The curse of the polls

Richard de Jongh

Two weeks before that momentous Wednesday 22 November, PVV leader Geert Wilders was still at about 10 percent of the vote. A “normal” percentage for his party. It’s the unhappy electorate he can -you might say- firmly count on and which has always accounted for approximately fifteen seats. A substantial number, but fairly unsignificant because the PVV has always been on the political sideline. Fifteen seats is too few to make a strong impact on government policy.

And then…

… November came and suddenly Wilders moved up to 25 percent in the polls. How did that happen? One can think of a political reason: the VVD had lifted Wilders’ quarantine, in a way, by suggesting that they were open to discuss government participation. There is also a campaign-related reason: in his television appearances, Wilders was on a roll and outperformed his political competitors with a mixture of charm, firmness and good will.

But was that all? Don’t polls and repeated polls and journalistic attention to those polls have an autonomous and self-reinforcing effect? Isn’t there something like a bandwagon effect: the brass band passes by on Main street, the happy crowd follows the music and all those followers automatically attract more followers, until a procession forms that suddenly turns out to be good for not fifteen, but 37 seats in the House of Representatives?

An adjacent effect can be added to this. Only a limited number of party leaders are invited to the major electoral debates on radio and television, for the sake of content, because it’s quite tricky having twenty politicians debating at the same time. Who are invited? You’d probably say: representatives of the five or six largest political groups in the House of Representatives. That’s a more or less objective benchmark, but in the world of television, other rules apply: those of the polls. Anticipating the voter’s vote, the outcome of the election, party leaders are invited who are highly ranked in the polls. An example, and no more than that: in recent years, D66 has been the second largest political group in the House of Representatives. The polls showed a dramatic decline of the party. Based on those polls, the party was excluded from the major debates. Wouldn’t that have a self-reinforcing effect?

The Election Law does not regulate polls. However, there is a gentleman’s agreement that polling companies do not release poll results on election day, during polling station opening hours. In civilized countries around us, such as France, polls may not be published up to a week before the election. Spain, Portugal and Poland also have restrictions. Apparently, there is wide consensus on the assumption that polls have an autonomous effect on what voters favor.

I don’t think you need to be a statistician or an actuary to reasonably assume that poll results reinforce the results of subsequent polls, which in their turn have an accelerating effect, both ways, negative and positive, on the outcome. That’s how, to his own surprise, Wilders suddenly ended up with 37 seats.

It’s not about Geert Wilders, but about the question whether, at important moments in a democracy, we should allow such external effects without any restrictions. It really surprises me that there is no sign of a public debate about that.

Meer columns