How accurate are the Dutch election polls?
The Dutch Tweede Kamer elections are two weeks away. Based on the most recent polls, our previous publication with three scenarios for the likely result is still current. For our assessment of the election outcome, we use the weighted average of six polls, as compiled by Tom Louwerse from Leiden University in his ‘Peilingwijzer’. How well does this aggregate perform?
Figure 1 shows the margin of error for the ‘Peilingwijzer’ published the day before the 2012 elections. It immediately becomes clear that polls certainly have to be interpreted with caution: in 2012 voter shares of four of the five largest parties fell out of the upper and lower estimates. But polls were not that wrong: the maximum deviation, for the Socialist Party (SP), was 2.9 percentage points. That corresponds to about four out of the 150 seats in Dutch lower house.
Rather than pollsters having incorrectly gauged voter sentiment, the difference in the 2012 opinion polls can be partially explained by the fact that the night before election day there was an important television debate between candidates. This final debate could have persuaded some voters to turn their allegiance, while making decided voters out of floating voters.
Floating voters may have in fact played quite a big role in 2012. According to one polling agency, four weeks before the 2012 elections some 75 percent of respondents said they had no or only weak preferences. Indeed: in the weeks following that survey, the political landscape in the Netherlands dramatically changed (see figure 2). Following lackluster performances in public debates by its party leader, the Socialist Party’s (SP) lost most of its support in the run-up to election day, mostly to the benefit of the Social Democrats (PvdA).
So what about the upcoming elections?
Two weeks ago, the same polling agency estimated that this time only 53 percent of respondents said they have no or weak preferences regarding who to vote for on March 15th. Whether it’s correlated or not: in the past weeks among the five largest polled parties, there have been no large shifts, apart from stagnating support for Wilders’ PVV (see figure 3).
This is not to say that shifts as seen in 2012 may not still take place this time. There are three important debates left in the next two weeks (see table 1). So far, poll leaders Mark Rutte (VVD) and Geert Wilders (PVV) did not bother showing up to any of them. When they do appear, voter sentiment might change significantly.