11

I have just taken part in a draw and it has left me wondering whether I made the right decision to enter the draw as an individual or whether I should have entered as a group, and if so, what size group would have been optimal.

For this draw my partner and I signed up as individuals. Neither of us wanted to deprive the other of a place if we didn't get in, but since I got a place and she didn't, and I now know the mechanics of the draw, I'm prompted to question our choices.

The set up is as follows.

  • A game has 30 places for male players and 30 places for female players.
  • 100 players have signed up for the game (half male and half female), so not everyone can play.
  • Players can either register as individuals, or register in a group, such as a couple, family or fraternity of friends. No-one can register more than once.
  • Individuals are assigned a single draw number. If an individual is selected before all places for their gender are filled they get in the game, otherwise they are placed on a wait list.
  • Groups are also assigned a single draw number. If a group is selected before all places required for that group are taken they all get in the game, otherwise they are all placed on a wait list.

For example, let's say the distribution is as follows:

  • 35 Male
  • 35 Female
  • 11 couples (FM)
  • 2 fraternities of four (MFFM)

This means that there are 83 draw numbers, but each draw could allocate 1, two or 4 places, or add that number of players to the wait list.

As an individual, do I have the same chance of getting a place as I would if I were registered as a couple? What about if I registered as part of a larger group?

Looking at this naively, I would assume that since there are 30 places for 50 players of each gender, both my partner and I would have a 60% chance of getting a place, a 36% chance that both of us would get a place and a 16% chance that neither of us would get a place.

Even more naively I assumed that if we registered as a couple, we would still have a 60% chance of getting a place, and thus 40% chance of not getting a place, but is that true?

† This is a simplification, the actual draw also had a small number of gender neutral players and places.

  • 2
    Please confirm that a player can register in only one category. (That is, "register as .. OR'' is an exclusive OR.) – mlc Mar 22 '17 at 08:48
  • Good point @mlc I have edited my question to confirm that is the case. – Mark Booth Mar 22 '17 at 09:03
  • I don't understand what happens with the wait list. Does this simply mean: you are out of contention for this game, but considered for the next? Or is it something else? – Ingix Mar 22 '17 at 09:30
  • It doesn't matter in the context of the question @Ingix but in practical terms, sometimes people drop out. Real life gets in the way and some time in the 9 months between when they sign up and when the game runs, they decide they can't make it. When someone drops out, the next person on the wait list gets their place. – Mark Booth Mar 22 '17 at 12:08

1 Answers1

3

You personally have a higher chance of getting a place if you enter as an individual. To see this, imagine that the other entries are fixed, and you either enter two separate individuals, or a couple and a blank piece of paper in the draw (this is to keep the number of things in the draw the same, but if the blank paper gets drawn out it just gets ignored). Now we order the things in the draw, and condition on knowing what order everything will be drawn except which of those two is which (so you know the positions of the two pieces of paper relevant to you, but not which is which, and you know the position of everything else). Now given this information, and whether you entered as a couple or an individual, you can tell whether you get a place in $0$, $1$ or $2$ of the $2$ possible arrangements. But this number will always be at least as large if you enter as an individual as it is if you enter as a couple, since if when the first (say) piece of paper is drawn there is enough room for a couple, there is certainly enough room for you as an individual.

edit: The only difficult case above is as follows (thanks to ingix for pointing this out). Is it possible that one of you (say, male) being chosen first prevents the other from being chosen second, but owing to a different subset of the groups drawn in between being given places, you would have gotten through as a couple in either position? Actually it isn't. Suppose that from the first paper one of you gets a place as an individual male. In order for this to change which groups get through between the two papers, there must be a group which would have gotten through had the first paper been blank, but didn't because you took an extra male spot. In that case, if that group did get through it would use up the last male slot, so you wouldn't get through as a couple.

This gives you at least as high a probability of getting a place as an individual. Provided there are some people entering as individuals, there are some arrangements of the others where you do strictly better as an individual.

For the example entries you suggest, I wrote a python program to simulate the draw. It suggests that the individuals have roughly a $61.5\%$ chance, the couples $56.8\%$ and the larger groups $55.9\%$ (based on a million runs).

from random import shuffle

draw=[(0,1)]*35+[(1,0)]*35+[(1,1)]*11+[(2,2)]*2
success={(0,1):0,(1,0):0,(1,1):0,(2,2):0}
runs=1000000
for i in range(runs):
    shuffle(draw)
    m,f=0,0
    for x in draw:
        if x[0]+m<=30 and x[1]+f<=30:
            m+=x[0]
            f+=x[1]
            success[x]+=1
for x in success:
    print("{}\t{:.1f}".format(x,100*success[x]/(draw.count(x)*runs)))
  • 1
    I don't doubt your simulation, but your mathematical argument about entering as an individual being better is not as straightforward as you may think. It is correct when your paper is the first being drawn, then this position is the first where a difference happens between entering as an individual and entering as a couple, and being individual is clearly better. But when your are the second paper, the first difference happened earlier: when the first paper was drawn your partner entered the game (individual) or nothing happened (couple). (To be continued) – Ingix Mar 22 '17 at 10:21
  • 1
    Your partner entering may have caused a different set of couples/groups of 4 being accepted or rejected into the game, which may in the end cause yourself to be rejected in the individual case. I can't construct such an example with the given set of groups, but it is possible if the groups can be imbalanced gender-wise: I assume that 'you' are male and your partner is female. Start with 27F and 28M already selected into the game, then comes your first Paper ($P_1$), then a group FFFM, then MF, then MF, then your second paper ($P_2$). (Again to be continued) – Ingix Mar 22 '17 at 10:33
  • Ah, yes, you're right. I was assuming all groups are balanced, so that the male partner getting a place first can only make things easier. But maybe even that isn't as obvious as I first thought. – Especially Lime Mar 22 '17 at 10:39
  • OK, it turns out all my counterexamples don't seem to work. But I still stand by my original statement that your argument needs more to back it up. – Ingix Mar 22 '17 at 10:40
  • True! I'll try to come up with a way to fix it, thanks. – Especially Lime Mar 22 '17 at 10:43
  • I think it now works. – Especially Lime Mar 22 '17 at 11:03
  • Yep, good work! – Ingix Mar 22 '17 at 11:45
  • Any chance that you could post your code in a gist? After posting my question, I knocked up a quick monte-carlo simulation and I got 60% +- 1% for every option. Also, I don't see the point of the blank pieces of paper. Certainly the real draw didn't use them. There were as many 'draws' as there were individuals or groups. – Mark Booth Mar 22 '17 at 12:15
  • In the real draw @Ingix there were non gender balanced groups, there was a FMMM for instance. Most of the couples were FM, but i think there were also FF and MM couples. Certainly I wouldn't want to exclude such a possibility from the analysis, but as with Neutral gender, I was simplifying for the sake of getting a more general answer. Maybe I should go back and update with the specifics and see if that changes the chances in my simulation. – Mark Booth Mar 22 '17 at 12:19
  • The point of the blank piece of paper is basically that if you perform a draw, then want to know the effect of taking out your two individual entries and replace them with one joint entry, keeping everything else the same, you still have the choice of which of the two positions you originally had to put your joint entry in. Each choice should be equally likely. I'll post the code in my answer as it's not very long. – Especially Lime Mar 22 '17 at 12:52
  • The argument does work fine with groups that aren't evenly divided between the genders. The only complication is: what happens if the draw ends without all of the places filled? Do the remaining places get filled from the waiting list in some way? I've not counted that possibility in my analysis. – Especially Lime Mar 22 '17 at 12:58
  • The nice thing about the python code being in a gist is that you can run it easily with python anywhere *8') – Mark Booth Mar 22 '17 at 13:00
  • Ah, I'll post my code later. I handle waitlisting differently to you, which may explain the difference in results. – Mark Booth Mar 22 '17 at 13:03
  • I don't handle waitlisting at all. If you're giving the waitlist priority for the next draw, then it makes sense that you will get 60% either way, but I assumed you wanted to know how your chances of getting a place on the first draw changed, so I'm just simulating the first draw 1M times. – Especially Lime Mar 22 '17 at 13:31
  • What I mean is that you only update the m & f variables if success is incremented, I always update the m & f variables, but only add a success if everyone can get in. – Mark Booth Mar 22 '17 at 14:16
  • Ah, perhaps I'm working on the wrong model then. I assumed that if there was one female place left, and the next two entries drawn are a couple (MF) followed by an individual female, the couple would get added to the waitlist and then the individual would get the place. (In other words I assumed waitlisting does not reduce the number of places remaining, which is why I don't change the variables when that happens.) – Especially Lime Mar 22 '17 at 14:35