Don't trust the French polls

 

Segolene Royal went on television, chatted with a few housewives, hugged someone in a wheelchair, and a day later has reclaimed pole position in the first-round polls. What does this tell us?

 

It tells that this election is wide open, and that the polls are currently highly sensitive to insignificant and symbolic events. This is, sort of, an endogenous process where the hypothesis of a candidate doing well creates meia headlines of a candidate doing well, which in turns creates the reality of a candidate doing well – and vice versa. Such endogenous processes matter especially during the early stages of campaigns, but usually do not last forever (remember Howard Dean in the last US presidential election). Royal has finally turned an important corner. She could not afforded another month of bad publicity, gaffes, resignations from senior staff. She needed some good news, and this was it. But whether Sarkozy or Royal win this election in two months time, is ultimately not decided by a TV chat show, but an informed electorate. And the numbers are telling us that the electorate has yet to make up its mind.

 

There is one baffling statistic that one should keep in mind when considering the two candidates. At this stage in the campaign, Royal has managed to persuade fewer Socialists to back her than Lionel Jospin did during his presidential bid in 2002, when he got knocked out in the first round. What appears to be a solid united left-of-centre vote in favour of Royal, is in fact much more fragmented. The same goes for Sarkozy and the Gaullists. So if you see a second-round poll that say Sarkozy 51 – Royal 49, just beware that we suffering what is known in statistics as selection bias. Even though pollsters select pseudo-random groups sample, the basis for the published results are only those members of the sample who have made up their mind. But this subselection is no longer random. They are the party loyalists, and the policy wonks. So this is an election where the candidates have still everything to play for, and given the nature of the candidates, and we may still be in that precarious position a week or so before the first round of the elections themselves.

 

There is another baffling statistic, concerning the UDF candidate Francois Bayrou. While he trails both Sarkozy and Royal in the first round polls, despite advancing 4 percentage points in the last poll, he is leading both of them in a second-round contest. In other words, if faced with a choice between Bayrou and Sarkozy, or between Bayrou and Royal, the French would in both cases prefer Bayrou. But if faced with a choice between all three candidates at the same time, the French currently rank Royal ahead of Sarkozy, and Sarkozy ahead of Bayrou. Now again, remember there is a problem of selection bias, and we should remind ourselves that Bayrou is presently the darling of the press, something which may not last once people take a closer at this programme (and discover that he has not got one yet).

 

But if all these polls are even remotely correct, this could promise to become a most interesting campaign. If Royal’s campaign were to collapse – something we can never exclude – Bayrou may turn to be an extremely dangerous challenger to Sarkozy. Sarkozy will manage to unite the centre-right only if confronted by a tax-and-spend Socialist, such as Royal, but not by a consensual liberal such as Bayrou. The UDF man is presently drawing much of his support from the disgruntled left, from those Socialists who hate Ms Royal so much that they would vote for anybody but her. Just image what could happen if Bayrou managed to crack open the right in the way he cracked open the left. Just as there is an anybody-but-Royal vote to be had on the left, there is an anybody-but-Sarkozy vote to be had on the right. 

 

It is for this reason that we should no longer simply look at this election as a two-horse race, but as a three-horse race (I would ignore Le Pen this time). For the moment, we have to treat Bayrou as a serious challenger. While in my view none of these candidates will solve the country’s economic and political problems (see also my last FT column), as far as an election contest is concerned, this is very entertaining.

---