Modeling and Software “To Do” List
Proposed New Student Assignment System, Fall 2003
Bruce Wicinas
8/26/03
LIST OF TASKS |
What has been done since 8/19 |
|
1 |
Reactivate the software and models from last year. |
Mostly done, though not all of last year's elements have been verified yet. |
2 |
Address “neighborhood schools” question. That is, what schools would be severely out of whack if they were enrolled as neighborhood schools? |
The demonstration we prepared last fall was pretty convincing we thought, so I have not done new work regarding this. |
3 |
Consider a very different but simpler scheme of “socio-economic boundaries” (such as the division proposed by Terry Doran.) |
I looked at the census income and education data relative to T.D.'s suggested division. I did not prepare any written comment on this because the boundaries T.D. proposed are so much at odds with the census data. Perhaps it's worth the work to try it but I'm doubtful about its outcome. The board may be reassured by seeing such a trial even if obviously unviable. |
4 |
Construct a two-factor socio economic model (Income and Parent Education), and do a full series of runs. |
I computed the two-factor socio-economic model and generated a corresponding “map”. It turned out to be identical to the map we produced last year for the three-factor model. This is apparently because last fall I severely discounted the third factor of our three-factor model, reducing its consequence to nil. Whatever the reason, the outcomes we presented last fall for the three-factor model are identical to what we will obtain by the two-factor model. |
5 |
Examine the “division” of the city produced by the socio-ec model. Are there parts of the boundary which are difficult to justify – because they separate regions which scarcely differ with regard to the socio-ec measures or because they classify parts of the city differently than is intuitive? Is it viable to alter the boundaries by hand a little to make them “simpler”? |
I am examining the boundaries corresponding to income and to educational level. I am looking for places where the boundary separates regions which hardly differ. I am doing this by visual comparison of maps of the boundaries, maps displaying the income and education values for each area, and a map indicating the number of K-5 students in each area. This study is still in process. |
6 |
What is the composition of our present schools in terms of the proposed two-factor socio-economic model? |
Rough version prepared for 8/26. |
7 |
Verify the validity of the socio-economic test runs performed previously (check our past work more carefully than time permitted previously.) |
This should be done before the plan moves close to final adoption, but I did not work on it in the past week. |
8 |
Study the performance of the plan in terms of “percent who got their first choice.” |
I have not yet done new work on this. |
10/2/2003 Bruce Comments
• I noticed that the number of EC=7 students has increased significantly since 1999. This undercuts the viability of our scheme.
• Note that this year's K outcome is the most out-of-whack in several years. I'd attribute that, not to incompetence or lack of commitment, but to spreading of our people resources too thinly, AND to the absence of anyone in upper administration who really understands the system, has any hands-on involvement, and hence can provide the cooperation and resources it needs.
• It looks like EC=1 kids may be draining out of the north zone
• A scheme using three “zones” computed from two soc-ec factors performs pretty well. Alternately, a scheme using three “zones” made from THREE factors, two soc-ec and one race, performs pretty well. It's not clear that the “with race” scheme performs better than the “without race” scheme, by the measures we've tried. So it's a political/legal decision as to whether to use race, NOT a viability-of-outcome decision. The two-factor, “without race” scheme seems quite viable. And as soon as you put race into it then you engage some parts of the community in evasive behaviors, not to mention you force a legal challenge.
• The third factor – that which introduces race, needs to be a “gradiant” like our soc-ec factors. Otherwise it is impossible to “combine” it with other “gradient” factors like income and education. So the factor I propose, and have tested, is “percent non-white” of our BUSD population. This “percent non-white” factor should be compared to the census data to test for agreement.