Latest Polls and Latest Odds

[ayrmad]

Your having a laugh using a word like gullible when you've been like a big cod on many issues during this campaign.

You spent days trying to convince us all that we wouldn't get into the EU, ffs.

Your claim that I spent last week trying to convince people an I Scotland wouldn't get into the EU.

Utter lies and a claim you have now been forced to admit is indeed utter bollocks.

As I said if you are going to make stuff up at least try to make up something semi credible

Another week off to a bundle of fail for you.

The village idiot never seems to do too bad against the might of Glasgow Uni, #prayforourunis

[H_B]

This is really dreadful stuff.

A bullshit claim you've had to admit you completely fabricated and then this litany of fail.

I'd take a deep breath and step away from the keyboard if I were you.

You've had a complete disaster

[ayrmad]

^^^taking defeat gracefully

[Mr Bairn]

As long as the weighting is correct?

Yes. The weighting is used to gain a more accurate cross-section of society. To an extent they can achieve this by screeing out people pre-survey, but this is not a foolproof method.

[Jamaldo]

^^^taking defeat gracefully

Would you and everyone else like to join me in putting this utter w**k on ignore?

[ayrmad]

Yes. The weighting is used to gain a more accurate cross-section of society. To an extent they can achieve this by screeing out people pre-survey, but this is not a foolproof method.

Who's weighting is gaining a more accurate cross-section of society?

[ayrmad]

Would you and everyone else like to join me in putting this utter w**k on ignore?

I know he's a pain in the tits but this place is actually too quiet when you ignore the 2 tedious pedants and their apprentice.

[Jamaldo]

I'm surprised no unionists have commented on the Corby mock referendum and how utterly desperate it makes Blair MacDougall look.

Amazing that a side that is so far ahead has to champion a mock vote that didn't even take place in Scotland and which has absolutely no relevance.

[Mr Bairn]

Who's weighting is gaining a more accurate cross-section of society?

I haven't read enough on their respective methodologies, but the most accurate should have 0.94 males to every female, 96 white people to every 3 Asians and 1 black person, just over 10% of the sample should live in Glasgow, about the same number should be gay people, 7% of the sample should be unemployed etc. You catch my drift.

[Ira Gaines]

I see Confidemus' ignoring of HB went well.

[ayrmad]

I haven't read enough on their respective methodologies, but the most accurate should have 0.94 males to every female, 96 white people to every 3 Asians and 1 black person, just over 10% of the sample should live in Glasgow, about the same number should be gay people, 7% of the sample should be unemployed etc. You catch my drift.

It won't matter how much you've read up on their respective methodologies, they don't have any previous referendums to test it against.

[Mr Bairn]

I see Confidemus' ignoring of HB went well.

Know what's pretty tragic? Confi is always accusing H_B of dedicating his life to this place. Well H_B averages 14 posts per day, Confi averages 9. Hardly a huge differential. At his current rate of posting confi will have 32000 posts by the time he's been here a decade.

[Mr Bairn]

It won't matter how much you've read up on their respective methodologies, they don't have any previous referendums to test it against.

They do have mathematics to test it against though.

[Jamaldo]

I haven't read enough on their respective methodologies, but the most accurate should have 0.94 males to every female, 96 white people to every 3 Asians and 1 black person, just over 10% of the sample should live in Glasgow, about the same number should be gay people, 7% of the sample should be unemployed etc. You catch my drift.

When did simply asking the referendum question become so out of fashion?

You'd think they were deliberately drowning the polling procedures in bullshit.

[Mr Bairn]

When did simply asking the referendum question become so out of fashion?

You'd think they were deliberately drowning the polling procedures in bullshit.

Because if you just picked a random sample and asked them the question, then the sample wouldn't be representative. It would be like picking 1000 Dundonians and expecting to achieve a representative result.

[ayrmad]

They do have mathematics to test it against though.

If they had maths to test it against you'd expect them all to come to the same conclusion regarding methodology.

FTR, I do know what a confidence interval is, I was just wanting you to use the word outlier, and you're correct if the methodology on weighting is correct, a bigger % could be outliers if they're wrong.

[renton]

I haven't read enough on their respective methodologies, but the most accurate should have 0.94 males to every female, 96 white people to every 3 Asians and 1 black person, just over 10% of the sample should live in Glasgow, about the same number should be gay people, 7% of the sample should be unemployed etc. You catch my drift.

ICM, Panelbase, Survation and TNS all weight by recalled 2011 vote (on top of weighting by age group, male and female) Yougov weight by 2011, but don't ask for a recalled vote but rather by inforation YG held over from immediately after that election. Lately they've also gone to the effort of splitting down that SNP vote into SNP 2010 SNP 2011 and Labour 2010 SNP 2011. Ispos-Mori wieght by some other demographic profile that I'm uncertain of.

Remember also, that how the pollsters collect the data must affect their performance. Online pollsters may well end up asking the same people over and over (or at least there is a risk of that), face to face (or even phonebank) polling is less anonymous and may lead to false answers from those being polled being uncomfortable in airing their views. Landline phone banking may lead to oversampling one demographic to the expense of others, meaning that you'd radically upweight the other demographics, introducing error from small sample sizes.

[Mr Bairn]

ICM, Panelbase, Survation and TNS all weight by recalled 2011 vote (on top of weighting by age group, male and female) Yougov weight by 2011, but don't ask for a recalled vote but rather by inforation YG held over from immediately after that election. Lately they've also gone to the effort of splitting down that SNP vote into SNP 2010 SNP 2011 and Labour 2010 SNP 2011. Ispos-Mori wieght by some other demographic profile that I'm uncertain of.

Remember also, that how the pollsters collect the data must affect their performance. Online pollsters may well end up asking the same people over and over (or at least there is a risk of that), face to face (or even phonebank) polling is less anonymous and may lead to false answers from those being polled being uncomfortable in airing their views. Landline phone banking may lead to oversampling one demographic to the expense of others, meaning that you'd radically upweight the other demographics, introducing error from small sample sizes.

I think the "red-nats" is quite a good idea. It's clear that many many people voted SNP for the first time in 2011, and this is just a theory of mine but I think that group might be one of the most split down the middle.

I know that most of my family are in that group, and are no voters, but many first time SNP voters would have implicitly became yes supporters at that point. I feel like the "red-nats" are a very important swing group in this referendum.

Now I'm not saying that yougov polling is necessarily right, but making concessions for "red-nats" is a sound principle imo.

[renton]

I think the "red-nats" is quite a good idea. It's clear that many many people voted SNP for the first time in 2011, and this is just a theory of mine but I think that group might be one of the most split down the middle.

I know that most of my family are in that group, and are no voters, but many first time SNP voters would have implicitly became yes supporters at that point. I feel like the "red-nats" are a very important swing group in this referendum.

Now I'm not saying that yougov polling is necessarily right, but making concessions for "red-nats" is a sound principle imo.

I'm not so sure, It involves breaking it down into smaller subsamples and weighting them to meet a chosen composition. It can't help but introduce a greater uncertainty in the final numbers. The other pollsters all show around 15-17% SNP voters wanting to vote no, so it's not like the others are sampling nothing but rabid nats, so why was the 'correction' necessary - what made YG feel that it was a necessary correction to make? Certainly Kellner laid out his theory on the YG blog and equally Survation responded robustly to that.

As you say, they may be right, they may be wrong - but I'm struggling with the logic of introducing a greater uncertainty into your numbers.

[Confidemus]

I see Confidemus' ignoring of HB went well.

I'm in your top 5 worst posters, I don't know why you're bothered..