Exit Poll Help/Explanation Please?

[TakemetoPluto]

Now I am not an expert regarding politics, this election or any prior, or brexit. I have my own views, and I'm not placing this thread to state who's right or wrong, for whoever they have voted for. Yet I'm confused and would like some help if someone can explain it in easy terms.

My local newspaper has printed this in regards to the exit poll

The exit poll was conducted by Ipsos Mori at 144 polling stations, with 22,790 interviews.

So my question is if anyone can help is, is the exit poll based on what 22,790 people have voted for, over 144 polling stations?

I'm confused if only 22,790 people have been 'interviewed' and obviously the number of people who will have voted today, is considerably higher than 22,970, how can such assumptions be made, until the ballot is counted.

Like I said I'm just confused, and if someone could kindly explain to me, in easy terms, I would appreciate it.

Whatever is decided obviously is what is going to occur, I just wanted an explanation off people who understand better than I do!

BBC explanation is really helpful
BBC News - Exit polls: How accurate are they?
www.bbc.co.uk/news/election-2019-50716626

I also forgot to say the local newspaper's statement could be incorrect or false, but just wanted clarification as I'm confused 🤔

They key thing with the poll is that it tends to use the same constituencies each time so it can track trends or swings, which is what tends to make it so accurate.

Here's an explanation from Times journalist Esther Webber- I found it useful

twitter.com/estwebber/status/1205102714612518912?s=19

it is only ever a prediction, but it is based on experience, and they chose a variety of constituencies across the country.

They have been known to be very wrong, but they have got better over the years.

The BBC explanation is good. It boils down to long term tracking and some seriously detailed analysis tools.

They are historically pretty accurate, a bit out on % but right in results.

Thanks @Seeline I read that before posting, but no figures were included regarding how many people had been 'interviewed'.

I understand the press releasing the exit poll and what it shows, I was just asking if they had released the exit poll based on 22,970 people's votes, and if so how can they base the outcome they have on such a small number of people, compared to the actual amount of people who have voted.

What I'm trying to say makes sense in my head but probably isn't coming across that way written.

It's extrapolation and some clever statistical manipulation. Heavy mathematics with a decent track record.

It’s all to do with statistics and margins of error.

We may as well come clean, @TakemetoPluto it's witchcraft

See this is why it's making no sense to me! 😂 Witchcraft on the other 🖐️ 🧙‍♀️

On a serious note though, if the statistical figures of people who were registered to vote in the UK was around 44 million prior to the deadline and then an extra 3 million people registered to vote before the deadline.... How is 22,970 people's votes in this election, providing the exit polls results, when the amount of people voting previously is considerably higher.

I know I sound thick (I will happily admit I'm not greatly academic) but 48 million registered people, say only 22 million actually voted, how can 22,970 people's votes, show the outcome? I'm really confused

Because the constituencies picked to be sampled are very carefully chosen. Those are not random choices, just the voters selected are random.

You need to just accept that superior beings, known as Political Analysts or Statistical Genii have grown a machine that takes words from very few people and turns them into a fairly accurate prediction of the voting public, no matter what the yearly variations.

Now, just hope that the next poster is not an acolyte of one of those Jedi Math Masters, because your brain, my brain, most posters brains will ache trying to follow the logic, new to us terminology, aka speaking in tongues.

I dabbled in social statistics at Uni, I know how powerful that magic is

@seeline thank you, I have now just googled they are 650 constituencies within the UK and as much as I get they are carefully picked to base the polls on I am now even more confused. I did say I wasn't very bright!

@CuriousaboutSamphire I do take your word, as you are much brighter than I am, and went to Uni 😂

I have no shame in admitting given they are 650 constituencies, only 144 of them chosen, and only 22,970 I still cannot thathom how a machine can base anything on those figures.

Thank you for making me laugh! 🍷

Yes I am not bright, and the machine isn't usually wrong but those figures, places, number of people, its made me more confused than when I started the thread! I don't understand how a landslide can be predicted, when it's not half of the constituencies being chosen, and based on the not confirmed 22 million people who I said may have voted, only 22,970 people is around 1%?

Am I really thick or am I making a small amount of sense? 😂

Don't trust me too much. I was never inducted into the inner circles of Statsticking.

I remained very much a novice in the arcane skill of number manipulation. As far as I can tell, it really is sorcery!

Apparently two thirds of the total number of constituencies don't change their vote, so it tends to be the marginals that are targeted.

Although when they held the referendum, the result went "leave" unexpectedly at the end of the night. So, nothing's written in stone.

You're thinking in straight lines, like 1 person asked = however many to make them proportionally important.

But that isn't how it works. Each person , each group of people, and many other variables mean different things in different parts of the analysis.

They have decades of variables plugged into the machine. It gets more accurate each year, the more unusual the year the bigger the effect on the analysis.

Statsticking made me laugh, a little too much!

It's just all numbers and it's not making any sense to me! See what the morning brings!

@singlenotsingle yes you are right, it was very much expected one way, at the end of voting, and remember the outcry the following morning.

So say you have 100 people. 40 of them always vote Conservative, 40 of them always vote Labour. The last 5 elections have depended on what the remaining 20 people vote. Who do you ask for their opinion?
Now apply some serious statistics and maybe you reduce that to 5 of them.
So why would you ask anyone but those 5? That's what the exit polls do on a constituency level.

My polling station was an exit poll station and was last time too. Overheard them telling the person in front of me it was every 14th voter I think.
I then read the BBC explanation as I couldn't get how it could possibly accurately predict, and still didn't properly understand it . Complex maths and probability clearly isn't my strong point

The problem with predicting the referendum was that there were no previous results to guide the exit pollsters.

Yes- few constituencies change colour in general elections, so the exit poll only has to look at those most likely to change each time. A random selection of voters there will give an indication as to which colour each individual constituency will turn. Those can then be added/subtracted to the seats always retained by each party to predict the outcome.