Talk:Swing (politics)

Untitled
Trying to properly characterise the dispute between what I'm calling conventional swing and Butler Swing rather than have a POV page that asserts that conventional swing is useless. --Po8crg 18:15, 10 February 2006 (UTC)

Usefulness as a measure of degree of change
I've amended it to say that it's used as a measure of the degree of change, rather than a that it is useful for that purpose. Clearly the media have some reason for quoting conventional swing, and I think this article should explain why they use it, rather than asserting it's a complete waste of time and then saying "but everyone uses it anyway". Po8crg 11:10, 12 February 2006 (UTC)


 * Unfortunately this happens to be true. David | Talk 21:35, 16 February 2006 (UTC)

You're making this article much better - good work! It's worth bearing in mind though that even Butler himself is happy to talk about "swing" when it involves parties other than simply Labour-Tory. E.g. his co-authored book on the 1997 election says (p.69) "... Liberal Democrat gains on overwhelming swings at Newbury and Christchurch" Mark 86.53.37.182 20:45, 16 February 2006 (UTC)

Why swing must be Conservative-Labour only: an objective view
The problem is this. If you have a measure which people use to say "10% swing to Labour in this constituency, 10% swing to Liberal Democrat in that one" then that is a statement which equates the two measures. It implicitly states that the two constituencies, albeit in different directions, had the same degree of voter change. However, the voters do not so think it. It is much easier to get voters to convert in large numbers between the Conservatives and the Liberal Democrats or between Labour and the Liberal Democrats or between any party and the Scottish National Party or Plaid Cymru, than it is to get voters to convert in large numbers between the Conservative Party and the Labour Party.

This appears to be because voters largely continue to understand that the only two parties which might actually form a government are Conservative and Labour. A crude mathematical way of trying to produce near equivalence in terms of the degree of voter change would be to half the 'swing' if it isn't Conservative-Labour, and that would be approximately the same degree of voter change than if it was. This is because voters who change from Conservative or Labour to a third party understand that they are in effect withholding their support for their original potential governing party, without transferring it to the other, and therefore only going halfway.

What this means is that it becomes quite pointedly misleading to readers ever to use swing in a context not involving both the Conservatives and Labour. It also is entirely misleading in terms of likely election outcomes because, for whatever reason, the 'uniform swing' (or variations of it such as uniform regional swing) have never applied other than between Conservative and Labour.

Come up with some way of resolving these difficulties and it would be very handy. But in the meantime, let us not mislead readers by using the exact same term to refer to two entirely different concepts. David | Talk 21:34, 16 February 2006 (UTC)


 * I agree that it's easier to get big swings to the Liberal Democrats than to either of the larger parties (though, interestingly, not so much away from the Lib Dems). But it doesn't represent the "degree of voter change"; it represents the number of voters who changed.  It's a matter for the reader to interpret the fact that big swings between Labour and Conservative are far more significant than the reverse.  Of course, this is the page where we should be giving that guidance on interpretation so people can understand that swings between one pair of parties are not comparable to those between another pair.


 * I don't agree that uniform swing doesn't apply when parties other than Lab and Con are involved. If you take the 2005 General Election as an example, Lab-Con swing has a standard deviation of about 1.5%, where both Lab-LD and Con-LD swing have a standard deviation between 2% and 2.5% (and yes, that's original research).  Given the much smaller sample sizes for the swings including the LDs (because the LDs have a lot more third places than the other two), I think the predictive power is best described as "similar".  Swings involving other parties than those three have sample sizes in single figures, so it's hardly surprising that they have no predictive power.  I think your problem is that you're expecting swings involving a third place party to mean something and therefore you include them in your "uniform swing" calculation.  Throw all of those constituencies out of your sample, and you see meaningful uniform swings.  The only reason you can get away with including them in Lab-Con is that there are relatively few of them, and so they are swamped by the Lab-Con seats.


 * Finally, there are some elections where swing is almost meaningless. Littleborough and Saddleworth by-election, 1995, is the classic example.  What happened is that the Tory vote collapsed catastrophically, with part going to the Lib Dems and part to Labour.  If swing is meant to represent how many voters (net) changed their opinions since the last election, then no two-party swing makes any sense for L&S. Po8crg 07:19, 17 February 2006 (UTC)


 * They perhaps are different concepts, but they are both called 'swing'. I'm not sure if the swing used in this article is the one most commonly referred to.  Paulleake 19:28, 18 March 2006 (UTC)


 * Much as I'd like to be able to say that we don't use the same term to refer to two different concepts, the fact is that Wikipedia is not the determinant of English usage. Paul Leake is right - there are two different concepts which are both called 'swing'. Surely our job here is to differentiate them.Richard Gadsden 19:23, 2 May 2006 (UTC)

Swing Analysis
I have added a example of a simple swing analysis for the Ukraine Parliamentary Elections 2006 to 2007. The analysis is straight forward as all parties represented contested the election and the electoral districts remained the same. The use of Swing charts can convey a lot of information one graphic. Most importantly it shows the movement in voter support as the percentage of votes change from one election to another. — Preceding unsigned comment added by ElectAnalysis (talk • contribs) 17:31, 4 February 2008 (UTC)

2007 Results 2006 Results

Names
Is there a reason why "Swing Analysis" is capitalized? It is not usual to capitalize nouns that are not proper names. For instance, no one would normally write of "Budgetary Analysis" or "Pearson's Chi-Square Test" with capitals (except in the proper name "Pearson"). Even "Butler Swing" is unusual; the normal spelling would seem to be "Butler swing". Zaslav (talk) 04:51, 31 July 2009 (UTC)

Two-party swing.
I have some problems with this paragraph.
 * It is claimed to be particularly Australian when it is widely used in the UK unless it is something other than what I'm thinking of.
 * It is not explained. (If it is what I am thinking of then the swing from party A to party B is the average of swing from A and swing to B as defined in the previous section.)
 * The criticism is opinion and therefore needs referencing and I doubt that it all can be referenced. For a start it is claimed to make certain assumptions e.g. of a two party system when the very calculation above allows for extra parties. --Peter cohen (talk) 14:00, 8 May 2010 (UTC)

Swing in ridings with multiple candidates (2+)
Hello, I've been calculating the swing for some Alberta elections by riding, as being the change in vote given to the elected candidate vs. how much the person running under that party banner received the last time. I don't think this is correct. Does anyone know how to calculate the swing for a riding like Calgary-McCall? I'll ask on the Canadian wikiproject too, but I figured here might be good because that information could be included in this article. Thanks, Ajraddatz (talk) 00:50, 5 April 2016 (UTC)