Categories
politicalbetting.com » Blog Archive » Trying to explain why the Ashcroft National Poll appears to
politicalbetting.com » Blog Archive » Trying to explain why the Ashcroft National Poll appears to be so erratic
At the excellent PB gathering in Ilkley last night the big topic of conversation was the extraordinary Ashcroft national poll which saw the 2% CON lead of last week become a 7% LAB one.
0
This discussion has been closed.
Comments
2 hours 2 minutes 2 seconds
The whole concept of weighting is fraught with potential for bias due to the the large adjustments to the raw data. It is the trend over time with the same methodology that is most useful. As yet Ashcroft has no clear trend.
http://www.telegraph.co.uk/news/matt/
Speculation that the fall has been caused by OGH changing from Marks and Spencer Y fronts to Calvin Klein boxers is rife in the world of Bedford haute couture.
You really are a sad little person, what will you do after a YES vote.
And the Stones' drummer Bill Wyman had that publicised affair with a girl that began when she was thirteen.
Little movement on the subsidiaries except on Taxation where Cons move from +4 to +10.
The YG polls (published in July) 2010 splits, show that both the Cons and LDs are retaining more of their supporters and losing less to UKIP. Also the LDs are losing a few less to Labour.
Whilst this Nelsonian approach has happened for decades, who gave the guidance to the police; as a time when the police pick and choose which laws they uphold is a very dangerous time and is the start of a police state.
1 hour 1 minutes 1 second
Since it looks like most of us will be over 100 yrs old by the time there is a future YES vote, I would guess that we would be reading the congrats from our King.
Nick Bennett, a member of the Williams Commission, will soon become Wales' public services ombudsman.
Published in January, the Williams commission said the 22 local councils should be cut to as few as 10.
Mr Bennett said he was "disappointed" a cross-party deal could not be reached by Easter - the date set in the Williams report for ministers and councils to agree a way forward - and urged politicians not to treat the report's recommendations as a "political football".
"Currently the people of Wales are being let down," said Mr Bennett.
"Too often services are poor and patchy. Perhaps you can have excellence and mediocrity or even failure within the same local government structure.
"We've got to be more ambitious as a nation and we've got to insist upon better."
Mr Bennett said the "status quo is not an option".
http://www.bbc.co.uk/news/uk-wales-28197270
Early days, but when the 2015 F1 Drivers' market goes up one thing I'll be looking for are the odds on Bottas, depending how Williams continue to develop. With Rosberg, Hamilton, Alonso and Vettel (and maybe Ricciardo) having undoubtedly short odds, I'm hoping the odds might be tempting on Bottas.
Incidentally, I hope the young maths student whose name escapes me delurks. He talked a lot of sense.
It seems more likely to me that the problem here is weighting. They are getting 500 random responses but are not weighting them back to the actual voting of the last election or alternatively getting a fully representative mix of social groups (IPSOS). The result is a much greater volatility than we see in other polls who will smooth out that any particular group of respondents happens to have more self employed or unemployed or tories or whatever than national make up or the previous election indicates.
Weighting is tricky, especially when you have a new force like UKIP and a long track record which allows the results to be refined against previous election results clearly helps. What I find odd is that although Lord Ashcroft's polling (in this form) is new my understanding is that he has an established pollster doing it for him so he is hardly starting from scratch.
As things stand these polls look like a waste of money to me telling us nothing other than the make up of the panel this week. That is a pity and I would hope that Lord Ashcroft would look to get better value for his money. These polls should be a better contribution to the debate than they currently are.
LAB 330 (329) CON 264(268) LD 32(29) UKIP 0(0) Others 24 (Ed is crap is PM)
Last weeks BJESUS in brackets
BJESUS (Big John Election Service Uniform Swing)
Using current polling adjusted for 300 days left to go factor and using UKPR
standard swingometer
On June 24th, the CBI Distributive Trades Survey reported that retail sales growth had slowed badly in June, setting off alarm bells:
the balance of retailers that reported sales were up year-on-year crashed to a seven-month low of +4% in June from +16% in May and +30% in April.
The CBI said a fall in clothing and food sales contributed to a disappointing month on the high street, and cautioned that slow wage growth continued to weigh on consumer sentiment.
Some commentators even suggested that the downturn might delay a BoE rate rise.
With the early retail sales forecasts/surveys it is always worth waiting to get a broad picture from a number of sources.
We now have the Visa Consumer Expenditure Index in for June and the alarm can be put on snooze. Yes there has been a slowing in the rate of growth in June, but this may only be month-to-month volatility combined with the annual baseline, June 2013, recording an above trend high. The second quarter of 2014, taken as a whole, recorded the highest annual growth rate since 2010.
All the same there was growth in only four of the eight spending sectors compared to seven out of eight last month, so it will be prudent to wait for the July figures before reaching any final conclusion on trends.
The tables:
Unfortunately I had a bit of a medical incident yesterday afternoon and couldn't go out in the evening.
I hope that all who did go enjoyed a touch of Yorkshire.
A clear majority also expect to see profitability to improve although I have some trouble in seeing how when there are rising input pressures and price increases in output seem extremely difficult to put through.
The key for the economy as a whole is that optimism has to be converted into investment. Although there has been some good news here there are still problems with credit availability and increasingly concerns about the state of some parts of the world economy, especially the EU.
As I have said before we need a long sustained period of above trend growth to make the spending cuts that the next government (of whatever complexion) will have to introduce less painful.
5 minutes
The breaking news is that WIND is reporting to the JNN the contents of the latest ARSE 2015 General Election and "JackW Dozen" Projections. (Change from 24th June Projection) :
Con 307 (-5) .. Lab 282 (+8) .. LibDem 30 (-2) .. SNP 8 .. PC 2 .. NI 18 .. UKIP 1 (-1) .. Respect 0 .. Green 1 .. Ind 0 .. Speaker 1
Conservatives 19 seats short of a majority
Labour 44 seats short of a majority
......................................................................................
"JackW Dozen" - 13 seats that will shape the General Election result :
Bury North - TCTC
Pudsey - TCTC
Broxtowe - Likely Lab Gain
Warwickshire North - Likely Lab Gain
Cambridge - Likely LibDem Hold
Ipswich - TCTC
Watford - TCTC
Croydon Central - Likely Con Hold
Enfield - TCTC
Cornwall North - TCTC
Great Yarmouth - Con Hold (From Likely Con Hold)
Vale of Glamorgan - Likely Con Hold
Ochil and South Perthshire - Likely Lab Hold (From TCTC)
Changes From 24th June - Great Yarmouth moves from Likely Con Hold to Con Hold and Ochil and South Perthshire moves from TCTC to Likely Lab Hold.
TCTC - Too Close To Call - Less than 500 votes
Likely Hold/Gain - 500 - 2500 votes
Gain/Hold - Over 2500 votes
.......................................................................................
WIND - Whimsical Independent News Division
JNN - Jacobite News Network
ARSE - Anonymous Random Selection of Electors
That's the chances of future Tory majorities seeping away right there.
Clearly the Green single seat projection is for Brighton Pavilion and it is a "Likely Green Hold" at the top end of the range noted below.
Brighton Pavilion was not included in the "JackW Dozen" because as an "anomaly seat" it will have no bearing on the overall landscape of the General Election.
Given that the LD’s will have lost half their seats, I’d expect the latter, although how easy will that be to sustain with fixed term Parliaments?
It'd be interesting if you labelled the swing required for each seat in your next update
http://www.theguardian.com/business/2014/jul/08/interest-rate-rises-chambers-of-commerce-economic-recovery
Of course a rate rise would derail the current recovery. The current recovery is entirely due to the low interest rate hence its brittle nature.
Talk of a rate rise massively premature.
edit: age as well
The dead tree press numbers have little or no bearing on election outcomes in the 21st century.
Have you passed your Economics prelims while you have been away?
Your conclusion at least seems much more on target than before.
Coalitionista yes .... Tory no.
so net effect would depend on what you think the net TV and internet effects are
e.g. if you thought the BBC had a big anti-right effect and the papers a big pro-right effect then the decline of the papers would have a net anti-right effect
Ed Miliband Will Never Become Prime Minister
JacW 22/6/14
BJESUS Ed is crap is PM interesting just under 10 months ahead.
Who will be proved right?
I notice ARSE moved closer to BJESUS and a little closer to the possibility of EICIPM in past 2 weeks?
.......................... JackW .................... Fisher
Con ...................... 307 ....................... 305
Lab ...................... 282 ....................... 288
LibDem .................. 30 ......................... 28
Others .................... 31 ........................ 29
Total .................... 650 ....................... 650
They can't both be wrong, surely?
It appears George Osborne and George Carney have finally got round to listening to the Warwickshire Malcontents.
Dozens of banks are poised to enter the market in the wake of barriers to new entrants being lifted, according to regulators.
The Financial Conduct Authority (FCA) and the Bank of England’s Prudential Regulation Authority (PRA) said five times as many businesses are currently applying for banking licences as were granted them last year.
And before the cynics start questioning the "five times" statement, there were 5 applications for new [retail] banking licences last year (3 granted) compared with 25 this year.
See; http://bit.ly/1pUP6vd
Of course, I could add historical stats for shares of the UK retail banking market taken by new entrants or start up banks, but let's not spoil the morning. Mr. Brooke should be permitted to get his pre-holiday tan by basking in the glory of George's sunshine.
Somewhat more seriously though, I do like the acronym!
What concerns me is the consistency or rather lack of consistency in the application of the law. For example drivers at 35mph are often prosecuted but publicans who serve people who are very intoxicated are not and we see the effects on our streets and in A&E.
If a law is a bad law, then surely it is up to our MPs to get it changed or if a law is not enforced then to ensure it is enforced. This should come from the home secretary as we do not want to see inconsistencies on law enforcement across the UK or private interpretation by Chief Constables.
24.6.14 LAB 330 CON 263 LD 33 UKIP 0 Others 24 (Ed is crap is PM)
BJESUS (Big John Election Service Uniform Swing)
Using current polling adjusted for 316 days left to go factor and using UKPR standard swingometer
BJESUS prediction GE2015 - To be released at 8.50 every Tuesday till 5/5/15
1.7.14 LAB 329(330) CON 268 (263) LD 29(33) UKIP 0(0) Others 24(24) (Ed is crap is PM)
Last weeks BJESUS in brackets
BJESUS (Big John Election Service Uniform Swing)
Using current polling adjusted for 309 days left to go factor and using UKPR standard swingometer
8.7.14 LAB 330 (329) CON 264(268) LD 32(29) UKIP 0(0) Others 24 (Ed is crap is PM)
Last weeks BJESUS in brackets
BJESUS (Big John Election Service Uniform Swing)
Using current polling adjusted for 302 days left to go factor and using UKPR standard swingometer
BJESUS consistently right or wrong
However, it's mildly flattering to note that denizens of academia are appearing at the side lines waving their imitations in direction of my ARSE.
http://www.theguardian.com/business/2014/jul/08/marks-and-spencer-results-clothing-sales-agm-business-live#block-53bbac6fe4b0e7b8fe27820c
On topic: Mike is confusing two different points when he writes "if accuracy was down to sample size alone then how come YouGov’s 6,000 sample final online poll fared so badly in the GE2010 accuracy table behind five pollsters with 1,000 sample phone polls?"
The sample size determines the random statistical error of a poll, in other words the volatility from one poll to the next arising purely from the fact that your sample might have randomly happened to pick up a few more or a few less of the supporters of a particular party. The random error in the Ashcroft polls is high because the sample size is small - less than 500, after adjusting for Don't Knows etc. (And it is made worse, as Oblitus has pointed out previously, by the fact that some important groups are heavily weighted up, which means the real sample size is effectively even smaller than it appears at first sight). Therefore, irrespective of any other considerations, you would expect the Ashcroft polls to vary randomly by several percent from one poll to the next, even if actually nothing whatsoever has changed. That is exactly what we see.
There is a totally separate set of errors, related to a systematic bias. If, for example, you carried out very large polls by telephoning people at their workplace, you might get consistent results from week to week (because your sample size is very large, the random error would be small), but they'd be consistently wrong because your sample is biased towards people of working age who are in employment. Pollsters therefore try to to minimise systematic bias through various sampling and weighting techniques, but it's not an exact science. So the fact that YouGov's last poll wasn't particularly accurate could mean that their weighting were slightly wrong or that their sampling technique under-represented some significant voting groups, which systematically (rather than randomly) skewed the result slightly. Of course it might also be the case that some other pollster equally had weightings slightly wrong but, by chance, the random error compensated for this in their last poll.
We see a good example of systematic bias in the current UKIP figures, where some pollsters consistently have much higher figures than others. Clearly, therefore, there must be a systematic error in at least some of the polling methods used, even if we can't be sure which if any is right.
It's a pity that the Ashcroft polls do have such relatively small sample sizes. He might get better value for his money if he did the polls less often but with bigger samples.
May 2014 grew by 2.3% over May 2013.
Apr 2014 grew by 2.9% over Apr 2014.
This mainly explains why May 2014 over Apr 2014 shows a decline of 0.7% (with manufacturing down 1.3%).
Variations caused by changing baseline comparators are not covered until the second year of Ben's economic degree course.
LAB 342 CON 248 LD 31 Other 29 (ukpr)
Ed is Crap is PM Less than 10 months to go
Polls are not elections.
It really has made my day as the Guardian don't even understand the figures.
"Total factory production across the UK fell by 0.7%"
No.
What fell by 0.7% (on a month on previous month basis) was Total Production including the four subsectors, Manufacturing; Mining and quarrying [the North Sea bit]; Electricity, gas, steam & air conditioning; and , Water supply, sewerage & waste management.
If the Guardian had understood the figures they would have been banging on about a 1.3% month on previous month fall in "total factory production across the UK".
Even Ben got that bit right.
What neither mentioned was that the 0.7% fall was increased by a 0.1% downward revision to growth in Apr 2014.
"Total factory production across the UK fell by 0.7%"
No.
What fell by 0.7% (on a month on previous month basis) was Total Production including the four subsectors, Manufacturing; Mining and quarrying [the North Sea bit]; Electricity, gas, steam & air conditioning; and , Water supply, sewerage & waste management.
If the Guardian had understood the figures they would have been banging on about a 1.3% month on previous month fall in "total factory production across the UK".
Even Ben got that bit right.
What neither mentioned was that the 0.7% fall was increased by a 0.1% downward revision to growth in Apr 2014.
Genuine question. If the figures include water supply, sewerage & waste management, where does handling the winter-spring floods come in?
A fine looking bunch of Tykes, but where were all the Laydeez?
That doesn't however stop the applicants themselves announcing if they so wish.
I did read somewhere that the Burnley chap got his knuckles rapped by the regulators though but it wasn't for the kind of mistake that this would cause the BoE to stop his application.
I'm surprised they don't just take that market down and put up a winner without Mercedes and Top 3 markets.
Williams have a good chance of overtaking Ferrari for third.
http://ukpollingreport.co.uk/blog/archives/8892
09/06/2014 08:35
Enoch Powell's uncannily accurate prediction of the euro crisis in 1978. (via @WhiteWednesday): #HeWhoMustNotBeNamed pic.twitter.com/793AuZDZeH
"Total factory production across the UK fell by 0.7%"
No.
What fell by 0.7% (on a month on previous month basis) was Total Production including the four subsectors, Manufacturing; Mining and quarrying [the North Sea bit]; Electricity, gas, steam & air conditioning; and , Water supply, sewerage & waste management.
If the Guardian had understood the figures they would have been banging on about a 1.3% month on previous month fall in "total factory production across the UK".
Even Ben got that bit right.
What neither mentioned was that the 0.7% fall was increased by a 0.1% downward revision to growth in Apr 2014.
Well there's statistics and statistics. There really should be one national measure of growth and output, or lack of, that easy to understand and explain to the uninitiated, like me.
The impact would have mainly been on February 2014 with some carry over to March. So the current bulletin won't be impacted, although next months figures will have a preliminary Q2 on Q1 comparison.
In general the impact of the floods was not significant enough for the ONS to call them a "special event" but I would expect you to be right in speculating that the water etc subsector would have been most affected.
Here is the full blurb on the sub-sector from this bulletin:
Water & waste management
Water supply, sewerage & waste management output increased by 2.3% between May 2013 and May 2014 and contributed 0.2 percentage points to production growth, with increases in all of its subsectors (see Figure 4). The largest contributor to the increase was the waste collection, treatment & disposal activities subsector, which increased by 3.6% and contributed 0.1 percentage points. The second largest was sewerage output, which increased by 2.2% and contributed 0.1 percentage points. This was followed by increases in water collection, treatment & supply, which increased by 0.4%; and in remediation activities & other waste management services output, which increased by 1.6% and both had negligible contributions.
Water supply, sewerage & waste management output decreased by 1.1% between April 2014 and May 2014, with decreases in all of its subsectors (see Figure 5). The largest contributor to the decrease was in the waste collection, treatment & disposal activities subsector, which decreased by 1.4% and contributed 0.1 percentage points. This was followed by decreases in sewerage output, which decreased by 1.3%; water collection, treatment & supply, which decreased by 0.3%; and remediation activities & other waste management services, which decreased by 3.4% and all had negligible contributions.
So a similar trend to the headline figures, with an annual rise and monthly fall. The water subsector only accounts for 8% of the total Production Index of Production, with IoP only accounting for 15% of total output. So differential shit flows tend not to discolour the headline GDP headline figures.
LAB 33% CON 27.67%%
The median and modal averages are both LAB 33 CON 29
These Brasenose fellas...
It's the immigration statistics that are truly dodgy, Special K!
http://www.bbc.com/news/uk-northern-ireland-28206581
I read your post after making my previous comment to you on this topic.
I am reluctantly adding the House to the credits given to Brasenose.
Whatever happened to Nicomachos/G.? He needs to return otherwise the PB lefties will claim a sample of one is not very reliable. At least Grandiose can call in Dave for support.
Agree about your two examples, though - the "35 is/is not speeding" unofficial rule needs to be standardised or abolished, and the drunks in pubs question has both interpretation and enforceability issues.
Either way I'm not concerned by the drop, it is a statistical phantom. There is enough data from other firms and companies showing massive growth in manufacturing and production.
Have you seen this man .... recently escaped from Bedford and now spotted in the Ilkley area sporting a body hugging yellow lycra suit and wanted for spreading alarm and despondency among Conservative supporters nationwide.