Common Lottery Algorithm
![[Post New]](/jforum/templates/default/images/icon_minipost_new.gif){kind=icon}
Anonymous
The schools don't need to see your rankings. The algorithm does. |
Anonymous
I believe you're right, but don't confuse your tracking number with your random lottery assignment. No one knows the random assignment except the algorithm. |
Anonymous
Random selection comes into play within the preferences. That random selection is based on the randomness assigned by the algorithm. Again, as I said above, the schools have nothing to do with this. It's all the folks running the lottery according to that nobel prize winning method mentioned in the Washington Post article. |
Anonymous
| ^ Understood. The #s I used above are the random assignments that the schools nor the applicants would see. |
Anonymous
| It is pretty ridiculous that the common application for the lottery is now underway and there is still this much confusion over how it works. Does DCPS intend to put out more information so people don't have to be guessing at how it works based on how it works in other cities or how anonymous people describe it on a message board? |
Anonymous
| All of this information is out there. The preferences, the algorithm, the schools, how it works. They dont lay it all out on a platter but if you research the company running the lottery you can find it. It isn't easy to understand, I mean the guy who figured this all out is a nobel prize winning economist. |
Anonymous
| I actually don't think it's that complicated, it's just the misinformation that's confusing matters. |
Anonymous
I am the PP to whom you are responding, and thank you for posting this! Very helpful. I did some follow-up research and found that you are right--the first assignment is just a temporary assignment and in subsequent rounds people are reordered. Preference (like siblings, boundaries, etc.) matter most, but after that, your ranking does make some difference. Here is an article on how Boston and NYC do it, with the methodology described on p. 9: http://www.educationsector.org/sites/default/files/publications/ChoiceMatching.pdf Specifically, below is the methodology. Some schools in DC, like those in NY, do rank students (e.g., SWW high school); certainly, all schools have preferences, even brand-new charter schools, who have a founders' preference. Thus, it seems like this system would be able to work within the constraints of ranking/preferences inherent to the DC system, while still using the student ranking secondarily. And the method below is also "strategy-proof," meaning that people have no incentive to misrepresent their preferences. "The new mathematical formulas for matching students and schools that Harvard economist Al Roth and his colleagues created for the New York and Boston school systems differ from the troublesome models they replaced in subtle but important ways. "The priority matching strategies that both cities abandoned— but that many other school districts still use—begins, reasonably enough, by trying to match students with their first-choice schools. Students listing a school as their first choice are placed on a list. In some districts, a student’s rank on the list is merely a matter of random selection. In other places, including Boston, students are moved up the list if they live near the school or have a sibling already in the school. Students are then assigned to the school from the list until it is full. The assignments are final. If there are more students on the first-choice list than there are seats in the school, as is routinely the case, unmatched students are then added to the lists of their second-choice schools. But because they are ranked below students who have made these schools their first choice, they often fail to get seats in these schools either. As this process repeats itself, many students fail to get assignments in any of the schools they’ve selected under the priority matching model. "The key difference between the priority matching model and the algorithm that Roth and his colleagues introduced in New York and Boston is that under Roth’s model (also called the Gale-Shapley algorithm, for mathematicians David Gale and Lloyd Shapley) school assignments are temporary—that is, they are deferred—until the computerized matching process is completed. "The process begins the same way as the priority matching model. Every student is given a ranking by every school. In Boston, the ranking is based on a combination of sibling and walk-zone preferences and a random lottery number that’s given to every student in the school choice system—primarily the city’s kindergartners and sixth- and eighth-graders. In New York as well, each of the city’s nearly 80,000 eighth- graders is given a random lottery number, but, in deference to a recent past where individual high schools had much freedom in selecting their students, New York allows for certain schools to rank students. Depending on their method for selecting students, some schools are allowed to “screen” students based on the student’s academic record and their attendance at school-based fairs. At the city’s roughly 200 “educational options” schools (a vestige of the city’s earlier experience with school choice), schools can rank students based on where they fall in the citywide distribution of scores on the seventh-grade reading test (every educational option school must have a bell-curve-like distribution of high, middle, and low achievers). These schools can express their priorities for individual applicants to fill half their seats, but both the ranked and unranked halves must meet the bell-curve distribution. "Next, students’ rankings and schools’ priorities are loaded into computers. Students are then matched to their first-choice schools. If the schools are filled with higher-priority students, the unmatched students are moved by the computers to the pool for their second-choice schools. But, because all seat allocations under the Gale-Shapley model are temporary, the second choices of unmatched students are compared to the schools’ first-choice matches. The computers then reshuffle the assignments to the schools to give seats to students who have listed the schools as their second choice, if those students rank higher in the citywide priority lists than students who have selected the schools as their first choices. These bumped students are then added to the pools of their second- choice selections. "This process continues until student choices are completely exhausted or all schools are full. Only then are the matches finalized and sent out to students. Because it’s centralized and computerized, the entire process takes only a few minutes once students’ preferences and schools’ priorities are entered." |
Anonymous
|
But this is NOT the methodology that DC is using. Theyare NOT using this Boston/NYC model. Schools' priorities are finite in this system, and not set by the individual student (their address, their academics), rather preferences are a pool of students with a category (sibling, in-bound, etc.)
For this reason, the algorithm used is not nearly as complicated, and there's no methodology for a parent to use to increase chances at any given school. Your preferences are already defined, after that the assignments are based upon your draw. Bottom line, your statement that "after preferences ranking matters" is not true in the model that DC is using this year. |
Anonymous
| Or I should state - your ranking doesn't have an effect on your placement. It has an effect in that they go down the list so your first school is the first to give a seat if available, but your ranking DOES NOT INCREASE YOUR PROBABILITY OF GAINING A SEAT. |
Anonymous
Academics (special high schools) and address (dcps proximity) are certainly factors in your lottery placement in the algorithm. |
Anonymous
It most certainly does. You will not steal a seat from a higher preferenced child if that match is more equitable for both parties. That's the whole point of deferred placement! Youre given a seat but it can be taken away and given to someone else if the algorithm deems it as a better match - and that match takes into account both child's assigned preferences, whether they are location based, merit based or sibling based. |
Anonymous
Based in my conversation with the DME office--where they have stated they are using an algorithm like the one in Boston, NYC, New Orleans, Newark, and Denver--they are indeed using this system. Preferences matter first, but then after that the algorithm uses the students' rankings alog with the random lottery number in multiple rounds to make matches. |
Anonymous
But I should say that this new system does not change that you should simply rank the schools by preference -- that is your best "strategy." |
Anonymous
| So if my child has no preferences anywhere we apply, the best we can hope is an early lottery number and enough open (non IB, non Sib) spaces at one of our top 3-5 schools? And if our lottery number is at the end of the pack, we will basically be shut out of all but choices 11 or 12 and at the end of most wait lists? |