Improving road safety


The National Audit Office has just published a report for the Department for Transport, Improving road safety for pedestrians and cyclists in Great Britain.

Here are the headline figures:

ModeUK fatalities per 100 million passenger km
average 1997–2006
Bus or coach0.029
Pedal cycle3.404

The most obvious problem with this report is the persistent confusion between absolute casualties, and casualties normalized by population and by distance travelled. In some cases I simply can’t tell which is meant, for example on page 10:

“The Department has targets to reduce by 2010 the numbers of people killed or seriously injured by 40 per cent, of children aged 0 to 15 by 50 per cent and slight injury rates per 100 million vehicle kilometres by 10 per cent compared with the average between 1994 and 1998.

Which of these three targets are absolute and which normalized to population and distance? I haven’t a clue. This matters because effective strategies are going to differ according to which measure is being targeted. If the target is simply to reduce the absolute number of casualties from cycling, the simplest approach would be to ban cycling altogether: the target would be achieved automatically. If such a ban proved politically impossible, then making cycling substantially more inconvenient (for example by imposing a registration scheme or other tax) would probably have a similar effect. But can this really be the kind of “safety” that the Department for Transport has in mind?

This confusion between absolute and proportional numbers comes to the fore in the section on international comparisons, where this claim is made:

“The United Kingdom was fourth highest out of 14 European nations in 2006 for the least number of cyclist deaths per head of population.” [page 12]

And backed up by this table on page 36:

RankCountry2006 cyclist fatality rate
(per million population)
13Czech Republic10.7

This seems to me to be a really misleading table to include in the report. Because these numbers are not normalized to distance travelled, the table is primarily telling you about the amount of cycling in each country, not the relative safety of cycling in that country. People in the UK hardly cycle at all—the report estimates an average of 39 miles per year per person—so it is not surprising that the number of fatalities is low too.

And where are the Netherlands? Using the figures from 2007 (see below) of 147 fatalities and a population of 16.4 million, I get a figure of 9.0 fatalities per million population. So why aren’t the Netherlands in there between Belgium and Estonia? (This could be because the Netherlands were dilatory in reporting road casualty figures for 2006 to the European Road Safety Observatory—in ESRO’s Annual Statistical Report 2008 the latest figures for the Netherlands are from 2003. But it would have still been fairer to include the most recent datum from that country, with a note about what year it’s from.)

The appropriate table would have shown the data normalized by distance travelled. Here’s my best attempt to generate such a table:

Country Year of fatality statistic Number of cyclist fatalities 1 km cycled per person per day (2000) 2 Population in millions (2007) 3 Cyclist fatalities per 100 million km cycled
Czech Republic200711610.3


  1. Number of cyclist fatalities — European Commission. EU energy and transport in figures, 2009.

  2. km cycled per person per day — Figure 2 of John Pucher and Ralph Buehler (2008). “Making cycling irresistible: lessons from the Netherlands, Denmark and Germany”. Transport Reviews 28:4:495–528. DOI: 10.1080/01441640701806612.

    I’m not particularly happy with using these statistics because they are so out of date, but they are the best I can find for the moment. If you know of a better or more up to date source, please let me know. Pucher and Buehler source them from EU energy and transport in figures, 2000, and that’s not online. More recent issues of the same publication which are online don’t report this statistic. (However, the estimate for the UK—0.2 km per person per day—is a good match for the National Audit Office’s figure of 0.17, so maybe they are not too bad.)

    Update: I e-mailed John Pucher to ask if he knew of more recent statistics for distances walked or cycled in the EU countries, and he wrote back, “The [European Commission] no longer publishes those estimates of distances walked and cycled, unfortunately. So the only alternative is to to get the info from individual countries, which we did for [the Netherlands], [Denmark], Germany, UK and USA, but it’s too much work to do it for all countries.”

  3. 2007 population — Eurostat. European Union: Total Population.

You can see that the normalized figures are not nearly so flattering to the UK. Is this just plain incompetence, or a deliberate attempt to mislead?

So much for the statistical side of the report. What about the recommendations?

Well, I was going to tell you about them, but I fell asleep. There’s absolutely nothing concrete to get a grip on, nothing that will result in any action, nothing that will make any actual difference to someone cycling or walking. Here are some highlights (page 7):

Most of the recommendations are completely non-actionable—“considering”, “engaging”, “educating” and “influencing” are the key verbs—and the report admits that the Department’s education campaigns have unmeasurable results (“there is no direct evidence of the contribution that the Think! campaign has made to reducing casualties”).

So thank you, National Audit Office. Next time I get run down by a bus, I will remember how you recommended that the Department for Transport develop an explicit strategy for developing key indicators for assessing how well it works with other bodies.