Comparisons with other estimates
There are several other estimates of the partisan effects of the boundary changes. These include Martin Baxter’s estimates at www.electoralcalculus.co.uk and Lewis Baston’s estimates at the ERS, the main political parties will also have their own estimates. The most important alternate figures are those compiled by Colin Rallings and Michael Thrasher of Plymouth University for the media guide, which will be used by the mainstream media for all their election coverage and are regarded as the “official” figures. All methods should result in broadly similar figures, but inevitably some differences will arise due to projects using figures from different local elections and handling non-contested seats and independents in different ways.
As of February 2007 the only complete detailed estimates are available from Martin Baxter and Rallings & Thrasher. Differences in who notionally holds seats in the two projections are detailed below, though obviouly these are only the most cosmetically obvious differences; there may be more larger differences between the projections in seats where it doesn’t lead to a seat changing hands.
Rallings and Thrasher
Seats held by Labour on Wells projections and the Conservatives on Rallings & Thrasher projections
Finchley and Golders Green (link)
Seats held by the Conservatives on Wells projections and Liberal Democrats on Rallings & Thrasher projections
Somerton and Frome (link)
Seats held by the Liberal Democrats on Wells projections and Labour on Rallings & Thrasher projections
Oxford East (link)
Rochdale (link)
Seats held by the Conservatives on Wells projections and Labour on Rallings & Thrasher projections
Rugby (link)
Gillingham and Rainham (link)
Portsmouth North (link)
Ealing Central and Acton (link)
Baxter
Seats held by the Conservatives on Wells projections and Liberal Democrats on Baxter projections
Solihull (Wells/Baxter)
Somerton and Frome (Wells/Baxter)
Guildford (Wells/Baxter)
Seats held by the Conservatives on Wells projections and Labour on Baxter projections
Somerset North East (Wells/Baxter)
Ealing Central and Acton (Wells/Baxter)
Seats held by the Liberal Democrats on Wells projections and Labour on Baxter projections
Oxford East (Wells/Baxter)
Leeds North West (Wells/Baxter)
Seats held by the Liberal Democrats on Wells projections and Conservative on Baxter projections
Taunton Deane (Wells/Baxter)
Seats held by Labour on Wells projections and Conservative on Baxter projections
Finchley and Golders Green (Wells/Baxter)
Seats held by Labour on Wells projections and the Liberal Democrats on Baxter projections
Hampstead and Kilburn (Wells/Baxter)
Total Changes
Overall the Wells projections see the Conservatives gaining 16 seats, Labour losing 12, the Liberal Democrats gaining 1 and Plaid Cymru losing 1. The Rallings and Thrasher projections increase the number of Conservative seats by 12, reduce Labour’s seats by 7 and reduce Plaid Cymru’s by 1. They do not project any overall change in the number of Lib Dem seats. The Baxter projections see the Conservatives gaining 13 seats, Labour losing 10, the Liberal Democrats gaining 2 and Plaid Cymru losing 1.
For reference, the methodology used by Rallings and Thrasher is almost identical to that used here. What differences there are are very minor and technical, but for those who are interested when using local election results from all-out elections Rallings and Thrasher use the average vote received by each party’s candidates in a ward, I used the highest vote recieved by each party’s candidates. Secondly, in uncontested wards I gave the party of the winning candidate a vote equal to the share of the electorate who voted for that party equal to the average share of the electorate who voted for them in all the wards they won in that local authority, Rallings & Thrasher gave the winning candidate a vote equal to the share of the electorate who voted for that party in their best ward in that local authority. For parties who didn’t contest a ward, Rallings and Thrasher gave them a share of the electorate equal to the share of the electorate who voted for them in their worst performing ward in the local authority, I gave they a share of the electorate equal to the average share of the electorate who voted for all their losing candidates in that local authority. I ignored all independent candidates, when in some cases where they were obvious proxies for main political parties Rallings and Thrasher reallocated them. In London I used London assembly constituency results, while Rallings and Thrasher used local authority results throughout.
I assume the methodology used by yourself and by Thrasher & Rallings is alot more sophisticated than that used by Martin Baxter. It is useful that he reveals his calculations but i’m afraid by doing so he removes any credibility they have.
Take this example for the Knowsley seat http://www.electoralcalculus.co.uk/index.html?conlist_f_k.html
Apparently more than half of those who voted for ‘others’ were contained in one ward (Stockbridge). Meanwhile nobody in that ward voted Conservative or Lib Dem. In other wards (Cherryfield and Kirkby C) every single vote cast was for Labour.
Ive looked around a number of constituencies and found they almost all contain such frankly ludicrous results.
Indeed even in the example he provides here http://www.electoralcalculus.co.uk/index.html?conlist_f_k.html for the imaginary Easthampton such an anomaly is created whereby all the votes for ‘others’ are contained within one ward which also has no Conservative voters in it. Clearly alot more massaging of the base data is required to prevent these kind of outcomes.
Another criricism I would make of his methodolgy which may or may mot also apply to yours is that he only uses local election data from a single year. In some council areas which only vote every four years this is perhaps understandable but in many such as the metroplitan boroughs there are elections almost every year and often the support for the various parties fluctuates widely from one year to the next. Therefore it would provide a more reliable starting point to take the average vote share for each party in a ward over a number of recent elections rather than just one. This is what I do when calculating how individual wards would have voted in general elections.
Do you have any plans to publish your own calculations? It would be interesting to see how you have calculated the results for each ward and indeed what those results are. In particular i’d be interested ofcourse to see how those ward results compare with those I have come up with in those areas i’ve covered, although most of them are based on old wards since they are the building blocks on which the existing constituencies were built.
Furtehr to my point above I have just taken a look at taunton on that site to see how he could possibly have come to the conclusion that the removal of the West Somerset wards would translate Taunton Deane from a Lib Ddem to a Conservative seat. In fact he does have it as a Lib Dem seat but with a reduced majority (of 14)
This is because according to his calculations the Conservatives did not receive a single vote in the Dulverton and Quarme wards in 2005. I’m sorry to say so, as he has clearly put alot of effort into this project, but he must really go back to the drawing board.
I had a look at Stockton South and Stockton North. In both seats the only change is from one to the other and yet both see a narrowing in the Labour majorities, how is this possible? Is it a flaw in the methodology?
Yes that seems rather implausible. It is possible to reduce the percentage majority in both, if say a ward was moving from |Stockton South to Stockton North which was more heavily Labour than the average for Stockton South but not as Labour as the average for Stockton North – that would reduce the percentage lead in both seats. Obviously though the numerical majority could not fall in both. Looking at the Stockton results in that seat the diference is too great to beaccounted for by my explanation so I suggest it is just plain wrong.
Does anybody know if the R&T figures are online yet, as although i’ve got the book, it would be interesting to directly compare the two sets of figures. Also are Martin Baxter’s figures final or would he revise them?
Thanks.
I believe there are some wards in Knowsley that are generally uncontested (I’m pretty sure it’s the only Metropolitan in England that has uncontested seats) so technically the Tories and LibDems score zero in them. However it is unrealistic to assume that they genuinely have zero support to Labours 15,000 or whatever the electorate is
Rallings and Thrasher usually use the lowest vote which a party received in that borough to predict votes in wards where a party didn’t stand.
Andy, it was more a response to PW’s first post – a possible explanation for why EleCalc treats some wards as 100% Labour in MB’s calculations
A very useful comparison. Thanks.
The one that really jumped out at me was the Baxter projection for the Taunton Deane constituency. I think we are all unanimous that the wards that leave the Taunton seat are far more Conservative than those which remain, so I just can’t understand how Baxter has the new Taunton Deane as MORE Conservative than on the present boundaries.
With such a shocking mistake, it would cause me to doubt the other notional results.
Please could somebody confirm whether or not the figures provided by Nick Whyte are the same as the R&T figures for Northern Ireland?
I wonder which notional results (if any) the next Times Guide to the House of Commons is likely to publish?
I assume the Times Guide will refer to the Thrasher & Rallings figures as these are the ‘official’ ones jointly sponsored by the BBC, ITN, Sky News and the Press Association.
I can confirm that the figures porovided by Nick Whyte and those by Thrasher and Rallings are the same for Northern Ireland.
Pingback: YouGov poll the Sun 19/4 - Lib Dems take the lead
Pingback: A note on electoral constituency boundaries | British Politics and Policy at LSE