Categories
politicalbetting.com » Blog Archive » Estimating the “house effect” for each pollster. How much d
politicalbetting.com » Blog Archive » Estimating the “house effect” for each pollster. How much do they differ from the overall average for each party.
All of us who follow the polls closely know that some firms will be particularly beneficial to one party or another and generally produce some of the worst figures for another party.
0
This discussion has been closed.
Comments
Labour Party (Margaret Beckett)
Green Party (Caroline Lucas, Natalie Bennett)
Progressive Unionist Party (Dawn Purvis)
SDLP (Margaret Ritchie)
Plaid Cymru (Leanne Wood)
Communist Party of Great Britain (Nina Temple)
Peace and Progress Party (Vanessa Redgrave)
Respect (Salma Yaqoob)
and some other woman leaders of parliamentary groups (e.g. SNP) and/or joint leaders or spokespersons (e.g. Jenny Jones, Sarah Parkin, Shirley Williams)
and perhaps one or two in other countries as well
http://www.guide2womenleaders.com/woman_party_leaders.htm
My guess is that the best firm for the SNP might be Populus.
The worst is crystal clear: YouGov. Their weighting are just barmy, which was on display yet again in their last Scottish Euro poll, just a day before polling stations opened, which had SLAB in the lead. Oops!
That makes it all the more interesting that since the Euros YouGov have been showing some excellent SNP performances in the sub-samples, which pre-Euros almost invariably had huge SLAB leads.
"Hague said it would be a 'summit like no other' to end 'mass crime'"
London is the perfect place to discuss gang-rape.
http://www.theguardian.com/society/2013/nov/26/gangs-sexual-violence-warzones
"Sexual violence in parts of UK 'as bad as in warzones'"
On line:... Plus 1.8
By phone: Minus 2.5
A spread of 4.3 in the one company. Strange that, unless they ask very different questions,
What we know in practice is that ICM is right. Even on the Euros, which was not the polling companies' finest hour for understandable reasons, they were closest on the graphic the OGH put up.
This is interesting because their Labour score is 0.9% below the average. If the base line is ICM them Yougov overstate Labour by a full 3%. Puts crossover into perspective doesn't it?
Of course when we start adding in the takings of prostitutes and drug dealers in September we will find that we actually did this at the back end of last year.
The UK has been one of the slowest to achieve this for 2 main reasons. Firstly, a financial recession hit us harder than any of the other big boys given the importance of that sector in our economy so we had further to come back.
Secondly, the UK is fighting against a fairly strong headwind in falling north sea output. The contrast with the US where shale output has been ramped up over the same period is stark and has underlain the apparent difference in performance to a very large extent.
Still, no point in being cry babies about it. Even Salmond cannot wish north sea oil back into the ground (although he had a go claiming huge coal and gas reserves in Scotland this week without any reference to whether they are economically recoverable). We need to adapt to a post north sea world and get on with our own fracking soonest.
The mother of a disabled child has accused the First Minister of spearheading a “disgusting” smear campaign against her after she hosted the launch of the Unionists’ campaign for the final 100 days of the independence referendum contest.
Clare Lally said she felt sickened after one of Mr Salmond’s spin doctors contacted the Telegraph questioning her claim at Monday’s Better Together event that she was a “normal” mother.
http://www.telegraph.co.uk/news/politics/10891229/Alex-Salmonds-office-implicated-in-smear-attack-on-mother-who-spoke-up-for-the-Union.html
"There were three concerns about the general election anchoring method. Firstly, it was harsh on the Liberal Democrats, who were over-estimated by pollsters ahead of 2010 but have been scoring very low in the polls ever since they lost over half their general election support after joining the Coalition. The negative public views of the Liberal Democrats, and their very different political position as a party of government, make it less likely that the current polls are over-estimating their underlying support.
Secondly, a general election anchor provides little guidance on UKIP, who scored only 3% in the general election but poll in the mid-teens now, but with large disagreements in estimated support between pollsters (see discussion of house effects below).
Thirdly, the polling ecosystem itself has changed dramatically since 2010, with several new pollsters starting operations, and several other established pollsters making such significant changes to their methodology that they were equivalent to new pollsters as well.
We have decided that these concerns are sufficiently serious to warrant an adjustment to our methodology. Rather than basing our statistical adjustment on the last general election, we now make adjustments relative to the “average pollster”. This assumes that the polling industry as a whole will not be biased. "
http://blog.policy.manchester.ac.uk/featured/2014/06/polling-observatory-37-no-westminster-polling-aftershock-from-european-earthquake/
2010 General Election :
Con .. -0.1
Lab .. +1.0
LibDem .. +0.5
The 2010VI for the Cons showed that they lost less to UKIP and so their vote recovered. It is their best result since EU14.
Labour still have the largest share of LD2010 VI.
Also, for many YouGov polls, for the region Midlands/Wales, the Cons and Labour have been neck and neck (usual caveats in place). As in Wales Labour is very likely to outvote the Cons at the GE, what does that say about the Midlands?
Got a potential bet or two in mind for Austria, but need the markets to get going before I can see if the odds are there.
1. Is being different from average necessarily bad? Can one not look at polling performance over time to see who is the best? The average can be more wrong than any one pollster.
2. Is it possible to adjust polling outcomes with this to produce a more accurate prediction?
"Welsh youngsters need to show they have the skills they need to succeed anywhere in the world, the first minister will tell head teachers later.
Carwyn Jones will give a keynote address at a Cardiff conference on the Pisa international education tests.
Wales has been the worst performing UK nation in the three-yearly tests prompting a drive to improve standards.
The international organisation behind the tests has criticised Wales for a lack of long-term vision for education.
The concerns raised by the Organisation for Economic Co-operation and Development (OECD) in April were first highlighted in a report for the Welsh government more than six years ago, unpublished at the time but recently passed to the BBC...
Once again, Wales was ranked bottom of the four UK nations and fell further behind other countries, with east Asian cities and nations coming out on top.
Reforms including statutory reading and numeracy tests had already been introduced in Wales following previous disappointing Pisa results.....
The conference will also hear from OECD director for education and skills Andreas Schleicher, who runs the Pisa tests.
He has previously claimed that the impact of poverty on education has been overstated, pointing out that maths results for the poorest children in Shanghai matched those of the wealthiest UK pupils."
http://www.bbc.co.uk/news/uk-wales-27785129
So that destroys the argument loved by Labour that children's performance should be judged by those who get free school meals.
2. Return of the Hapsburg dynasty - 66/1
3. The waltz banned as anti social - 33/1
4. Austrian corporal becomes elected dictator - TSE - Evens
5. Conchita Wurst revealed as love child of Peter the Punter - 4/6 fav
Is it me, or is there lots of sport on, or coming?
As well as the excellence of F1, there's the rioting of the World Cup (with a side-order of Blatter lunacy), Wimbledon, the rugby tour in New Zealand and then the Commonwealth Games.
UKPR on Labour's uplift.
http://ukpollingreport.co.uk/blog/archives/8862#comments
However, I reckon that - outside of four or five seats - they're going to accrue the most votes in places where it doesn't matter: that is, they'll get big second places in Labour heartlands, and they'll get a lot of angry Tories in the shires.
I also suspect that Farage will lose where he stands, simply because the seat he stands in will be opinion polled to death, and - unlike with Diane James, for example - there will be tactical voting against him.
Question: if UKIP gets four seats (which is a perfectly sensible guess), and Nigel Farage is not one of them, can he sensibly remain as leader?
http://www.dailymail.co.uk/news/article-2654679/Miliband-A-political-freak-Prescott-He-touched-wife.html
Airbus said the cancellation had followed "ongoing discussions with the airline in light of their fleet requirement review".
In 2007, Emirates placed 50 orders for the A350-900 and 20 for the A350-1000, with deliveries due from 2019.
UK engine maker Rolls-Royce said the cancellation would lead to a £2.6bn hit to its order book.
http://www.bbc.co.uk/news/business-27791448
Is this a sign of retrenchment or of increased competition from Ethiad?
You are coming across as whiny.
Any comments on the narrowing of Labour's lead in the latest YG poll?
Mr. Jim, well, I think Wimbledon's rather good, but I'm not taken by the World Cup.
YES remain dead in the water, completely becalmed.
Tick tock ....
Electoral calculus estimated that tactical voting cost the Conservatives 26 seats in the 2005 election.
http://www.electoralcalculus.co.uk/conlabgap.html#tactical
Re: YouGov. You should know better Charles.
Just one poll.
Labour still in the high thirties.
Tories at the top of their range.
MOE.
Look at the trend.
Don't forget US Open, which starts today and the First Test tomorrow!
The recent introduction of the Ashcroft polls is a further complication, as they will modify the average used to calculate against.
The ‘house effect’ strikes me as a little odd as it appears to be anchored to an arbitrary reference point that in itself could be flawed. – I’ll stick with the gold standard ICM, thanks.
I would have thought it was the sort of thing you would do...
Edited extra bit: Mr. Fett, not into golf.
Mr. W, anyone who thinks Caesar was a better general than Alexander is too mad to hold political office.
I'm not into golf.
Mr. W, I believe that Mr. Eagles' claim that Caesar was a better general than Alexander disqualifies him from holding political office in Austria on grounds of insanity.
I'm not into golf.
Mr. W, I believe that Mr. Eagles' claim that Caesar was a better general than Alexander disqualifies him from holding political office in Austria on grounds of insanity.
That said, the differences between ICM and the average are very interesting, and heartening for the Coalition parties.
The implications for 2016 are far reaching.
http://survation.com/new-constituency-polling-in-south-thanet/
Doesn't tactical voting require a 2 horse race?
Weirdly, my earlier disappearing post not only appeared a bit late, it did so twice.
Let's hope Rod Crosby didn't see your post.
We'll be subject to misogynistic rants about witches.
Then things will move on to birther stuff about Obama.
And we'll segue naturally from there into holocaust denial.
A normal day on PB.
However "impossible" is a substantial stretch in betting terms.
Only 1 draw in last 10 (back to 2009) at Lords - that was vs Lanka but a day lost to rain.
7/1 from Stan James on Plunkett being top England wicket taker looks excellent also.
Let's go for "highly unlikely".
http://www.nationalreview.com/corner/380057/looking-back-tea-leaves-cantor-john-fund
http://www.nationalreview.com/corner
Today the Labour Market for June 2014.
Summary:
Employment: UP to 30.5 million
Total Unemployment DOWN to 6.6%
Youth Unemployment DOWN by 1.3%
Economically Inactive DOWN to 8.8 million
Key Findings from ONS Bulletin
• Comparing February to April 2014 with November 2013 to January 2014, there was a large increase in employment and a large fall in unemployment. There was a further fall in the number of economically inactive people aged from 16 to 64. These changes continue the general direction of movement over the past two years.
• There were 30.54 million people in work, 345,000 more than for November 2013 to January 2014 and 780,000 more than a year earlier.
• 72.9% of people aged from 16 to 64 were in work, up from 72.3% for November 2013 to January 2014 and up from 71.5% a year earlier. The latest figure of 72.9%, for February to April 2014, is 0.1 percentage point lower than before the downturn of 2008-09.
• There were 2.16 million unemployed people, 161,000 fewer than for November 2013 to January 2014 and 347,000 fewer than a year earlier.
• The unemployment rate was 6.6% of the economically active population (those in work plus those seeking and available to work), down from 7.2% for November 2013 to January 2014 and down from 7.8% a year earlier.
• There were 8.82 million economically inactive people (those out of work but not seeking or available to work) aged from 16 to 64. This was 80,000 fewer than for November 2013 to January 2014 and 178,000 fewer than a year earlier.
• 21.8% of people aged from 16 to 64 were economically inactive, down from 22.1% for November 2013 to January 2014 and down from 22.4% for a year earlier. The latest figure of 21.8%, for February to April 2014, is the lowest since 1990.
• Pay including bonuses for employees in Great Britain for February to April 2014 was 0.7% higher than a year earlier, with pay excluding bonuses 0.9% higher.
Isn't it time taxpayers took out "key individual insurance" on George Osborne?
Weird pay growth figures though. Nice to see the revision up of last month's pay figures to 1.9% though.
What we do know, roughly, is why the polls differ: sometimes they have similar raw data and then adjust them differently (compare YG and ICM, for instance). The assumptions which drive those differences may be valid sometimes and invalid at other times. For example, ICM assumes that half the uncertain 2010 LibDems will vote LibDem. That might have been right when in opposition. Does it still apply now? Perhaps it's more, or less? The other big difference is that pollsters vary wildly in weighting for stated certainty to vote - some only count those who say they're absolutely certain, some don't even ask, some weight on a sliding scale, some adjust by what voters did last time. Which assumption is right, in marginal seats in 2015? Only the brave would claim to know for sure.
IMO the only safe thing is to look at the trend within one pollster. Same assumptions, same sampling method. If anything is happening (and usually it isn't!) that's where it will show.
My betting antennae always twitch in a sporting contest when "impossible" is invoked. Even "highly unlikely" raises an eyebrow in modest interest.
Unfavoured would fit the bill suitably.
Immigration reform is now off the agenda. Right wing economic policies are on. Ted Cruz is a live candidate again. It's a blow to a number of others; I should think Bush, Rubio, and Christie will be badly hampered.
Beyond that, it won't help the GOP in 'must-win' States like Florida, Virginia and N Carolina when the 2016 Presidentials roll round.
Apparently Nancy Pelosi could hardly contain her mirth when the result came in.
This also caused a big fall (£0.8 bn) in NI and PAYE Income Tax receipts in April 2014 when compared to April 2013 in the last Public Finances Bulletin. The fall occured at the same time as all other economic growth linked receipts rose above OBR's forecasts.
We need a few more months of stats to see what the underlying trend on income related tax receipts and wage level really is.
"Is anyone still foolish enough to believe Rubio will be in the running?"
Well he's 11/2 favorite with the bookies and second fav on Betfair.
That adds up to a lot of fools.
Difficult to say what the true position is.
This confirmed (but not quantified) in main body of ONS bulletin:
The single month growth rate for total pay for April 2014 (minus 1.7%) was the lowest since March 2009. This reflects an unusually high growth rate for April 2013, due to some companies which usually paid bonuses in March paying them in April.
I'm sure @Morris_Dancer will be pleased that paleo-F1 was so lucrative
Still some short term headlines for Mr Which? magazine.
All the adjustments in the world can not make up for an inadequate sample in terms of size or randomness, and those of us in online panels are not typical of the wider public.
That said the trends over time probably have some meaning within any particular methodology, but surely we are an over polled society. Politicians should lift their noses to the horizon rather than obbsess over minutiae.