Independence - how would you vote?

[Baxter Parp]

Your HNC isn't going to impress anyone here I'm afraid,

Neither is your BA in Hotel Management.

[Baxter Parp]

That's just the point though. A spectrum of established pollsters, both those affiliated to and those not affiliated to, Panelbase's commercial competitors, have raised their eyebrows at their decision to base 2014 polling on those certain to vote at Holyrood elections. What they're saying, politely, is that Panelbase "doesn't know what it's doing".

Could we sample this wide spectrum please?

[vikingTON]

That's just the point though. A spectrum of established pollsters, both those affiliated to and those not affiliated to, Panelbase's commercial competitors, have raised their eyebrows at their decision to base 2014 polling on those certain to vote at Holyrood elections. What they're saying, politely, is that Panelbase "doesn't know what it's doing".

What should the established basis for poll weighting be, and why?

To fall into a valid double negative, there is absolutely no basis by which to claim that Panelbase are 'doing it wrong', whereas other polling companies are 'doing it right'. There is no established precedent for this referendum. It reflects neither UK nor Scottish election results.

[Ad Lib]

I did this shit at college. Polling companies adapt their game depending on what they're working on, just as players adapt to different pitches. You're wrong, just accept it and move on.

I'm quaking in my boots at your Higher Modern-Studies-esque difficulty research methods experience.

The precise point being made here is that Panelbase have not adapted their game to deal with the referendum. It's a poll where there are only two groups, which aren't based on broad policy platforms but single-issue positions, the issue is of a kind that lends itself to higher turnout than elections generally, and where many who would not normally vote (as party politics apathists) will more than likely participate. It is therefore really really stupid to continue to poll based on certainty to vote in Holyrood elections. It doesn't take a Masters in statistics to work that one out.

At absolute best, Panelbase have turned up with a badminton racquet to Wimbledon.

[H_B]

I'm confused. 16% of respondents in their latest poll didn't vote in the 2011 election. That doesn't sound much like "excluding Scots who think that the Scottish Parliament is a complete and utter waste of time and is a glorified town council".

"As you can see, there isn’t actually any obvious trend in the polling, each company’s figures seem to be roughly steady. The main variation there is actually between Panelbase, who do regular polling for the Sunday Times, and the polling done by Ipsos MORI, TNS BMRB and other companies, with Panelbase tending to show a much tighter race than the others. One can only speculate what the reasons might be. Panelbase take the slightly strange decision of only including people certain to vote in a Scottish Parliament election in their samples for referendum voting intention, which could have an impact."

http://ukpollingreport.co.uk/blog/archives/category/scotland

[Baxter Parp]

I'm quaking in my boots at your Higher Modern-Studies-esque difficulty research methods experience.

The precise point being made here is that Panelbase have not adapted their game to deal with the referendum. It's a poll where there are only two groups, which aren't based on broad policy platforms but single-issue positions, the issue is of a kind that lends itself to higher turnout than elections generally, and where many who would not normally vote (as party politics apathists) will more than likely participate. It is therefore really really stupid to continue to poll based on certainty to vote in Holyrood elections. It doesn't take a Masters in statistics to work that one out.

At absolute best, Panelbase have turned up with a badminton racquet to Wimbledon.

And the evidence is? That's a lot of definitive statements you've made there, chum.

[Ad Lib]

Could we sample this wide spectrum please?

I don't believe the word "wide" was in my post, champ.

What should the established basis for poll weighting be, and why?

To fall into a valid double negative, there is absolutely no basis by which to claim that Panelbase are 'doing it wrong', whereas other polling companies are 'doing it right'. There is no established precedent for this referendum. It reflects neither UK nor Scottish election results.

It has to take into account an expressed certainty to vote in the referendum, rather than a completely unconnected aspect, certainty to vote in Holyrood elections. Which is why all the other pollsters have opted not simply to do this little platform shimmy.

[xbl]

Talk about squirming and climbing down. First I was a liar, now I am a liar or stupid. When asked for specific examples the best you can come up with is 'everything you say' is dishonest - ergo I can't come up with anything specific because my accusation was totally baseless. Post two things I've said side by side and show the thread how I am being dishonest. Put up or shut up.

I think you should stick to your ' team: St Mirren ' patter pal, you're hopelessly out your depth when you say anything further.

Actually, I said the same thing in a previous post. So maybe you are that stupid?

[Ad Lib]

And the evidence is? That's a lot of definitive statements you've made there, chum.

See H_B's post. Panelbase only take into account those certain to vote in a Holyrood election. That's pretty self evident because, you know, it's what they're doing.

[Baxter Parp]

"As you can see, there isn’t actually any obvious trend in the polling, each company’s figures seem to be roughly steady. The main variation there is actually between Panelbase, who do regular polling for the Sunday Times, and the polling done by Ipsos MORI, TNS BMRB and other companies, with Panelbase tending to show a much tighter race than the others. One can only speculate what the reasons might be. Panelbase take the slightly strange decision of only including people certain to vote in a Scottish Parliament election in their samples for referendum voting intention, which could have an impact."

http://ukpollingreport.co.uk/blog/archives/category/scotland

...says YouGov's Anthony Wells.

[SpungoMcGoo]

"As you can see, there isn’t actually any obvious trend in the polling, each company’s figures seem to be roughly steady. The main variation there is actually between Panelbase, who do regular polling for the Sunday Times, and the polling done by Ipsos MORI, TNS BMRB and other companies, with Panelbase tending to show a much tighter race than the others. One can only speculate what the reasons might be. Panelbase take the slightly strange decision of only including people certain to vote in a Scottish Parliament election in their samples for referendum voting intention, which could have an impact."

http://ukpollingreport.co.uk/blog/archives/category/scotland

Yet that claim seems to be flatly disproved by the poll's published demographics.

[Baxter Parp]

I don't believe the word "wide" was in my post, champ.

I believe that's irrelevant, can we sample it? Actually, it would be nice if you could provide any evidence to support that post at all, frankly.

[SpungoMcGoo]

I wonder if the Sunday Times know all the polls they pay Panelbase for are totally invalid? Someone should probably tell them.

[Ad Lib]

Yet that claim seems to be flatly disproved by the poll's published demographics.

No it doesn't. It's not based on people who voted in the 2011 election. It's based on people who say they are certain to vote in the 2016 election.

[Ad Lib]

I believe that's irrelevant, can we sample it?

Feel free. Ample quotes and persons referred to already on this thread, from Curtice to UKPollingReport.

[vikingTON]

It has to take into account an expressed certainty to vote in the referendum, rather than a completely unconnected aspect, certainty to vote in Holyrood elections. Which is why all the other pollsters have opted not simply to do this little platform shimmy.

No, that's not a valid answer. I asked for the basis of weighting poll ie how many respondents to be included from each camp, on a perceived basis of support among the electorate. This is normally the last election: a basis that led many polling organisations to grievously underestimate the SNP's lead prior to 2011. So... what are pollsters using to weigh their responses this time?

Put simply, there is little reason to trust the methodology of any pollster. But rather unsurprising to see our resident law trolls attempt to smear one poll as invalid while not questioning any others.

Nobody's impressed.

[Baxter Parp]

Feel free. Ample quotes and persons referred to already on this thread, from Curtice to UKPollingReport.

If you're not prepared to back up what you post, just say so, chum.

[Ad Lib]

No, that's not a valid answer. I asked for the basis of weighting poll ie how many respondents to be included from each camp, on a perceived basis of support among the electorate. This is normally the last election: a basis that led many polling organisations to grievously underestimate the SNP's lead prior to 2011. So... what are pollsters using to weigh their responses this time?

Put simply, there is little reason to trust the methodology of any pollster. But rather unsurprising to see our resident law trolls attempt to smear one poll as invalid while not questioning any others.

Nobody's impressed.

The weighting should take into account the expressed level of certainty by respondents that they will vote in the referendum. Just like Holyrood election polls should take into account the expressed level of certainty by respondents that they will vote in the Holyrood Election. And the UK election polls should do the same for likelihood to vote in the UK elections.

Not having a Masters in statistics, I'm not going to tell polling companies the precise weighting they should give to particular demographics, though each sampling method has its flaws and benefits. What I am competent enough to say, is that basing voting in a referendum on the certainty to vote in a particular type of domestic party-political election has demonstrable flaws that you don't need to be a stats geek to recognise from the outset. And the evidence of the polling results suggests a correlation between Panelbase's outlandish approach to this and the near total disconnect its results have from those of the other pollsters.

[H_B]

Put simply, there is little reason to trust the methodology of any pollster.

Of any single pollster in isolation, yes. So it makes sense to consider all of the data from all independent pollsters, not just cherry pick one you happen to like.

The overwhelming evidence from all available independent polling data is that Yes are so far beind they can't see No from where they are.

[Ad Lib]

If you're not prepared to back up what you post, just say so, chum.

I just did, HNDParp.