Talk:Opinion polling for the 2014 Scottish independence referendum/Archive 1

Archive 1

New polls, 20 April 2014

Two new polls to be added: Yes 39 No 42 and Yes 38 No 46

http://www.heraldscotland.com/politics/referendum-news/latest-polls-yes-38-39-no-42-46-dont-know-14-19.1397985959 — Preceding unsigned comment added by 31.51.190.114 (talk) 12:16, 20 April 2014 (UTC)

• We need to wait for the tables to be published. Jmorrison230582 (talk) 12:23, 20 April 2014 (UTC)

TNS poll, April 2014

Latest TNS not included?

• Yes, it is (Yes 29%, No 41%, Don't know 30%). The reason it is not at the top of the table is because, although it was released more recently than some other recent polls, its fieldwork was completed by 2 April and it was not released until earlier this week. Jmorrison230582 (talk) 12:08, 18 April 2014 (UTC)

Would it be worth a placeholder? And a date as next to 2014? Keep up good work, this is my definitive source for polling info. Steve Sayers. — Preceding unsigned comment added by 80.42.6.74 (talk) 07:59, 21 April 2014 (UTC)

That's a good idea; polls that have been reported in the media, but have not yet had detailed tables published, could be added to the table with the text hidden (as above) until the tables become available. Jmorrison230582 (talk) 08:41, 21 April 2014 (UTC)

Financial Times poll

The FT has been running a tracking poll, but it's behind a paywall here if anyone has access. Tiller54 (talk) 12:56, 23 April 2014 (UTC)

It's just an average of the last several public polls to be released. It can be slightly misleading due to the significant differences between polling companies. ICM and Panelbase have been more favourable to yes, YouGov and MORI to no and TNS has a much higher "don't know" % due to the question they ask. Panelbase released 3 polls in a short period of time recently, which skews the FT poll of polls towards the Panelbase / ICM view of the race. Jmorrison230582 (talk) 13:02, 23 April 2014 (UTC)

Ah I see, thanks. The source I got it from made it sounded like they were conducting their own polling. Tiller54 (talk) 21:31, 25 April 2014 (UTC)

Progressive Scottish Opinion poll, May 2014

The latest OP in the table (YouGov/Sunday Times) gives the No campaign a 20% lead, but the link seems to bear no resemblance to this figure. Is the link wrong, or is data in the table on this page just made up? 86.31.187.167 (talk) 17:18, 11 May 2014 (UTC)

You are correct, the tables given relate to a different opinion poll. There is a Scottish independence poll out today showing a 34-54-12 split, but it was done by Progressive Partnership for the Sunday Mail newspaper [1]. I have hidden the result because I cannot find tables for this poll. Jmorrison230582 (talk) 18:03, 11 May 2014 (UTC)

• Update: Progressive Partnership (also called Progressive Scottish Opinion) are not British Polling Council members. I think we need to separate out their results from BPC members because it is unlikely that the raw data for their polls will be published. I have created a new section for these polls. Jmorrison230582 (talk) 09:18, 12 May 2014 (UTC)

Jmorrison230582

I agree with you there. Kezzer16 (talk) 21:48, 13 May 2014 (UTC)

I think they should definately be included, but perhaps highlighted. To hide them disturbs the time spans, they polled 1300 people and apparently did one last November. If a National a Newspaper commissions and publishes then I think this page should also, by all means colour the cells or mark them in some way, but don't miss any data, that's my opinion anyway 😃 Steve Sayers

The PSO polls have been included, but in a different section because they will not release detailed tables. I don't see any reasonable justification for "highlighting" their polls (as opposed to anyone else's). Jmorrison230582 (talk) 12:58, 19 May 2014 (UTC)

Highlight as a visual marker to indicate diff in poll status. Ie nit member of x etc. I would have thought an inclusive master table would give a more complete picture. Steve Sayers — Preceding unsigned comment added by 80.42.1.16 (talk) 22:39, 19 May 2014 (UTC)

If a poll for a national provided yes, no and don't know percentages is that not a reason fir main table inclusion? Stevenxlead (talk) 22:44, 19 May 2014 (UTC)

Not if we can't see what it was based on. Jmorrison230582 (talk) 05:25, 20 May 2014 (UTC)

new poll to add

New poll has Yes 43%, No 46% and if don't knows excluded, Yes 48%, No 52%.

http://www.panelbase.com/media/polls/F5852YesScotland.pdf — Preceding unsigned comment added by 109.153.18.136 (talk) 13:32, 15 June 2014 (UTC)

 Fixed Thanks for the link. It has been added now. Jmorrison230582 (talk) 15:58, 15 June 2014 (UTC)

No sample size

I propose removing all polls / results for which there is no sample size. This is particularly for the schools and colleges, but could apply anywhere. Without a sample size or number of people voting, the percentages are meaningless, so there is no reason to include the 'results'. EddieHugh (talk) 21:26, 18 June 2014 (UTC)

I would prefer removing all of the school / college / university surveys. They're pretty much all unscientific (even the ones with relatively large samples) and not necessarily representative of opinion in the institution concerned, let alone the wider population. I think it is giving undue coverage to a relatively fringe concern. Also, part of the point of regular public opinion polling is that you can see the movement (or lack of movement) over time, which can't be ascertained from one-off surveys or straw polls. Jmorrison230582 (talk) 22:17, 18 June 2014 (UTC)

I thought that I'd oppose your suggestion, on the grounds that we should report those mock referendums that all eligible to vote in the real thing could vote in, that a sample size is given for, and that only students could vote in, as they'd contain useful information. However, after looking at the sources for the first ten in the first table, none meet those criteria, so I suspect that very few if any will, so support your suggestion. Best to wait around a week for others to comment. EddieHugh (talk) 00:16, 19 June 2014 (UTC)

There was no further input so I have removed the tables of mock referendum and survey results. I have left in the references which were given from reliable sources (items that were not self-published, i.e. twitter or blogs) if it interests people that much. Jmorrison230582 (talk) 10:54, 27 June 2014 (UTC)

Argument between pollsters

Peter Kellner (YouGov) arguing that his results (showing a higher % for No) are correct: [2]

Survation have written a response justifying their results (showing a lower % for No): [3]

This looks like a notable dispute between the pollsters to me; the standard margin of error (+-3%) does not explain the discrepancy between YouGov (15-20% No lead) and Survation (6-9% No lead). Jmorrison230582 (talk) 21:36, 3 July 2014 (UTC)

Lord Ashcroft poll

What about this poll? Should it be included? [4] Laurenisme (talk) 19:27, 11 July 2014 (UTC)

Yes, I think so, under "Two option polling by other organisations" (along with the Progressive Scottish Opinion results). This is the full report written by Lord Ashcroft [5]. The only other caveat I would attach is that the referendum question was 26th (out of 26) in the sequence asked by Lord Ashcroft, which makes it less valuable as a resource on that specific question. Jmorrison230582 (talk) 20:45, 11 July 2014 (UTC)

Removal of visual guides to polling results for School, College and Universities

See section at link http://en.wikipedia.org/wiki/Opinion_polling_for_the_Scottish_independence_referendum,_2014#School.2C_college_and_university_surveys

The table with green/red backgrounds was removed and replaced with a few dozen links to various articles. Why should people have to link into this when others are visible ? Why is having this deemed such a bad thing ? It doesn't clutter the page. It makes information easier to extract in a visual way. I edited it back to how it was but Jmorrison230582 disagreed and put it back. Please explain. — Preceding unsigned comment added by Gavman99 (talk • contribs) 11:52, 14 July 2014 (UTC)

See the section above "no sample size". The mock polls are unscientific and were being given undue weight within the article. Jmorrison230582 (talk) 11:55, 14 July 2014 (UTC)

Removing the links would make sense. The sentence is just "Schools, colleges and universities across Scotland have conducted polls and mock referendums to gauge the opinion of pupils and students" – a single source for that broad comment would be fine, or no source at all. EddieHugh (talk) 12:09, 14 July 2014 (UTC)

British Election Study poll

Shall we include this poll by BES? [6] Laurenisme (talk) 18:50, 5 August 2014 (UTC)

I couldn't find a link to supporting tables. I've read on a blog site that the BES fieldwork was done by Yougov. Jmorrison230582 (talk) 11:05, 6 August 2014 (UTC)

Okay, I guess we can leave this one. Laurenisme (talk) 15:58, 6 August 2014 (UTC)

ICM/Guardian Poll from 5 August

Isn't the table a bit misguiding because of rounding? It seems to indicate that the lead decreased (from 6% to 5%), while in fact it increased (from 5.3% to 5.7%). Not statistically significant, but it certainly gives a different impression to a visitor just looking at the tables briefly. Filipvanlaenen (talk) 17:52, 8 August 2014 (UTC)

As you say, it's just a quirk of rounding. No need to get terribly stressed about it. As all polls are at least +-3% (this one is a bit higher because it is only a sample of 512), it would be a bit dodgy to go to one decimal place. Jmorrison230582 (talk) 18:05, 8 August 2014 (UTC)

Panelbase data for 2014.09.02 - 2014.09.06

Where's the actual data? What's the sample size? — Preceding unsigned comment added by 68.189.212.85 (talk) 01:21, 8 September 2014 (UTC)

Latest TNS

Aren't those two TNS currently at/near the top of the table the same poll? Akerbeltz (talk) 15:11, 9 September 2014 (UTC)

Yes. I'll sort that out now. Clyde1998 (talk) 15:19, 9 September 2014 (UTC)

I've noticed that TNS poll has Yes 38%, No 39%; with Yes 1% lead -- shouldn't it be other way around, i.e. No 1% lead. Am I missing something? 213.106.115.108 (talk) 15:53, 9 September 2014 (UTC)

According to the above conversation - about only showing voters who are very likely to vote - it should show Yes 41% (364); No 40% (361); DK 19% (174) (taken from page 3 of the TNS poll data), with a Yes lead of 1%. Clyde1998 (talk) 16:15, 9 September 2014 (UTC)

Thanks for that. It doesn't show those figures anymore, someone changed them. I'm unfamiliar with editing, can you may be correct them? 213.106.115.108 (talk) 17:00, 9 September 2014 (UTC)

Certain/Very likely to vote

To confirm, are we posting the opinion poll data as total number of people who took part in the poll or people who a certain or very likely to vote? It should really be those who are certain/very likely to vote that we list on the table. Clyde1998 (talk · contribs)

I think that's what has been done so far, and that's why I just corrected the numbers for the TNS BRMB poll. Maybe it should also be stated explicitly in the introduction? Filipvanlaenen (talk) 20:56, 8 September 2014 (UTC)

I'll put something about that now then. Clyde1998 (talk) 15:18, 9 September 2014 (UTC)

As far as I'm aware not all polling organisations use the "Certain/Very Likely to Vote" parameter, so I assume we are using the Yes, No and Undecided percentages after weighting, only. I.e. the headline figures in each report. For reference the previous TNS poll doesn't use the results from page 3, it used the headline figures on page one of the report. I'm sure Jmorrison230582 will be able to clear this up, soon hopefully as there is some edit warring going on and I don't want to have to request the page be locked. I'll leave it as it was originally (38 Yes, 39 No, 23 DN) until this is cleared up on here. Cheers, VanguardScot 20:29, 9 September 2014 (UTC)

He's on holiday. =D. Most of the pollsters do use the certain/very likely to vote parameter, only TNS don't - thus they're not comparable with the other opinion polls. The main point for using the parameter is to have consistency on the list. The only reason why the undecided voters are so high in these polls is because 25% won't (or are unlikely to) vote in the referendum. We (and the pollster for that matter) shouldn't include people in their headline figures who won't vote. Clyde1998 (talk) 20:41, 9 September 2014 (UTC)

The TNS polls have always been input using the headline figures on page one. That will continue unless you can gain consensus to change it. The conversation you had on here before you changed it was not a consensus. I doubt you will get a concensus on changing it as the headline figure is always the one the media use, and so it should also be the one Wikipedia uses. Cheers, VanguardScot 20:44, 9 September 2014 (UTC)

The media tend to report the numbers without don't knows, but some media sources are using a 41%-41% figure - Herald, Sky News, Mirror, Telegraph and ITV. Eitherway, how would you go about getting a consensus on this? Clyde1998 (talk) 20:53, 9 September 2014 (UTC)

New poll

There is a new poll: http://barker.co.uk/scotlandpoll

Weighted Results: “Should Scotland be an independent country(1,000 people)?

   53.9%  ‘Yes’
   46.1% ‘No’

Could somebody add it?145.52.141.54 (talk) 11:25, 10 September 2014 (UTC)

it has already succeeded145.52.141.54 (talk) 11:58, 10 September 2014 (UTC)

Before this turns into an edit war, let's take it to talk. So this poll does not go in the top section, clearly, as it's not a member of the polling council. But does it merit inclusion further down perhaps? I've not heard of this guy but that doesn't mean much as I'm not a polling expert! Akerbeltz (talk) 12:25, 10 September 2014 (UTC)

Right, looking at his LinkedIn, he's a web analytics consultant. Hm. Akerbeltz (talk) 12:28, 10 September 2014 (UTC)

"The poll was displayed to web users as they browsed media, mobile, arts & entertainment websites." The sampling method (among other things) means that it is of little value and should not be anywhere on the page. EddieHugh (talk) 12:31, 10 September 2014 (UTC)

Daily Record Poll

There is a new poll by the Daily Record, however I can't as yet find a link to the full poll results. The news article can be found here:

http://www.dailyrecord.co.uk/news/independence-referendum-exclusive-daily-record-4196976

The results are yes 42.4%, no 47.6%

Does anyone have the link to the full results?

Adam282 (talk) 17:33, 10 September 2014 (UTC)

ICM (Online) Poll

link to whatscotlandthinks.org

Yes 49%; No 42%; DK 9%. Should I put it on the list? Clyde1998 (talk) 18:58, 13 September 2014 (UTC)

Once the tables are out it's fine to add it. With the link to the tables as the reference. Cheers, VanguardScot 19:59, 13 September 2014 (UTC)

Panelbase out tonight too lots of polls this evening — Preceding unsigned comment added by 95.172.231.235 (talk) 23:06, 13 September 2014 (UTC)

I have added ICM/Telegraph as the online article contains the minimum data set, Panelbase isn't up yet and S Times is paywalled, so will have to wait.Saxmund (talk) 10:12, 14 September 2014 (UTC)

Rest of UK poll

The first line of this page states that "This page lists public opinion polls that have been conducted in relation to the Scottish independence referendum, 2014, to be held on 18 September 2014". As such, should we not also be listing polls about the attitudes of the rest of the UK towards Scottish independence?

One such poll has been released today by TNS:

http://www.tns-bmrb.co.uk/news/english-and-welsh-sentiment-against-scottish-independence-hardens-but-a-lar

Please advise.

Adam282 (talk) 17:00, 16 September 2014 (UTC)

See Section 5. Add what's necessary. EddieHugh (talk) 17:20, 16 September 2014 (UTC)

Latest Opinium/Telegraph Poll (16/09/2014)

The Undecided figure at 8% doesn't make sense. Surely it should be 10% or am I missing something? I changed it to 10% but it went back to 8%, maybe as a result of an automatic calculation when other data was changed? 86.29.194.205 (talk) 22:04, 16 September 2014 (UTC)

Unlike the other polls there is a 1% "Will not vote" so does not go into the table. The table on the PDF file shows Yes 43%/ No 47%/ Will not vote 1%/ Don’t know 8%. For some reason there is 1% missing, but that is perhaps because Opinium have rounded their sums in a way that it doesn’t come to 100% Maybe someone can look into it better and adjust the figures better, but until then we will leave the "Don’t know" at 8%. — Preceding unsigned comment added by 217.39.36.64 (talk) 22:41, 16 September 2014 (UTC)

OK, I'll put it back if not already so. Maybe a footnote or something on it? (I don't know how to do that) 86.29.194.205 (talk) 22:43, 16 September 2014 (UTC)

I'm one of the people who put the value back to 8%. Values need to be calculated based on the statistics, and rounded appropriately. I've been fixing other polls in the past few days, and noticed that there are several people that revert them, because they do not know what numbers to use for the calculations. Numbers do not always add up to 100%, because for this table we are rounding to the nearest percentage point, so if three numbers are for example 45.4% 48.3% and 6.3%, they will be represented in the table as 45%, 48% and 6% which add up to only 99%. I hope you understand and will be more careful in the future. There has been some discussion of closing this article to people without a username, because there needs to be some accountability. Wikipedia gives you great power because many people look up these statistics in here, but with great power you have to use great responsibility. Please make responsible changes, double check etc. If you do not understand the results go back to the poller's table and make an effort to understand.--Gciriani (talk) 23:55, 16 September 2014 (UTC)

Fuck off, you arrogant, 'know all' git. I suggest the article is closed to people who are obvious arseholes. That would rule you out immediately. I hope you understand. 86.29.194.205 (talk) 08:15, 17 September 2014 (UTC)

Calm Down; swearing is not necessary. His comment made me angry as well, but you and I done nothing wrong and came to the right conclusion, without anyone such as (talk) interfering and what seems to be shallow and empty threats. — Preceding unsigned comment added by 217.39.36.64 (talk) 14:13, 17 September 2014 (UTC)