Savanta/ComRes

[jamie]

Dec 15, 2022 17:18:17 GMT woollyliberal said:

Dec 15, 2022 17:10:57 GMT jamie said:

So the Labour lead suddenly sees a 10% decline, then next week sees a 5% rise? Yep, totally expected.

Go easy. Sample based statistics are not precise. 95% confidence interval means that 1 poll in 20 will be off by more than 3%. The last five Savanta polls are 45%, 48%, 42%, 47%, 46% for Labour. The average of these is 46%. The 42 is just outside the margin of error and you should expect 1 in 20 polls to do that. No big deal.

I’m not criticising the pollster (it’s good when they don’t massage their data to give the ‘expected’ result), more the reaction people give to polls that show a surprisingly large swing that is usually followed by a correction towards the mean. People should be a bit more cautious and not over-interpret single shock polls. 

[clyde1998]

Dec 15, 2022 17:30:49 GMT jamie said:

Dec 15, 2022 17:18:17 GMT woollyliberal said:

Go easy. Sample based statistics are not precise. 95% confidence interval means that 1 poll in 20 will be off by more than 3%. The last five Savanta polls are 45%, 48%, 42%, 47%, 46% for Labour. The average of these is 46%. The 42 is just outside the margin of error and you should expect 1 in 20 polls to do that. No big deal.

I’m not criticising the pollster (it’s good when they don’t massage their data to give the ‘expected’ result), more the reaction people give to polls that show a surprisingly large swing that is usually followed by a correction towards the mean. People should be a bit more cautious and not over-interpret single shock polls. 

Indeed - it's part of the reason there's a level of scepticism about polling: a lot of people only to hear about polls when there's 'surprising' findings; often those findings are as a result of statistical variation rather than any actual change in opinion.

[timmullen1]

[andrewp]

[The Bishop]

Cheers, just the latest YouGov to do now and we are up to date

[andrewp]

[andrewp]

[batman]

www.jewishnews.co.uk/voters-back-starmer-over-corbyn-ban-new-poll-shows/?fbclid=IwAR0rFOsJUVsTbHcA02pNFoX7Li4h-PGo1MM_cQMazPshlgzqw6aCmAONlfM - 55% believe that Starmer was right to prevent Jeremy Corbyn from standing as a Labour candidate, 23% think he was wrong.

[robert1]

The poll, conducted over the weekend (17-19 February), shows Labour on 45% of the vote and the Conservatives on 31%, the former seeing no change from the previous week, and the latter up three points.

(sorry can't get the graphics to copy!) Apologies

LD, Reform and Grn all down one. No other movement

[andrewp]

Here’s the latest Com Res mentioned by Robert

[cogload]

Hmm. Why you can't shout herding can you?

[LDCaerdydd]

This is only the second poll this year that puts the Conservatives above 30% (the other was Deltapoll on 8th Jan)

[batman]

not sure they are TBH.

[The Bishop]

Not only does this poll show the Tories doing much better than in all other current surveys, it also - uniquely - claims that Sunak is still ahead as "best PM".

Similar scepticism is warranted as with the latest People Polling's figure for Reform IMO.

[mattbewilson]

Outliers are an important part of polling trends

[The Bishop]

As long as they are actually recognised as outliers, but we all know that is far too often not the case.

[andrewp]

[Devil Wincarnate]

How on earth do Reform poll so well?

[andrewp]

Westminster voting intention:

LAB: 45% (+2)
CON: 30% (-2)
LDEM: 9% (-)
REF: 5% (-)
GRN: 3% (-)

via @savanta_UK, 10 - 12 Mar

[kevinf]

Mar 11, 2023 15:38:06 GMT Devil Wincarnate said:

How on earth do Reform poll so well?

It hasn’t translated into votes anywhere outside Derby. Yet.