The Great Expectations Game

by Stephen Tall on June 6, 2009

Earlier this week, I blogged about the ‘expectations game’, the way in which post-election analysis can be spun, and speculated that the Lib Dems were being set up for a fall. In fact, though, the reporting of the party’s performance has been generally fair. For instance, here’s Tony Travers in today’s Guardian:

The national equivalent vote share put the Conservatives on 38%, the Liberal Dems on 28%, Labour on 23% and “others” on 11%. Compared with 2008, the Tories are down five points, Labour down one point, the Lib Dems up one point and Others up by 5 points. In short, both Labour and the Conservatives have taken a political hit compared with their position last year, while other parties, notably the Lib Dems, have generally gained.

Or Gary O’Donoghue for the BBC:

Taking Bristol is a major plus for the Lib Dems, particularly as the city has three Labour MPs and the party’s whole strategy has been aimed at hoovering up in the wake of Labour weakness. But the party will be concerned about the Tory gains in the South West, and will hope that a projected national share of 28% will bolster numbers elsewhere.

Unfortunately, though, the usually indispensable and impartial PoliticsHome.com seems resolutely determined to undermine its reputation, with a bizarrely one-sided analysis which concludes that the biggest disappointment of these elections was … the Lib Dems’ performance. Apparently it was ‘much worse than expected’, while Labour’s dismal showing qualified only as ‘worse than expected’.

Yes, you read that right: PoliticsHome reckons that Labour, who have won 23% of the vote and 176 councillors to the Lib Dems’ 28% of the vote and 473 councillors, performed less disappointingly than the Lib Dems. Go figure.

Their evidence for this dodgy conclusion? Well, the website asked its Phi100 panel of ‘influential insiders’ (declaration of interest: I’m a member) for their assessment of success criteria for the parties. The panel reckoned the Lib Dems should poll between 0% and 5% ahead of Labour, and would make net gains of 25-55 seats. The reality is the Lib Dems polled at the top end of the panel’s expectations, 5% ahead of Labour, but are currently showing a net loss of 4 councillors compared with 2005, when the seats were last contested.

Interestingly, the Tories’ 38% projected share of the vote was at the very bottom end of the panel’s expectations – a 6% drop from 2008 – though the fact the Tories made big overall gains suggests their vote is becoming much more evenly spread and better targeted than in former years, when the party was stacking up votes in safe seats.

Now, as I’ve already stated today, there were undoubted disappointments for the Lib Dems in these elections, chiefly in Devon and Cornwall, to a less extent in Somerset. But overall, measured by projected vote share, these were the most successful local election results for the Lib Dems in our history. If scoring 28% of the projected national share of the vote counts as ‘much worse than expected’, can I order the same for next year, please?

It is perhaps predictable that the usually sane Tim Mongomerie on the wholly partisan ConservativeHome.com should try and spin this as ‘the decline of the Liberal Democrats’ (they’re welcome to their delusion). But it is surprising that PoliticsHome.com – which, despite the fact that it is funded by ConservativeHome’s main investor Stephan Shakespeare, is normally an entirely impartial source of news – has allowed itself to torpedo its own reputation for balance.

Update: PoliticsHome’s managing director Freddie Sayers has emailed me with this reponse, which I’m happy to publish here:

Without doubt, Labour’s result was disastrous. But as we say above, this is not an analysis of the positions of the parties – purely how the results fared against expectations. And the only real surprise was that the Lib Dems lost rather than gained.

We prioritise the seat numbers over the vote share because the vote shares are much smaller ranges – they were individual choices for the panel, not the range between a lowball and highball prediction. In the case of both the Lib Dems and the Tories the predicted range was correct, but it is less remarkable.

Enjoy reading this? Please like and share:

No comments

Are they not just reflecting the panel’s views? Sometimes the panel has views favourable to the LibDems does it not?

by Paul Walter on June 6, 2009 at 2:01 pm. Reply #

“Most successful local election results for the Lib Dems in our history.”

Well I thought we got 30% in 2004 actually, but leaving that to one side, I hope the successes of yesterday are built on – (they certainly should be in four years time if there is a Tory government!) but I also hope that the reasons for our failures in the south-west are analysed honestly and not just spun away or swept under the carpet.

by jim on June 6, 2009 at 2:11 pm. Reply #

A very succinct summary, I think. I too read the bizarre conclusions drawn on ConservativeHome.com, and wondered whether they were starting to be more afraid of the Lib Dems. Simon Hughes on Radio 4, (I think it was him!) put across a great argument about the decline of Labour which was, as expected, refuted by Diane Abbott, but I feel that both Labour and the Tories have increased their derogatory comments about the lib dems recently as they are seeing the pattern that is emerging across the country. Bring it on, I say!

by Rachel Smith on June 6, 2009 at 2:15 pm. Reply #

One could alternatively conclude that the “insiders'” expectations of vote share were wildly inconsistent with their expectations of seats, and that therefore they hadn’t got much of a clue what they were talking about.

by Herbert Brown on June 6, 2009 at 2:17 pm. Reply #

These results are good news for the Tories because it means they have learnt how to fight the Lib Dems and win back seats. It’s potentially bad news for us because despite our good vote share we seem to have reached a critical mass and are chasing seats around the country – winning from Labour only to lose to the Tories in other areas. The good news is we’re not being crushed by the dreaded two party squeeze but in the long term this isn’t good enough if we’re determined to overtake Labour in the Commons.

Overall good – but we have a lot of work to do to translate the growing goodwill of the country into tangible seats and real influence.

by Letterman on June 6, 2009 at 2:18 pm. Reply #

Looking at results in various places I have an interest in, it occurs to me that Gordon Brown may have been laying a Machiavellian trap for the Tories in the long term by holding the County elections on the same day as the Euros. The result has been that UKIP have done very well at a local level for the first time (I know they only won 6 seats, but their share of the vote in many seats was impressive). Building a political party from the ground up is a long, hard slog, but getting a good vote at elections is an encouragement to the activists and a catalyst for more activity. When Cameron’s chickens start coming home to roost in three or four years’ time there will be a larger, stronger alternative for the disaffected right wingers in the Conservative Party to support as a result of Thursday’s election.

by tony hill on June 6, 2009 at 3:31 pm. Reply #

er…from yesterday’s 34 councils you got 1…sure, Tory performance (+7) was not entirely fabulous (38% share as you spoint out)…but 1 council is not ‘go back to your constituencies and prepare for Government’ is it? The map doesn’t look very yellow this morning and giving the beating the Tories got over expenses I think the point ConHome was making – a fair one – was that the Lib Dems might have been expected to do better?

by Domesday on June 6, 2009 at 3:43 pm. Reply #

I, too, find the results disappointing. The obvious question is this: If the voters wanted to give Labour a kicking, why did they shun the trouser-press-and-HobNob Lib Dems and instead embrace the flipping moat-and-duck-island Conservatives?

by Paul Griffiths on June 6, 2009 at 4:48 pm. Reply #

Tories traditionally do well in shire elections. And most local elections were in the shires. Domesday, you should look at what councils weren’t up for election i.e. a lot of cities in the north, London and Scotland and Wales. These are areas where Tories do poorly generally compared to LDs and Labour, but since there’s a Labour backlash, expect LDs to make big gains in those areas come general election time.

by Alex on June 6, 2009 at 6:55 pm. Reply #

Aye, the Tories did incredibly well in these elections, overall, last time. In 2005, when they were the same day as the General Election.

In Devon (because I’ve spent a chunk of time today going through the result), the BBC Politics Show report said the Lib Dems were defending a large number of very marginal seats that they’d be very lucky to hold on to. Um, they lost those seats, plus one to the Greens in a weird vote split in Totnes town.

But having looked at it, it’s fairly clear that this time the result is skewed not by General Election voting patterns, but by EU election voting patterns—a chunk of our vote in one formerly held ward went to UKIP, giving the Tories the seat with little gain in votes from what I can see.

I really dislike the way the national media completely ignores marginal seats, swing and other issues for locals, but concentrates on them heavily for nationals.

A large number of seats in Devon, as in Torbay, are tight marginals, they were before, they are now.

These were shire counties, where the Conservatives should’ve been, given their national rankings, getting closer to 50%. They didn’t manage 40%. That’s really bad for them.

I’m not complacent about the way we beat Labour—shire counties are bad for them across the board. I’m really happy for Burnley, Hywel’s team deserve a lot of praise for that lot.

To say we did exceptionally well isn’t true, but the results were good, better than I’d expected in the SW all things considered. A pretty good vote to hold the seats we’ve got there, and possibly challenge in Totnes still.

Cameron’s crowing, but 38% in the shires is terrible. Truly terrible. Sure, they gained seats, but if they can’t get above 40% in what should be their heartlands, how are they going to win the GE?

by MatGB on June 6, 2009 at 10:17 pm. Reply #

“Cameron’s crowing, but 38% in the shires is terrible. Truly terrible. Sure, they gained seats, but if they can’t get above 40% in what should be their heartlands, how are they going to win the GE?”

?

I’m sure 38% isn’t the raw percentage of votes cast on Thursday, but a projection of national support, taking account of which councils were up for election.

Though, even taking that into account, 38% is scarcely a strong performance by the Tories.

by Herbert Brown on June 6, 2009 at 11:38 pm. Reply #

38% is not a strong performance and Cameron is not crowing (show me where he is). 38% will not win a GE. The Tories are hoping, and it’s a sensible hope looked at objectively, that the significant turnout for UKIP will revert to the Tories for the GE in order to get Labour out.

In the context of these elections (expenses, simultaneous Euro election and a county election where in most counties the Tories were unassailable), then 38% is not bad but it is way overstating the case to say it’s truly terrible. 43% delivers a GE win: the difference of that 5% is not truly terrible for the Tories.

by Domesday on June 7, 2009 at 7:21 am. Reply #

The quality of much of the analysis (in press coverage and BBC coverage of the locals) has been very poor. Lib Dem Voice readers might want to take a look at Rallings and Thraser at http://tinyurl.com/lxyd7z. I don’t claim it is comprehensive or truly balanced … but it really is much better informed and reasoned.

by Ed Randall on June 7, 2009 at 8:56 am. Reply #

For a Party whose only medium term chance of power is in local Government, it’s amazing how blase Liberal Democrats are about losing councils which they hold. I would have thought that the actual results are far more important to LibDems than “projected shares” but there we are.

It should be pointed out that in 2005 the LibDems got 27-28% “projected national share” in these elections. On the same day they got an ACTUAL national share of, er, 23%. Something to mull over before ascribing some great national significance to the projected results.

When you consider that LibDems would expect to gain relatively from the lower turnout in local elections (and from the presumed problems that Conservatives have from UKIP) and they really didn’t do very well.

Has anyone calculated the ACTUAL changes in share of the vote from 2005?

by greg on June 7, 2009 at 8:59 am. Reply #

“It should be pointed out that in 2005 the LibDems got 27-28% “projected national share” in these elections.”

I don’t think we did. I’ve not seen any calculations for a projected vote share in 2005 and AFAIK no-one bothers when there is a combined general/local.

The 28% figure from 2005 is the actual vote share on the day.

by Hywel on June 7, 2009 at 9:19 am. Reply #

“It should be pointed out that in 2005 the LibDems got 27-28% “projected national share” in these elections. On the same day they got an ACTUAL national share of, er, 23%. Something to mull over before ascribing some great national significance to the projected results.”

It’s well known that people vote differently in different elections, and that the Lib Dems normally do better in locals than in nationals.

This projection, as I understand it, is an attempt to compensate for the fact that elections weren’t held across the whole country. But not to compensate for the other differences between local and other elections.

Even so, if the projection is done sensibly, it should be possible to compare levels of support from year to year, and in particular to look at the changes since 2005.

by Herbert Brown on June 7, 2009 at 9:24 am. Reply #

Hywel:
“I don’t think we did. I’ve not seen any calculations for a projected vote share in 2005 and AFAIK no-one bothers when there is a combined general/local.

The 28% figure from 2005 is the actual vote share on the day.”

Admittedly nothing is made very clear about these comparisons, but I don’t think that’s correct.

For example, in an article in the Independent analysing the results, John Curtice (who made the projection and is professor of politics at Strathclyde) says:
“True, at 38 per cent, the [Conservative] party’s projected national share of the vote was as much as seven points above the equivalent figure for 2005 …”

That figure of 31% is the same one that has been quoted elsewhere in the comparisons with 2005.

It would be incredibly stupid to go to the trouble of making a national projection for one year, and then compare it with raw figures from another, which would be very unrepresentative because of the omission of the metropolitan authorities. I hope academic standards haven’t fallen quite that far.

by Herbert Brown on June 7, 2009 at 9:39 am. Reply #

Maybe but:

1)if the PNVS in 2005 was 28% and on the same day the actual vote share was 23% then that suggests the formula/model is deeply flawed!

2) The HoC Library paper has a table of PNVS figures for each year (from Rallings & Thrasher) and uses the actual vote share for 1997, 2001 & 2005.

3) R&T elections summary pages report that the actual share of votes case in 2005 was 28%

by Hywel on June 7, 2009 at 9:49 am. Reply #

Hywel

(1) None of these projections is exact, but I can easily believe 5% more people vote Lib Dems in local than do in general elections.

(2) The projections given by the BBC are by John Curtice, not Rallings and Thrasher, so I don’t see the relevance of that.

(3) And do they say the Tory and Labour shares were 31% and 33%?

Incidentally, it’s interesting to see what Rallings and Thrasher say in that article for which Ed Randall provided a link:
“In an exclusive analysis for The Sunday Times, we have examined the votes cast by more than 4.5m voters in more than 1,200 local divisions and wards contested by at least the three major parties in order to calculate how they would have fared if these elections had taken place in every part of the country.

This analysis has the Tories on a national equivalent vote of 35% (a four-point increase since the local elections in 2005). This is below the symbolic 40% mark, but still with a comfortable lead over Labour who are at a historic low of 22% (down 12 points). The Lib Dems scored 25%, down two points from 2005.”

Surely they are comparing like with like? I note that they are comparing with a Lib Dem figure of 27% for 2005, which is slightly lower than the raw figure you quote.

by Herbert Brown on June 7, 2009 at 10:08 am. Reply #

My understanding of PNVS was that it extraplolated local election results into a national vote share at a general election.

If it is actually a figure of the national vote share in local elections if every local authority was voting then any comparison with the 2005 23% figure is completely bogus. But in that case why bother with the calculation? Why not just compare actual votes cast from 2005-9?

I’ve not read Ed’s link – but my experience is that R&T use a much bigger sample than the BBC

by Hywel on June 7, 2009 at 10:25 am. Reply #

Hywel

I repeat the quotation from the article by Rallings and Thrasher:
” … we have examined the votes cast by more than 4.5m voters in more than 1,200 local divisions and wards contested by at least the three major parties in order to calculate how they would have fared if these elections had taken place in every part of the country.”

So they are not trying to correct for differences between voting in local and general elections – only for the fact that the local elections didn’t cover the whole country. As I understand it, the same is true of John Curtice’s projections.

by Herbert Brown on June 7, 2009 at 10:51 am. Reply #

By the way, I checked the Rallings and Thrasher page showing the raw votes in 2005:
http://www.research.plymouth.ac.uk/elections/elections/2005%20results.htm

Hywel pointed out that the raw Lib Dem vote then was 28% – suggesting that the 2005 figures that have been used for comparison with the 2009 projection by the BBC – CON 31%, LAB 33%, LD 28% – were the raw votes rather than projected ones.

But according to Rallings and Thrasher the raw votes in 2005 were CON 40%, LAB 25%, LD 28%. So evidently the fact that the Lib Dem percentage was 28% in both raw and projected was just a statistical fluke.

Clearly the 2005 figures used for comparison are indeed a projection, not the raw percentages.

by Herbert Brown on June 7, 2009 at 11:17 am. Reply #

Mark

On the question of what to compare with, the answer to Hywel’s “Why not just compare actual votes cast from 2005-9?” is obvious – the actual votes cast will fluctuate from year to year according to the political complexion of the areas that happen to have local elections.

But if the projection method is valid, then obviously one can make a valid comparison with any previous year.

I can see that comparing with last year has the virtue of producing a favourable headline for the Lib Dems (!).

But really if you are trying to work out what will happen in a general election, the obvious comparison to make is with 2005. And of course that comparison is not so reliant on the validity of the projection methos, because the same local authorities were voting then.

by Herbert Brown on June 7, 2009 at 11:27 am. Reply #

“the actual votes cast will fluctuate from year to year according to the political complexion of the areas that happen to have local elections.”

By and large they are the same now as in 2005 (eg all of Cornwall votes – just for a differen authority)

You could adjust for that by just comparing areas that voted in both years.

Yes the political dynamic is different – but that is the flaw in using local elections to project nationally.

by Hywel on June 7, 2009 at 12:11 pm. Reply #

‘Steady as she goes’. This is what I read from the local results. Yes we could have done better, yes we could have swept away by the Tories.The Euros always seem to produce unusual results some of which would have rubbed off on the local elections on the same day.
Looking long term we have firm foundations to advance from – things are coming together with Clegg & Cable etc

If next year the Tories win the GE & Labour are destroyed – we will have opportunities to advance on the Tories again – but for now – lets go for Labour – hard.

by Greenfield on June 7, 2009 at 12:27 pm. Reply #

Hywel

Perhaps I misunderstood what you meant by comparing “actual votes cast from 2005-9?”

I agree there’s no great problem in comparing 2005 directly with 2009, but it’s the intervening years you can’t meaningfully compare with 2009 in this way (including Mark’s favourite comparison with last year!).

And of course comparing locals with nationals in problematic anyway. But without this kind of projection it’s worse than that – it’s meaningless. Witness the difference between the Tories’ 15% lead in raw local votes in 2005 and Labour’s two-point lead in the projected version.

by Herbert Brown on June 7, 2009 at 12:37 pm. Reply #

Mark

Surely, if you want to know how things have changed since last year, you compare with last year, and if you want to know how things have changed since the last general election, you compare with 2005.

There’s no “right” year to compare with. The comparisons with different years are just giving you different information.

by Herbert Brown on June 7, 2009 at 4:35 pm. Reply #

Leave your comment

Required.

Required. Not published.

If you have one.