Friday, March 7, 2014

Let's play rounders!

I've just noticed something rather peculiar.  Of the seven BPC pollsters that have published referendum polls so far, there are five that provide actual numbers (as opposed to just percentages) in their datasets.  All five are currently showing a slightly lower No lead on their actual numbers than in their published figures.  There's nothing sinister about that - it's just an unlucky coincidence caused by the effect of rounding.  But it does mean that the Poll of Polls has been overstating the No lead by a smidgeon.

ICM -

Published figures :

Yes 37%
No 49%

NO LEAD - 12%

Actual figures before rounding :

Yes 37.4%
No 48.7%

NO LEAD - 11.3%

*  *  *

Panelbase -

Published figures :

Yes 37%
No 47%

NO LEAD - 10%

Actual figures before rounding :

Yes 37.0%
No 46.7%

NO LEAD - 9.7%

*  *  *

Survation -

Published figures :

Yes 38%
No 47%

NO LEAD - 9%

Actual figures before rounding :

Yes 37.7%
No 46.6%

NO LEAD - 8.9%

*  *  *

Ipsos-Mori -

Published figures :

Yes 32%
No 57%

NO LEAD - 25%

Actual figures before rounding :

Yes 32.1%
No 56.6%

NO LEAD - 24.5%

*  *  *

TNS-BMRB -

Published figures :

Yes 29%
No 42%

NO LEAD - 13%

Actual figures before rounding :

Yes 29.1%
No 41.8%

NO LEAD - 12.7%

*  *  *

Scot Goes Pop Poll of Polls - 

Average based on published figures :

Yes 34.6%
No 48.9%

NO LEAD - 14.3%

Average based on actual unrounded numbers (where available) :

Yes 34.6%
No 48.6%

NO LEAD - 14.0%

You might be wondering why I don't simply use the actual unrounded numbers for the Poll of Polls in the first place.  It's simply a question of practicality - the headline percentages are often published several days before the datasets turn up.  I had assumed that it wouldn't make any difference, because the rounding effects in different polls would always cancel each other out, but clearly that isn't necessarily always going to be the case.

*  *  *

You may already have seen this, but Women for Independence are running an Indiegogo fundraiser with a target of £20,140.  I know it's difficult to decide which of the many pro-independence causes are most worth helping, but the way I'm looking at it is that Yes have three key challenges that need to be met - 1) reducing the gender gap in referendum voting intentions (which is where the Women for Independence fundraiser comes in), 2) nurturing an alternative media to at least partly balance out the severe anti-independence bias of the traditional media (which is where the Wings over Scotland fundraiser comes in), and 3) convincing traditional Labour voters that a Yes vote is a vote in favour of their own values (which is where the Labour for Independence fundraiser comes in).

7 comments:

  1. Happy to bung £50 quid to the ladies.

    Go Lassies Go!

    ReplyDelete
  2. Why do these polls have no relevance to what folk are finding 'on the streets'?
    I've always been a Yes, and talk about the issues with anyone that is interested.
    Even folk that I had down as a certain No are now telling me they are voting Yes.
    Is it because i'm a manual worker, and work and socialise with folk that mostly don't have internet access, and think the news is something you wrap your mock chop supper in?

    ReplyDelete
  3. Juteman : Maybe a gender thing as well - is it predominantly men that you speak to? Even YouGov are currently showing that the race is a virtual dead heat among men.

    In terms of the internet, I'd guess that people who get their information from online sources are if anything more likely to break for Yes. It's people who are totally reliant on TV and the print press that we really have to worry about.

    ReplyDelete
  4. Maybe it's a gender thing, as I work in a men only environment. The 'discussions' can be quite aggressive, and folk know what is going on, internet or not.
    My female family members mostly work in the health side of things though, and everyone seems to be coming forward for a Yes.
    I'm seriously thinking that the polling companies are just an extra arm of the British State.

    ReplyDelete
  5. Hello James.

    Thought you might like an update on the Stormfront lite PioliticalBetting and it's racist posters. Now the incompetent moderator in chief is getting upset about me linking the Daniel Morgan case despite me being proved right yet again on it's seriousness and consequences.

    TSE was of course one of the main right-wing twats who thought phone hacking was a "non-story" which is why even reputable links on it have been banned on PB. There never was any legally problematic posts on it. He just didn't like being reminded how utterly wrong he and the herd were and still are.

    ReplyDelete
  6. I started an internet meme! Huzzah!!

    Does Old Grey Hair Smithson still have a large wager on Ally Carmickel being the next libdumb leader?

    ReplyDelete