I had an odd discussion about voting systems

[andrewducker]

Over on Twitter I was having an odd argument about voting systems, where I really didn't understand the other person's argument. Probably because trying to put across a significant point in 140 character is. both hard and frustrating. So I offered to stick up a post about it, so we could continue the conversation over here. That being the case,

Comments

[steer]

Note that what I am about to say here is in no way a defence of FPTP. STV is a superior system to FPTP. If the choice was there tomorrow and it was STV vs FPTP I would vote STV. Because I support something does not mean I support any argument for it. My pet peeve is people making erroneous arguments for stuff I support.

OK -- so I start from the standpoint if you're going to look at a news item and go "hooray, this supports my viewpoint" you have to ask yourself the question "would you have accepted this argument if it was made in favour of something you disagree with". He're we're told "Electoral system X was used and 76% of people got their first preference". In general if you are told "X is good because Y=Z" you have to ask
(a) does Y=Z actually represent a "good" result
and
(b) is this a representative comparison of X versus some alternative
and
(c) is Y what you should look at anyway.

Continued next post.

[steer]

Taking these in turn, firstly is "76% of people got their first preference" actually a large number of people to get their first preference. No, it's not. It's actually very easy to have a high proportion of people get their first preference results in most elections. This was for the Scottish local elections of 2012 I am guessing -- if not some of the rest of this looks a bit silly but the general point applies. I am presuming their result means that person got their first preference in that ward. In most places this meant three or four candidates to each ward I think. Assume 3 for pessimism. If you simply threw the votes away and elected 1 SNP, 1 labour and 1 tory you get 76% without looking at any votes. That particular election had actually a quite diverse vote split too. If you looked at (say) the 2010 general election then you would get 88% of people getting first preference by giving every seat a tory, labour and LD candidate irrespective of the votes. The 76% figure seems high because we're not used to analysing

[andrewducker]

So they are comparing "there was one person elected, you got the one you wanted most" with "there are three or four people elected, you got one of the ones you wanted".Except that a lot of the time people don't. As I said in paragraph four. Because you frequently end up with very similar distributions of votes across areas. And thus all three of them voting in the party with 40% of the vote, and three SNP councillors elected. Which is one of the reasons why Scotland's local elections got more diverse after STV, and we went from lots of councils that had one party in control to the vast majority of them being either minority or coalition

[steer]

You seriously don't consider this a dishonest comparison? I am gobsmacked. So if it was the other way round, they compared the number of people who got their first pref in a single member STV election with the number of people who got their first pref in a multi-member FPTP (which, by the way, optimises first pref wins by design) then you'd say "oh, good point, nice graphic, FPTP must have some good points I hadn't previously grasped".

I'm sorry but I view it as amazingly dishonest.

If you're taking this as an argument for multi-member versus single member then... well, it's an argument... it's not a great argument.

You keep coming across as if systems can only have one advantage. And that's simply not true.

Yes, but on the other hand, you're looking at something that STV is terrible at and being mislead by an abuse of stats into believing it is good at it. STV is bad at getting people their first choice candidate. This is by design.

what system you're thinking would produce it that's actually being suggested by anyone?The

[mummoth]

We had a referendum on STV in my province in 2005. It lost by a narrow margin. I voted in favour of STV.

We've got a federal election coming up, under FPTP. What a lot of people are worried about is that the vote will be split between the smaller parties and our current government willb e re-elected. The Green Party is very small, so small that a vote for them is considered a 'throw away' vote... but in talking to people there's quite a bit of support for them. They'd have a much better chance in STV.

I can't find the exact figures, but the percentage of seats our government has is much higher than the percentage of votes they received, because of the way ridings have been divided... supposedly the government has no say in that, but they are super shady so I think they have manipulated things in their favour. I think STV will reduce if not eliminate the skewed results this creates.

[markama]

very well said

[kalimac]

I didn't even try reading all of steer's comments. Because the problem is very simple. The reported superior voter satisfaction of the result is a function of it being a multi-member constituency with a single vote, not of its using a preferential system of voting. If you had a large multi-member constituency with FPTP voting instead of STV, assuming that each voter got only one vote, you'd still have a high percentage of voters getting someone from their first-preference party elected.

Conversely, in a single-member constituency, AV [the single-member equivalent of STV] would not increase the percentage of voters getting their first preference elected; in fact it might decrease it because they wouldn't be forced into tactical voting.

[steer]

Pretty much agree with it all here.

In a single member constituency no system can beat FPTP by the measure "most voters getting their first preference (*) candidate elected" because that is what FPTP does. (*) Preference by vote here -- it may be you don't vote for the person you actually wanted but that is what the chart here is measuring.

[andrewducker]

I don't think there's any disagreement about that.