Wednesday, December 1, 2021

This rips up everything we thought we knew: Ipsos-Mori TELEPHONE poll shows massive majority support for Scottish independence

Before today, there had been eleven polls since the Holyrood election showing a No lead (albeit for the most part a slim No lead), one showing a tie, and just one showing a slim Yes lead.  The big majorities for independence that we got so used to in 2020 had begun to look, at least for the time being, like a thing of the past.  And yet, all of a sudden, today brought word of a poll that wouldn't have looked at all out of place one year ago.

Should Scotland be an independent country? (Ipsos-Mori / STV, 22nd-29th November 2021)

Yes 55% (+5)
No 45% (-5)

Can one extremely good opinion poll negate the previous thirteen? (Or, to be more precise, eleven of the previous thirteen?)  In usual circumstances, the answer would be no - the suspicion would be that this is probably an outlier result caused by random sampling variation.  At the very least, we'd be cautiously waiting for one or two more polls before jumping to the conclusion that the new trend is real.  But with Ipsos-Mori, in general the only firm to conduct Scottish polls by telephone, it's a different story.  There's still no consensus on whether telephone polls are superior to online polls, or vice versa, and the only reason the vast majority of polls are conducted online is to sharply reduce costs.  If the only telephone poll you have shows a picture that is wildly out of line with the sea of online polls, there are very good reasons for seriously considering the possibility that the telephone poll is actually the accurate one - or closest to being the accurate one.

To be clear, there's no great surprise in Ipsos-Mori showing a Yes lead - in recent years they've been on the Yes-friendly end of the spectrum, so a slim No lead with Panelbase or YouGov might be expected to translate into a slim Yes lead with Ipsos-Mori.  But what is genuinely startling is that Ipsos-Mori are showing a completely different trend from the online firms.  The No leads in the online polls have generally represented little or no change since the latter stages of the Holyrood campaign, whereas today's poll suggests there has been a massive Yes resurgence since the election.  That seriously calls into question the story we have been telling ourselves, and more importantly the story the media have been telling us, about the state of public opinion over the last few months.

Having said all that, the slight health warning that needs to be added is that data collection by telephone is not the only thing that sets Ipsos-Mori apart from other polling companies - they also have a different attitude to weighting by recalled vote.  So in theory that could be an alternative explanation for the disparity.

As Marcia has pointed out in the comments section below, there is a clear Yes majority in today's poll even if Don't Knows are left in: Yes 52%, No 43%, Don't Know 4%.  It's to be hoped that the SNP leadership don't repeat the mistake of last year by treating this Yes lead as a precious piece of china that might break if anyone so much as sneezes, and instead use it to loudly demonstrate that there is a strong appetite for a choice on independence in the near future.


  1. The poll has 52% Yes No 43% before undecideds removed. Very healthy 9% lead.

  2. Recently, during the SNP's online conference, there has been a slight increase in the emphasis of determination to hold indyref2 by the Scottish government. Could this have influenced the increase in support for independence?

    It seems to me indyref2 has been in the news more than usual recently too.

  3. I don't see anything in the results that explicitly says they weight by recalled vote - do you know what they do James?

    I'm puzzled by the data they give, e.g. on pp 21 and 27, for Past vote (Indyref 2014).

    On p21 for All:-
    471 Yes, 464 No, and 152 "Not vote/too young/not Scot'"

    On p27, the final results for those 9/10 likely to vote:-
    406 Yes, 398 No, and 85 "Not vote/too young/not Scot'"

    So p27 shows a sizeable majority Yes for recalled vote. Although it's more equal Yes/No on p21, the figures still seem odd given the 45/55 result in 2014.

    So why do the recalled data for All on p21 seem to be pretty evenly split Yes/No? Does the explanation lie in the 152 listed as "Not vote/too young/not Scot'"?

    1. They don't weight by recalled vote (or they didn't the last time I checked, anyway). That's the point I was making in blogpost. Other firms do weight by recalled vote, for the most part.

    2. This is an important but oft overlooked point. Since we know that older voters are much more likely to vote 'No' than younger voters, weighting by recalled vote effectively only gives a measure of how many people have changed their mind, it does not give a true reflection of current voting intentions. As a not insignificant proportion of the older voters of 2014 are 'ahem' no longer around in order to vote, and a significant number of those too young to vote in 2014 are now able to, it can be argued that not weighting by past recall is the correct thing to do.

      Some years ago, I made a simple spreadsheet which calculated the expected Yes vote if nobody changed their minds but the young continued to vote Yes at the same %age level that they did in 2014. Demographic change alone led to a predicted Yes vote passing the 50% mark about now. In other words, recalled vote should be fairly evenly split, due to demographic changes over the last seven years.

    3. Thanks James. The other question I asked was (in a nutshell) -
      In the Ipsos-Mori sample, why do the recalled votes not align roughly with the actual vote in 2014?

      The result for all those in the sample was close to 50/50 (51/49 after weighting), which seems a bit too far away from 45/55.
      Or am I missing something?

    4. "As a not insignificant proportion of the older voters of 2014 are 'ahem' no longer around in order to vote, and a significant number of those too young to vote in 2014 are now able to"

      Do young people not get older, though?

    5. They do, but they don't change their minds. Or at least I don't think they do, on this subject at least. It is said that people's party voting patterns drift to the right as they get older and it was an open question as to whether people's support for independence drifts from Yes to No as they get older, but my personal feeling was that this would not happen. I think maybe the evidence is beginning to support my theory.

    6. Sorry Stavaiger for raining on your parade that thinks the demographics will definitely move towards yes. Your analysis does not take in to account young people leaving Scotland and old English no voters migrating to Scotland.

    7. Sorry for raining on YOUR parade, but actually it does. The only two assumptions in the prediction were as follows:
      1) That the youngsters entering the voting pool would support Yes at the same level as the youngsters of 2014, as there was no way to predict the actual level at that time (no crystal ball). Subsequent data could support or undermine this assumption, I simply haven't looked for it, but as far as I know, Yes support among the young has remained high.
      And 2) That people do not drift 'naturally' from Yes to No as they get older.

    8. Sorry again Stavaiger but it is not raining here.

      "The only two assumptions were...."

      and that is what I mean.

    9. When extrapolating to the future, assumptions have to be made, unless of course you can sell me a crystal ball.

      I see your point, but I doubt that cross-border migration will have had much effect over the past seven years.

    10. Stravaiger - apologises I just noticed I had been accidentally omitting the R in your name. No offence intended. I have posted a more detailed reply to you but it has not appeared. Not sure why.

    11. There's nothing from you in the moderation queue.

      As for misspelling names, you're in good company, because our former resident winter sports enthusiast always used to spell the name of the former Green MSP for Lothian as "Andy Whightman".

    12. Ok Jaimes 😇 thanks for that. I do lose a lot of posts which I just tend to think you don't care much for and don't publish - your blog your right to do so but a bit more of a problem if it is for technical reasons.

      Is it possible SNP HQ spent the missing £600k on new spoiling technology 😂😂?

  4. Great news, James. Thanks for letting us know.

  5. Any breakdown by national self identity (british or scottish or Welsh or Irish or Polish or Greek or English or Lithuanian etc)? Or by birthplace?

  6. Reply to George S. Gordon

    If Stravaiger's analysis is correct, you'd expect the recalled vote to split about 50:50 now.

    1. The demographic change seems a possible explanation, but I was hoping for James to answer too. Presumably the history of polls would confirm the trend suggested by Stravaiger?

  7. Nasty WGD numpty DrJim says "she's the only political leader in the Britland isles who has a plus rating ....."

    Sorry nasty Jim just look up at the table posted above on WGD and you will see Sturgeon is +20 and Sarwar is + 17. So Jim are you just a plain numpty or are you so desperate to big up Sturgeons rating you will deliberately lie.

    1. I could also have pointed out to numpty Dr Jim that Harvie +12 and Slater +10 have a plus rating as well. Numpties tend to have a fixed mindset that facts cannot penetrate.

    2. Now that original post by nasty numpty Dr Jim has been up since 7.24pm last night. Plenty of WGD numpties will have read it so will any of them have corrected nasty Dr Jim? You are joking of course. Numpties cannot think anything less than super positive about Sturgeon even when the fact is staring them in the face. Only one comment and that is by the mad liar Skier who says - ...she's still well into positive territory and there's not a politician in Europe that wouldn't kill for how she scores in public opinion."

      Ya numpty Skier she is only 3 points in front of Sarwar.

      As an aside Skier has given his Ski slope graph of voting yes polling and it shows if we wait until something like 2040 yes will be at 120% and surely even Skier would say it is save to have a referendum? Well maybe most people but Skier - nope too risky for him.

      In summary, the WGD site is just a mirror image of Britnat propaganda sites. Accuracy, integrity, objectivity nae chance on WGD.


    The WGD site which boldy proclaims don't post anything unless it supports Scottish independence and nae slagging off Nicola provides a platform for poster Union Man who is a self proclaimed ideological supporter of the Union. Union Man goes on to tell independence supporters which type of independence supporters should be cast in to the wilderness and ignored - code for Alba.

    Indyref2yespleasenicola says he agrees with Union Man.

    So there you have it WGD is a site that hates other independence supporters so much that it will give succour to people like UNION MAN. Of course it could be the mad liar Skier in disguise.