, Issue 04, Page 24

Related Stories
You must be logged in to leave a comment. Login |  Register
Ground Rules for Posting
We encourage lively debate, but please be respectful of others. Profanity and personal attacks are prohibited. By commenting, you are agreeing to abide by our user agreement.
All comments are public.

Back to Top Back to Top

Most Popular Stories

Viewed

Emailed

Commented