champagne anarchist | armchair activist


Can Twitter predict the new Dutch trade union president

Number of tweets in which candidates are mentioned

According to an American study, you can predict the outcome of elections by simply counting how often the names of the candidates are mentioned on Twitter. Members of the Dutch union confederation FNV are currently voting for their new president (it has been claimed this is the first time in the world union members get to directly elect their confederation president). Would it be possible to predict who will be the new FNV president using Twitter?

Since last Friday, I’ve been collecting the tweets containing the term ‘FNV’; so far, there are over 2,500. In those tweets, the incumbent Ton Heerts is mentioned 204 times, whereas his challenger Corrie van Brenk is mentioned 146 times. In short, if Twitter is a good predictor (which of course is a matter for debate), the contest is tighter than one might have expected.

The graph above shows the results for the days for which complete data is available. On Saturday, Van Brenk got some attention because something she had said had been fact checked (and found to be correct). On Sunday, Heerts was mentioned because he appeared on a TV show hosted by Eva Jinek. On 1 May, it was officially announced who the candidates are and they had a debate.

Update - Updated to include 13 May, the final voting day. In sum, Van Brenk was mentioned 497 times and Heerts 631. It has since been announced that Heerts has won the election (of course, this doesn’t necessarily mean that the method is sound; in order to make such claims one would need to evaluate a fair amount of predictions).
Influences reflected in the graph include: Factcheck confirms Van Brenk statement (27 April); Heerts in Eva Jinek TV show (28 April); candidates officially announced (1 May); debate in Buitenhof TV show (5 May); problems at tax authorities that Van Brenk’s Abvakabo FNV had warned about (6 May); Van Brenk interview at (9 May); Van Brenk in radio show (10 May); Heerts at presentation of initiative to train technical staff (13 May); EenVandaag TV show poll predicts Heerts will win (13 May).
The graph may not be visible in older versions of Internet Explorer.


I collected tweets using the Twitter Streaming API (the ‘firehose’), in the way described here. I prepared the data using Python and analysed it using R (find the code on Github). The graph was created with D3.js.
I looked into how influential twitterers are (how many followers, how often listed) and into their backgrounds (e.g., do they mention ‘fnv’ in their profile). The most important finding is that twitterers who mention Van Brenk, more often mention ‘abva’ or ‘akf’ in their profile - not surprising since Van Brenk is currently president of Abvakabo FNV, the public sector union affiliated to the FNV.
The American study on Twitter as a predictor of election outcomes was done by DiGrazia c.s. and can be found here. Some remarks on their study:

  • Yes, twitterers are only a small part of the population and no, they’re not representative of the entire population. Likely, Twitter is dominated by a small, active incrowd. It’s also correct that tweets mentioning a candidate need not endorse them; they may as well be critical. Despite all this, DiGrazia c.s. found that mentions on Twitter consistently predict election outcomes. Perhaps they are an indicator of something else - e.g. media attention or how actively people are campaigning for a candidate.
  • Of course, this method doesn’t provide any certainty on who will win. It’s possible for a candidate to get almost 100% of the tweet share and still lose (at least, that’s what the scatterplots of DiGrazia c.s. suggest).
  • It’s unclear to what extent the conclusions of the American study can be generalised to other situations. It’s therefore a bit of a gamble to use this method to predict who will be the next president of the FNV.

German coalition parties hardly ask any questions

Nice: der Spiegel has launched a data blog, Datenlese. One of the first posts analyses questions asked by members of the lower house, the Bundestag. I thought it might be interesting to compare these findings to the Dutch situation. Unfortunately, der Spiegel doesn’t appear to publish the actual dataset they use in their analysis (unlike, for example, the Guardian Data Blog, which usually provides a spreadsheet with all the relevant data). [Update: the author kindly provides a link to the dataset here]

However, the Bundestag does publish statistics of parliamentary initiatives, as does the Dutch Tweede Kamer. A few conclusions:

  • The Bundestag asks about 75 questions per month. The Tweede Kamer more than three times as many, even though the Bundestag has four times as many members.
  • Written questions are primarily a tool for the opposition, but more so in Germany than in the Netherlands. In Germany, only 1% of questions are asked by members of coalition parties. In the Netherlands, 17% (or even 33% if former quasi-coalition party PVV is included).
  • In Germany, most questions are asked by the left-wing party die Linke and by the green party. Far fewer questions are asked by the social-democrats. A spokesperson told der Spiegel that the party knows from its experience as a former government party that questions ‘can paralyse the entire apparatus’. In the Netherlands, the social-democrats asked the largest number of questions in 2011. This hasn’t always been the case: when the social-democrats were still in government, they asked fewer questions and the left-wing SP headed the list.


Statistics of the current session of the Bundestag can be found here (pdf). Apparently, there is a distinction between ‘small’ and ‘large’ questions; the latter resulting in a debate. The number of large questions is very small; like der Spiegel I focused on the little questions. The Tweede Kamer is quite a bit slower than the Bundestag in publishing its statistics; I used the figures for 2011 published here.

US Congress’ interest in the world: the role of elections, trade and oil

Graphs may not be visible in older versions of Internet Explorer.

Number of times countries are mentioned by year

Select country:

Codeyear offers a course on how to use the API of the Sunlight Foundation to search transcripts of the US Congress. I used this approach to find out how often foreign countries are discussed in Congress. A simple inspection of the total frequencies suggests two conclusions:

  • Interest in foreign countries rose under the ‘Bush doctrine’ and fell since the start of the current economic crisis;
  • There are often peaks during odd years. Plausibly, Congress focuses more on domestic issues in even years, when there are elections for Congress.

Of course, the pattern may be different for individual countries (use the selector under the line graph to see data on individual countries). For example, interest in Afghanistan took off after 9/11; for Hong Kong it peaked in 1997 (transfer of sovereignty); for Serbia, Kosovo and Albania in 1999 (NATO bombing campaign); for Tunisia in 2011 (Arab Spring) and interest in Austria took off in 2009 (well that’s actually a mistake: in 2009, somebody named Steve Austria joined the US House of Representatives, boosting the number of times the term ‘Austria’ appears in transcripts).

Number of times countries are mentioned by population, GDP and trade

The scatterplots illustrate how the total number of times a country has been discussed in Congress over the period 1996-2012 is associated with population size, GDP and the amount of trade between that country and the US (note that the scales are log scales, a feature of D3.js; unfortunately I didn’t manage to get readable values on the x axis). Population, GDP and trade are correlated, so figuring out what exactly drives US Congress interest in a country remains an interesting challenge.

Interest in countries is also related to the presence of natural resources: for countries without oil, the median number of times they were discussed in Congress is 331; for countries with oil it is 900.

Of course this is just an exploratory analysis. An analysis at country/year level might yield more specific conclusions. If you want to do your own analysis, download the data here (country/year) and here (country).


I searched transcripts of the US Congress using the Capitol Words api of the Sunlight Foundation, using country names as search terms. Of course, this method isn’t perfect. I had to remove country names that can’t be distinguished from names of US states (Georgia, Mexico). Afterwards, I realised that I should also have removed Austria, because of confusion with a representative with that name.

Because this is just an exploratory analysis, I took a rather pragmatic approach to selecting background information on countries. For GDP, I used data from the World Bank; data on population, trade (2009) and oil reserves are from Wikipedia. For the scatterplots, I removed countries with incomplete data.