Although I'm not qualified to judge I trust that you are - I'm not going to
argue statistics with you. I barely passed high-school math. ;^)
My problem is with using this context to denounce the whole system.
(Especially any system as complex as this - and I'm not actually for charter
schools, by the way). I know you personally didn't do that, but Times
article seemed to (as much as I personally agree with the conclusions).
I think we need to explore multiple angles before making the decision. If
it is true, for example, that more kids in charter schools are "problem"
kids then comparing points in time would not paint the whole picture
regardless of the quality of the study. Rather a study comparing
improvement rates from a baseline might be helpful (in other words although
one group of kids may be less able than the other they may also have
improved more overall if their starting point was lower).
I'm also interested in administrative information. For example Charter
school kids may lag behind public schools, but perhaps the cost is half that
of public schools (I doubt that greatly, by the way). If the lag is slight
that may be an acceptable situation. I'm also greatly interested to
discover if there are social differences between the schools: is there more
violence in one than another? More sex? More crime? More or less
extra-curricular participation?
In short I wasn't saying "we need more than one study to measure this"
(although that doesn't hurt) I was saying "we need more than one metric by
which to judge success or failure".
(Of course the irony here is that perhaps the biggest failure in drafting
the NCLB Act was to focus pretty exclusively on only one metric.)
Jim Davis
_____
From: Larry C. Lyons [mailto:[EMAIL PROTECTED]
Sent: Wednesday, August 18, 2004 10:53 AM
To: CF-Community
Subject: Re: NYT Op-Ed Piece on Charter Schools Performance
If the sample size is large enough, then I think you can. The
variability of the study results is a function of the size of the
study's sample. You will see a far more reliable estimate of what is
occuring in the population when you sample 10,000 than when you have a
sample size of 100. Its one of the basic assumptions of statistics.
larry
On Wed, 18 Aug 2004 10:42:08 -0400, Jim Davis
<[EMAIL PROTECTED]> wrote:
> Or, more importantly, a single study.
>
> Jim Davis
>
> From: Andy Ousterhout [mailto:[EMAIL PROTECTED]
> Sent: Wednesday, August 18, 2004 9:52 AM
> To: CF-Community
> Subject: RE: NYT Op-Ed Piece on Charter Schools Performance
>
>
> That nail may have been burying a live person <g>
> Again, this is why you can't rely on a single paper to kill any idea.
>
>
>
_____
[Todays Threads] [This Message] [Subscription] [Fast Unsubscribe] [User Settings] [Donations and Support]
