I think that may just be a statistical blip that can disappear with more polls. For example, the line from Techne would be straight up and now in thirties, I’m predicting next Saturday’s Opinium to be 31+. We haven’t had a Kantor for a while.
Your worm that turned can be gone next time you look. A case of Ephemeral Worms.
Have we returned to analysing hypothetical polls published at some vaguely defined point in the future?
Well. To be bluff. Yes.
You see different polls whose methodologies favour different results report at different times, maybe not even spaced but in clusters. People polling for example had a reform at 8 and others at six from their method, therefore a con of 21 last week, even though it sits away from the polling average or herding they are not remotely outliers for the firm, it’s what people polling been doing since they appeared, so it’s not an outlier because it’s in sequence to their method. Similarly Kantor, Techne and Opinium tend to produce higher Con share and they feed into this graph too - so where you have a statistical blip at the moment, next Techne 30 or 29, Opinium 31+ and a Kantor in 30s will make the blip vanish forever - the more recent ticks on the graph can be ephemeral.
We've had four polls in a row where that company's swing towards the Tories has slowed virtually to a halt, or started reversing. That might be a short term trend, but it's still a real trend, rather than a reporting quirk.
I understand your point about Opinium giving a higher Tory figure, but even their last two polls showed movement back towards Labour.
It’s a mixed picture right now, still to early to say the new leader bounce has peaked and gone into reverse - and rather shockingly didn’t really get anywhere in the first place.
It is beginning to look like that though. 😕
Definitely too early - but I don't see where it's mixed. If we had a selection of polls showing the lead coming solidly down, then that would be mixed, but they pretty much all seem to be slowing/reversing.
It's also a reasonable bounce - not many leaders start from 30%+ down, so it was always going to take time. He was always up against it, and I can't imagine people will be on a budget that raises tax and cuts spending.
I suspect they're looking to hang on till hopefully Ukraine win, and we see fuel prices and inflation fall, and they will claim it as a win for "grown-up politics" or some other nonsense.
It’s mixed because I’m not looking at gap between parties - if Starmgasm belches back to libdems 6% it stole the lead drops even if Tories rise not one iota. I am looking at the Tory share for their recovery, and the last three Techne each show an increase as one example of how it’s mixed, and the down tick does not tell us it’s mixed because of things like people polling con 21 and low yougov that didn’t move built in.
Isn't the Tory share doing exactly the same?
It was going up with when Rishi took over with just about every polling company.
Now it's going down, static, or the rise has slowed almost to a halt with every poll from the last week and that's from half a dozen different pollsters. People polling was down to 14pts before Rishi, so 14-20 was a jump, while 20-21-21 is a stalling. Techne has shown a 4pt jump, then a 3pt jump, and now a 1pt jump. That's technically three increases in a row, but it's also a clear slow down that every other polling company has reflected recently.
But a slow down is not a stop. Speed is not of the essence here, just direction of travel.
It’s still a mixed picture on polling, a good 7 days for Tory share and todays little kink in graph vanishes. you seem to be edging toward admitting it.
Have I missed something? I thought you were suggesting that the Rishi bounce was continuing, when we've had six polls in the last week or so, which all show it stalling.
We've gone from a steady increase, reflected across all the polls, to a week where the Tory vote change for each polling company is -2, -2, -1, +1, 0, 0. That's not mixed, that's pretty consistently showing that, for now, the Tory share is no longer going up.
It may go up again if they have a good week, but that doesn't make this week disappear.
I understand what you are trying to say, but this is where you are wrong.
The way you apply your sequence to your paragraph implies they are all apples of the same variety, but each one is of a different variety, any moment we can have a Kantor variety, Opinium variety. YouGov tend to be random could show +4 next regardless of anything else.
I think I am trying to take a longer bigger picture than you are. I think I am trying to be careful not to draw conclusions a honeymoon is over too soon as next set of polls could make mockery of that - yes it can make the kink in todays graph disappear into a smooth upward one. Over generous to the Tories my approach maybe right now, but only on basis of trying to be right and fair.
Opinium were included in that list. YouGov went from 19% before Rishi, to 23 then 24. They're the only polling company, apart from the 6 I mentioned, that I can see that have polled regularly through Truss/Rishi, who haven't polled in the last week or so, but I guess you could stick them in the same box of apples as they had a 4% honeymoon bounce, followed by a 1% change. There aren't lots of other polling companies that have polled throughout that we can identify any short term trends from.
If six polling companies all pretty much follow the same pattern, then it's a real thing, not a statistical quirk. It's also not going to make that bend into a smooth curve no matter what happens.
You can argue that Rishi might do something to restart his recovery - but it will still have stalled in the last week.
So, what are you saying? The honeymoon is over?
I was replying to your post that said, "I think that may just be a statistical blip that can disappear with more polls".
It's not a statistical blip, and it won't "disappear into a smooth upward one".
Whether the honeymoon is over is a different question
The way the graph works, yes it can change shape at the live end.
With those polls I'm pretty sure it can't move to a smooth upward curve. That levelling off bump - let's call it Gavin Williamson peak - will still be there, even if Rishi's mountain is ultimately a lot higher.
Gavins Peak? 😆
One year in, and I’m embroiled in a proper PB argument. Because I’m still convinced three thirty fives tomorrow, and drawing straightest line through the last 8 goes straight up.
I've just tried it with a loess macro in Excel and I can't get rid of Gavin's peak without adding some crazy figures.
Even Kwasi couldn't help, and believe me he was keen to see any sign of Gavin disappear.
If Gavin’s Peak is here to stay, it means it’s not really that mixed a picture on polling this week either. History is in the week leading up to the budget, Tory share already took fall back regardless where it then went after. And this has been a history defining conversation.
Just thought I’d pop in and find out if WW3 has broken out since PB is more likely to tell me quickly and reliably than any MSM. It seems not so on that happy thought night all.
Comments