some random exit polls/election day polls for other ballots

[nodealbrexiteer]

Apr 18, 2021 0:09:26 GMT Andrew_S said:

Not quite the same thing, but just after 10pm in 1992 Peter Snow went through his battleground with the forecasts for the 94 target seats, and you can seat what was expected to happen in the seats displayed on the big screen, such as Edinburgh West turning gold for the LDs (despite being on the Labour target list). But that was of course after the adjustment was made putting the Tories narrowly ahead by 301 to 298.

Incidentally, I have Sky's 1992 election show on video, but Sky don't allow their election shows to be uploaded to YouTube as far as I'm aware.

If you can view if it would be nice to hear what forecasts they made before the first result and based on what data thanks!

The ICM exit poll has wrongly been attributed as partly being sponsored by BSKYB

[nodealbrexiteer]

Apr 17, 2021 23:19:27 GMT Davıd Boothroyd said:

That explains why David Dimbleby remarked (in the documentary Swing Time) about the fact that a big picture of Neil Kinnock was being displayed as the forecast winner of the election, until a few minutes before going on air when the 10pm numbers showed the Conservatives narrowly ahead.

The data that changed the forecast to Con largest party was received at 9.45pm, ten minutes before the programme went on air and when it did at five to ten the Kinnock photo was still showing so frantic voices were telling Dimbleby in his earpiece the forecast was about to change and he may have only  heard the new forecast seconds before Big Ben struck 10pm!

[nodealbrexiteer]

few extra bits on ITN 1992 exit poll and updated forecast from the Political Communications book:

They had 41 special seats(or 'handset' as they called them). Without handsetting the conservatives would have been nearer to a majority but still short but at the expense of the Liberal Democrats who'd have been underestimated. Whereas BBC's special seats error knocked 9 seats off the Con estimate there have been no precise figures suggested for how many seats the ITN forecast for the handset seats knocked off their total but a figure is suggested of 13 in the article The Opinions Polls and the 1992 General Election-thanks to hullenedge for sending me a link to the full version of it.


Chris Long of ITN retrospectively showed how much quicker a forecast based on real results(starting with the result of the 1987 election instead of the exit poll) would have forecast a Con majority. Apparently such a forecast was run in the background alongside the one based on the exit poll and it was due to the disparity between the two forecasts that ITN decided to pull the exit poll from their system not long after 1am.

This method would have hit the final Con seat total with the declaration of the second marginal at 00:19 according to this article(ITN forecast methodology at this time was based on declarations from marginals so only changed when one was declared) .

How this alternative forecast would have developed is show below:

The first figure shows the amount of seats already declared(so Basildon declared 4th),then the name of the seat, then we have the time of declaration, the swing in that seat and the forecast Con and Lab totals. Initial forecast was Con 366, Lab 233(presumably state of those parties at dissolution):

3,Basildon,23:34,1.3%,Con 352, Lab 247
16,Pendle,00:19,4.5%,Con 336, Lab 263
19,Hyndburn,00:24,4.3%, Con 331, Lab 268
25,Nuneaton,00:28,6.5%, Con 319, Lab 280
43, Putney,00:39,0.6%, Con 329, Lab 270
56, Lincoln,00:46,4.8%, Con 329, Lab 270
60,Delyn,00:49,3.0%, Con 333,Lab 266
84,Bristol East,6.8%, Con 328, Lab 272
88,Coventry SW,1.7%, Con 334, Lab 266
95,Wolverhampton NE,4.2%, Con 332, Lab 268

[nodealbrexiteer]

The article below indicates the rogue ICM 1992 exit poll handset just the 17 Northern Ireland seats and about 12 mainland seats(see pages 193 and 194):

www.electionscentre.co.uk/wp-content/uploads/2015/07/20150706093835OpinionPollingandtheAftermath.pdf

[nodealbrexiteer]

More on special seats at other elections-I am still looking for the 2010 information, it's in a book on that election:

1997-BBC aimed to minimise the number of special seats as a reaction to 1992-they were all 18 Northern Ireland seats, 5 Welsh seats where Paid had a serious chance,2 seats in GB where a major party wasn't standing-Tatton and West Bromwich East,3 seats lost by Con to Lib Dem in by elections with very large swings-Christchurch, Eastleigh, Newbury-28 in total. There were also 25 'not so special' seats whose results weren't used to predict other seats but whose results could be predicted from non special seats-the 25 seats were so defined to include those where a major party candidate was opposed by a rebel candidate from the same party.

2005:(Constituencies in Northern Ireland, which are not contested by the main British parties, were not covered by or predicted from the exit poll. In addition four ‘special’ seats in Great Britain were not predicted from the exit poll: these were the Speaker's seat, one seat that has been held since 2001 by an Independent Member of Parliament and two constituencies where there the seat had changed hands as a result of a by‐election. Predictions for these seats were made separately, using subjective probabilities.)

[nodealbrexiteer]

1983-BBC-49 special seats
1-Seats where parties other than Con,Lab,All,SNP had a chance of winning-17 Northern Ireland,4 in Wales for Plaid Cymru.
2-4 constituencies where recent by elections had been held and where the Alliance performance in the by election could distort the result(eg Darlington and Bermondsey).
3-4 seats fought by the founder members of the SDP who were expected to gain personal votes.
4-4 seats where there had been a dispute over the notional 1979 results.
5-8 seats where a candidate of a major party was opposed by a rebel candidate of that party either in 1979 or 1983(5 Labour rebels and 3 independent Liberals opposing the SDP.
6-A residual group where special factors were thought likely to operate such as personal votes for particular candidates.

[nodealbrexiteer]

February 1974-BBC-40 seats-12 Northern Ireland, seats where Welsh and Scottish Nationalists were though strong(11-probabilty of winning though to be greater than 0.3),unique or personality seats and those where boundary changes made the notional result highly unreliable.

October 1974-BBC 32 seats including those seats in Wales where the Nationalists were thought to have a chance of winning.

[nodealbrexiteer]

Breakdown of BBC special seats February 1974-the 12 Northern Ireland seats all went to others whilst the forecast gave 6 others and 3 Lab, 3 Con in Northern Ireland(the figures in brackets are what the forecast would have been if the 12 Northern Ireland seats were forecast correctly):

Con Actual 5 Predicted 9.6(6.6)
Lab Actual 11 Predicted 14.6(11.5)
Lib Actual 2 Predicted 2.1
Nationalist(*) Actual 8 Predicted 6.1
Other Actual 14 Predicted 7.6(13.6)

(*)-Nationalists won an extra seat not in the special category)

Breakdown of BBC special seats October 1974:

Con Actual 5 Predicted 4.9
Lab Actual 4 Predicted 2.9
Lib Actual 7 Predicted 6.0
Welsh Nationalists Actual 3 Predicted 1.9
Others Actual 14 Predicted 14.3

Incidentally the 1997 special seats had a net error of just 1 seat, Martin Bell's seat being allocated to Con merely for reasons related to the BBC's battleground display.

[nodealbrexiteer]

2010 special seats-18 Northern Ireland,5 mainland seats-Current speaker,former Speaker plus 3 seats won by independents in 2005-Bethnal Green and Bow, Blaenau Gwent,Wyre Forest

[nodealbrexiteer]

Aug 28, 2019 21:50:45 GMT nodealbrexiteer said:

The BBC/October 1974 Scotland (computer based) prediction based on their exit poll is interesting:


Con 15.6(actual 16),Lab 38.5(40),Lib 3.4(4), SNP 13.5(11)-remembering that the forecasts are sum of the probabilities of each party winning each seat in Scotland hence the forecasts are not whole numbers!-this is from the article below:


www.jstor.org/stable/2345211?read-now=1&seq=21#page_scan_tab_contents

Bumped this up to illustrate how the computer based exit poll prediction was less wild on the picture of the SNP gains compared to the cruder swingometer based suggestions of even more gains based on the Scotland alone exit poll.

[nodealbrexiteer]

www.facebook.com/tvark.org/videos/election-83-bbc/1961803040528053/

Close of polls 1983 David Dimbleby gives the BBC forecast and a range of seats

[nodealbrexiteer]

www.tvark.org/?page=media&mediaid=53447

ITV equivalent clip for 1983 with Alastair Burnet

[Andrew_S]

Apr 28, 2021 22:04:28 GMT nodealbrexiteer said:

www.facebook.com/tvark.org/videos/election-83-bbc/1961803040528053/

Close of polls 1983 David Dimbleby gives the BBC forecast and a range of seats

One of the best ever forecasts that wasn't based on modern exit poll techniques.

[nodealbrexiteer]

Apr 28, 2021 22:34:16 GMT Andrew_S said:

Apr 28, 2021 22:04:28 GMT nodealbrexiteer said:

www.facebook.com/tvark.org/videos/election-83-bbc/1961803040528053/

Close of polls 1983 David Dimbleby gives the BBC forecast and a range of seats

One of the best ever forecasts that wasn't based on modern exit poll techniques.

Possibly followed by BBC 1979. I do wonder if they'd had reasonable success in 1987 using the same poll technique if they may have used it in 1992-even bigger disaster then than it actually was probably!

[nodealbrexiteer]

more details of BBC/Gallup election day poll 1983:

4141 voters sampled(target was 4500), each interview lasted 30 minutes with fieldwork on 8th and 9th June 1983. 3174 voters in England and Wales,967 in Scotland(downweighted for overall GB sample). A quota sample controlled by age, sex and housing. Weighted oversample of 862 voters in Labour held marginals and 738 in Conservative/Liberal marginals.

Information from Britain at the Polls 1983, a write up of some of the data was done by Professor Sir Ivor Crewe who designed the survey in The Guardian a few days after the election.

This line from my post in 'Interesting electoral facts needs a slight correction:

Britain at the Polls 1992 erroneously published the BBC forecast shares as Con 39%,Lab 39%,Lib Dem 18%(and I believe the Birmingham Post a few days later may have published 40% Con,Lab 40% too)

It should read(having checked Britain at the Polls 1992): Britain at the Polls 1992 erroneously published the BBC forecast shares as Con 39%,Lab 39%,Lib Dem 17%(and I believe the Birmingham Post a few days later may have published 40% Con,Lab 40% too).

Apologies!

[nodealbrexiteer]

May 9, 2021 18:57:34 GMT nodealbrexiteer said:

more details of BBC/Gallup election day poll 1983:

4141 voters sampled(target was 4500), each interview lasted 30 minutes with fieldwork on 8th and 9th June 1983. 3174 voters in England and Wales,967 in Scotland(downweighted for overall GB sample). A quota sample controlled by age, sex and housing. Weighted oversample of 862 voters in Labour held marginals and 738 in Conservative/Liberal marginals.

Information from Britain at the Polls 1983, a write up of some of the data was done by Professor Sir Ivor Crewe who designed the survey in The Guardian a few days after the election.

This line from my post in 'Interesting electoral facts needs a slight correction:

Britain at the Polls 1992 erroneously published the BBC forecast shares as Con 39%,Lab 39%,Lib Dem 18%(and I believe the Birmingham Post a few days later may have published 40% Con,Lab 40% too)

It should read(having checked Britain at the Polls 1992): Britain at the Polls 1992 erroneously published the BBC forecast shares as Con 39%,Lab 39%,Lib Dem 17%(and I believe the Birmingham Post a few days later may have published 40% Con,Lab 40% too).

Apologies!

The poll used data from Lab held seats in the North, South of England and the Midlands in seats where the Conservatives were no more than 12.5% behind in the 1979 notionals for tactical vote information and Con/Lib seats in the South of England(as defined by the 1979 notionals)

[nodealbrexiteer]

some miscellaneous updated results based forecasts(more to come in time when I can get a good look through the various election night programmes):

1979:BBC after 22 results Con 337, Lab 271, Lib 9,SNP 3,Others 15,after 135 results(02:05am approx)-343/269/7/2/14,half hour later after 203 results same predictions,about 3.15am after 325 results same forecast as last 2,after 448 results about 4.15am 330/277/11/3/14,same after 510 results,after 515 results 337/271/10/3/14,after 519 results same as after 515,after 526 results 338/270/10/3/14,(Euro elections projection Con 47,Lab 30,Northern Ireland 3,Highland and Islands 1),after 553 results 339/270/10/2/14,after 571 results 339/269/11/2/14,after 592 results 339/268/11/2/15,after 619 results 339/269/11/2/14 at nearly 3.50pm 4th May 1979.

[nodealbrexiteer]

1983:BBC after 1 result Con 390/211/Lib 23 SDP 3/23,after 3 results 389/213/25(22/3)/23.after 15 results 381/219/27/23,after 34 results 393/210/24(20/4)/23,after 94 results 398/209/20(17/3)/23,(PR after 395 results 280/193/166/11),after 383 results 394/217/18(15/3)/21,after 572 results 398/209/21(16/5)/22 and after 572 results.(PR after 579 results 279(-119),190(-19),167(+146),14(-8).)

ITN(based on video on Youtube) after 4 results still Con 383/Lab 227/All 17/Other 23,then Con 388,Lab 217,SDP 3,Others 23,then 381/224/22/23,then 390/213/25/22,then 391/211/26/22,then bumped up to 136,then 394/210,then Con 395,Lab 210,Lib 19,SDP 4,SNP 2,Plaid 3,NI 17.Then Con maj of 140

[nodealbrexiteer]

Oct 1974:BBC after 2 results Lab 332, Con 261, Lib 14, Other 28(not referred to on air at 23:23),an interview with Ron Hayward then refers to a Lab maj of 33,after 16 results 333/263/10/29,then after 20 results 329/267/10/29,after 52/53 results 325/271/10/29,then 324/272/9/30,after 361 results 325/272/10/28,then 322/275/11/27,at 02:56 321/276/11/27,then 321/275/11/28(lab figure not on screen),after 489 results 321/276/11/27(end of overnight coverage-03:59 approx),after 493 results 322/276/11/26,10:19 am 321/276/11/27,after 495 results 320/276/12/27,after 496 results 320/275/13/27,after 497 results 320/276/12/27 ,after 542 results 320/276/12/27(McKenzie commented the majority will be 1,3 or 5),after Bosworth stayed in Con hands the Lab maj was forecast as 3(presumably 319/277).

ITN(possibly 1.45am)-Lab 328, Con 267, Lib 11,Others 29

[nodealbrexiteer]

Feb 1974: BBC after 3 results Con 303, Lab 291,Lib 27-23:34(just before it was announced the prediction board shows Lib 32 and Con lead over Lab as 5),after 6/7 results Con 301,Lab 296,Lib 24,after 10 result Lab 325, Con 274,Lib 22?(obscured by Burnet's head),after 11/12 results Lab 325, Con 274, Lib 22,after 25 results 319/290/12,after 72 results 314/297/10,after 100 results 311/301/9,after 197 results 305/302/14,after 234 results Lab 302,Con 304,Lib 14(around 02:15),after 279 results Lab 306,Con 300,Lib 14(after this point Lab were always predicted to be the largest party),after 340 results 307/300/13,after 346 results 308/299/13(plus some implied ones possibly missed when showing studio shots),307/299/14,after 404 results 307/297/15,after 409 results 308/?/15,then ?/?/14-Lab lead over Con 13-? is for figures obscured by studio camera angles, after 428 results 308/297/14(03:33am),after 439 results 307/298/14,after 442 results 308/298/14,after 458 results 309/297/14,after 477 results 307/300/13,after 558 results 306/297/14,after 574 results 306/297/13,after 14:00 on 1/3/74 Lab 304,Con 299,Lib 12,Nats 7,Other 13,then Lab 300,Con 296,Lib 14,Ulster 12,Nats 10,Others 3,then 301/296/14/Ulster 12/Others 12,then Lab 301,Con 295,Lib 15,Nats 9,Others 3,UU 6,SDLP 2,Vanguard 3,DUP 1,mid afternoon circa 15:40-Lab 300,Con 297,Lib 14,Nats 9,Others 3,UU 6,SDLP 2,Vanguard 3,DUP 1(after 622 results),after 625 virtually same as after 622 result bar UU 7,SDLP 1.