Monday 21 February 2011

An unfortunate series of events (the perils of AV).

Ok, as requested.

There is an election.
For Simplicity only, lets say it involves three candidates.
One Labour,
One Liberal Democrat,
One Tory (what the hell, we've taken back the term ;) .

Results are as follows.
Labour 45%
Lib Dem 30%
Tory 25%
FPTP winner Labour. But this is an AV election, so onto the next round

The Tory Candidate is eliminated. In the spirit of the coalition (and to keep out Labour), 18 of the 25%  second pref Lib Dem and 2 of the 25% vote labour (SHOCK!), 5 were exhausted.

We now have a winner. On 48% of the total vote, the Lib Dem's against 47% of Labour.
Close , but it is obvious that the most popular person won right?

Lets have a look at the information that AV doesn't reveal then.

Lib Dems second pref'd 20 of their 30% voted Labour and 5 of the 30% voted Tory, 5 were exhausted

Labour second pref'd 15 of their 45% to the Lib Dems (Yellow Tories) and 5 of the 45% to Tory, 25 exhausted.

So how does that make the end result look?

Lib Dems AV winner has 30% first round votes and 33% of the second prefs. (63% of people preferenced them 1st or second)

Tory has 25% of the first round votes and 10% of the second prefs (35% of people preferenced them 1st or second)

Lastly we have Labour, FPTP winner. Has 45% of first preferences and 22% of the second prefs (67% of people preferenced them 1st or second)


This highlights the problem of allowing just 25% of people in a vote to second preference.

*Obviously I could have made these figures a lot more biased to prove my point even more but I think this is a quite fair representation of how it could look.*
The AV winner had neither the most support from the populace or the most concentrated support.
In short, it is perfectly possible for the person with the most votes in FPTP to be more supported in the voting populace than the winner under AV.

Which is one of the reasons I support FPTP, not that it'll always get it right, but because it processes the information given to it right.
 If 45% voted labour, that is what it will tell you. If 25% voted Tory, that is what it would tell you.

What AV would tell you under that election up there is that 48% of people support the Liberal Democrats and 47% people support Labour. It is a system that is designed to misinterpret the raw data given to it.

The End (or is it? DUM DUM DUUUUUUM)

7 comments:

  1. "it is perfectly possible for the person with the most votes in FPTP to be more supported in the voting populace than the winner under AV."

    Is it not also perfectly possible for the person with the most votes in FPTP to be the least supported candidate?

    ReplyDelete
  2. An unfortunate example, given that - in this example - the Lib Dems are the Condorcet winners. In a straight two-candidate fight, they would beat both the Labour candidate *and* the Tory candidate. Whichever way you try to spin it, more voters prefer the LD candidate to the Labour candidate, than the other way around (48% of people gave LD a higher preference than Lab, 47% gave them a lower preference). In a straight Con-LD contest, the LDs would win by a considerably larger margin (50-30).

    So AV gets the right result here. A Borda count could conceivably give you the wrong one.

    ReplyDelete
  3. Why has the condorcet method suddenly got this God like status amongst Av'ers?
    The lib dems winning would only please 63% of people , labour winning would please 67% of people. Any system claiming the lib dems should win...not so good

    ReplyDelete
  4. The problem with that argument is, as I said, that in a straight two-candidate fight between the Lib Dem and Labour candidates, the Lib Dem candidate would win. So your argument against AV... is also an argument against FPTP in a situation where only these two candidates were standing.

    As such, it makes your argument an incoherent mess.

    ReplyDelete
  5. Why would a conservative candidate not stand when possibly 1/4 of all people support him?

    ReplyDelete
  6. By giving as many options of differing political views, (which is why we have so many candidates on the ballot) we can actually find out who has the most real support.. which is why I am not in favor of suppressing everything down to two parties...which AV tends to do by forcing coalitions.

    ReplyDelete