Even Panelbase https://t.co/YOchFx7D1q
— Blair McDougall (@blairmcdougall) November 13, 2021
"Even Panelbase."
— James Kelly (@JamesKelly) November 13, 2021
McDougall is really demeaning himself with that tweet - it echoes the nastiest and most wretched of Brit Nat trolls who peddle conspiracy theories about Panelbase, which in fact produces independence results typical of most firms.https://t.co/ShcwmLBxwZ
In fact, if anything, Panelbase was (along with YouGov) on the No-friendly end of the spectrum until around 2018 or 2019. "Even Ipsos-Mori" might make a bit of sense, because Ipsos-Mori are perhaps the closest thing to a Yes-friendly pollster.
— James Kelly (@JamesKelly) November 13, 2021
But for whatever reason, the Brit Nat nutters which McDougall has shamefully associated himself with reserve their ire for Panelbase not Ipsos-Mori.
— James Kelly (@JamesKelly) November 13, 2021
Panelbase, *and* the old 2014 affirmation bias question... https://t.co/jchZDDaKsZ
— Alastair Cameron 🏴 (@HappyBritScot) November 13, 2021
Apologies, HappyBritScot, but I'm afraid your pathetic attempts to rig the indyref question are unlikely to gain any more traction with the Electoral Commission than they have thus far with polling companies (with one notable exception).
— James Kelly (@JamesKelly) November 13, 2021
Indeed. Yes is further behind with Panelbase than 7 years ago.
— Blair McDougall (@blairmcdougall) November 13, 2021
And here Blair is demonstrating himself to be either clueless or dishonest about polling methods. Pre-indyref Panelbase polls are not comparable with current Panelbase polls for one simple and obvious reason: after the indyref, weighting by recalled indyref vote was introduced.
— James Kelly (@JamesKelly) November 13, 2021
Why did it have to be introduced? To correct for the fact that the polls had systemically overestimated Yes. Doh!
— James Kelly (@JamesKelly) November 13, 2021
Scotland had its poll last year
— Wllliam Hayes PgDip (@WllliamHayes) November 13, 2021
This is perhaps my favourite ever tweet. In one sleepy part of Unionist Twitter, it will forever be 2015.https://t.co/QbK9tdF8fz
— James Kelly (@JamesKelly) November 13, 2021
If I'd known the villain of #Pounds4McDougallgate and his chums were going to get so over-excited, I would have made the obvious point last night - that the Scot Goes Pop / Panelbase poll actually has slightly older fieldwork than the Scotsman / Savanta ComRes poll, which showed a narrower No lead and no change whatsoever from the previous ComRes poll. It was also markedly better for the Yes side than ComRes polling back in the spring. So there are very good reasons for thinking that the Yes vote is holding up, and that the statistically insignificant 1% decrease with Panelbase that Blair is hyperventilating about is just meaningless margin of error noise.
Could someone clarify for me what the issue actually is? What would a poll of polls tell us about the independence / dependence split?
ReplyDeleteAny individual poll, depending on how the question is put, can skew things a little bit.
James, for your non-statistician user base, i.e. me, could you explain your concerns in non technical language?
Cause I'm not getting it.
Thanks.