Psephology for beginners

[andrewducker]

Here's some data from the Scottish Elections:

	Constituency%	List%	Difference%	Predicted seats (from list vote)	Actual seats	Difference
SNP	32.9	31.0	-1.9	40	47	7
Labour	32.2	29.2	-3.0	38	46	8
Conservatives	16.6	13.9	-2.7	18	17	-1
Lib Dem	16.2	11.3	-4.9	15	16	1
Green	0.2	4	3.8	5	2	-3
Total	98.1	89.4	-8.7	115	128	13

As you can see, tactical voting meant that for the constituency vote the larger parties were given a boost of a few percent. The Greens were the only party with successful candidates who actually got a boost in the Proportional part of the vote. In addition 11% of the population "wasted" their vote by voting for a party that didn't get any representatives at all. In fact as the total on the far right shows, 13 seats "should" have gone to other parties such as the Scottish Socialists, British Nationalists or Scottish Christians.

The system clearly favours the large parties. I'd be interested to know how the lists would have come out if they were country-wide rather than regional. I _think_ the end result would be more proportional, but I have other things to do tonight. Of course, depending on your opinions you might think that keeping the SSP,BNP, etc. out of government was a feature rather than a bug...

Comments

[nuttyxander.livejournal.com]

I think the result works pretty well to be honest, and it's clear that the SNP have gained in part thanks to the ability to convert the abilities of a good list MSP into excellent constituency candidates.

However, I am suspicious that the comparative speed with which counts were conducted in the west indicate a certain lack of examination of the spoiled ballots. If any valid list votes have been skipped on the grounds that they had no constituency vote on the same sheet then that is at least careless and at worst malicious.<conspiracy mode>

[nuttyxander.livejournal.com]

or even </conspiracy mode>

[davidcook.livejournal.com]

Small correction, assuming the BBC website is to be believed - SNP are on 47 seats, Labour on 46, Others on 3.

[andrewducker]

Cheers - fixed it.

And yes, Margo MacDonald isn't listed there, on the grounds that it made the maths more complex as she isn't a national party. The Greens are, though.

[heron61.livejournal.com]

So, (if it's not too personal) who did you vote for?

Also, I'm remain continually shocked that a party named the "anything" National Party is actually fairly progressive/leftist. Over hear, no one who was not a right-wing jingoistic nutcase would name a party the American National Party, which perhaps says a great deal about the US, where since the late 1960s the American flag itself has been a right-wing symbol. I very much envy you your country.

[andrewducker]

The SNP has, in fact, varied from being a right-ist party to being a left-ist one, over time.

Take a look at the Ideology section of
http://en.wikipedia.org/wiki/Scottish_National_Party

But yes, it is unusual. One assumes that if we ever get seperated from the rest of the UK they'll cease to exist and new parties will be needed...

[bohemiancoast.livejournal.com]

Bias against tiny parties is almost certainly a feature rather than a bug, in fact. The countries with PR systems that generate a seat or two for a host of tiny parties are generally held to expend a great deal of energy in trying to hang coalitions together with a resulting loss of energy for actual government.

[andrewducker]

It seemed likely that that was why they chose D'Hondt rather than Sainte-Laguë for their PR method. Seems unfair in some ways, but worth it in the long run.

[randomchris.livejournal.com]

I think your predicted seats should actually be from the average percentages of the two votes combined, rather than from the list vote alone. A party which doesn't make any showing in the constituencies will get a higher proportional share of the list vote. People vote tactically in a two-vote election; if it was list vote alone I think the standings would change noticeably.

This would give a prediction of SNP 32%, Labour 30.7%, Conservatives 15.3%, LibDem 13.7%, Green 2.1%, SSCUP 0.9%, Solidarity 0.75%, SCHP 0.65%, BNP 0.6%, Margo 0.5%. That'd be something like 41, 40, 20, 18, 2, 1, 1, 1 and 1. The numbers would probably be made up by a CLP, an SSP and an LLP, or whatever the other mob were. This would of course mean BNP representation. So I think the current system actually works quite well from that point of view.