More info on common lottery algorithm

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:I had this conversation with people from the Common Lottery helpline a month ago - I brought up that I noticed last year, because you could click on your DCPS desired school's admit and waitlist and then click on the random assigned numbers of the people ahead of you to see hwo they'd ranked their schools - and when I looked at most of the people who were in single digits, they were on single digits on ALL or MOST of their waitlists. Whereas many people who were #320 on one had sucky sucky 200s-300s numbers for all their choices too. It wasn't true 100%, but it was true enough at the many random people I looked at, and it was true of us (but we were lucky to have our numbers all be in the teens and 20s). The lottery staff said "No no that won't happen again this year", but really, it will, because preference people had preference before, and they do now, and there will be a computer attempt to go down each person's desired rankings, but at the end of the day the computer will use random lottery number, NOT desired school ranking, to choose between people who otherwise everything else is equal. With the exception of the number of people who previously would have applied to 20 charters and now have to rank most of them along with DCPS (and the double/triple spot-holding that that leads to), it really doesn't sound like it's that much different in terms of giving you a better chance at YOUR #1 ranked school. But my understanding from the school lottery folks was that you DID have a better chance under this model.

Thanks for this explanation, it is the first thing that seems reasonably clear to me. It also makes it abundantly clear that this system will be worse for individuals than applying to all of the school separately. At least in separate lotteries, I have an "new" chance of getting a good or bad lottery number for every school that we apply to. If I get shut out in one place, I could still have a shot somewhere else. This explanation makes it seem like with the common lottery, there will be one random good or bad number that will either give me an excellent shot at all of my choices or a terrible shot at all of my choices. The moment all the most coveted spots are filled, a few people will have the top spots on waiting lists for all the most coveted schools. By the same token, the same sad folks will be together in the bottom of all of the lists.

Hi, I'm the PP you're responding to. I think there's actually better news than this - if you keep reading the conversation past this post, it sounds like actually, even if the computer randomly assigns you #312 for the lottery and that stays your lottery number, you are not in fact #312 and everyone between #1-#311 has a better shot than you. From what I'm reading above, that is NOT how it will work.

If I understand the above, even if your random assigned number is #312, it's possible you could be #2 on the waitlist for your 1st or 2nd choice school and the person with random lottery #3 could end up #235 on a waitlist. It sounds like for each school, the computer looks at preference groups (i.e. everyone with sibling preference, or everyone with no preference at all) and within that preference group, for that school, does a random shuffle to establish the slots and waitlist. So no matter what your assigned lottery number (i.e. 312), you could end up #4 ont he waitlist for one school, number 87 on the waitlist for another school, and #452 for a 3rd school.

So if what's said above is true, actually your randomly assigned number does NOT condemn you to unifrom suckiness or awesomeness. It's a different shuffle for each school, even though your random number stays the same.

What you are reading above is not correct. There is no "random shuffle." That's just wishful thinking.

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:So do you get one, single randomly assigned lottery number that applies to all 12 of your selections? Or are you entered essentially into almost 12 separate lotteries for each school, with a random assigned number for that school?

Man would it SUCK to get a crappy number for all 12 choices. Wouldn't that essentially (unless you have both IB and sibling preference for your 1st choice or something) knock you out of the whole lottery because, at least for the more popular schools, there is guaranteed to be more people with better numbers who ranked it high enough to be competing with you?

Your lottery number is not the same for all schools, so your second scenario is not accurate. No worries there. But it's not exactly as simple as each school running a separate lottery, because rather than having the schools run the lottery and then choose the students, the students in the algorithm do the requesting of the spot. They are temporarily assigned and reassigned in many rounds. In the end, your relative waitlist number will vary in the schools you apply to because their "rank" of you depends first on preference (they will always choose someone who is IB over someone who is OOB, e.g.) and then, within that preference band, by random lottery number.

As PP said, all you need to do is list in the order of preference.

Sorry, should have been more clear - your lottery number ORDER is not the same for all schools. Meaning, within preference catgories, each school then randomly chooses an order for the people who have applied. So you could be without a sibling at TR and have a relatively good spot on the waitlist and without a sibling at MV and be in a terrible waitlist position.

The last part was already clear. Here is what's not:

Say I have NO preference for any school (and not applying to my IB school). And say my #1 choice is Ross (random name that came to me) and my #2 choice is Bancroft. Am I given one single lottery number which will apply to both Ross and Bancroft, even though who I'm up against in each is obviously diffferent? Or will the computer look at all the students with no preference requesting Ross, assign random lottery numbers to all those requests, assign/waitlist, and then if I don't get in to Ross, the computer then looks at all the no preference people requesting Bancroft, assigns a random lottery number to each person in THAT pool of requests, and assigns/waitlists for that pool, etc until it successfully matches me or runs out of options and I'm waitlisted everywhere?

The latter scenario would be an improvement over last year. Not sure though that the 1st scenario really adds much fairness though.

Here's how it would work.

Let's say your first choice is Ross and second choice Bancroft.

In round 1 of the lottery, you request Ross, and everyone else requests their first choice school. Each school then goes through and accepts or rejects applicants. (This is all done by computer, of course.) Each school has ranked each student according to preference first and then a randomly after that. So if you are OOB for Ross, but a lot of IB people put them as their #1 choice, then you are definitely not going to get in. But, let's say that Ross families put Ross further down their list -- let's say 4 or 5 -- and among the OOB families, you are pretty high on Ross's list --then maybe you get a TEMPORARY assignment at Ross in Round 1.

In round 2, all the people who weren't matched in round 1 bid on their next choice school. Let's say some of the IB ross people have put Ross as #2. Ross then considers all the people they accepted in round 1 and all the people who bid in round 2 and makes decisions. So, if you are IB for Ross, even though someone OOB was assigned in round 1, you trump them. The OOB person gets bumped and you get the spot.

So, then, let's say you don't get into Ross, in Round 3 you bid on your next choice school, which is Bancroft. The same thing happens--you are compared with all the people who have bid in this round and all the people who were temporarily accepted in prior rounds. Etc. etc. etc.

What this shows is that:

1) Schools will have randomly generated waitlists within each category. So, someone who is OOB w/sibling will definitely have a better waitlist # than someone who is with no preference, but within the no preference category, the waistlist will be random.

2) Each school has a different random lottery order within each category. So, at Ross as OOB you might have a relatively good # (say, the #2 OOB person, just from random order), and at Bancroft you could have a relatively bad # (say, the #120 OOB person, just from random order).

3) It is definitely possible because of the randomization of the order from the schools for mutually beneficial trades to occur. Example: I get into A, you get into B. I prefer B to A; you prefer A to B. The computer could trade us so we are both in our top-ranked choice. Not sure whether DC will do this.

4) It definitely makes sense to rank order in terms of your true preferences BUT

5) If you definitely want a spot, you need to think of schools where you are likely to get in. Think of schools that went through their whole waitlists last year, your IB school. My spouse and I thought we would never apply to our IB school but then were pleasantly surprised at the open house. Knowing more about how the lottery works, I would not waste an OOB spot on a school like Ross or even Bancroft--which didn't accept any oob students as of count day. I am using my spots carefully, and my #12 spot will go to my IB school.

Good luck!

This is not how DC works. There is no randomization by the schools. Your one lottery number is your number for every school.

You are mistaken. The above is EXACTLY how the deferred acceptance algorithm cited in the q and a works.

[Anonymous]

Anonymous wrote:
You are mistaken. The above is EXACTLY how the deferred acceptance algorithm cited in the q and a works.

Cite please?

[Anonymous]

This is what the FAQ says:

The matching algorithm assigns each student a random lottery number and attempts to match each student with his or her first choice first, then his or her second choice, and so on. A student’s ranking of schools (the order a student puts schools on his or her application) is critical because the matching algorithm attempts to match each student with his or her first choice first, then his or her second choice, and so on. However, when the matching algorithm is comparing two students who have applied to the same school, the decision is based on two criteria: (1) each student’s preferences at that school (e.g., sibling preference); and (2) if there is no difference in preferences, each student’s randomly assigned lottery number. The students’ rankings of the schools are not a factor at that point in the process. This is why the system is strategy-proof — and why students are best served by ranking schools according to their true choices (the school they most want to attend first, then their second choice, and so on).

It's pretty clear to me that there is only one lottery number, that both bolded sections are referring to the same thing.

[Anonymous]

Anonymous wrote:Thank you PP, the above is clear and very helpful.

The only thing that is very different about what you say and what most other people have been harping on, is your last bit of advice. Even though I do NOT at all believe that being strategic about where you apply (i.e. applying to schools you have a better chance at because, say, they went through most of hteir waitlist last year) is "gaming the system", a LOT of people here have been saying that that is somehow gaming hte system and not being "honest" about your "true preferences". So if you really really want Ross, most people are saying you should put Ross as #1 even if, as OOB with no sib preference, your odds are almost nil at getting in.

Me, I'd do what you suggest and combine our dream list with what is more likely. I would not do strict "just put it in true order" if my #1 choice is an OOB DCPS that NEVER takes OOB. Charters do seem different though, since Sib preference is the only preference category (right?), so seems like hte odds there would be generally much better? Or do I have that wrong?

I think that what I am saying is not that different from what other people are saying. That is, you need to choose your 12 schools with whatever criteria you want--one of which may be probability of getting in. For me, it is, because our priority is getting into a PK3 program, even if it's not a long-term solution (like Appletree). Once you choose your 12 schools, you should then place them in the order in which you truly prefer them. I have heard some people say that maybe you can "game" the system by putting a school that you desire less as a higher-ranked option to maximize the probability that you get in somewhere. This is not a good strategy--once your 12 schools are selected, you should rank them with your true preference. But you might be strategic about what choices you make.

Ross is a good example. I'd love my child to go to Ross, but we are OOB, and they took zero out of boundary people last year (even as late as count day) for PS3. They couldn't meet the in-boundary demand. For me, it's not worth putting it down as a choice. However, maybe for someone else, who has a good in-boundary option but would prefer Ross, it makes sense to apply even though they know they have virtually no probability of getting in. They know they can always fall back on a good IB option, so that makes sense for that family.

Hope that makes sense!

[Anonymous]

Anonymous wrote:

Anonymous wrote:Thank you PP, the above is clear and very helpful.

The only thing that is very different about what you say and what most other people have been harping on, is your last bit of advice. Even though I do NOT at all believe that being strategic about where you apply (i.e. applying to schools you have a better chance at because, say, they went through most of hteir waitlist last year) is "gaming the system", a LOT of people here have been saying that that is somehow gaming hte system and not being "honest" about your "true preferences". So if you really really want Ross, most people are saying you should put Ross as #1 even if, as OOB with no sib preference, your odds are almost nil at getting in.

Me, I'd do what you suggest and combine our dream list with what is more likely. I would not do strict "just put it in true order" if my #1 choice is an OOB DCPS that NEVER takes OOB. Charters do seem different though, since Sib preference is the only preference category (right?), so seems like hte odds there would be generally much better? Or do I have that wrong?

Same PP - clarification: I'm saying it seems like if I"ve got no preference of any kind for any of the schools I'm applying to, it seems like the charters are a better shot since there really is no preference other than sibling preference that trumps me (except then my random order number for that school), whereas with DCPS it's sibling AND IB that trumps me.

I think that's right, with the caveat that there are some EOTP schools that have room beyond their IB populations.

[Anonymous]

Anonymous wrote:

Anonymous wrote:What a colossal bummer. As someone who majorly struck out on the lottery last year, I fail to see how this improves the winners and losers situation that's prevailed in past years, other than making it occur on a larger scale. Some kids stand to win big by having great odds at getting into their top choice, or even one of their top choices, and some kids have awful odds of getting into anything.

Changing the lottery method doesn't change the size of the pie. There are about twice as many people in the lottery as there are seats for the, about half of the people don't get into any of their choices. The only way to keep that from happening is to have more seats.

The advantage of the common lottery is that everybody finds out sooner.

That is definitely true--the lottery can't in and of itself make more seats appear. And you do hear earlier.

However, it is also true that every year, there are schools that have PK3 and PK4 seats available. So there are definitely more seats than people who take them. The difference is that you might not be willing to go to the schools where the seats are available. (My IB usually has 1-2 PK3 spots available as of the start of the school year. We are EOTP but in NW.)

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:I had this conversation with people from the Common Lottery helpline a month ago - I brought up that I noticed last year, because you could click on your DCPS desired school's admit and waitlist and then click on the random assigned numbers of the people ahead of you to see hwo they'd ranked their schools - and when I looked at most of the people who were in single digits, they were on single digits on ALL or MOST of their waitlists. Whereas many people who were #320 on one had sucky sucky 200s-300s numbers for all their choices too. It wasn't true 100%, but it was true enough at the many random people I looked at, and it was true of us (but we were lucky to have our numbers all be in the teens and 20s). The lottery staff said "No no that won't happen again this year", but really, it will, because preference people had preference before, and they do now, and there will be a computer attempt to go down each person's desired rankings, but at the end of the day the computer will use random lottery number, NOT desired school ranking, to choose between people who otherwise everything else is equal. With the exception of the number of people who previously would have applied to 20 charters and now have to rank most of them along with DCPS (and the double/triple spot-holding that that leads to), it really doesn't sound like it's that much different in terms of giving you a better chance at YOUR #1 ranked school. But my understanding from the school lottery folks was that you DID have a better chance under this model.

Thanks for this explanation, it is the first thing that seems reasonably clear to me. It also makes it abundantly clear that this system will be worse for individuals than applying to all of the school separately. At least in separate lotteries, I have an "new" chance of getting a good or bad lottery number for every school that we apply to. If I get shut out in one place, I could still have a shot somewhere else. This explanation makes it seem like with the common lottery, there will be one random good or bad number that will either give me an excellent shot at all of my choices or a terrible shot at all of my choices. The moment all the most coveted spots are filled, a few people will have the top spots on waiting lists for all the most coveted schools. By the same token, the same sad folks will be together in the bottom of all of the lists.

Hi, I'm the PP you're responding to. I think there's actually better news than this - if you keep reading the conversation past this post, it sounds like actually, even if the computer randomly assigns you #312 for the lottery and that stays your lottery number, you are not in fact #312 and everyone between #1-#311 has a better shot than you. From what I'm reading above, that is NOT how it will work.

If I understand the above, even if your random assigned number is #312, it's possible you could be #2 on the waitlist for your 1st or 2nd choice school and the person with random lottery #3 could end up #235 on a waitlist. It sounds like for each school, the computer looks at preference groups (i.e. everyone with sibling preference, or everyone with no preference at all) and within that preference group, for that school, does a random shuffle to establish the slots and waitlist. So no matter what your assigned lottery number (i.e. 312), you could end up #4 ont he waitlist for one school, number 87 on the waitlist for another school, and #452 for a 3rd school.

So if what's said above is true, actually your randomly assigned number does NOT condemn you to unifrom suckiness or awesomeness. It's a different shuffle for each school, even though your random number stays the same.

That is correct. At least if you have bad luck in one school, you are not condemned to having bad luck at every school. The same, of course, is true for good luck.

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:I had this conversation with people from the Common Lottery helpline a month ago - I brought up that I noticed last year, because you could click on your DCPS desired school's admit and waitlist and then click on the random assigned numbers of the people ahead of you to see hwo they'd ranked their schools - and when I looked at most of the people who were in single digits, they were on single digits on ALL or MOST of their waitlists. Whereas many people who were #320 on one had sucky sucky 200s-300s numbers for all their choices too. It wasn't true 100%, but it was true enough at the many random people I looked at, and it was true of us (but we were lucky to have our numbers all be in the teens and 20s). The lottery staff said "No no that won't happen again this year", but really, it will, because preference people had preference before, and they do now, and there will be a computer attempt to go down each person's desired rankings, but at the end of the day the computer will use random lottery number, NOT desired school ranking, to choose between people who otherwise everything else is equal. With the exception of the number of people who previously would have applied to 20 charters and now have to rank most of them along with DCPS (and the double/triple spot-holding that that leads to), it really doesn't sound like it's that much different in terms of giving you a better chance at YOUR #1 ranked school. But my understanding from the school lottery folks was that you DID have a better chance under this model.

Thanks for this explanation, it is the first thing that seems reasonably clear to me. It also makes it abundantly clear that this system will be worse for individuals than applying to all of the school separately. At least in separate lotteries, I have an "new" chance of getting a good or bad lottery number for every school that we apply to. If I get shut out in one place, I could still have a shot somewhere else. This explanation makes it seem like with the common lottery, there will be one random good or bad number that will either give me an excellent shot at all of my choices or a terrible shot at all of my choices. The moment all the most coveted spots are filled, a few people will have the top spots on waiting lists for all the most coveted schools. By the same token, the same sad folks will be together in the bottom of all of the lists.

Hi, I'm the PP you're responding to. I think there's actually better news than this - if you keep reading the conversation past this post, it sounds like actually, even if the computer randomly assigns you #312 for the lottery and that stays your lottery number, you are not in fact #312 and everyone between #1-#311 has a better shot than you. From what I'm reading above, that is NOT how it will work.

If I understand the above, even if your random assigned number is #312, it's possible you could be #2 on the waitlist for your 1st or 2nd choice school and the person with random lottery #3 could end up #235 on a waitlist. It sounds like for each school, the computer looks at preference groups (i.e. everyone with sibling preference, or everyone with no preference at all) and within that preference group, for that school, does a random shuffle to establish the slots and waitlist. So no matter what your assigned lottery number (i.e. 312), you could end up #4 ont he waitlist for one school, number 87 on the waitlist for another school, and #452 for a 3rd school.

So if what's said above is true, actually your randomly assigned number does NOT condemn you to unifrom suckiness or awesomeness. It's a different shuffle for each school, even though your random number stays the same.

What you are reading above is not correct. There is no "random shuffle." That's just wishful thinking.

I am a statistician and know all about the deferred acceptance algorithm--you are incorrect. Please do some reading on this before you spread misinformation. And I don't know why you would think it to be "wishful thinking." It is simple statistics--doesn't make it any better for the chances that I, you, or anyone else will get in.

[Anonymous]

Anonymous wrote:

Anonymous wrote:
You are mistaken. The above is EXACTLY how the deferred acceptance algorithm cited in the q and a works.

Cite please?

The Q&A is in OP's post. There are lots of great academic papers on the deferred-acceptance method of school assignment; here is one: http://economics.mit.edu/files/6390

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:
You are mistaken. The above is EXACTLY how the deferred acceptance algorithm cited in the q and a works.

Cite please?

The Q&A is in OP's post. There are lots of great academic papers on the deferred-acceptance method of school assignment; here is one: http://economics.mit.edu/files/6390

This is why the other thread went to 85 pages. Lotteries in other places have nothing to do with DC. I've read the article, it's not about DC. Read about the DC lottery and then get back to us. A good starting point is the FAQ.

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:
You are mistaken. The above is EXACTLY how the deferred acceptance algorithm cited in the q and a works.

Cite please?

The Q&A is in OP's post. There are lots of great academic papers on the deferred-acceptance method of school assignment; here is one: http://economics.mit.edu/files/6390

This is why the other thread went to 85 pages. Lotteries in other places have nothing to do with DC. I've read the article, it's not about DC. Read about the DC lottery and then get back to us. A good starting point is the FAQ.

Yes, the FAQ that specifically cites this algorithm.

[Anonymous]

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:

Anonymous wrote:I had this conversation with people from the Common Lottery helpline a month ago - I brought up that I noticed last year, because you could click on your DCPS desired school's admit and waitlist and then click on the random assigned numbers of the people ahead of you to see hwo they'd ranked their schools - and when I looked at most of the people who were in single digits, they were on single digits on ALL or MOST of their waitlists. Whereas many people who were #320 on one had sucky sucky 200s-300s numbers for all their choices too. It wasn't true 100%, but it was true enough at the many random people I looked at, and it was true of us (but we were lucky to have our numbers all be in the teens and 20s). The lottery staff said "No no that won't happen again this year", but really, it will, because preference people had preference before, and they do now, and there will be a computer attempt to go down each person's desired rankings, but at the end of the day the computer will use random lottery number, NOT desired school ranking, to choose between people who otherwise everything else is equal. With the exception of the number of people who previously would have applied to 20 charters and now have to rank most of them along with DCPS (and the double/triple spot-holding that that leads to), it really doesn't sound like it's that much different in terms of giving you a better chance at YOUR #1 ranked school. But my understanding from the school lottery folks was that you DID have a better chance under this model.

Thanks for this explanation, it is the first thing that seems reasonably clear to me. It also makes it abundantly clear that this system will be worse for individuals than applying to all of the school separately. At least in separate lotteries, I have an "new" chance of getting a good or bad lottery number for every school that we apply to. If I get shut out in one place, I could still have a shot somewhere else. This explanation makes it seem like with the common lottery, there will be one random good or bad number that will either give me an excellent shot at all of my choices or a terrible shot at all of my choices. The moment all the most coveted spots are filled, a few people will have the top spots on waiting lists for all the most coveted schools. By the same token, the same sad folks will be together in the bottom of all of the lists.

Hi, I'm the PP you're responding to. I think there's actually better news than this - if you keep reading the conversation past this post, it sounds like actually, even if the computer randomly assigns you #312 for the lottery and that stays your lottery number, you are not in fact #312 and everyone between #1-#311 has a better shot than you. From what I'm reading above, that is NOT how it will work.

If I understand the above, even if your random assigned number is #312, it's possible you could be #2 on the waitlist for your 1st or 2nd choice school and the person with random lottery #3 could end up #235 on a waitlist. It sounds like for each school, the computer looks at preference groups (i.e. everyone with sibling preference, or everyone with no preference at all) and within that preference group, for that school, does a random shuffle to establish the slots and waitlist. So no matter what your assigned lottery number (i.e. 312), you could end up #4 ont he waitlist for one school, number 87 on the waitlist for another school, and #452 for a 3rd school.

So if what's said above is true, actually your randomly assigned number does NOT condemn you to unifrom suckiness or awesomeness. It's a different shuffle for each school, even though your random number stays the same.

What you are reading above is not correct. There is no "random shuffle." That's just wishful thinking.

I am a statistician and know all about the deferred acceptance algorithm--you are incorrect. Please do some reading on this before you spread misinformation. And I don't know why you would think it to be "wishful thinking." It is simple statistics--doesn't make it any better for the chances that I, you, or anyone else will get in.

I'm very happy for you. The FAQ makes it clear that there is no random shuffle. They may be using the term "deferred acceptance" improperly, but there is one draw, and that lottery number is used throughout.

[Anonymous]

I can't reveal my identity or the identity of the person I just spoke with but I assure you we have been intimately involved with this planning since day one.

The way we just discussed and confirmed it is that there is not just one lottery which would doom you to poor chances at 1-12 of your choices.

Rather there is effectively a lottery at each school/grade level.

just because you were last for your #1 doesnt mean you will be for your #2, you could be first.

Ranking your schools in order is the best way

And we see no need to rank your IB school if you already attend there or have sib preference and your older sib is enrolled. You have a slot already if you did an intent form.

The schools only release to the lottery the seats they have available after re-enrollment and preferences.

This is our understanding.

[Anonymous]

NP here, hoping to clear up some confusion. Some people are thinking that the "randomly assigned number" is their lottery number for each of their 12 picks.

But it's merely an identification. Think of it as "Student #312" who enters the lottery at each of the schools ranked by her family, not "Position #312" the spot in line for each of those lotteries. Otherwise, there'd be no way to track where a certain student ended up at each of the schools they selected.

For those lamenting the fact that your ranking does not influence whether or not you get a seat or your spot on the wait list, please understand that isn't the purpose of your choosing one school over another in any given order. Its purpose is to keep families from holding seats at multiple schools while they wait to get the seat they really want. For those using the example of Ross upthread, imagine if some of those inbound families ranked a charter above Ross and then got seats at those schools they preferred more. Maybe the chances of that are slim, but they're greater than they were last year.

And that's the reason it's important that every family rank their true preferences.