General article(s) on Polls

[Clark]

Labour votes in 2019 under Jeremy Corbyn - 10,269,051
Labour votes in 2024 under Keir Starmer - 9,712,011

[observer]

Jul 6, 2024 11:55:42 GMT Clark said:

Labour votes in 2019 under Jeremy Corbyn - 10,269,051
Labour votes in 2024 under Keir Starmer - 9,712,011

Ah yes, the surge of support for Labour

[Clark]

Jul 6, 2024 12:03:42 GMT observer said:

Jul 6, 2024 11:55:42 GMT Clark said:

Labour votes in 2019 under Jeremy Corbyn - 10,269,051
Labour votes in 2024 under Keir Starmer - 9,712,011

Ah yes, the surge of support for Labour

I agree - the Tories lost this election more than Labour won it

[johnloony]

Jul 6, 2024 11:55:42 GMT Clark said:

Labour votes in 2019 under Jeremy Corbyn - 10,269,051
Labour votes in 2024 under Keir Starmer - 9,712,011

Just needs another 8,932 to make it an average of 10,000,000

[Davıd Boothroyd]

Which MRP was best?

[nodealbrexiteer]

www1.politicalbetting.com/index.php/archives/2024/07/07/how-the-pollsters-did/

Political betting article on how polls did

[nobodyimportant]

Jul 6, 2024 13:05:38 GMT Davıd Boothroyd said:

Which MRP was best?

A ComRes by any other name would produce just as inaccurate results.

[nodealbrexiteer]

www.britishpollingcouncil.org/the-performance-of-the-polls-in-the-2024-general-election/

British Polling Council statement on the polls.

[hullenedge]

Jul 11, 2024 13:33:45 GMT nodealbrexiteer said:

www.britishpollingcouncil.org/the-performance-of-the-polls-in-the-2024-general-election/

British Polling Council statement on the polls.

Thanks for posting. A disappointing fail for the polls especially when the May locals signalled that Labour was losing support in some strongholds.

[nodealbrexiteer]

Jul 11, 2024 14:01:43 GMT hullenedge said:

Jul 11, 2024 13:33:45 GMT nodealbrexiteer said:

www.britishpollingcouncil.org/the-performance-of-the-polls-in-the-2024-general-election/

British Polling Council statement on the polls.

Thanks for posting. A disappointing fail for the polls especially when the May locals signalled that Labour was losing support in some strongholds.

The performance of the exit poll might mask it a little bit and for joe public they might see if as polls 'forecast' Labour landslide and that's what we got

[observer]

I don't think the polls overstated Labour support and understated Con support. A poll is only a snapshot and Labour support fell fairly rapidly. Certainly in the last days and even hours

[msc]

It's a chicken and egg issue imo. The lengthy sizeable poll lead allowed many voters to vote Green, independent, whoever safe in the knowledge their vote wouldn't keep the Tories in. Had it been a tighter race there would have been less churn. The general election followed the model projected by the last three rounds of local elections, in that where there was a candidate best suited to give the local Tory a kicking, they were usually voted for.

[nodealbrexiteer]

The final polls are treated even if the pollsters don't like it as forecasts and it does seem like the all too common Lab overstatement, Con understatement was present this time though msc point above is fair in terms of the polls themselves causing shifts.

[The Bishop]

Jul 11, 2024 16:29:18 GMT nodealbrexiteer said:

The final polls are treated even if the pollsters don't like it as forecasts and it does seem like the all too common Lab overstatement, Con understatement was present this time though msc point above is fair in terms of the polls themselves causing shifts.

Apart from in 2005, 2010, 2017 and 2019 you mean? And this year (as in 1997 and 2001) the overstatement was to a significant degree *caused* by the polls, which leaves you just with 1992 and 2015 really - and I still think that in the former, some were influenced by the polls too.

[nodealbrexiteer]

Jul 12, 2024 10:52:19 GMT The Bishop said:

Jul 11, 2024 16:29:18 GMT nodealbrexiteer said:

The final polls are treated even if the pollsters don't like it as forecasts and it does seem like the all too common Lab overstatement, Con understatement was present this time though msc point above is fair in terms of the polls themselves causing shifts.

Apart from in 2005, 2010, 2017 and 2019 you mean? And this year (as in 1997 and 2001) the overstatement was to a significant degree *caused* by the polls, which leaves you just with 1992 and 2015 really - and I still think that in the former, some were influenced by the polls too.

2005 was slight overstatement, fair point re 2010 and 2017 but it does seem too post 1959 been an issue, it's not something the pollsters can just say well it was late swing, differential turnout etc

[The Bishop]

I don't really count MoE differences like 2005 (and 1987 - 1983 was like 2017 in that it *under*stated Labour if anything)

Seriously though, its a problem if the pollsters findings actually influence how people vote! Its just something we (and they) have to live with I suppose.

[nodealbrexiteer]

Jul 12, 2024 10:59:50 GMT The Bishop said:

I don't really count MoE differences like 2005 (and 1987 - 1983 was like 2017 in that it *under*stated Labour if anything)

Seriously though, its a problem if the pollsters findings actually influence how people vote! Its just something we (and they) have to live with I suppose.

liked for last sentence. We do need a future election where the tories suffer a 1970,1992,2015 in reverse!

[The Bishop]

1983 was maybe the closest - Tories consistently polled in the high 40s or even 50%, but ended up at about 43%.

But as with this month, they were so far ahead of the second party (who as this time, got "crossed over" in some polls) that it didn't really matter.

[nodealbrexiteer]

Jul 12, 2024 11:36:17 GMT The Bishop said:

1983 was maybe the closest - Tories consistently polled in the high 40s or even 50%, but ended up at about 43%.

But as with this month, they were so far ahead of the second party (who as this time, got "crossed over" in some polls) that it didn't really matter.

I think some pundits were rather harsh on the pollsters for their last polls in Feb 74-it wasnt their fault the electoral system didnt give the largest party in votes the most seats!

[manchesterman]

I think pollsters now have very advanced means by which they can make MRP-style projections based on vote share reported back to them and the 'science' is far better than it was 30-40 years ago. What they missed out on this time was the fact that - acc to research I saw a few days ago and was posted elsewhere - around 50% of those who voted LD and Green only decided to do so in the last day or so. This implies they were telling pollsters maybe a week before the GE that they were 'likely' to vote Labour, but once it became clear that a Labour landslide was certain, there was an acknowledgment amongst a lot of voters that Labour needed to have some sort of opposition and switched accordingly.

So either 'soft' Labour support was diverted to those parties, as a 2nd preference if you like, or natural LD/Green voters were telling pollsters earlier in the campaign that they were likely to vote Labour to 'get the Tories out' but once they realised mission was going to be accomplished, they felt able to vote for their 1st preference. This meant Greens hit their target seats and LDs exceeded their targets, but as a result, Labour vote share fell from that shown in polling data only days before.

One of the next challenges for pollsters is how they accurately assess the 'weight' of tactical voting which may occur (and often late in the day). How do you calculate whether it will be a trickle or a tidal wave, as it seems to have been this time?