Polls

Politics homepage.

Blog homepage.

This is a hobby project. My hobbies do not include old IEs. Therefore this page does not work in IE8 or lower.

Here I study the polls.

My average

My weighted average of the polls is designed to flatten out trends, so that outliers are somewhat ignored, and only persistent trends accepted. It reports a shift of seats relative to current parliament.

My average of the polls
Party Seats Senate Party

The effective number of parties is a measure for the fragmentation of a party system. The highest effective number ever reached in actual elections was 6.7 in 2010.

See the party profiles for a description of most parties.

Current polls

Here’s an overview of the latest poll by each pollster.

Current poll overview
Pollster Date Weight
Een Vandaag
()
Politieke Barometer
()
Peil.nl
()
TNS-NIPO
()

The Weight value gives the relative weight of the poll in the calculation of the Now column above. However, the Now column also uses older polls that fall before the second column’s date, so the average isn’t dependent on these last polls alone.

Coalitions

For an introduction to Dutch coalitions I advise you to read this article series that I wrote for the 2010 elections. The details are slightly different today, but the broad overview is still valid.

  1. Overview and general introduction.
  2. The Left coalition.
  3. The Right coalition.
  4. The Centre-right coalition.
  5. The Purple coalition.
  6. The Centre-left coalition.

The coalition tables are automatically generated — they may sometimes show weird coalitions. Still, Dutch politics are in such a state of advanced chaos that even weird coalitions may come to look appealing.

Coalitions with at least a % likelihood according to
Coalition Type Likelihood Polls ch Seats Senate

A five- or even a six-party coalition is not as remote as it might seem. Not all parties have to send ministers to the cabinet — they can support government from parliament, like Wilders supported Rutte I (VVD+CDA). The left-wing equivalent would be a PvdA+D66+GL government.

Still, such a minority government would have to come to agreements with other parties that promise to support it. This combination of government parties and supporting parties would have to have a majority, and it is likely to be one of the coalitions mentioned here.

Source
Coalition Type Likelihood Polls ch Seats Senate

I add the likelihood of all coalitions that a party participates in to get at its likelihood to be in government. I do the same for all coalition types and sizes.

Coalition chances
Coalition type
Coalition size
Prime minister

Raw data

There are four pollsters in Dutch politics: the Politieke Barometer, Peil.nl, Een Vandaag, and TNS-NIPO. Here’s the raw JSON data; below are some nice tables.

2011-2012 data are here.

Politieke Barometer

The Politieke Barometer publishes its poll every two weeks on a Thursday.

I trust the Politieke Barometer more than the other pollsters. It has a better score than the others for the right block and the traditional catch-all parties.

Politieke Barometer polls are usually quoted in the press, but don’t get anything near the exposure of Peil.nl polls.

Last Politieke Barometer polls
Party Seats Party

Peil.nl

Peil.nl publishes its poll every week on Sunday.

Peil.nl is always out for sensational headlines. Protest parties SP and PVV usually poll better with Peil.nl than with the other two, and the same goes for the left block and for small parties.

Maurice de Hond, Peil.nl’s owner and a well-known political commentator in his own right, uses an open Internet poll to which anyone can subscribe. (I have.) This methodology is criticised time and again by the other two pollsters and political scientists, but if we compare his last polls to the election results he doesn’t do significantly worse than the other two. Part of the problem is that he has the best press contacts of the three, and his polls always draw headlines. Besides, if a TV programme needs a political pollster they always ask him. This won’t make him very popular among his colleagues.

Last Peil.nl polls
Party Seats Party

TNS-NIPO

TNS-NIPO publishes its poll about every two months.

Generally TNS-NIPO is closer to the Politieke Barometer than to Peil.nl, with maybe a tad advantage for the centre parties.

TNS-NIPO polls rarely garner much attention in the press, except in De Volkskrant, which is a partner in this series of polls. That’s probably due to the confused release schedule, and TNS-NIPO’s annoying habit to release some polls only together with the next one.

Last TNS-NIPO polls
Party Seats Party

Een Vandaag

Een Vandaag publishes its poll every month.

The poll is executed by Intomart GFK, which also polled in 2010 but not earlier, and did the worst job of the four pollsters. It seems even more left-minded than Peil.nl, which is no mean feat. Also, the single issue parties score better than in the other polls.

One incredibly annoying habit this poll has is giving the occasional seat to “other parties” — without making clear which parties. I assign these seats to the Pirate Party, which right now is the only Other party that stands a chance of entering parliament. Of course that’s incorrect, but I don’t know how else to deal with these seats.

Last Een Vandaag polls
Party Seats Party

Calculation methods

How do I calculate my average and the coalitions. (Warning: I have zero knowledge of statistics.)

Polls

I treat the polls as follows:

  1. I established weighting factors for the pollsters based on past performance. They are
  2. I also use a decay factor based on the poll’s age. This is an exponential decay function with N0 of 1, t in days, and λ of .
  3. The polls are sorted according to the periods defined in the first table ( days).
  4. For each period I take the polls of one pollster, multiply the individual party scores by the decay factor, and put them in a bucket by party.
  5. I use the combined scores as votes in an election, which yields a average pollster score for this period.
  6. Once I’ve done so for all pollsters I multiply the average pollster scores by the pollster weight and the decay factor of the most recent poll and put them in a bucket.
  7. I again use the combined scores as votes in an election, which now yields my final average.
  8. Election: this is an election according to the normal rules.
    In order to protect the small parties I use the system of highest remainders for rest seats, and I have no electoral threshold.

Coalitions

coalition relations:

Preferred partners
Possible partners
Unlikely or unwilling partners
Excluded parties

The most negative opinion prevails. So if the SP indicates it can work with the VVD, but the VVD says it can’t work with the SP, their relation is Excluded.

The script creates all possible coalitions and then rejects the following ones:

  1. Coalitions with PvdA, CDA, and VVD.
  2. Coalitions with less than seats.
  3. Coalitions of which a subset already has a majority of seats.
  4. Coalitions with parties that have excluded each other. (See sidebar for the current exclusion list.)
  5. Coalitions with two or more parties that have less than seats.
  6. Coalitions whose smallest party has less than seats, unless that party gives the coalition its majority in either parliament or senate.

The table shows the remaining coalitions.

Coalition likelihood

The likelihood of a coalition is calculated by the following formula that I tweaked by hand (there are few theoretical underpinnings here). I don’t doubt I’ll make frequent changes.

The formula is
1/SIZE * MAJORITY * SMALLEST * RELATION * SENATE

SIZE
the number of parties in the coalition
MAJORITY
the majority of the coalition (seats over 75)
SMALLEST
the size of the smallest party
RELATION
Each pair of parties has a relation from 0 (excluded) to 3 (preferred). The sidebar shows which parties have which relations.
The variable is the sum of the worst relations in the coalition.
(Coalitions in which any two parties have relation 0 are rejected automatically, so that value doesn’t occur.)
SENATE
Is 1 if the coalition has a majority in the senate, if it does not.

Once the likelihood of all coalitions has been calculated, the results are treated as votes in an election for 100 seats. This yields the percentages that are shown in the table.

Pollster quality

Here I compare the final polls of the pollsters to the election outcome. I hope this makes clear why I consider the Politieke Barometer the best pollster.

The average error averages all errors of each pollster, where the most recent error counts double.

Pollster errors

Lower numbers are better. The calculation is as follows:

  1. For each misassigned seat one point is given. In 2003 the Barometer gave the CDA two seats too few, the LPF one too few, the CU two too many, and the PvdA one too many. Thus it misassigned three seats and gets three points.
  2. If the pollster sees a party entering parliament but it wins zero seats, or if a party enters parliament but a pollster gives it zero seats, it gets an extra point.
    In 2010 Een Vandaag and TNS-NIPO both gave TON one seat, while it did not enter parliament. They both get a point because they were wrong about TON, in addition to the point for the misassigned seat.
  3. If the pollster incorrectly predicts the largest party it gets an extra point. In 2003 all pollsters showed the PvdA larger than the CDA, and they were all wrong and all get an extra point.

The PAM is the Politieke Aandelenmarkt (Political Stock Market), an initiative of De Volkskrant, which was supposed to give better results than the pollsters but didn’t. I played in 2006.

In 2010 I drew an average of the polls; I think I just averaged the numbers without weighting or other subtle tweaks. Since the pollsters were fairly close I also was fairly close. In 2012 I use the results of my poll average.

Poll data over 2002-2006 come from Cijfers.net.

2012

PvdA and especially VVD were underestimated by all pollsters — clearly a last-minute prime-minister race effect boosted them both, mostly to the detriment of the SP. Also, the PVV was overestimated for the first time, and by a lot less seats than it was underestimated in 2010. The theory that they lost last-minute seats to the VVD is defensible. So maybe the PVV is becoming predictable.

2010

In 2010 all pollsters underestimated Wilders’s PVV for the second time running, which is why it is a difficult party to predict. Some polls again missed the fact that the PVV would be the only extreme right party in parliament.

Also, the CU’s loss of one seat was not predicted (and Intomart even predicted a ridiculous 10 seats).

2006

All pollsters underestimated Wilders’s PVV, and only TNS-NIPO saw that it would be the only extreme right party — the others have one or two others win a few seats. Also, the SP polls diverged very widely.

2003

All pollsters but the PAM had the PvdA a tiny bit larger than the CDA, but in the elections the CDA became the biggest. This is a classic prime-minister-race effect.

2002

The big surprise was that the CDA grew a lot more than expected, and GL and CU lost one seat instead of winning a few. CU to CDA is a logical movement.

All pollsters got big newcomer LPF (List Pim Fortuyn) about right.

1998

Data as yet incomplete; Peil.nl is missing. And to be honest I’m not totally sure I got the very badly formatted data right, or whether it’s reliable. I mean, did they really predict 3 seats for Janmaat?