Is a 20,000 sample across 650 constituencies sufficient?

[MyNameIsArthur]

I was just watching Brexit : What The Nation Really Thinks on Channel 4 . A poll had been taken across the nation with a sample of 20,000 across all the constituencies. I did a quick calculation and that works out as an average of only 30 people per constituency. That doesn't seem like a very sufficient sample to me.

This explains the method: www.survation.com/wp-content/uploads/2018/11/Report-for-MRP-Estimates.pdf

If the sample is representative that's a huge sample! They usually only use a couple of thousand.

Thanks TalktoTrees

Those ‘graphs’ or whatever the hell they are, are as clear as mud.

I think in general that those charts show that if an area was strong 'leave' in 2016 it is a little less leave now, but areas that were not so strong have fewer wanting to leave now.
It certainly doesn't represent a 'landslide' remain.
It seems a bit 'unfair' to lump EU initiatives to reduce the destruction of the planet (attitudes to more efficient lightbulbs and emissions from discarded fridges and freezers for example) as being ''brexity' reasons.
Lower energy lightbulbs (which are a pain!) HELP you save electricity, and regulation on refrigerants helps the damage to the ozone layer, reducing global weather variations.

Anything over 5000 had a high confidence interval

Is it a similar method to the YouGov constituency poll in 2017? Because that was spookily accurate.

That poll was actually 1.5% accurate as an overall assessment of UK voting intentions compared to the actually result.

Which is what the 46/54 poll result was also based on.

People are allowed to change their minds, even if its probably too late but it does mean though is that the running sore that is the EU will continue on and on LOL!

General weighted polls only use around 1000. These are the type of polls you see everyday and give a fairly good result. 5000 gives a very good result so 20000 is massive and would have to be severely, severely out to be wrong. Incidentally, 20000 is the number they use in the GE exit polls and theyre always, always accurate.

Slightly off topic maybe, but I remember the first exit poll they ever did in a UK election. It was in 1970 and Labour was expected to win. It came out with a win for the Tories and was indeed accurate.

20000 is an enormous sample.

For the remain/leave question, because any rerun of the referendum will be national, it's likely that this poll is very accurate (assuming it's been conducted competently).

At the individual constituency level the results are going to be a bit less accurate, but a referendum doesn't work like a constituency election so I don't think that's an issue. And you'd be surprised how few participants are needed to get a good result if your sampling is done properly.

For me the interesting thing was the move away from leave in the west and north east. While wealthy Tory pensioners in the SE are still prepared to put empire ahead of prosperity, the "Brexit will hurt but it'll be all right in 50 years" message is a harder sell to people on low incomes.

I was thinking that 20,000 overall is a reasonable sample but if you are looking at the swing in each constituency, an average sample size of 30 in each constituency doesn't seem much. Im no statistician so I dont know. Its just what i thought when i watched the programme.

Was that a Gallup poll Peregrina?

Gallup poll - now that I don't remember - it was 1970. It was the very first exit poll, taken in Gravesend, chosen because it represented a typical town, and I recall being stunned by the result. I remember because it was the first election I was able to vote in. The New Statesman went to press before the result was out, and got its front cover completely wrong.

if you are looking at the swing in each constituency, an average sample size of 30 in each constituency doesn't seem much

You'd think not, but for the 1992 election I did a street poll in Accrington for MORI. That was only 10 people (of specific age/sex/social class) and by the time I finished it I had a horrible feeling Labour were going to lose after all.

I am in the YouGov polling set.
I do survey's most days.
They know a lot about me and thus if I get picked for one of those surveys I know that out of their 300,000 regulars I am 'representative' for one of the criteria

I am too, but I haven't had one for weeks. I'm 67, Labour/Remain voter in Tory/Leave constituency. Obviously I don't count

Giant kitten, I like you am an older voter and voted remain. I have heard Caroline Lucas and Vince cable talk about the intergenerational betrayal in the brexit referendum and I feel this encourages ageism. I know a lot of older people who voted remain and younger people who voted leave.

My DS goes on and on about how my generation has shafted his, not just over the Referendum. It's not us personally, so I wish he would save his anger for something more productive.

As one who is a feminist, starting from the early seventies, I feel it's time he started fighting a few battles for what he believes in, as we had to.

Giantkitten
Have you opted in to get every available survey?
I get political ones about once a week but the more I do, the more I get

Peregrina
I find the apathy of my children very depressing. They feel utterly powerless so do nothing.

@Ta1kinpeece as far as I know I have. How do I check? Until recently I was getting at least one a week, often 2 or 3, but they’ve dried up completely. The last one was 17th Oct (I always screenshot the points total page so I know where I’m up up)

Go into your account and check your preferences. Are you getting the "ratings" thing that comes up most days?
Opt in to it if not.

@Ta1kinpeece

So am I. I've got bored now I am no closer to £5 than several months ago. Takes aaaages to do.

Can't see preferences on my account, Ta1kinpeece

I was getting ratings, following on from ordinary surveys, but the surveys I was getting regularly were mostly about TV programmes (& as I virtually never watch TV that was bloody boring too!) rather than eg politics

Not sure.
It may just be that there are several other people in your postcode block who are the same demographic so you are sharing out the surveys between you.