Salonanarchist | Leunstoelactivist

Organisations demand raw data of controversial survey

In November, a survey was published which found that 80% of Dutch youth with a Turkish background would not have a problem with the use of violence in jihad and 90% would think that Dutch Muslims who fight in Syria are heroes.

While some were shocked by the findings, others expressed doubts about the methodology of the survey or simply thought the results were improbable. Among other things, the survey was not based on a random sample and non-response wasn’t reported. The researchers did try to recruit a sample that was representative of the wider population on a number of background variables through quota sampling (discussions in Dutch here, here and here; research method here).

A Motivaction spokesperson said the way in which the research had been done was «acceptable» from a social sciences perspective.

Now a group of Turkish organisations demands that Minister Lodewijk Asscher, who contracted the survey, order Motivaction to release all results so they can ask an independent expert to review the study. If necessary, they may go to court to get the data.

I don’t know how likely it is the organisations will get the original data and if they do, it may still be difficult to demonstrate sampling bias. Still, if this is a step towards introducing principles of reproducible research in contracted policy research, then that’s an interesting development.


Criticising charts

I missed this one: in October, Dutch economics journal ESB published an article that critically reviews all the charts in a report of the CPB (the semi-offical neoliberal economic institute that dominates Dutch policy debates). Authors Frank Kalshoven and Peter van Bergeijk find that on average, as many as four out of eight aspects of the charts have been done badly.

The authors invented a scale to assess charts, using the following criteria: the title describes what the chart shows; abbreviations and terms are explained; axis units are clearly described; axes are aligned; the source is explicitly mentioned; charts tell a clear story; charts contain little «noise» and there’s an explicit relation between panels in a chart.

One of the charts discussed is the one shown above. Among other things, the source is missing. Further, the y-axes of the bottom panels aren’t aligned, wrongly suggesting that taxes (bottom right panel) are often higher than collective expenditures, whereas in fact expenditures are higher than taxes (note that the government also has other sources of income).

Kalshoven and Van Bergeijk’s analysis seems to be strangely unconnected to the broader universe of data visualisation critique (interestingly, one of their sources of inspiration has - somewhat harshly - been described as «a horrible example of economists not recognizing that outsiders can help them»). Some of the most popular topics of chart criticism are missing from Kalshoven and Van Bergeijk’s article: use of colour; if and when it’s ok to truncate y-axes; legends versus labels; and if and how to use the area size of bubbles or icons to represent quantity.

Frank Kalshoven and Peter van Bergeijk, Datavisualisatie in de MEV onvoldoende. ESB 99, nr 4696. Online version here


As trade unions consider merger, the Dutch want their unions to take a much tougher stance

A large majority has voted in favour of the merger - A plan to create a new Dutch union with about 1 million members was put on hold in October, when the plan just failed to get a two-thirds majority at the convention of FNV Bondgenoten, one of the unions involved in the merger plans. A new vote will take place on 26 November.

Representatives of employers’ organisations expressed disappointment at the initial rejection of the merger. They had been hoping the merger would result in a stable trade union that will play a constructive role in the elaborate social dialogue institutions of the Dutch «polder model».

In fact, that’s exactly what Dutch unions have been doing over the past decades, as evidenced by their low strike rates. But with growing inequality and an erosion of the welfare state going on, doubts arise whether social dialogue is enough. Some groups of workers, like cleaners and health care workers, have successfully resorted to more assertive campaign methods to fight for decent pay and better working conditions.

Since 2007, researchers of the University of Tilburg have been asking a panel of about 6,000 respondents what they expect of unions. More specifically, they have asked respondents whether they agree that «Trade unions should take a much tougher political stance, if they wish to promote the workers’ interests». In the latest edition of the study, 44% (strongly) agree and only 13% (strongly) disagree.

If anything, support for tougher unions seems to have grown over the past years. Surprisingly, even among the self-employed and among people who voted for neoliberal parties like VVD and D66 in 2012, more respondents agree than disagree that unions should take a much tougher stance. High-income respondents are among the few groups that are not so keen on tougher unions.

Last weekend, chairman Ton Heerts explained the position of the FNV to the Telegraaf newspaper: «I think we’ve proven over the past year that it’s quite possible to combine substance, dialogue and action. With the current wave of right-wing policies, the emphasis will be more on actions. That’s fine.»

An earlier version of this analysis was published here


Second jobs - job erosion or appetite for consumption

Last year, a spokesperson of the German federal government suggested the explosive growth of Zweitjobs (second jobs) could have various explanations. Yes, people may be forced to take on second jobs out of financial necessity and because of the flexible labour market; but it could also have something to do with an increased «appetite for consumption» (Konsumlust). The suggestion immediately resulted in 2,000 sarcastic tweets.

The Netherlands has also seen a substantial growth in the number of people with second jobs, as new data from Statistics Netherlands illustrate. The chart shows the total number of employees (blue); employees with non-permanent jobs such as temp jobs and zero-hours contracts (red) and employees with a second job (green, all index 2002 = 100).

The green and red lines show a quite similar pattern. One might try arguing that the crisis caused a dip in the appetite for consumption, but more likely there’s a broader pattern of job erosion going on, temporarily slowed down when employers shedded their «flexible skin» (Dutch jargon for the precarious workers employers use) during the crisis.


Scooters often faster than cars

Minister Schultz wants to allow Amsterdam to ban scooters from cycle paths and make them use the road, wearing a helmet. This should make cycle paths safer for cyclists and reduce their exposure to air pollution. However, car and scooter lobbyists argue that the speed difference between scooters and cars is too large for scooters to ride safely on the road, with motorists driving 50 kmph.

So do motorists really make 50 kmph in Amsterdam? «Cycling professor» Marco te Brömmelstroet has tweeted a map showing rush hour speeds far below 50 kmph.

As part of its open data initiative, Amsterdam has released some 5 million speed measurements at the «Hoofdnet Auto» (the network of major roads for cars) during the month of January 2014. The histogram above shows that even at these main roads, the majority of measurements recorded a speed below 50 kmph, with a median speed of 31 kmph. Average speeds during afternoon rush hour were about 5 kmph lower than at night.

A 2011 study by cyclists’ organisation Fietsersbond found found an average speed for scooters on Amsterdam’s cycle paths of 36.9 kmph. The map shows roads where motorists drive on average at least 36.9 kmph (thin red line) or 50 kmph (thick red line). Note that the method by which the Fietsersbond measured scooter speed may be different from the method used to measure car speed.

There have been jokes that scooter riders don’t want to use the road because this would force them to reduce their speed. The data of the Amsterdam government show there’s actually some truth to this.

Scripts for processing the data can be found here.

Executive pay

Last weekend, Senate member Roger van Boxtel criticised trade union FNV’s new central wage demand of 900 euro (which would narrow the gap between low and high incomes), arguing that it’s «really too much». Van Boxtel himself, in his capacity as ceo of the Menzis health insurance company, got a 5,000 euro raise last year, resulting in a remuneration of 389,000 euros.

Over the past year and a half, high-paid executives of (semi) public organisations have been somewhat sheltered from public scrutiny. Because of the introduction of a new norm, the government has suspended its annual publication on excessive pay at (semi) public institutions.

Google Trends data show that there was a peak in searches for «top incomes» in January 2013, when the latest report (on 2011 incomes) was published. Interest in the topic remained, but hasn’t reached the January 2013 level since.

In the meantime, some efforts have been made to analyse data from the annual reports of the institutions themselves. Abvakabo FNV has published its annual Actiz 50, documenting excessive pay at health care institutions. And newspaper de Volkskrant has published an analysis of 119 (semi) public institutions.

The Volkskrant data contain about 40 board members receiving remunerations in excess of the current norm for newly hired executives (230,000 euros). And their list is far from complete: many of the highest-paying institutions in 2011 aren’t even included. Among them Roger van Boxtel’s employer, Menzis.

In short, we won’t have the complete picture until the government publishes a new report, perhaps in a few months.


Cycling against traffic #2

The other day I posted something about cycling against traffic which, it has been claimed, is allowed on 85% of oneway streets in Brussels. I tried to find out the percentage for Amsterdam using Open Street Map, but found that the relevant information is often missing. Or so I thought.

I posted a question on the OSM forum (here and here) and got various helpful answers. Basically, I shouldn’t have looked just for oneway:bicycle=no tags, but also for cycleway=opposite (and perhaps a few more). I was also directed to a web page where cycling tags can be shown on a map.

So does Amsterdam allow cycling against traffic on anyway near the 85% of oneway streets reported for Brussels, if you include the cycleway=opposite tags? Well, no. Then again, looking at a similar map of Brussels, it doesn’t really look like they do any better. Of course, one shouldn’t jump to conclusions:

  • It depends on the part of the city you look at. In Amsterdam, cycling against traffic is more often allowed in the city centre and some other parts like Oost; in Brussels is appears to be more spread out over the city,
  • Perhaps local Open Street Map contributors have different mapping habits.

That said, I was getting curious as to the basis for the 85% claim for Brussels. I found a report from 2010 published by cyclists’ organisation Gracq, which said that 75% of oneway streets in Brussels had sens unique limité (which is apparently a legal requirement on suitable oneway streets). Gracq had contacted local governments by telephone to collect the data.

Can Open Street Map and Qgis show where it’s ok to cycle against traffic

[Update here] - The Italian cities Milan, Bologna and Turin would like to allow cyclists to ride against traffic on some oneway streets. This would help promote environment-friendly modes of transport and it would bring Italian cities in line with many European cities, where this is already allowed. For example, Brussels allows cycling contromano on 85% of oneway streets, they argue.

I was intrigued by that percentage, and curious what the percentage for Amsterdam might be. My first hunch was that it might well be similar, because you sort of expect that cycling in both directions is normally allowed here. Then I realised that the exceptions to this rule include canals, where the streets usually are oneway for cyclists as well. That might cost us percentage points.

I reckoned it should be possible to find out more using Open Street Map, where streets have oneway and oneway:bicycle labels. Unfortunately, the oneway:bicycle information is often missing (dotted lines on the map). This includes streets along canals that are oneway for cyclists, but also streets in neighbourhoods such as the Oosterparkbuurt where cycling in both directions is allowed.

Of course, Open Street Map is a volunteer project, so if information appears to be missing, I guess that’s my responsibility as much as anyone else’s. So here’s my to-do list:

  • Try to find out if I interpreted the oneway:bicycle tag correctly,
  • Figure out how to edit Open Street Map,
  • Add some oneway:bicycle information.

At the very least, it will be a good opportunity to learn something about Open Street Map.

Incidentally, the Italian cities saw their request turned down by minister Maurizio Lupi. Cycling against traffic may work elsewhere, but «we’re in Italy, not Germany», he argues.


Not only am I basically new to OSM; I also don’t have much experience with Qgis, so this was a bit of a trial and error thing. First I tried to define specific types of roads based on this overview of types of oneway roads with cycle lanes. However, trying to create new attributes in Qgis based on these descriptions all but crashed my computer (for some reason using conditions containing AND in the field calculator seems to be problematic). Further, almost no roads in Amsterdam appear to meet these specific criteria.

So instead I took a more basic approach, looking for oneway=yes in combination with different values for oneway:bicycle. Out of more than 11,500 polylines with oneway=yes, 267 had oneway:bicycle=no and five oneway:bicycle=yes (Halvemaansbrug, a nearby bit of Kloveniersburgwal and three unnamed polylines).

Data based on a rectangle comprising the city of Amsterdam, downloaded on 20 September 2014.