Unified Lottery- Algorithm Question

[Anonymous]

Anonymous wrote:In the past, for the DCPS lotteries, each school was essentially considered it's own lottery with preference offered for in-boundary, proximity, and having sibling preference. I had hoped that as a unified lottery was pursued the lottery would begin providing a weighted preference based on how high you ranked a school--so if you're out-of-boundary for two schools and rank one as your top choice you'd actually have a bit more chance to get a spot at that top ranked school in the initial lottery pull. Trying to match families up with their top choice in the initial lottery will create more stability and help diminish the shuffle in late summer.

So will this new lottery incorporate some weighting that gives a family a better shot at getting a spot or will each school's lottery continue to be essentially stand-alone?

No - I asked this question.

[Anonymous]

Anonymous wrote:In the past, for the DCPS lotteries, each school was essentially considered it's own lottery with preference offered for in-boundary, proximity, and having sibling preference. I had hoped that as a unified lottery was pursued the lottery would begin providing a weighted preference based on how high you ranked a school--so if you're out-of-boundary for two schools and rank one as your top choice you'd actually have a bit more chance to get a spot at that top ranked school in the initial lottery pull. Trying to match families up with their top choice in the initial lottery will create more stability and help diminish the shuffle in late summer.

So will this new lottery incorporate some weighting that gives a family a better shot at getting a spot or will each school's lottery continue to be essentially stand-alone?

While there might be an intuitive attraction to the idea that somehow you can affect your outcome by putting one school first, I think you're mistaken in the belief that trying to maximize the number of people getting their first choice increases stability.

Imagine doing a lottery by hand. The parents have all filled out forms with their choices ranked one through six. You pick the forms out of a hat, look at the first place choice for that form, and put it into a box representing that school. Eventually you pick a form, and find that the box for the first-place box school is full. What do you do now? You could either:
A) Start a pile of people who didn't get their first place pick and set this form into it. As you go through the remaining forms, you only assign first place choices. Once you have finished with the original pile, you start over with the unplaced pile and try to place people in their second place pick. Then you do third place, then fourth, fifth and sixth.
B) Look at the second place school on this form, and if it's full the third place, as so on, through all six choices. You place the form in the highest-ranked school that has an opening, and then move on to the next form.

Method A is a method that maximizes the number of people getting their first place pick. Method B is the method the unified lottery uses. Under method B, the optimal strategy for a family is to rank the schools in their actual order of preference. If you do so, you are guaranteed to get into the highest-ranked school that your lottery number qualifies you for. Under method A, there is a big difference between getting into your first ranked school and not. As you make your rankings, you have to consider your preference, but also your chance of being successful. If you have a good lottery number but don't get into your first ranked school, it's equivalent to getting picked last in the lottery. Whatever your number two school is, it needs to be one that you could get into if you were picked last. Number three is even worse. This leads to people avoiding schools that they want, but worry they can't get into, and listing schools that they're not really interested in but think they can get into.

That is destabilizing.

Anyone who is interested in the lottery should read this report: http://dme.dc.gov/DC/DME/Publication%20Files/Policy%20Brief%201%20-%20DC%20Student%20Assignment%20and%20Choice%20Policy%20DRAFT.pdf

In particular, I recommend this paragraph:

For SY 2013–14, 2,994 students applied for 2,491 available K–12 OOB seats at 97 schools and 7 programs
within schools. In the initial lottery, 1,195 students (39.9%) secured seats. The remaining 1,799 students
did not secure desired seats despite the fact that a total of 1,296 seats were available in 60 schools and 2
programs within schools. This means that 52% of the available OOB seats went unfilled in SY 2013-14.
Thirty-two schools had seats available in at least one grade that attracted no applicants whatsoever, for a
total of 431 seats for which there appears to have been no interest.

The lottery in a nutshell: there are more kids than desirable seats, and a whole lot of seats that nobody wants. No amount of tweaking the algorithm will change that.