Why is predicting next UK General Election so difficult?

The polls are increasingly suggesting a Labour majority at the next UK General Election. However, here is the thing – never in the history of polling have polls been a less reliable indicator of the outcome. The truth it is phenomenally difficult to predict the next UK General Election. It is worth specifying why this is so.

Firstly, historically, it used to be relatively straightforward because the two main groups – Conservative-Unionist-NatLib on the centre right and Labour on the centre left – scored over 90% of the vote between them. It was relatively easy to take the spread of seats from the last election, calculate the “swing” in favour of one group or the other, and then apply that consistently across the seats to work out roughly how that translated into seats. Even this wasn’t perfect – as long ago as 1959, the first properly televised UK General Election, the swing was away from Labour in most places, but to Labour in Scotland and Lancashire (which was noted at the time and taken account of in calculations).

Secondly, it is now the case that even in a good year for the three main parties collectively, an eighth of votes cast in Great Britain are against them. With the growth of UKIP, this figure will only increase. That figure is not enough for any other party to win a seat except if it specifically targets one (e.g. the Greens in Brighton Pavilion), but it does make it hard to predict which of the three parties they are taking their votes from, and thus to predict a winner. Put another way, if the race is to 45-50% between two or three candidates, it is fairly easy to predict a winner from some basic polling; if the race is to 30-35% with up to 20% of votes cast constituting effectively a protest, it all becomes a lot closer. There aren’t, in fact, many three-way marginals, but it becomes quite possible that those that there are (e.g. Watford) will be won at under 30%. Predicting that reliably, even from accurate regional polling figures, is nigh impossible.

Thirdly, the loss of some parties’ votes may be more marked in some areas than others, and the scale of that difference may be difficult to pick up. For example, we can reasonably guess three things: a) the Liberal Democrat vote will decrease; b) it will decrease more in seats where they do not have an incumbent MP (and more so where than incumbent him/herself is not defending the seat than where he/she is); c) it will decrease more in regions which are not traditionally Liberal than in areas which always had a Liberal strain (e.g. the rural South West or the Scottish Highlands). All of this means that even if the Liberal Democrats lost more than half their vote, crashing to 8-12%, they could still retain more than half their seats (but it would be hard to predict which)… or not…

Fourthly, tactical voting is hard to call. In 1983, the Conservative vote share declined but their majority trebled because of a shift in votes from Labour to the then Liberal/SDP Alliance, which was rarely enough to elect a Liberal/SDP MP but often unseated the Labour one, leaving the Conservative elected thanks to the split. In 1997, however, voters essentially voted for whichever party was most likely to unseat the Conservative, giving the Liberal Democrats double the seats despite an unchanged vote share (the Conservatives only actually lost a quarter of their vote share, but more than half their seats).

After all this, the likelihood is that of all the votes cast for the three main parties, the Conservatives will probably have roughly the same share as in 2010 (maybe slightly more in fact), and Labour will have considerably more at the expense of the Liberal Democrats (although exactly how this works out will depend on region and who holds the seat). This would see the Conservatives lose some seats to Labour (as former Liberal Democrats lend their votes to the Labour candidate in Conservative/Labour marginals), but actually gain some from the Liberal Democrats (as Labour pick up Liberal Democrat votes allowing the Conservative to sneak through). Quite how many of each would happen is uncertain, as is the precise effect of a likely upturn in the UKIP vote (often Conservative, but often also Labour or just protest) and even the Green vote (an obvious destination for disaffected SDP-leaning former Liberal Democrats). In the end, there will be a fair degree of luck in simply just how the votes fall. It’ll be some election night!



One thought on “Why is predicting next UK General Election so difficult?

  1. Chris Roche says:

    This post from Chris Roche made no sense.

    Mr Roche – Please write sense and show respect, or frankly go elsewhere. – Ed

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: