Make elections work for you: check the polls, but follow the money

Tags

, , , , , ,

Chris Game

Spoiler alert: this is a blog about elections, but not local elections – mainly because it’s about election betting, and, with one conspicuous exception, which will be mentioned, our modern-day local election contests and candidates are rarely of sufficiently general interest to attract much serious fixed odds betting.

My prompt was the Conservatives’ recourse to their apparently hastily conceived Campaign Plan C – following the failures of A: negative, personal and increasingly counter-productive attacks on Ed Miliband; and B: daily, unexplained and increasingly implausible financial treats for everyone from NHS patients and rail users to volunteers and better-off housing association tenants.

Plan C involves drawing on – or alarmingly, in the latter case, ‘weaponising’ – the proven, if contrasting, electioneering skills of famous grey man, soapbox campaigner and former PM, John Major, and safe seat candidate and London Mayor, Boris Johnson.

Let’s start with Johnson, the opening paragraph’s ‘conspicuous exception’. You can currently get odds of 33/1 both on his being the next London mayor and next Deputy Prime Minister, and a very short 1/50 on his becoming MP for Uxbridge and South Ruislip – which in itself gives a small hint of the huge growth in recent years of so-called novelty betting in general and political betting in particular.

We wagered over £1 million on Prince George’s arrival in 2013 and could well top that this time. Current shortest odds are on a blonde female, eventually named Alice, born on April 25, with 50/1 on triplets and 1000/1 on the hapless infant being named Boris. But royal births are peanuts compared to political betting. Bookies reckon we laid out £30 million on the 2010 election, £50 million on the Scottish independence referendum, and predict this election will be the first UK non-sports betting event to break £100 million.

With William Hill’s taking bets online of £200,000 and over the counter of £50,000 on a hung parliament – at 2/9 and 1/8 respectively – it suggests some clearly see it as an easy earner, and they could be right. The odds on these big bets may be short – £50,000 at 1/8 would win just £1 for every £8 staked, or £6,250 plus the returned stake. But that doesn’t necessarily make them unattractive compared, say, to the 8/1 (£8 for £1 staked) you could get on an improbable coalition involving the Scottish Nationalists.

The attractiveness of particularly these short odds bets obviously depends on whether you think the bookies can predict the results of election races as skilfully as they can horse races and football matches. Put another way, and the main topic of this blog: who are generally more reliable – pollsters or punters?

Step up, John Major, the country’s most electorally successful living Conservative, thanks to his historic triumph in the 1992 General Election. Averaged out, the then four main final polls put Labour ahead on 39%, the Conservatives on 38%, and projected a comprehensively hung parliament. Next day, Labour managed just 35%, while the Conservatives won nearly 43% and a Commons majority of 21 seats. Major became the only UK party leader ever to win 14 million votes – nearly a third more than Cameron in 2010 – a hung parliament was postponed for another 18 years, and ‘shy Tories’ had arrived as a pollster’s nightmare. For a young and still mistrusted polling industry it was a humiliating setback.

It has, as the current campaign daily demonstrates, recovered, grown, and evolved methodologically almost beyond recognition. At the same time, particularly with the proliferation of smaller parties, both polling and seat prediction have become considerably more hazardous. All political pollsters, however, are parts of large commercial companies. Screw up, and their other clients immediately know, so generally they’re highly rigorous and pretty good – provided you judge them reasonably.

They’re not predictors or forecasters. They take time-specific opinion snapshots, with different interrogatory cameras – some using online panels, some random digit phone dialling – of what they hope are politically as well as demographically representative samples of the whole electorate. But because they’re samples, mostly of between 1,000 and 2,000, not much more than 19 times in 20 will any single response be within 3 per cent (plus or minus) of what it would be, had the whole population been surveyed.

This means two things: first, roughly every 20th finding or poll will be outside that +/- 3 per cent margin of error; second, that ‘rogue poll’ will invariably attract more media attention than the rest put together.

That’s almost certainly what happened in the last fortnight of the Scottish referendum campaign. Of nearly 40 polls published between June and the September 18 polling day, only two put the Yes vote ahead. The coverage given to particularly the first of these polls was enough, though, to prompt all three major party leaders to panic in concert, rush up to Scotland, and make desperate vows and commitments they’re still regretting.

It doesn’t, though, explain why the averaged five final polls put the No vote on 49.2%, with a lead of just 4.2% – when the actual result was 55.3% to 44.7%, and a No lead of 10.6%. Poor methodology, very late change, shy No voters, whatever – the pollsters got it wrong.

And they got it as badly wrong – or, in fairness, their Israeli counterparts did – in this year’s perhaps most publicised elections: those in March to the Israeli Knesset, called early by Prime Minister and Likud Party leader, Benjamin Netanyahu.  The averaged seat projections of the final seven polls published before Israel’s five-day pre-election poll ban gave Likud 22 seats and its main opposition, Isaac Herzog’s two-party Zionist Union 26 – prompting newspaper headlines like the Daily Telegraph’s ‘Bye-bye for Bibi?’.

The three exit polls suggested that the four-point gap had been closed in the final few days’ campaigning, but none came anywhere near the actual result of a six-seat lead (30-24) to Likud/Netanyahu. In what is a 120-seat legislature, with 10 parties having at least five seats, forming a government is tricky, but, in these matters anyway, Israelis are more patient than we are, and the permitted 42 days have been extended to allow Netanyahu until May 6 to name his.

Those ‘Bye Bye, Bibi’ headlines, however, weren’t the only ones. Gambling sites and the more refined ‘prediction markets’ were giving Netanyahu “an 81% chance of being re-elected”, and offering the equivalent of fractional odds of 1/6 on his winning and 4/1on losing.

It could have been a replay of the Scottish referendum. Over that final ten days, while English politicians and pollsters were over-reacting, the betting odds, overwhelmingly predicting No throughout the campaign, hardly wobbled: around 7/2 against Yes, and 1/4 for No. Indeed, one firm paid out a six-figure sum on a No bet three days before polling day.

These cases of the betting industry having a better sense (no apologies; pun deliberate!) of what’s actually happening aren’t the exceptions that prove the rule; they are the rule – a rule, moreover, that’s logically to be expected. Pollsters ask about our voting intentions and opinions, whereas bookies and bettors focus only on results and outcomes. Above all, though, they back their judgement with their money. So watch the polls carefully, as the bookies do, but if in doubt, then, as Americans might say, follow the frogskins (greenbacks for the alliteratively inclined).

At the time of writing – Wednesday 22nd – there have been 11 new polls since last Thursday’s BBC1Opposition Leaders’ debate. Five put the Conservatives ahead in percentage votes, five Labour, and one had them tied on 34 per cent.  There have also been seven poll-based seat forecasts: five showing the Conservatives ahead, two Labour.

No division among the bookies, though. The best seat-number odds being offered on Labour by any of Oddschecker’s 24 bookmakers were 21/10, while the best on the Conservatives were 1/2. Next PM, though, is very different: Miliband was 3/4 and shortening; Cameron 11/8 and drifting – rather like his party’s campaign.

Chris Game - pic

Chris Game is a Visiting Lecturer at INLOGOV interested in the politics of local government; local elections, electoral reform and other electoral behaviour; party politics; political leadership and management; member-officer relations; central-local relations; use of consumer and opinion research in local government; the modernisation agenda and the implementation of executive local government.

Follow

Get every new post delivered to your Inbox.

Join 2,518 other followers