I had a really good look at the YouGov stuff last week, to see if there was anything that supported the theory of it being due to LabourLosingWomen.
I don't think it was anything notable or out of the ordinary or significant shift shown last week.
See my post on page 4
Sun 11-Feb-18 10:39:51
Its about what YouGov is showing and what it is not showing. There is only one trend there, which should be noted: How volatile it is and how Labour's vote isn't as solid as the Con one. Beyond that it only shows which voters are showing up as particularly tricky and have no confirmed loyalty.
I think the point is never simply read the headline figure. Its pretty meaningless. The devil is in the detail, I'm afraid.
The yougov polling is showing something different to what everyone is reporting it is because too many reporters haven't got a clue about stats or polling methodology.
Here's the single biggest key point, well put:
twitter.com/keiranpedley/status/964496258068107264
This also touches on the same thing.
As does this tweet.
Jane Green @ProfJaneGreen
Women consistently say 'don't know' more regularly than men; the most consistent gender gap. Sometimes the don't knows split in one direction more than others (#ge2017 to Labour) making this very significant.
The accuracy of the final poll before an election, isn't about those headline numbers that generate all the headlines for weeks in advance. Its about how they classify all those don't know and how accurately they assess likelihood to vote. YouGov nailed it at the election when others didn't.
This latest headline YouGov figure does not include those who are don't knows. This is (obviously) really important, and frankly is a statistical abomination.
20% of women currently saying 'don't know' is a huge number. You also have 9% of women currently saying they will not vote at all (which is the same as men so perhaps not as relevant)
They MIGHT be more likely to vote Labour. But I'd be extremely weary of the assumption that women will definitely behave the same way on the next election day compared with the last one. Why? Because of the groups which are most likely to be the don't knows and the stay at homes.
Because the don't knows are so big, it could hide all manner of trends.
The 'don't know' option is much higher in the 18 - 24 age group and the 25 - 50 age group too (men and women combined) compared to the 65+ group. Younger people are less confident of the way they will vote.
Also notable in this weeks poll is that the 25 - 50 age group (for both male and female) has a significantly higher number of people saying they won't vote at all, than any other group. (14% compared to 5% 65+, 5% 50 - 64 and 8% 18 - 24) At the last election the biggest age group who didn't vote were the 18 - 24s. This is a big and significantly difference between last week and this.
Not only that, I've looked at the past three YouGovs for this 12/13th Feb / 5th/6th Feb / 28th / 29th Jan and 7th / 8th Jan:
18 - 24 - 8 / 9 / 18 / 19
25 - 50 14 / 16 / 12 / 12
50 - 64 5 / 7 / 5 / 7
65+ 5 / 6 / 5 / 7
So young people are more likely to vote than they were a month ago, whilst the middle aged is the least likely to vote. Looking back further the 25 - 50 group seem to be the difficult one consistently. The young intention is up and down.
The long and short of it, is that the Labour vote is much softer than the Conservative one. Women and the 25 - 50 groups are the most undecided (and perhaps are the most volatile decisive groups particularly as both are more spread out and evenly distributed than other age groups / social class etc etc).
Women and the 25 - 50 groups are notable for four reasons: they are parents / they are the core of the work force / and they are the group settling down and looking to buy a house or have bought one recently / Remain supporters. Plus the 25 - 50 group isn't so concentrated in particular areas like the youth or older vote so it has more of an effect.
I would not like to second guess group either tbh because they are the ones torn in different directions most. The classic centrists! Most of the pollsters got it wrong for the last election, when they tried to call it.
Given that only 72 votes across a few key marginal seats, made the difference between a tory majority and a hung parliament, anyone drawing any conclusions about who currently might win an election and who is currently leading without counting 20% of women and 9% of men in the figures, is talking bullshit to say the least.
The poll doesn't say nothing, but it certainly isn't saying anything clear or straightforward either.
(Sorry for boring everyone to death. That's just my take on the YouGov figures).