The Futility Monster

He'll pointlessly derive more enjoyment out of your resources than you

Roll-y Poll-y

Posted by The Futility Monster on April 24, 2010 @ 23:42

There is a real danger in my poll projection methodology at the moment. The huge highs recorded in the Lib Dem share are estimating that the figure for “Others” could go as low as 6%, which is never going to happen. At the last election the three main parties achieved 90% of the British vote. That will not be lower this time, especially with the rise of the Nats.

So there is going to have to be some revision to my methodology. Again. Before that, the new numbers…

CONSERVATIVES: 33%

LIB DEMS: 29%

LABOUR: 28%

No changes from last time, because of the methodology change.

I have decided, because of the huge volume of polls coming through at the moment, on average three a day, to close the polling window down to just 10 days. 10 days of polls is roughly 30 polls, which is sufficient to get a good view of what’s going on. And, in light of the major developments in recent weeks, it’s important the averaging becomes a little more sensitive. 10 days will achieve that.

I have also decided that it would be wrong to use maxima and minima. Labour are not going get 23%, whcih is what the minimum would predict. It is not going to happen. The Lib Dems, much as they usually outperform the polling, are not going to get 34% this time.

So now we’re all on medians. All the above numbers are now the medians spread across all results from members of the British Polling Council.

It looks and feels a little closer to the truth. A major drop for the Tories, and a major rise for the Lib Dems. As the election enters its closing stages, this index will be regularly updated.

What do these results mean in terms of seats? Using our old, discredited friend – UNS – we get…

  • ukpollingreport.co.uk – Conservatives: 247 seats; Labour: 276 seats; LD: 96 seats: LABOUR 50 seats short
  • Electoral Calculus –  Conservatives: 258 seats; Labour: 265 seats; LD: 95 seats: LABOUR 61 seats short

Unfortunately, I can’t even try it with Andy Cooke’s version, principally because his system was designed in the bad old days when the Lib Dems knew their place.

So we’re in bizarre territory. Potentially Labour coming third in share of the vote, and yet getting the most seats. And thus getting the first crack at forming a government. God bless FPTP!

But UNS can’t possibly be a useful guide in such a close election. It cannot cope with a genuine third party “surge” like this. It’s going to make some of these predictions look very bad come results day. Unless, that is, the Lib Dems do fail to consolidate their vote in the key marginals, and instead lift up everywhere, potentially gaining very little as a result. That is my worry.

And yet, these numbers sort of feel right. Perhaps they may be transposed; maybe the slightly greater swing to the Tories in the key marginals will mean the Tories get roughly what we think Labour will get in such a close scenario. Either way, both parties are way, way short of a majority.

All this can change so easily. A couple of percent either way will make the outcome more decisive, and bring in lots of other scenarios, potentially involving other parties than the Lib Dems.

But if we get stuck here, there will only be one winner: the Yellow Peril. The Lib Dems will get the choice of government. Or neither, precipitating another election. That would be highly dangerous. I sense a lose-lose situation for them.

So many scenarios. So much still to play for…

Advertisements

One Response to “Roll-y Poll-y”

  1. […] Roll-y Poll-y […]

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s

 
%d bloggers like this: