Oxford Election Results Prediction

[woollyliberal]

I've been following this for a few weeks. They take the UKPR polling average for the week and feed it into their model of how polls typically change in the run up to an election. The predition they make is quite interesting. This week's probabilities are

49% hung parliament (25% Con largest, 24% Lab)
27% Con majority
23% Lab majority

Last weeks was 30% Con 20% Lab.

electionsetc.com/

[sdoerr]

So their prediction is for the Conservatives to be the biggest party, but 27 seats short of a majority. Yet the outcome with the highest probability is a Tory majority (27% probability, versus 25% for 'Hung, Con largest'). How does that work then?

-- 
Steve

[woollyliberal]

There's a description of their method here:

users.ox.ac.uk/~nuff0084/ge15forecast/

[greenchristian]

Jul 25, 2014 14:47:49 GMT sdoerr said:

So their prediction is for the Conservatives to be the biggest party, but 27 seats short of a majority. Yet the outcome with the highest probability is a Tory majority (27% probability, versus 25% for 'Hung, Con largest'). How does that work then?

-- 
Steve

Presumably they are averaging out the possible results. The modal result is not necessarily the mean or the median.

[woollyliberal]

This week's prediction is hardly changed from last week as the polls have hardly changed. 50% likelihood of a hung parliament, Con the largest party.

Con 299
Lab 295
LD 28
Others 28

[Deleted]

Aug 1, 2014 16:46:42 GMT woollyliberal said:

This week's prediction is hardly changed from last week as the polls have hardly changed. 50% likelihood of a hung parliament, Con the largest party.

Con 299
Lab 295
LD 28
Others 28

Did they give a breakdown for the Others? It would be crucial factor in any coalition talks with numbers like that. A separate TNS survey says a third of Conservative voters would like a coalition with UKIP should there be a hung parliament. I would probably include myself in that group, but it doesn't seem a possibility with those numbers.

[greatkingrat]

Sounds like they have just used the current Others figure - 18 NI, 6 SNP, 3 PC, 1 Grn.

[Deleted]

It would be absolute bloody chaos if those were the results.

[woollyliberal]

With 18 Northern Ireland MPs, Their prediction of 28 probably means single figures for UKIP. I'd have thought that if the Tories are close enough that a UKIP coalition would make the difference, they' do prefer to go it alone.

[Pete Whitehead]

Its not too far off the result I expect although I think the LDs will be 5-10 higher with the other two parties correspondingly lower.

[Deleted]

Aug 1, 2014 19:49:15 GMT woollyliberal said:

With 18 Northern Ireland MPs, Their prediction of 28 probably means single figures for UKIP. I'd have thought that if the Tories are close enough that a UKIP coalition would make the difference, they' do prefer to go it alone.

18 NI, 1 Green, 2 PC, 1 Speaker, 6 SNP = 0 UKIP?

[justin124]

In addition the SDLP would take the Labour whip effectively giving Labour 298.

[No Offence Alan]

Aug 2, 2014 18:12:44 GMT justin124 said:

In addition the SDLP would take the Labour whip effectively giving Labour 298.

And Sinn Fein (5/6 MPs?) don't attend, so bringing the "winning post" a bit nearer.

[Deleted]

Could a grand coalition be a possibility in these circumstances?

[woollyliberal]

Aug 4, 2014 17:55:08 GMT @barctom said:

Could a grand coalition be a possibility in these circumstances?

322 MPs just about gives you a voting majority. The Tories need to find 23 others and Labour 27 in order to cross the line. With the Lib Dems predicted to get around 28 MPs and the others combined (less SF) making 23, there are only a few coalition options avaialble:

Con + LD (maj 5)
Lab + LD (maj 1)
Con + everyone else (maj 0)

A grand coalition of either Con or Lab plus LD plus at least one other party would be messy but would give a slightly more comfortable majority. Would the comfort be worth the pain?

I get the feeling that both Con and Lab would prefer to go it alone and some in the Lib Dems would be happy to let them. I'd imagine a minority government would be able to do almost nothing for fear of being outvoted by the majority of the house.

[woollyliberal]

This week, con are moving forwards. 30% chance of an overall majority. Still 49% chance of a coalition. Labour majority the least likely outcome on just 21%.

The Tories are likely to be 27 seats short of a majority and Labour 30 seats. With the Lib Dems likely to get 28, 2015's likely result is looking a lot like 2010.

[Philip Davies]

Aug 1, 2014 19:50:34 GMT Pete Whitehead said:

Its not too far off the result I expect although I think the LDs will be 5-10 higher with the other two parties correspondingly lower.

Do you mean 5-10 seats higher or 5-10% of the vote higher? I suspect the former!

[Pete Whitehead]

Yes it was seats. The precentages of the vote weren't presented here

[woollyliberal]

Two lots of results since I was away
15th/22nd Con 35.9/35.9 Lab 32.2/32.2 LD 12.5/12.5 % of vote
15th/22nd Con 301/301 Lab 292/293 LD 28/28 seats

51% probability of a hung parliament, 53% probability of Con being the largest party.

It is worth noting that the margins of error on these figures tighten slightly every week, as you would expect from a prediction.

[Deleted]

I'm not quite sure how it deals with Scotland, but given that 'Others' remain the same I think the model could be gifting Labour seats like Argyll, Inverness, East Dunbartonshire and the like.