bias in the electoral system since 1945

[hullenedge]

Oct 1, 2020 10:22:38 GMT nodealbrexiteer said:

Oct 1, 2020 10:01:20 GMT hullenedge said:

From Britain Votes, Sir JC's calcs, the %spread of Con/Lab seats at equal shares (Tory advantage of 28 seats, 293-265):-



More marginal seats for Labour but also more strongholds.

the 80-100 to figure for Lab is not good, they need more in the middle of these spreads

On the plus side for Labour these strongholds prevent their total collapse/replacement by another party of the left and also ensure a 'safety net' in v.bad years (1983). A wise leader/party would use these seats to bring talent into parliament.

[nodealbrexiteer]

Of course the workings of the electoral system in some elections may have changed the course of the ensuing parliament. I'd say 1950 as the bias denied Labour a working majority it might have expected from the cube law and which would have allowed them to carry on until the mid 1950s at least,1951 Conservatives get in with few votes than Labour, Feb 1974 similar to 1951 with Lab as the beneficiary, Oct 74 Lab get similar lead to the Tories in 1955 but a fraction of the majority setting up an unstable parliamentary situation until 1979.

Also 1992 a 7.6% Con lead giving only a 21 majority easily eaten away by defections and by election defeats and 2010 as in normal circumstances the Conservatives might have expected a majority from a 7.3% lead

[nodealbrexiteer]

My rather verbose thoughts on the reasons for changing bias election to election post war:

1945-Lab bias with out of date boundaries-noticeable smaller electorates.
1950-1959-Clear Con bias in all cases due to Lab piling up votes in safe seats but in 1950 to 1955 due to rural seats being allowed to be smaller and in 1950/1 turnouts lower in Con seats. Some Lab counteraction to bias due to lower turnouts favouring them from 1955 and lower electorates from 1959(also Lab domination in Scotland 1959-2010).1955 bias little less as lower swing to Cons in marginals
1964-mild Con bias due to lower turnouts in Lab and increasing constituency size variations.
1966-Lab bias due to similar reasons as 1964 though not as much as could have been in 1970 as Con vote much more efficiently distributed in 1970 than 1966
Feb 1974-Lab bias though notionals showed con bias-more efficient distribution of Lab vote and tactical voting depriving Con of some seats that might have kept them largest party.
Oct 1974-Con bias-mild due to their vote being more efficiently distributed due to lower swings in Con/Lab marginals
1979-Lab bias-lower swings in Lab marginals,ageing of boundaries,lower swings to Con in seats with lower electorates
1983 and 1987-Con bias due to most detailed redistribution to date,Labour piling up votes in safe seats,3rd party votes in con seats-all partially offset by lower turnouts and lower electorates in Lab seats and 3rd party wins
1992-significant Lab bias due to wider gap in turnouts/electorates and bigger swings to Lab in marginals
1997-Lab bias more than in 1992 due to larger swings in Con seats and ongoing big gap in turnouts between Con and Lab seats plus more 3rd party wins at Con expenses
2001-Lab bias widened still further due to larger turnout drop in Lab seats, boundaries based on 1991 data,increase in 3rd party votes in Lab seats and drop in Con ones(due to less Eurosceptic votes in Con seats and increase in Lib dem/some far left votes in Lab seats),plus Lab holding on to marginals and getting swing in their favours whilst safer seats lost chunks of their majorities
2005:Reduction in Lab bias but still large due to some tactical unwinding or marginals swings favouring con for first time since 1987
2010:Reduction in Lab bias as Con knocks out Lab's efficiency advantage and do better than uniform swing though clear bias still there due to the other factors
2015-2019-flipped back to Con bias-incurred in first election due to Con holding on in marginals(often getting swings in their favour) reducing issues of 3rd party wins due to SNP wins at Lab expense and Con gaining Lib Dem seats.Maintaining the bias in latter 2 elections as Lab continue to pile up votes in seats they hold and also Cons gaining in old Lab heartlands which means Con vote now more concentrated in seats with less voters and lower turnouts.

[islington]

Or, to put it another way: swings, roundabouts.

[nodealbrexiteer]

What seat distribution/majority would have resulted if the swing had been uniform from the previous election:
October 1974:Lab maj of 17 or 326 seats
1979:Con maj of 65-Con 350,Lab 258
1983:No figures in Nuffield study but article by Clive Payne suggests Con maj of 134-Con 392,Lab 213,Alliance 19,Nationalists 7
1987: Con maj of 98-374 seats, Lab 237,Alliance and Nationalists together 22
1992: Con maj of 71
1997:Lab Maj of 131-Lab 395,Con 208, Lib Dem 28
2001: Lab maj of 145-Lab 402, Con 180(181 with Tatton), Lib dem 48
2005:Lab maj of 94-Lab 370,Con 184, Lib dem 62
2010:Con short by 35-Con 291,Lab 264,Lib Dem 64,Other 31
2015:Con short by 6: Con 320,Lab 243,SNP 55, Lib dem 10, Plaid Cymru 3, green 1
2017:Con short by 3: Con 323,Lab 259,Snp 37,Lib Dem 9,Plaid Cymru 3,green 1
2019:Awaiting the Nuffield study for firmer figures but Stephen Fisher suggests Con maj of 58-Con 354,Lab 211,Lib dem 17,SNP 46,PLaid Cymru 3,green 1

[The Bishop]

1987 was basically spot on for the majority!

Striking how poorly the Tories did in the marginals come 1992, though. And to a significant degree Labour in 2005 too, a lot of "unwind" of the pro-Labour bias there.

[nodealbrexiteer]

Jan 28, 2021 12:22:16 GMT The Bishop said:

1987 was basically spot on for the majority!

Striking how poorly the Tories did in the marginals come 1992, though. And to a significant degree Labour in 2005 too, a lot of "unwind" of the pro-Labour bias there.

Unwind yes but I think Sir John Curtice pointed out as there were something like double the amount of Lab marginals that would go on a 3-6% swing compared to those that would fall on a 0-3% swing any random variation around the national 3% swing in that election would almost certainly mean Lab would lose more seats than uniform swing would suggest. 

[nodealbrexiteer]

see p144 of the article below for uniform swing projections from 1979-1983 and 1983-1987:

pdfslide.net/reader/f/statistical-methods-for-election-forecasting-in-the-united-kingdom-197090

[nodealbrexiteer]

www.academia.edu/18128419/Unequal_and_unequally_distributed_votes_The_sources_of_electoral_bias_at_recent_british_general_elections?email_work_card=view-paper

[nodealbrexiteer]

ur.booksc.eu/book/19970949/81d074

Proportionality and exaggeration in the British electoral system

1986 article by Michael Steed and John Curtice

[nodealbrexiteer]

The new Nuffield study gives the following seat deviations from uniform swing Con(+14),Lab and Lib Dem each 6 under, Nationalists 2 under