How good a pointer are local by-election results?
November 23rd, 2004
-
Be wary of the party spinners
As we get nearer the General Election pundits and punters will be looking beyond the opinion polls to find pointers as to what is going on.
One source that’s readily available are the results from the handful of local council by-elections that take place up and down the country almost every Thursday. For many years the results were aggregated and issued monthly as a sort of poll.
-
These are real electors putting real votes in real ballot boxes and so, the argument goes, they are as good, if not better, a test of public opinion than the polls which only survey 1,000 to 2,000 people.
But be careful of all the party machines which seek to reinforce the morale of activists by selecting specific results and then extrapolating national trends.
We wondered whether the head of Populus, Andrew Cooper, might have been on the receiving end of such “spin” when he made this observation on the site yesterday “..local by-election results as well as national polls seem to suggest that Labour’s vote is hardening: if you average recent local by-elections in which all 3 main parties stood Labour lead by about 5% over the Tories, which wasn’t true at this point in the last Parliament.”
We follow the by-election results closely and could not work out where Andrew was getting his information from. If Labour was doing so well, the predictable response came, how come that it’s been the Tories and to a lesser extent the LDs who have been winning seats?
The Tory activist and regular contributor to the site, Sean Fear, then produced the following analysis of 24 wards where all three parties have stood since the start of October, and also stood in the previous election.
The big problem is that there are so few seats that fit the bill and it is hard to draw conclusions. Council by-elections usually have very low-turnouts and are probably a better indicator of the strength and morale of local party machines than they are of public opinion.
-
In the year before the last General Election, it was possibly to look at the local by-election results and mistakenly conclude that the Tories and Lib Dems were doing much better than they did in the General Election.
So far in 2004 the Tories have net gains of 15, the LD s have net gains of 4, and Labour is down 14 seats. We think this is pretty meaningless and we should be wary of those who seek to extrapolate large conclusions that they say apply to the nation as a whole.
General Election seat markets:- Total Labour seats , Total Conservative seats , Total Lib Dem Seats.
Mike Smithson
MessageSpace Advertising
Slightly O/T: did anyone see Michael Howard on Newsnight last night? I\’m no fan of the Folkestone Vampire, but Paxo and crew did their best to portray him in a most unflattering light, shooting his visit to Falmouth from bizarre camera angles and vox-popping with a grey-haired rentatory mob some of whom it seems had been dragged off the street to make up the numbers.
I didn\’t see it… though to be fair, they probably try to tilt the vox pops against most politicians they cover. Violent agreement wouldn\’t make very good television.
Presumably Falmouth and Camborne is regarded as winnable by all three parties, since he made a point of going there?
BV - in 2001 it was 18500 Lab, 14000 Tory, 11500 LD, which makes it a fairly close 3-way contest. Winnable for the Tories but any Lab -> LD switching could see them take it.
Well, yes. The results do indicate certain trends. Plainly, Labour support has risen from the Summer, which mirrors national opinion polls. However, I don\’t think they\’re particularly useful as a predictor of national vote shares - though they can be useful at predicting how the vote will go in a particular locality.
One can say with some certainty that the Lib Dems will do well in Hornsey and Wood Green for example. Two recent poor performances in by-elections in Leicester South, OTOH, suggest they will struggle to hold that seat.
I think it was more of a general Cornwall thing as he kicked off in South East Cornwall which is also fairly tight Lib Dem-Tory.
Howard did in fact refer to local by elections in response to questions about why the Conservatives weren\’t doing better. Paxman\’s response was to give him a look which said about the same as Mike\’s article.
I thought Paxo was on hilarious form, clearly having a miserable time in the rain amid a group of third division local journos and enjoying a bit of random verbal violence on Howard, Tory activists and the general public.
\”What do you mean you don\’t know who Michael Howard is?!? He\’s the leader of the Conservatives!!\” Jeremy shouted at one bemused local. \”That\’ll be why then\” came the inspired response - get that man in Parliament right away!
Apparently he asked one member of the public 14 times, and he still didn\’t know who Howard was.
Funny you should mention it, but that did incident did come up in conversation between Paxo and Howard!
James is right - he was on absolute top form: \”He\’s the leader of your party and you don\’t know why you\’re here???\”
How long will it be before Paxo becomes enshrined as part of the British Consitution? He\’s the official media debunker-in-chief taking over from the late Vincent Hanna
Ah yes, the late great… remember his appearance in an episode of Blackadder III?
He and Andrew Rawnsley managed to be astute, serious and also very funny on \”A Week In Politics\” on C4. I think it was Rawnsley who really riled Charles Kennedy by referring to the (even then, former) Social and Liberal Democrats as the \”Salads\”.
I know this isn\’t really the place for it, but my favourite moment from last night was the anuguished incredulity in Paxo\’s voice as he screamed, \”YOUR DOG\’S A CONSERVATIVE??!?!?\” at a rather posh activist lady of a certain age who had decided to deflect his questions by pointing out that it was her pooch and not her sporting a blue balloon. It is a great shame he doesn\’t get out on the road more often.
Mention of Vincent Hanna is a great blast from the past. There is absolutely nothing today to hold a candle to \”A Week In Politics\”. When you compare the smug banter of Andrew Neil with Portillo and Abbott and the self-conciously \”whacky\” Mark Mardell on whatever their prog is called with the classic Hanna/Rawnsley double act you almost want to cry! It\’s like comparing the Chuckle Brothers with Laurel and Hardy.
So there are other people out there who enjoyed A Week in Politics as much as I did! Good to know also that other people are irritated by Mark Mardell\’s rather forced whackiness (including the absolutely wince-inducing impersonations of different accents, including the worst impersonation of a Cockney accent since Dick van Dyke). If you wanted to start a new version of A Week in Politics, how about Andrew Rawnsley and Michael Crick?
Paxman was, indeed, on top form last night (at times, it verged so much on self-parody that I was reminded of The Day Today).
Vincent Hannah is very sorely missed …..as much humour in real life as on tv….I once Mc d a folk evening at a Lib Assembly and he came along and sang….told us in advance he would allowing us to sell 20 tickets to his BBc colleagues who already knew he had a good singing voice….
Only saw him angry once and that was when an idiot friend mentioned to him a story that had been in Private Eye suggesting he might have enjoyed the overnight company of a prominent(and still prominent)female Labour politician…..angry because he was very happily married and detested the said individual
not that I like to gossip or anything…
Mike - you have assumed that the next GE will be a 3 horse race and ignored the effect of \”others\”.\”Others\” depending who they are have an effect on the 3 main parties that is out of proportion to the number of votes they(\”others\”)receive.Certain \”others\” take the vast majority of their votes from only 1 of the 3 main parties.
Mike -\”We think this is pretty meaningless and we should be wary of those who seek to extrapolate large conclusions that they say apply to the nation as a whole.\”
You are wrong - the byelections results in each GE year got it right.
1979 Tory 45% Lab 38% LD 14%
1983 Tory 39% Lab 36% LD 20%
1987 Tory 38% Lab 32% LD 27%
1992 Tory 46% Lab 30% LD 20%
1997 Tory 31% Lab 30% LD17%
2001 Tory 32% Lab 41% LD 18%
They might not have got the right % but they got the winner.
Sorry 1997 Lab44%
There is a difference between seats won at local byelections and the percentage of the votes cast. The latter reflects the strength of the local organisation, while the former - especially in closely fought contests - is to some extent a matter of luck.
So I was interested to see the percentages calculated by Vino and to take Sean Fear´s figures a bit further.
For 2004 (so far) we have the following percentages (I think):
Tory 35.3%
Labour 30.2%
Lib Dem 26.8%
Applying these percentages to the Martin Baxter Uniform National Swing model, we have
Con 269
Lab 278
Lib Dem 68
Others 7.7%
(which probably under-rates the seats to be won by Lib Dems, because it over-rates the seats won by \”Others\”)
But Vino´s argument is that this calculation at least gets the parties in the right order (except that these are 2004 results and not 2005 ones).
And then again the National Uniform Swing model fails to take into account local strengths and weaknesses.
So its rather difficult to forecast, innit?
Vino - to point 14 - if as you say the local election results showed the General Election winner then the current forecast would be for a Tory victory. I remember before the 1997 election people pointing to the then Tory strength in local by-elections as a reason not to believe Labour\’s big poll leads. They were wrong.
But were they talking in 1997, Mike, about percentages or seats? I cannot for the life of me remember…..
I think they wre talking about votes.
Back in 1997, the Conservatives were at least 10% behind Labour in local by-elections - only because the seats that were coming up had last been fought in 1994, 1995 and 1996 when they were even further behind, were the Conservatives gaining seats.
Greetings night owls….dont you guys sleep?
Personally I would like to think that local by-elections are an accurate guide to the next general election….thats because the only one that Plaid has contesed in Wales since June we got 70% of the vote…..
Vino - a good point, and one that has been debated here before. The general consensus seems to be that UKIP will take votes form the Tories, except in the West Country where they will also hit the Lib Dems. Respect will take votes from Lib Dems and Labour. So potentially the two could cancel each other out, except in the West where LD/Con marginals mnight go blue.
Respect will take votes overwhelmingly from Labour…………mainly in England…in Wales they will make little impact and I dont know if they contest elections in Scotland. Given the disproportionate amount of people in the armed forces in Scotland and Wales the war will be as important as in communities with substantial muslim voters.
However not sure if you are including nationalists in your \”others\”
Mark - slap my anglo wrist! I\’d not considered the Nationalists as I was looking at the UK as a whole. And me born in Wales, with a Scots-born, Irish-parented wife too.
Steve
Yes well its over 100 seats we ware talking about….crucial to Labour if not the Tories….I have posted before about thow the election will be different in Wales…I assume the same is true for scotland….
While Respect will take voters who previously voted Labour, they will take voters whose primary aim is to protest about the war. Therefore they will take votes away from the Lib Dems in the context of the GE. In the few Lab/Lib marginals a strong Respect challenge is good for Labour.
So LDs don\’t want Nuff Respec\’ in da Area.
Something else I forgot to mention was the potential role of the BNP - they may have an effect in the Penines
Respect only got 1.5% in the European elections, so one should not exaggerate its impact. Outside the East End, and a few constituencies with large Muslim populations, I think its vote will be derisory.
I think the Green Party is likely to pull more votes away from Labour and the Lib Dems than Respect, and the Greens have more broadly based support.
Will the BNP pull votes from labour or the Tories? Presumably Labour as they are seen as a recipient for white working class protest votes, and these are more likely to come in Labour-represented urban areas.
There are a few marginal seats, like Keighley, Pendle, Calder Valley, where the BNP may affect the outcome. They take support from all parties (and usual non-voters) but probably more from Labour overall.
anecdotal eveidence is that around half the BNP vote comes from otherwise non-voters….
Think respect will do well in Leicester South and East……..bits of Birmingham…other than that around 2 % as Sean suggests….the serious players for others are Plaid SNP Greens and UKIP……
If Labour vote is down in Wales both Lib Dems and Plaid will increase if the patten that has occurred at most other elections is followed….
The thinking is that the BNP take their first votes from previous non-voters, then from the Tories and then from Labour. What often happens when Labour heartlands swing to the BNP is that a lot of people who have not voted for a long time vote.
There was a study a few months ago of BNP voters in Burnley, Stoke and Calderdale which indicated that of those who had previously voted for a party, the largest element had voted Labour (albeit there are many more Labour voters in Burnley and Stoke than voters for any other party).
Certainly, the BNP does motivate people who don\’t normally vote.
On local elections/local by elections, does anyone have a view on whether swings are more volatile than they used to be? It may very well just be me, but there seems to be a lot of swings going all over the place (10% Con to Lib Dem here, 15% Lib Dem to Con there and 3% Con to Labour somewhere else). In that context, surely some of these very small monthly and quarterly samples (and quite a lot have to be ruled out as non-comparable) are massively skewed by a couple of \”big\” results in large wards/divisions which may be down to very localised factors?
Again, is it also just my impression or is there evidence that the link between council results and general election results within a single area is a lot less strong today? Thinking back to 2001, I believe that the Lib Dems won one of about ten County Council seats in North Norfolk on the same day they won the General Election, and the opposite occurred in Totnes (something like that anyway). I am sure there are many other examples.
I would like to think people are getting better at distinguishing between the national and local (and European) scene but this is all impression and I wonder whether anyone has figures to confirm or refute it.
My impression is that that was true in the 80s when Labour controlled many councils but quite obviously wasn\’t doing well at national level.
I believe that one of the indicators as to whether a seat is winnable is the local govenment presence in that seat (Mr Fear uses this link often), but given the difference in turnout between local elections and national elections its hard to make links I would postulate. For example, a significant tory presence in local government may be just down to the fact that its supporters are more likely to turn out in local elections (hopefully! :))
I think local elections always have produced massive swings one way or the other. Given the small size of the electorate in many wards, a 15% swing may just mean that 100 voters have switched from one party to another.
Your point about one or two huge swings distorting the whole picture is quite correct. That\’s why vote shares in local by-elections are not a good indicator of each party\’s share of the vote nationally, although they can indicate certain trends.
I agree that more people are splitting their votes. In 2001, the Conservatives won one Parliamentary seat, yet 121 seats in County Council elections held on the same day.
The alternative explanation is that there is a large body of basically Tory support who cannot summon the enthusiasm to vote Tory at national level (or vote Blair), but are far more motivated to provide their support in local elections. It is one reason I think people should be a bit wary of assuming (like in the US!) that larger turnouts are necessarily bad for the Tories, considering the low historical base (59%) from which we start. There must be a large amount of truth in the theory that the main feature of the 1997 election was a large number of moderate Tory voters simply staying at home.
Alex - I find this hard to believe given the level of turnout in local elections is below 40% and the GE was 60% *at its lowest*
Also have to remember that this years local elections coincided with the Euros…..there was a large turnout from anti-euro people in the local elections…
It seems clear to me that Labour are unlikely to lose the GE outright and events in Jan March will shape the size of their majority…..Iraq, the Hunting ban and subsequent campaign….Council tax bills(we have just had rebanding in Wales) all will have an ipact…the boost Labour will get from the Queens speech will wear off after christmas….
Vino, unfortunately the local election results for the year before General Elections dont seem to work so well as a predictor.
1996 - Lab 43% Con 29% LD 24%
2000 - Lab 30% Con 38% LD 26%
(Those figures are from a parliamentary paper and are adjusted to reflect which council areas had/did not have elections)
I think that Alex - 37 - is right in saying that people should be a bit wary of assuming (like in the US!) that larger turnouts are necessarily bad for the Tories. On General Election day in 2001 I lost a bit of money after I mistakenly interpreted the very low turnout that was apparent by late afternoon as being good for the Tories.
Back to Paxo: evidently the Tories shared my views about Monday\’s Newsnight. Great television though! http://news.bbc.co.uk/1/hi/uk_politics/4045203.stm