United Kingdom Referendum on European Union Membership

[Serenity Now]

Quite a lot of Corbyn nominations in all regions, I note.

Not to mention where the majority of Kendall nominations are.

[Serenity Now]

On the subject of the potential geography of the result of the referendum, the other day I made the map below.  It shows a very crude regression model of the following variables:

• UKIP 2015 General election vote share (dependent variable)
• Conservative 2015 GE vote share (independent variable)
• Turnout at the 2015 GE (independent variable)
• 'White British' in 2011 Census (independent variable)
• Degree or equivalent in 2011 Census (independent variable)
• Aged 60 and over in 2011 Census (independent variable)

Using this, I divided the Parliamentary Constituencies into quartiles in order to show a hypothetical 'tie' situation:

Dark green = 25% most likely to vote to remain.
Green = the 25% which are likely than the median constituency to vote to remain.
Red = 25% which are less likely than the median constituency to vote to leave.
Dark red = the 25% most likely to vote to leave.

I ran this very arbitrary regression model for a laugh, just to see what the outcome would look like.  It may prove primarily useful for comedy value but I thought it might provoke some interesting discussion nonetheless.

Please feel free to criticize the hell out of this model (I can see a few howlers) as I may be able to change it into something more credible.  There a things said in previous posts pointing out potential problems (Sibboleth's comment about the geography of the UKIP vote, for one thing).

For extra 'fun', I can easily recreate this model to assume different national outcomes upon request, just give me the desired yes/no percentages.



Link to bigger version in gallery:
https://uselectionatlas.org/FORUM/index.php?action=gallery;sa=view&id=14405

[Serenity Now]

@Bro-mentum

Hey, just curious: Why do you think that Glasgow NorthEast/East & Banff and Buchan will vote to leave?

I don't think this, the regression model detailed above did.  If any areas in Scotland did vote to leave, I doubt it would be these ones.  Also, please note this is how areas would vote compared to average in a 'tie' type situation.  If the outcome is a 'remain' vote nationally I suspect it would be by a wider margin and therefore said constituencies would probably vote remain.

[Serenity Now]

On the subject of the potential geography of the result of the referendum, the other day I made the map below.  It shows a very crude regression model of the following variables:

• UKIP 2015 General election vote share (dependent variable)
• Conservative 2015 GE vote share (independent variable)
• Turnout at the 2015 GE (independent variable)
• 'White British' in 2011 Census (independent variable)
• Degree or equivalent in 2011 Census (independent variable)
• Aged 60 and over in 2011 Census (independent variable)

Using this, I divided the Parliamentary Constituencies into quartiles in order to show a hypothetical 'tie' situation:

Dark green = 25% most likely to vote to remain.
Green = the 25% which are likely than the median constituency to vote to remain.
Red = 25% which are less likely than the median constituency to vote to leave.
Dark red = the 25% most likely to vote to leave.

I ran this very arbitrary regression model for a laugh, just to see what the outcome would look like.  It may prove primarily useful for comedy value but I thought it might provoke some interesting discussion nonetheless.

Please feel free to criticize the hell out of this model (I can see a few howlers) as I may be able to change it into something more credible.  There a things said in previous posts pointing out potential problems (Sibboleth's comment about the geography of the UKIP vote, for one thing).

For extra 'fun', I can easily recreate this model to assume different national outcomes upon request, just give me the desired yes/no percentages.



Link to bigger version in gallery:
https://uselectionatlas.org/FORUM/index.php?action=gallery;sa=view&id=14405

How did you make this (as in what program did you use)?

I used Excel 2010 for the regression model (using the 'Data Analysis' add-on) and ArcMap to map the results.

[Serenity Now]

For the record, results won't be reported by constituency, but by local authority.

Yes, I should have stated this in my post, given I'd assumed that would be the case (as it was with the Alternative Vote referendum).  However, I didn't have the data I wanted to use by local authority.  I'll probably have another go at a model using variables which are available by local authority (eg Census data).

[Serenity Now]

Here's a more useful map than the last one, which shows how MPs have said they'll vote (Great Britain only, I'm afraid):



Map created using data from: http://gu.com/p/4hxmj/stw

Green = remain
Red = leave
Grey = undeclared

Two seats are excluded: one neutral (Speaker's seat of Buckingham) and one vacant (Sheffield Brightside and Hillsborough).

Bigger version in gallery: https://uselectionatlas.org/FORUM/index.php?action=gallery;sa=view&id=14419

[Serenity Now]

On the subject of the potential geography of the result of the referendum, the other day I made the map below.  It shows a very crude regression model of the following variables:

• UKIP 2015 General election vote share (dependent variable)
• Conservative 2015 GE vote share (independent variable)
• Turnout at the 2015 GE (independent variable)
• 'White British' in 2011 Census (independent variable)
• Degree or equivalent in 2011 Census (independent variable)
• Aged 60 and over in 2011 Census (independent variable)

Using this, I divided the Parliamentary Constituencies into quartiles in order to show a hypothetical 'tie' situation:

Dark green = 25% most likely to vote to remain.
Green = the 25% which are likely than the median constituency to vote to remain.
Red = 25% which are less likely than the median constituency to vote to leave.
Dark red = the 25% most likely to vote to leave.

I ran this very arbitrary regression model for a laugh, just to see what the outcome would look like.  It may prove primarily useful for comedy value but I thought it might provoke some interesting discussion nonetheless.

Please feel free to criticize the hell out of this model (I can see a few howlers) as I may be able to change it into something more credible.  There a things said in previous posts pointing out potential problems (Sibboleth's comment about the geography of the UKIP vote, for one thing).

For extra 'fun', I can easily recreate this model to assume different national outcomes upon request, just give me the desired yes/no percentages.



Link to bigger version in gallery:
https://uselectionatlas.org/FORUM/index.php?action=gallery;sa=view&id=14405

How did you make this (as in what program did you use)?

I used Excel 2010 for the regression model (using the 'Data Analysis' add-on) and ArcMap to map the results.

fantastic work! but you've a fairly obvious outlier, what on your model tips Thanet South into Remain?
I also can't see Colchester, Chelmsford and Saffron Walden voting In when the rest of Essex is so heavily out, which it will be (ditto for Norfolk North, interesting pattern of residual LibDemmage though)

Yes, that's very strange about Thanet South: I'm wondering if that was an error with the data matching using the ONS codes for the constituencies (I've found mistakes like this before in various datasets of election results).  It could be to do with the relatively high turnout there.  The Lib Dem vote clearly confounds this model too.  The more I look at it the sillier it appears, although I think it does nail it for some areas and probably the two lighter shades give an indication of areas that could go either way..

[Serenity Now]

The following document should be helpful for people interested in how the count will be carried out, and result declared:

http://researchbriefings.parliament.uk/ResearchBriefing/Summary/CBP-7588

Basically, the count will start once polls close at 10pm, and results will be declared (for counting areas) throughout the night.

The list of counting areas expected to complete the count before 1.30am is interesting:

Sunderland (North East): 12.30am
Wandsworth (London): 12.30am
City of London (London): 12.45am
Foyle (Northern Ireland): 12:30am
Hartlepool (North East): 1.00-1.30am
Newcastle upon Tyne (North East): 1.00am
Oldham (North West): 1.00am
Swindon (South West): 1.00am
West Tyrone (Northern Ireland): 1:00am
Belfast East (Northern Ireland):   1:15am
Belfast South (Northern Ireland): 1:15am
Belfast West (Northern Ireland): 1:15am

We'll have a good indication of how Northern Ireland has voted, and as for English local authorities, we've got a major northern city (Newcastle), a socially mixed London Borough (Wandsworth), three (ex-)industrial northern towns (Oldham, Hartlepool and Sunderland) and a 'swing' area in the south of England (Swindon).

[Serenity Now]

Below are a few demographic maps which may be useful to refer to when watching the results come in. I don't wish to imply that these groups are likely to show a particular preference at this stage (except for maybe students and the over 60s) but they could turn out to be relevant.

I'll add some more later (I'll do requests).

Over 60s (as % of voting age population)


Full-time students (as % of people aged 16 to 74)


English only identity (as % of all usual residents)


British only identity (as % of all usual residents)


No passport (as % of all usual residents)

[Serenity Now]

A map from Yovgov,  Blue Leave and Red Remain



I'm surprised that the welsh valleys are leaning leave, and North East of England is leaning leave!?

It's a shame tho, no break down of Northern Ireland

It's also a bit weird that they're using parliamentary constituencies, rather than local authorities, but I guess they're basing this on voting patterns data.  Like I did with that very crude and crappy (but fun) map I made a while back.

[Serenity Now]

A map from Yovgov,  Blue Leave and Red Remain



I'm surprised that the welsh valleys are leaning leave, and North East of England is leaning leave!?

It's a shame tho, no break down of Northern Ireland

It's also a bit weird that they're using parliamentary constituencies, rather than local authorities, but I guess they're basing this on voting patterns data.  Like I did with that very crude and crappy (but fun) map I made a while back.

Just from looking at the Lib Dem areas I'd say they're defo basing this on General Election voting patterns (which probably isn't the best thing to do given this referendum cuts across those lines).

[Serenity Now]

PEOPLE NOTHING YOU WILL HEAR BEFORE 10PM GMT ON THURSDAY IS LIKELY TO BE ANY GOOD (and not much in the few hours after will be worth much either).

This. However, it would be hilarious if Plymouth and Clacton were somehow 70% and 52% Leave respectively.  Of course we won't know the result for the latter, but for the local authority it is located in instead (Tendring).

[Serenity Now]

some are saying that although they're counting by those they'll declare it by EU Parliament regions which would just be odd.

Where have you heard that? I have only heard that results are to be declared by Local Authorities.

As far as how results might go, Sunderland will be the first to declare, as always. Sunderland is relatively eurosceptic, I can't remember where I found it, but someone has calculated, based on the general Euroscepticism of the various parts of the UK, that Sunderland will probably be about 6% more "leave" than the country.

Although that clearly has to taken with a pinch of salt, and the margins of error are massive. It means we might be able to get some sort of picture relatively early on.

I really recommend reading the Chris Hanretty article which Parochial Boy has linked to.  It's very informative.

[Serenity Now]

For 'fun', here's the map of the 2014 European Parliament election.  Purple is UKIP, red is Labour, blue is Conservative, Yellow is SNP, Gold is Lib Dem and Green is Plaid Cymru.

The top three parties nationwide were UKIP (26.6%), Labour (24.4%) and the Conservatives (23.0%).  Turnout was a measly 35.6%.



Here are some fun questions to consider:

• Are all of the UKIP areas likely to vote Leave?
• Which UKIP area is most likely to vote to Remain, if any?
• Which Labour areas are likely vote to Leave?
• Which Conservative areas are likely to vote to Remain?
• Are all SNP and PC areas likely to vote to Remain?

I don't wish to suggest that this map has any bearing on how the referendum  result will look, particularly as it reflects pluralities for multiple parties in a relatively obscure election on a crappy turnout, rather than a 'yes/no' question.

[Serenity Now]

Ipsos-MORI's final poll is 52 L 48 R. Not sure what the figures are when undecideds are included so can't include in the table.

I think it's the other way round? Unless I'm thinking of a different poll..