Hey there! Sign in to join this conversationNew here? Join for free

MHoC General Election March '13 - Post-results analysis Watch

    • Wiki Support Team
    • Welcome Squad
    • Thread Starter
    Offline

    18
    ReputationRep:
    Wiki Support Team
    Welcome Squad
    (Original post by TopHat)
    This is not actually true. Statistically speaking, it's most likely that any extra seats would go to the biggest parties, but it's not an absolute. Consider the following result:

    Alpha Party: 79 votes
    Bravo Party: 79 votes
    Charlie Party: 65 votes
    Delta Party: 53 votes
    Echo (Indy): 30 votes
    Foxtrot Party: 12 votes
    Golf Party: 6 votes

    This produces (before eliminations) 13 Alpha seats, 13 Bravo seats, 10 Charlie seats, 8 Delta seats, 4 Echo seats, 1 Foxtrot seat and no Golf seats.

    Once you remove Echo's 3 seats and redistribute them, the results are 13/13/10/9/1/2/1 - here, you can see all of the smallest parties actually gained, rather than the large parties. D'Hondt statistically is not certain to give extra seats to the large parties, just likely.

    Say we compare this with holding a by-election for excess seats using FPTP or even multivote. Look at all the three by-elections, one of which (the first) under multivote and the next two FPTP. Four of the five candidates elected from these by-elections were sponsored by either the largest or second largest bloc. By-elections are even more biased in favour of large parties than D'Hondt is.

    If you want to adjust the system so that small parties don't get hit quite so hard by skewed results, the answer is not to use by-elections but to change the selection algorithm from D'Hondt to Sainte-Lague. We wouldn't have to change any of the election process at all other than how seats are distributed at the end.

    It just goes from quot = V/(s+1) to quot = V/(2s+1). Because the divisor has changed so it "fragments" faster, it is statistically more likely that any extra seats will be picked up by smaller parties. If we thought it necessary, we could even go further and try V/(3s+1) or (4s+1).

    What happens with proportional systems is that you have decide what happens to the "spare votes". If absolutely everyone in the election got a vote percentage that was exactly equivalent to a multiple of 2% of the vote, D'Hondt, Sainte-Lague, and all the other variations would give out seats on a one seat per 2% basis - they'd work exactly the same.

    Unfortunately, people don't get nice multiples of 2%, they get weird and wacky stuff like 2.37%. That means the system you are working with has to decide how to hand out seats to the extra portion without losing too much proportionality. If you have 8 people entering, and they all got the exact worse amount possible (which would be (2n+1)% of the vote), there are 4 seats that can't be allocated with exact proportionality. The closer each party is to 2%, the smaller the number of seats that can't be allocated properly. However, we then make that worse by adding in independents, because all of the spare votes from the independent all also go into this pot.

    This where all of the methods start to differ. D'Hondt doesn't cut your votes up into chunks very frequently, which means if you are a big party and you get close-ish to your next seat, it'll favour you over a small party who is also close-ish. This isn't an absolute, but it is just statistically more likely to do so.. Sainte-Lague cut up votes more frequently, and can "spot" better when small parties have done well. Again, it's not absolute, just favourable. Similarly, the higher "k" gets, the more the system tends towards spotting the results of smaller parties.
    Sorry yes, that's what I meant. I do not know why I used the word "always" there... that's what happens when I don't do a proofread of my posts!

    If you have to re-allocate k seats then for small values of k it is most likely that only the largest parties will stand to benefit with the smaller parties not benefiting, this is what we saw in this election. For larger values of k, it is more likely that the small parties will start to benefit though not completely certain. Overall it seems that it is more likely that it helps the largest parties and this is down to the nature of the quotient as you've said.

    As for using Sainte-Lague, I've read a little bit about it in the past and liked it (for this House). I suggested by-elections last night without really thinking about it, to be honest I just wanted to get the ball rolling. In retrospect, I'm pretty sure I don't like my by-election suggestion - it would still probably end up benefiting more established parties and would mean a whole lot of extra election work for everybody. Sainte-Lague is a decent suggestion, I haven't run through the numbers to get a visual on how it would change things if it had been used in the past, but I think it's worth considering.
    Offline

    3
    ReputationRep:
    Great analysis Jarred, pleased with our result
    Offline

    2
    ReputationRep:
    To be honest I'm surprised that the Libers get the votes we do. In a world where people are constantly encouraged to vote away their rights in exchange for a share of the states credit card it is nice to see that not everyone is a freedom hating sheep who can be bought with empty promises and objects they already own.
    Offline

    16
    ReputationRep:
    Well done to the Greens. I voted & joined 2 days ago but I'm claiming my influence as integral.
    Offline

    1
    ReputationRep:
    (Original post by Jarred)
    Sorry yes, that's what I meant. I do not know why I used the word "always" there... that's what happens when I don't do a proofread of my posts!

    If you have to re-allocate k seats then for small values of k it is most likely that only the largest parties will stand to benefit with the smaller parties not benefiting, this is what we saw in this election. For larger values of k, it is more likely that the small parties will start to benefit though not completely certain. Overall it seems that it is more likely that it helps the largest parties and this is down to the nature of the quotient as you've said.

    As for using Sainte-Lague, I've read a little bit about it in the past and liked it (for this House). I suggested by-elections last night without really thinking about it, to be honest I just wanted to get the ball rolling. In retrospect, I'm pretty sure I don't like my by-election suggestion - it would still probably end up benefiting more established parties and would mean a whole lot of extra election work for everybody. Sainte-Lague is a decent suggestion, I haven't run through the numbers to get a visual on how it would change things if it had been used in the past, but I think it's worth considering.
    To be honest, the House is so small and we elect so few seats (only 50) that the difference between Sainte-Lague and D'Hondt is absolutely minimal. After some back of the hand calculations, k would have to be quite big (roughly speaking I think k would need to be about 4.7 before it started making even an appreciable difference, and would reach peak efficiency at 6.8). If you want I'll do a quick breakdown of the results of previous elections under k=4.7 and k=6.8 (bearing in mind we're at k=1 now).

    The other thing we could do is find some way to make independent candidates fairer, both for themselves and for parties.
    • Wiki Support Team
    • Welcome Squad
    • Thread Starter
    Offline

    18
    ReputationRep:
    Wiki Support Team
    Welcome Squad
    (Original post by rsplaya)
    So were UKIP leading before the election wax extended and a mass pm went out? Seems like a conspiracy to me.
    Supposing that the mass PM did go out but the election were not extended (though I expected it was less likely to go out without an extension), then UKIP would still have seen their lead diminish from what it was before the mass PM in time for the original closing time. It was not the extension that did it, it was the increased turnout, which is something we always get because we always send out a mass PM - so there's no conspiracy here

    (Original post by Abiraleft)
    Cool beans.

    Not that it mattered in my particular case, but could I just note that I never received the supposed mass PM, so it's possible that other people didn't either. I'm also hearing that some people received it a couple of times over, so I'm guessing there were some technical issues. :dontknow:
    I received two PMs myself. The first one contained a broken link and then it was resent out with the right one.
    I'm not sure why some people didn't get the PM but I think it is probably a technical matter as you say. If not that then it's probably something to do with usergroups. Perhaps your case is down to you being a Section Mod? Just a theory, I don't really know how it works. From what I can tell, sending a mass PM to everyone is quite a lot harder than the sort of mass PMs I send out to single usergroups.

    (Original post by Tuerin)
    There do seem to have been quite a few people who didn't get a PM, though.

    I'm also still very shaky about PM votes going through the Speaker; surely the mods should handle them?
    In the past it was handled by the mods, though one of these was the Deputy Speaker, TCIT. From what I can tell, the only reason we did this was because the Speaker himself wouldn't have been able to edit the poll due to it being in Current Affairs. TCIT was however a Section Mod (or at least some kind of mod) in D&CA so was able to edit the poll.

    This time what we did was make people send their votes to both Toronto and myself, and about an hour or so before the election I verified the votes I received with the ones he received. Thankfully our numbers matched, he confirmed that he had received the same votes, and I edited the poll in the final couple of minutes before the poll ended.

    If anything, this is a much more secure arrangement than what we had before. It's much easier for one moderator to conspire alone than for two to do it.

    The mods aren't without bias anyhow. The Section Leader for example is quite active in the Socialist subforum - and I would of course trust them with the poll and 100% doubt they'd ever let politics get in the way of fairness, but in theory they're no less likely to abuse the system than I or Toronto. The difference is that the mods are appointed and we're elected, furthermore elected on the promise that we relinquish political activities when conducting Speakership business which is not required of the mods. So in fact, there really is less of a chance that we'll abuse it than any one else.
    Offline

    12
    (Original post by Keckers)
    To be honest I'm surprised that the Libers get the votes we do. In a world where people are constantly encouraged to vote away their rights in exchange for a share of the states credit card it is nice to see that not everyone is a freedom hating sheep who can be bought with empty promises and objects they already own.
    PRSOM
    Offline

    3
    ReputationRep:
    (Original post by Jarred)
    In the past it was handled by the mods, though one of these was the Deputy Speaker, TCIT. From what I can tell, the only reason we did this was because the Speaker himself wouldn't have been able to edit the poll due to it being in Current Affairs. TCIT was however a Section Mod (or at least some kind of mod) in D&CA so was able to edit the poll.

    This time what we did was make people send their votes to both Toronto and myself, and about an hour or so before the election I verified the votes I received with the ones he received. Thankfully our numbers matched, he confirmed that he had received the same votes, and I edited the poll in the final couple of minutes before the poll ended.

    If anything, this is a much more secure arrangement than what we had before. It's much easier for one moderator to conspire alone than for two to do it.

    The mods aren't without bias anyhow. The Section Leader for example is quite active in the Socialist subforum - and I would of course trust them with the poll and 100% doubt they'd ever let politics get in the way of fairness, but in theory they're no less likely to abuse the system than I or Toronto. The difference is that the mods are appointed and we're elected, furthermore elected on the promise that we relinquish political activities when conducting Speakership business which is not required of the mods. So in fact, there really is less of a chance that we'll abuse it than any one else.
    That makes a lot of sense, thanks.
    Offline

    1
    ReputationRep:
    Excellent analysis, Jarred! Thoroughly enjoyed my first TSR election. Thanks folks! I look forward to some fantastic debates.
    • Wiki Support Team
    • Welcome Squad
    • Thread Starter
    Offline

    18
    ReputationRep:
    Wiki Support Team
    Welcome Squad
    The final seat allocations have changed slightly since posting this thread. TehFrance was banned before the original seat allocations were made so we have taken the decision to re-allocate the seat in the usual D'Hont fashion and as a result, the Liberals now have 5 seats, taking the one previously erroneously assigned to TehFrance.
    For an official statement on this, please click here.
    Offline

    2
    ReputationRep:
    Without breaking any site rules, what can you tell of us the reason for tehFrance being banned permanently from this area of the site as opposed to the rest of it?
    (Original post by Jarred)
    X
    Meant to post in Ask the Speaker, but will jsut edit in a quote rather than moving it.
    Offline

    3
    ReputationRep:
    (Original post by Jarred)
    x
    tehFrance you beautiful implosion
    Offline

    2
    ReputationRep:
    I'm getting tempted to start a Spoilt Ballot party
    • Section Leader
    Offline

    20
    ReputationRep:
    Section Leader
    (Original post by Tuerin)
    There do seem to have been quite a few people who didn't get a PM, though.

    I'm also still very shaky about PM votes going through the Speaker; surely the mods should handle them?
    Why would a mod be more trustworthy than the Speaker, whom we ourselves elected to represent us and handle MHoC matters?
    Offline

    3
    ReputationRep:
    (Original post by miser)
    Why would a mod be more trustworthy than the Speaker, whom we ourselves elected to represent us and handle MHoC matters?
    I believe Blair was voted in 3 times; I wouldn't say he was entirely trustworthy in office despite us having elected him to represent us several times.

    My original contention was that a mod (preferably one with no stated political interest on TSR) would be a more sensible choice, only because the Speaker had previously belonged to a party themselves and so there always remained the possibility of them abusing the position of relaying PM votes to favour their past connections.
    This wasn't a personal attack on Jarred, who has since explained the reasoning behind it (which I accepted some hours ago).
    Offline

    12
    Bleedin' heck! Permanent MHoC ban! Ouch, well done Liberals then I suppose.
    Offline

    17
    ReputationRep:
    Bleeding heck indeed :holmes:
    Offline

    12
    (Original post by nixonsjellybeans)
    Bleeding heck indeed :holmes:
    I'd be interested to know what happened.
    Offline

    0
    ReputationRep:
    Whilst I'm not in the business of speculation tehFrance did claim to be Morgsie a while back.
    Offline

    12
    (Original post by The Mad Dog)
    Whilst I'm not in the business of speculation tehFrance did claim to be Morgsie a while back.
    I don't believe you.
 
 
 
  • See more of what you like on The Student Room

    You can personalise what you see on TSR. Tell us a little about yourself to get started.

  • Poll
    Brussels sprouts
  • See more of what you like on The Student Room

    You can personalise what you see on TSR. Tell us a little about yourself to get started.

  • The Student Room, Get Revising and Marked by Teachers are trading names of The Student Room Group Ltd.

    Register Number: 04666380 (England and Wales), VAT No. 806 8067 22 Registered Office: International House, Queens Road, Brighton, BN1 3XE

    Quick reply
    Reputation gems: You get these gems as you gain rep from other members for making good contributions and giving helpful advice.