New Scottish Constituency Boundaries 2005

In December 2003, the Boundary Commission for Scotland published its final recommendations for the new Scottish constituencies boundaries. These boundaries will take effect at the next General Election.

The number of Scottish consituencies has decreased from 72 to 59. This reduces the over-representation of Scotland at Westminster following the creation of the devolved Scottish Parliament in Edinburgh.

The Boundary Commission's changes have been widespread and almost all Scottish seats have been altered in some way.

I am very grateful to Raymond Baxter for the preparation and construction of these data.

Boundaries' Impact on 2001 results

If the 2001 election had been run on the new boundaries the results would have been.

Party	Gains	Losses	Change	Old Seats	New Seats
CON	0	1	-1	1	0
LAB	1	11	-10	56	46
LIB	0	1	-1	10	9
SNP	0	1	-1	5	4

The one seat to "change hands" is Galloway and Upper Nithsdale held by Peter Duncan (CON). This old seat has much in common with the new seat of Dumfries and Galloway. However, Labour would have won this new seat narrowly in 2001, had the new boundaries been in force then.

Disappearing seats

The total number of seats is decreasing by 13. Although there is no hard-and-fast definition of a "vanishing seat" the following seats are particularly broken-up and dispersed under the new boundaries.

Old Constituency	Party	Member (as at 2001)
Aberdeen North	LAB	Malcolm Savidge
Cunninghame South	LAB	Brian Donohoe
Dumbarton	LAB	John McFall
Dunfermline East	LAB	Gordon Brown
Edinburgh Central	LAB	Alistair Darling
Glasgow Govan	LAB	Mohammed Sarwar
Glasgow Kelvin	LAB	George Galloway
Hamilton North and Bellshill	LAB	John Reid
Hamilton South	LAB	Bill Tynan
Linlithgow	LAB	Tam Dalyell
Renfrewshire West	LAB	James Sheridan
Tayside North	SNP	Peter Wishart
Tweeddale Ettrick and Lauderdale	LIB	Michael Moore

The current MPs named above may well be selected by their parties for a neighbouring new seat, but that information is not available at this time. Please contact the party concerned for details of their candidates.

Calculation details

The following assumptions were made to calculate the 2001 implied results.

The Boundary Commission for Scotland's final recommendations of 11 December 2003 are assumed to be adopted unchanged.
Old constituencies are assumed made up of whole new wards. This is not strictly true, but we do not think it introduces significant error.
The Speaker of the House of Commons, Michael Martin, is deemed be a Labour MP. The Unionist votes in his old seat of Glasgow Springburn are assumed to represent Conservative voters.
The calculation makes use of local government elections of May 2003 which show party support ward-by-ward. Where a party had no candidate in a particular ward, a subjective estimate of likely support for that party in that ward has been made. Typically a party does not stand where it would only receive small support, so this should not materially change the winning party.

The calculation is based on allocating votes in old constituencies into new ones. If the old constituency is completely contained within a new constituency, then its votes are simply transferred into the new one. If the old constituency is split between two or more new constituencies, then the calulcation is performed as follows.

Let the wards in the old constituency be called $i$, where $i=1,\ldots,N$. Let $j$ be the index of a particular party. Let $E(i)$ be the 2001 electorate in that ward. Let $E=E(1)+\ldots+E(N)$ be the total electorate in the old constituency.

Let $G(j)$ be the absolute votes cast for party $j$ in the old constituency at the June 2001 General Election. Let $L(i,j)$ be the absolute votes cast in ward $i$ for party $j$ at the May 2003 local elections. Let $S(j)=L(1,j)+\ldots+L(N,j)$ be the total votes cast in the old constituency for party $j$ at the May 2003 local elections.

Then our formula for the implied votes in ward i for party j at the 2001 General Election is

$$ V(i,j) = L(i,j) + \left(G(j) - S(j)\right){E(i)\over E}. $$

Or, in the less typical case where $S(j)$ is larger than $G(j)$, we use a multiplicative formula which avoids the risk of negative implied votes,

$$ V(i,j) = L(i,j){G(j)\over S(j)}. $$

This formula has the desired property that $V(1,j)+\ldots +V(N,j) = G(j)$, so that the total number of party $j$ votes allocated from the old constituency equals the votes actually cast for that party.

A new constituency will then receive an additional $V(i,j)$ votes for party $j$ if it contains ward $i$.

There is a worked example of these calculations for the new seat of East Dunbartonshire.

External link

Boundary Commission for Scotland