Understanding the Massive Flaws in UK Polls
The above polls are from Wikipedia Opinion Polling for the 2019 United Kingdom General Election.
I removed columns for Area (all Great Britain) and poll size. I also removed columns for UKIP, Change UK, and Other (all polling zero) to present a better looking chart.
To understand one of the major flaws in the polls, we must first understand the population.
UK Percentage Makeup
All of the above polls were conducted in Great Britain, the political entity. This is the first major mistake because some of the political parties exist in only one country.
- SNP: Scotland Only - The Scottish National Party is a social-democratic political party in Scotland. The SNP supports and campaigns for Scottish independence within the European Union.
- Plaid Cymru: Wales Only - The Party of Wales, is a social-democratic political party in Wales advocating Welsh independence from the United Kingdom within the European Union.
- DUP: Northern Ireland Only - The Democratic Unionist Party (DUP) is a unionist political party in Northern Ireland favouring British identity.
- Sinn Féin: Northern Ireland and Ireland - Sinn Féin (pronounced shin-FAYN not sin-fine, and commonly shortened to SF) is the political branch of the IRA an Irish republican party, operating in both Northern Ireland and the Republic of Ireland.
The pollsters avoided issues with DUP and Sinn Féin by not polling Northern Ireland. But the big problems are Wales and Scotland, especially the latter.
Sinn Féin Impact on the Majority
Sinn Féin MPs did not show up in in the last Parliament. They have pledged once again to sit out.
I calculate 8 Sinn Féin MPs, none of which will sit. This affects the number needed for a majority.
Martin Baxter, CEO of Electoral Calculus figures things this way.
Of the four speakers and deputy speakers, two are Conservative and two are Labour so that always cancels out. For that reason I count Lindsay Hoyle as Labour, because he is balanced by a Conservative deputy speaker. Deputy speakers stand under a party banner and are contested in their seats.
If Sinn Fein do not attend and if they had won 8 seats, then a party would need 322 seats (including speaker and deputies) to have a majority of 2. But there is no absolute guarantee that Sinn Fein will not attend, so to be certain a party needs 326 seats to have a majority.
SNP and Plaid Cymru Discussion
- SNP is polling 3-4%
- SNP currently holds 35 of 59 seats in the UK parliament on that 3-4% of the vote.
- How? 4% of Great Britain is almost 50% of Scotland!
It makes no sense to report SNP in polls or even take polls in Scotland, but they all do. Moreover, some analysts believe the SNP will take every Tory seat in the next election, a pickup of 13 seats to 48, all on that same 4% of the vote.
Wales is not as lopsided, but Plaid Cymru, polling about 0% is likely to win 2-4 seats.
Percentage of Undecided Voters
- The YouGov Nov 5-6 Poll says 10% will not vote, a whopping 17% don't know who they will vote for, and 2% refused to answer.
- The PanelBase Nov 6-8 Poll says <1% will not vote, 10% don't know, and <1% refused to answer.
- The percentage of undecided voters as polled by ComRes, DeltaPoll, Opinium is a mystery. None of them answered my question when I inquired about undecided voters.
- George Pinder, Senior Research Executive at ICM did answer. Pinder says 14% are undecided.
Undecided and Nonvoters YouGov, Panelbase, ICM
- YouGov says 17% of the voters are undecided
- PanelBase says its 10% of voters are undecided.
- ICM, by email, says 14% of voters are undecided
- YouGov says 10% won't vote but PanelBase says <1%
- I did not ask ICM how many would not vote.
- 1% of PanelBase seems unbelievably small.
- YouGov translates a raw Tory 25% to a net 39%
- YouGov translates a raw Labour 18% to a net 26%
- YouGov translates a raw Lib Dem 12% to a net 17%
- PanelBase translates a raw Tory 34% to a net 40%
- PanelBase translates a raw Labour 24% to a net 30%
- PanelBase translates a raw Lib Dem 13% to a net 15%
- I just have an overall number from ICM. I do not have raw numbers to show what happened at ICM.
Not Confidence Inspiring
The extra questions in all the polls were along the lines of: Who is doing a good job? How did you vote last time? Is Boris Johnson a racist? Is Corbyn anti-Semitic? Who will make a better leader?
Those questions are anemic. They provide little or no information on voting intention.
We are supposed to assume the pollsters will be accurate. And perhaps they are, but this is not confidence inspiring.
Badly Needed Question for Undecided Voters in England
- Are you primarily undecided between Labor and Tory? Leaning which way?
- Are you primarily undecided between Labour and Liberal Democrat? Leaning which way?
- Are you primarily undecided between Tory and Brexit Party? Leaning which way?
- Other - Leaning which way?
There is a vast difference between the eventual resolution of Labour-Tory uncertainty than Tory-Brexit Party uncertainty.
The impact between Labour and Liberal democrats will be region-dependent.
Regional parties exacerbate the problem. In terms of figuring a Tory majority, it does not matter at all if someone is torn between SNP and the Liberal Democrats.
Undecided Scotland votes are most likely irrelevant. If not, a separate poll on Scotland-only can sort it out with this question: Are you primarily undecided between SNP and Tory? Leaning which way?
Thus, Great Britain-wide polls add to the uncertainty confusion.
Three Major Flaws
- Not breaking out Scotland and Wales independently
- Not disclosing the percentage of undecided voters or the methods used in resolving the allocations
- Not asking badly needed questions to help factor the nature of the voter indecision.
I am nowhere near that total.
For now I am sticking with my 351 seat projection as noted in UK Election Math: What are the Odds of a Hung Parliament?
Election in 4 Weeks!
Mike "Mish" Shedlock