UK parliamentary boundary review

[YL]

The process for another attempt at reviewing the Westminster constituency boundaries in the UK is starting, and unlike the last two attempts this one is likely to get through.

Legislation since the last two attempts means that the number of seats will remain 650.  With a few exceptions which only apply to a handful of areas, there is a 5% tolerance on electorate, giving a minimum electorate of 69724 and a maximum of 77062.

The exceptions:
- Four island areas are protected from the process: Orkney & Shetland, Ynys Môn and Na h-Eileanan an Iar will remain as they are, and the Isle of Wight will be divided into two constituencies.
- There is an upper area limit, and constituencies which approach it are allowed to have lower electorates.  This is only relevant in the Scottish Highlands, and maybe not even there.
- Northern Ireland has a little more flexibility because of its small size, with constituency electorates allowed to be as low as 5% below the regional quota, so its minimum is 68313.

Numbers of constituencies are allocated to the four parts of the UKby Sainte-Laguë, and the Boundary Commission for England allocates the English quota among the English regions in the same way.  This gives

England 543 (+10)
Eastern 61 (+3)
East Midlands 47 (+1)
London 75 (+2)
North East 27 (-2)
North West 73 (-2)
South East mainland 89 (+6)
Isle of Wight 2 (+1)
South West 58 (+3)
West Midlands 57 (-2)
Yorkshire and the Humber 54 (no change)
Scotland 57 (-2)
55 excluding Orkney & Shetland and na h-Eileanan an Iar
Wales 32 (-8)
31 excluding Ynys Môn
Northern Ireland 18 (no change)

Initial proposals from the various Boundary Commissions are expected in a few months' time.

I recommend Kevin Larkin's Boundary Assistant (a UK counterpart to Dave's Redistricting) for playing with possibilities.  At the moment it has ward electorates for the date used (March 2020) in Wales, Northern Ireland and most of England; a few areas of England, mostly in London, need updating for new ward boundaries.

[President Punxsutawney Phil]

Anything update-worthy on this front?

[CumbrianLefty]

Don't think much has happened on the legislation front since the parameters were set.

But people are already starting to suggest ideas

[Former President tack50]

If there was a tool like DRA for the UK we could have some fun drawing maps or hilarious gerrymanders

Sadly, no such tool exists

[MaxQue]

If there was a tool like DRA for the UK we could have some fun drawing maps or hilarious gerrymanders

Sadly, no such tool exists

https://boundaryassistant.org/

[Former President tack50]

If there was a tool like DRA for the UK we could have some fun drawing maps or hilarious gerrymanders

Sadly, no such tool exists

https://boundaryassistant.org/

Huh, that is a very interesting and fun tool! Sadly, it has no election data of any kind which is half the fun of DRA

[President Punxsutawney Phil]

If there was a tool like DRA for the UK we could have some fun drawing maps or hilarious gerrymanders

Sadly, no such tool exists

https://boundaryassistant.org/

Huh, that is a very interesting and fun tool! Sadly, it has no election data of any kind which is half the fun of DRA

combine it with electoralcalculus if need be.

[Alcibiades]

If there was a tool like DRA for the UK we could have some fun drawing maps or hilarious gerrymanders

Sadly, no such tool exists

https://boundaryassistant.org/

Huh, that is a very interesting and fun tool! Sadly, it has no election data of any kind which is half the fun of DRA

That’s because there are no official general election results released in the UK for anything below the constituency level. However, as TimTurner says, Electoral Calculus has ward-level estimates using regression modelling. Anyway, due to tactical voting, nominal results under new boundaries are often not particularly useful guides to how a seat might vote.

[JerryArkansas]

If there was a tool like DRA for the UK we could have some fun drawing maps or hilarious gerrymanders

Sadly, no such tool exists

https://boundaryassistant.org/

Huh, that is a very interesting and fun tool! Sadly, it has no election data of any kind which is half the fun of DRA

That’s because there are no official general election results released in the UK for anything below the constituency level. However, as TimTurner says, Electoral Calculus has ward-level estimates using regression modelling. Anyway, due to tactical voting, nominal results under new boundaries are often not particularly useful guides to how a seat might vote.

If I recall, in the last review that went trough, a few seats that were using that regression model said to be Lib Dem seats were easily held by the conservative.  York Outer comes to mind in that regard.

[beesley]

If I recall, in the last review that went trough, a few seats that were using that regression model said to be Lib Dem seats were easily held by the conservative.  York Outer comes to mind in that regard.

Oh, yes - though to be fair, the Conservative only won by about 2000 or so votes (with some big local LD campaign), and the LDs also lost a nearby seat with a substantial majority. I think the Conservatives doing very well in open seats that election (Dartford, Cannock Chase, Winchester, Carlisle etc.) was a big part of that.

[/quote]

If there was a tool like DRA for the UK we could have some fun drawing maps or hilarious gerrymanders

Sadly, no such tool exists

Gerrymandering in the UK is a lot harder than the US - the building blocks are much bigger (in Scotland, London and Yorkshire especially). So you can probably make a gerrymander by turning safe seats into more marginal ones and pack the others, like the 2016-2020 NC Map, but you would never be able to make the current Maryland Map or seats like TX-21. Still, see how much fun you can have with it.

[Oryxslayer]

If there was a tool like DRA for the UK we could have some fun drawing maps or hilarious gerrymanders

Sadly, no such tool exists

Gerrymandering in the UK is a lot harder than the US - the building blocks are much bigger (in Scotland, London and Yorkshire especially). So you can probably make a gerrymander by turning safe seats into more marginal ones and pack the others, like the 2016-2020 NC Map, but you would never be able to make the current Maryland Map or seats like TX-21. Still, see how much fun you can have with it.

Don't be so naive. Countries with population flexibility, unlike the US's strict adherence to equal pop per district, don't need to draw tentacles to gerrymander. In fact a tentacle would do a disservice to the state mappers legitimacy, Instead, one simply has to ensure that the seats which you know from the get go will be won by the opposition have more population per seat than those seats you will be won by the governing party. Another trick is to cut or add seats to the count, depending on if the governments areas are growing compared to the opposition. Cutting seats, like the 600 map, would hurt incumbents, but produced maps more favorable to the Tories.

[beesley]

If there was a tool like DRA for the UK we could have some fun drawing maps or hilarious gerrymanders

Sadly, no such tool exists

Gerrymandering in the UK is a lot harder than the US - the building blocks are much bigger (in Scotland, London and Yorkshire especially). So you can probably make a gerrymander by turning safe seats into more marginal ones and pack the others, like the 2016-2020 NC Map, but you would never be able to make the current Maryland Map or seats like TX-21. Still, see how much fun you can have with it.

Don't be so naive. Countries with population flexibility, unlike the US's strict adherence to equal pop per district, don't need to draw tentacles to gerrymander. In fact a tentacle would do a disservice to the state mappers legitimacy, Instead, one simply has to ensure that the seats which you know from the get go will be won by the opposition have more population per seat than those seats you will be won by the governing party. Another trick is to cut or add seats to the count, depending on if the governments areas are growing compared to the opposition. Cutting seats, like the 600 map, would hurt incumbents, but produced maps more favorable to the Tories.

You're right, but I dispute that any of that makes it easier to gerrymander here, and in any case would be unachievable now due to the principles of the Boundary Commissions.

[EastAnglianLefty]

If there was a tool like DRA for the UK we could have some fun drawing maps or hilarious gerrymanders

Sadly, no such tool exists

https://boundaryassistant.org/

Huh, that is a very interesting and fun tool! Sadly, it has no election data of any kind which is half the fun of DRA

That’s because there are no official general election results released in the UK for anything below the constituency level. However, as TimTurner says, Electoral Calculus has ward-level estimates using regression modelling. Anyway, due to tactical voting, nominal results under new boundaries are often not particularly useful guides to how a seat might vote.

The quality of Electoral Calculus' ward-level estimates is in any case usually pretty poor, especially in areas where the major parties at Westminster level aren't the major competitors in local elections.

[IceAgeComing]

Yeah I'd not trust Electoral Calculus's numbers - from memory their notional numbers were very different from the other calculations in 2010 (that do things in a more complex way) and based on the swings in them they were very far out.

Its come increasingly common to split wards when drawing up seats in the UK - especially in Scotland but also in some other places that have large populations and relatively small councils either elected by thirds or with three councillors per ward.  The main factor has been the shift to the 5% threshold - in Scotland especially that gives you very narrow leeway without splitting wards, and the results are never satisfactory because of that fact.  The pre-2015 review had some monstrosities in it (remember Mersey Banks?) and I think that experience made the Commissions a lot more willing to split wards to avoid those: better to split a few wards and get seats that all make sense over not splitting them and having a few really dodgy seats that are just the orphaned wards that you can't get anywhere else.

[beesley]

(remember Mersey Banks?)

Good grief yes. Especially when it was suggested they should name it 'Mersey Banks and Weaver', as if that somehow justifies a seat containing both Ellesmere Port and Widnes.

[President Punxsutawney Phil]

I get that Electoral Calculus is flawed. But is there any alternative?

[IceAgeComing]

Local knowledge, hunches, intuition and an awareness of the fact that whatever methods you use; none of them will be perfect.

The other issue with using pure ward numbers is that tactical voting is a big thing; and so you can't assume that a ward that moves from a safe seat to a marginal or vice versa; or from a Labour/Conservative to Lib Dem/Conservative marginal will behave anything like it did in the last election.

[afleitch]

When I collated the ward results for Holyrood in Scotland in 2007 (the only time they were published) and engineered 'notionals' based on local elections held the same day there were notable differences, even in places where all parties stood. In that election, votes for MSP's were actually much more evenly spread across seats in both very urban and very rural areas. Only in seats where you had a very ABC1 heavy bastion did you find significant deviation from the rest of the seat.

[beesley]

Local knowledge, hunches, intuition and an awareness of the fact that whatever methods you use; none of them will be perfect.

The other issue with using pure ward numbers is that tactical voting is a big thing; and so you can't assume that a ward that moves from a safe seat to a marginal or vice versa; or from a Labour/Conservative to Lib Dem/Conservative marginal will behave anything like it did in the last election.

I think the best example is some of the proposed SW London seats. If you take away a Labour ward from Putney and add two wards from Richmond Park, suddenly it's a lot better for the Tories because Labour have such a low total in Richmond Park given the Lib Dem strength. But in a GE a big chunk of the new Lib Dem vote would coalesce around the Labour incumbent.

[afleitch]

To be fair, notional results always state that they can't take account of voters voting differently if say a safe Labour ward moves into a Tory/Lib Dem marginal.

[vileplume]

Local knowledge, hunches, intuition and an awareness of the fact that whatever methods you use; none of them will be perfect.

The other issue with using pure ward numbers is that tactical voting is a big thing; and so you can't assume that a ward that moves from a safe seat to a marginal or vice versa; or from a Labour/Conservative to Lib Dem/Conservative marginal will behave anything like it did in the last election.

I think the best example is some of the proposed SW London seats. If you take away a Labour ward from Putney and add two wards from Richmond Park, suddenly it's a lot better for the Tories because Labour have such a low total in Richmond Park given the Lib Dem strength. But in a GE a big chunk of the new Lib Dem vote would coalesce around the Labour incumbent.

Richmond Park's probably not the best example for this as it's easily the most right-wing Lib Dem held constituency and consequently a rather large minority of the Lib Dem vote in this seat would prefer the Tories to Labour. If you're looking for a constituency where the Lib Dem vote would heavily prefer Labour, Bath would be a good bet.

A gerrymander that switched Roehampton for Barnes would, short of a Tory collapse, wreck Labour's chances of holding Putney especially if turnout among Corbyn friendly groups (young renters) falls under Starmer, which is a distinct possibility.

[YL]

The Boundary Commission for England announced today that their initial proposals will be published on 8 June, and published the "Guide to the Review".  We also now have all the ward electorate data, although we don't yet have electorate data at sub-ward level if splitting a ward is being considered (which it probably will be in a few places).

I'll go through some of the regions (maybe all of them) with some thoughts.  Of course most of my plans are rather unlikely to end up actually being proposed.

Let's start with the South West.  It gets 58 seats, up 3.  The entitlements of the counties and unitary authorities are:

Cornwall 5.95
Scilly 0.02
Cornwall and Scilly 5.97

Devon (county council) 8.52
Plymouth 2.60
Torbay 1.40
Devon total 12.53

Dorset (unitary) 4.08
Bournemouth 3.92
Dorset total 8.00

Somerset (county council) 5.80
North Somerset 2.24
Bath & NE Somerset (BANES) 1.93
Somerset total 9.97

Bristol 4.57

Gloucestershire (county council) 6.59
South Gloucestershire 2.90
Gloucestershire total 9.49

Wiltshire (unitary) 5.15
Swindon 2.12
Wiltshire total 7.27

With the 5% tolerance this allows 6 seats for Cornwall (no change), 13 for Devon (up 1), 10 for Somerset (up 1), 8 for Dorset (no change), 7 for Wiltshire (no change) and 14 for Bristol and Gloucestershire treated together (up 1).  Devon is on the small side for 13, though, so you could consider crossing its boundaries somewhere, and likewise Wiltshire is on the large side.

BANES could be separated out from the rest of Somerset, as it is now, and the two Dorset unitaries could be treated separately, which they're not now (as they didn't exist when the current boundaries were drawn).

[YL]

Let's start with Cornwall.  One rule is that "minimal change" tends to be favoured: if no changes at all or only minor ones are needed, that's what will usually happen, unless there's some very good reason.  Cornwall is a case in point: its six existing seats are either within the legal range or not far outside it, and so not much is going to change.



1. St Ives
2. Camborne & Redruth
3. Truro & Falmouth
4. St Austell & Newquay
5. North Cornwall
6. South East Cornwall

[Geoffrey Howe]

Is it possible to engineer a Labour seat in Cornwall? I think Truro, Falmouth and Redruth are more amenable.

[President Punxsutawney Phil]

Is it possible to engineer a Labour seat in Cornwall? I think Truro, Falmouth and Redruth are more amenable.

It would be easier if you halved the number of seats in the House of Commons. That way seats would be larger and it would be possible to link them all up.

edit: it seems possible to link up even with current sizes. interesting. I didn't realize how close they were to each other.