Who’s afraid of “Safety in Numbers”?

The following essay is based on a review of “Is it safe in numbers?” by Christie and Pike (in Injury Prevention August 2015 Vol 21 No. 4 276-277 – see the reference to it here ) . It indicates certain attitudes and beliefs about human behaviour amongst “road safety” researchers and professionals – attitudes and beliefs which we think it important to criticise.

What is “Safety in Numbers”?

The “Safety in Numbers” (SiN) thesis is associated with Jacobsen, and argues that, as Christie and Pike note:

“Based on data from the Netherlands, Denmark, USA and UK, Jacobsen’s paper in 2003 1 identified the non-linearity between the number of cyclists and pedestrians and the risk of injury from being hit by a motor vehicle. In other words, the more people walked and cycled, the fewer the number and rate of traffic collisions and injuries experienced by cyclists and pedestrians—a non-linear relationship. Jacobsen, termed this relationship, ‘Safety in Numbers’ (SIN), which was shown at different levels of scale, whether at an intersection, a city or a country. More recent work has since shown SIN to occur in other countries such as Australia” (*)

I am not here restricting myself to, or defending, the work of Jacobsen. I am, however, defending the idea that people adapt to perceptions of risk (risk compensation, henceforth RC) and that SiN is a manifestation of this. In fact, a reduction in casualty rate per motor vehicle at a roundabout with the increase in flow of motor vehicles through it was noted in 1962 in a paper quoted by the father of road safety studies, Reuben Smeed, shortly afterwards. For his and other illustrations of what RC is about, see Chapter 2 here  and of course the work of John Adams , particularly “Risk and Freedom” and “Risk” downloadable from his site.

RC is the causative mechanism for numerous phenomena. Some of these are the effects of “road safety” interventions which have adverse effects on those of us outside cars: the iconic case of seat belts, for example:  . In presentations I often refer to the comments of Alec Issigonis about the lack of “safety features” in his Mini Minor: “If I had wanted to build a safe car, I would have put nails in the dashboard”, generally to be greeted with laughter. The effects of the use of more crashworthy cars on those outside them have been anything but laughable.

Sometimes there are beneficial types of adaptive behaviour to changes in the transport mix – albeit not ones created by “road safety” or other transport interventions. These are the ones referred to by Christie and Pike (increased numbers of people walking and cycling). Consider the following two cases. Firstly the case of an increase in cycling in the first decade of the current administration of London’s transport affairs through the GLA and TfL:

Cycling in London 2000 – 2010

Looking at the annual reports by Transport for London for the first decade of his century, we see a decline in reported Killed and Seriously Injured (the most reliable casualty statistics) cyclists, moving back up to around the 2000 level in 2009. Over this time cycling approximately doubled in London, indicating something like a 40% cut in the KSI rate for the average cyclist.

What’s the explanation for this? There have been similar casualty rate declines for other modes, but these are claimed to be due to highway and vehicle engineering. There has been little of engineering with significant benefit for cyclists – the much-vaunted segregated cycle tracks promised in the Mayor’s Vision for Cycling are only being completed now, in mid-2015. Policing of London’s drivers is minimal at best, with a flurry of activity occurring at the end of 2013 after a spate of cyclist deaths.

There has been an increase in cyclist training, with some evidence that recipients are less likely to be involved in collisions – but the vast majority of London’s cyclists have not had this training. There have been advertising and publicity campaigns asking cyclists and drivers to behave in certain ways: this form of “road safety” intervention is the least proven of all types.

There is one possible area where some interventions by Transport for London may have been effective: collisions where cyclists are killed under the wheels of lorries (HGVs). While the Killed statistic is the most reliable of all casualty statistics, in London it is difficult to use, as it is small for cyclists – the annual number is typically between 9 and 18 and is difficult to use in statistical analysis. Nevertheless, the number of cyclists killed in this way stayed more or less the same during the first decade of this century despite there being an increase in lorry traffic (and even more so in the construction industry, whose vehicles are particularly implicated) in the areas where most London cycling occurs – inner and central London. Even more striking, the increase in cycling in inner London was greater than the London average: other things being equal, one might have expected the number of cyclists killed in this way to be 3 – 4 times higher than it was in this decade. (Of course, as will be pointed out below, this is not any reason for complacency – the lorries issue is one which should be addressed forcefully, as we argue here ).

What’s the reason for this? TfL during this decade made efforts to alert lorry drivers, a small fraction of cyclists have been warned not to undertake lorries, and operators were urged to use additional mirrors. But would these efforts have resulted in a massive cut in the death rate amongst cyclists? My argument would be that there has been substantial publicity (particularly in London’s Evening Standard and the London BBC and ITV news programmes), and likely informal discussion among the lorry driver community.

It is crucial to remember that this need not involve what is often referred to as “respect” for cyclists – as in so-called “mutual respect” advertising campaigns – but simple awareness of cyclists. Lorry drivers may still (or even more so) regard cyclists as a hazard or menace: the point is that they appear to be more willing to watch out for them.

Of course, outside the community of professional lorry drivers, we would expect the SiN effect to be less, as indeed it appears to be with a less dramatic fall in the cyclist casualty rate.

Of other explanations the main one raised is of course cycle helmet wearing, with use apparently increasing during this time period. On this I would refer to the work by the Cyclists’ Touring Club and other evidence collected by the Bicycle Helmet Research Foundation.

In addition, we have to ask what the beneficial effect of cycle helmets would be on the Serious Injury figures. (As already noted, the annual number of cyclist deaths is difficult to use in statistical analysis, and the majority of these either involve lorries or high speed collisions with motor vehicles where helmets will have little benefit).

Even where a helmet may have had a beneficial effect after a collision, the collision should still be analysed as a Serious Injury (SI) for reporting (on the STATS 19 form) purposes. One reason for this is that where the helmet liner has been damaged (supposedly preventing a head injury from being Serious rather than Slight) the injured person should still be under observation, and thus more likely to be recorded as a Serious Injury. Another is that other injuries may fulfil the SI criteria. All of this applies even without any adaptive (risk compensatory) behaviour by either helmeted cyclists or other road users towards them.

Removal of pedestrian guard railings in London

This is the second case: For the evidence see the debate with John Adams in Local Transport Today 26/08/2011. Essentially, the pressure put on motorists to have to watch out for crossing pedestrians and be more careful has had some beneficial effect. Of course, there are always other factors – like having sufficiently slow speeds – but there is still the sort of effect that SiN is all about.

What we think of SiN

Our reading of SiN as a form of adaptive behaviour/RC is, like RC, something which we think of as a plausible reason for some phenomena. It is important to state what we DON’T think about it.

Firstly: “Implied in the SIN concept is that cyclists and pedestrians travel in proximity with each other and that the protective effect is similar to the protective effect of animal herds”. We have no evidence of this assumption. The “critical mass” sometimes referred to in discussions about SiN is the effect of increasing numbers impressing greater awareness on other road users, specifically drivers.

Secondly, it is not sufficient to argue that all we have to do is to increase the numbers of people walking and cycling – nor has anybody ever suggested this to be the case. The road danger reduction agenda argues that we require a shift in our everyday assumptions about transport and safety. This may be manifested through technological changes (as in highway and vehicle engineering, or in aids to law enforcement) or simply in the ways people behave.

UK cyclists riding in in countries like France and Spain point out significantly different driver behaviour towards them: the idea that UK drivers (with basically similar genetic make-up) could not change, albeit over some time, to behave the same way is nonsense. That does not mean that the usual publicity campaigns will be effective, or that relying on drivers to change their behaviour because they have also cycled will be effective. It just means that change occurs – often entirely irrespective of any official “road safety” intervention.

SiN is at least a partial explanation for some beneficial changes, and can be part of wider moves to reduce danger at source.


Don’t panic!

By contrast, the reaction of some such as Christie and Pike is hyperbolic. The attention of public health promoters was apparently “grabbed” by the concept of SiN, allowing them to make “clarion calls” resulting in :

“… a paradigm shift among planners and engineers who could think about pedestrian and bicycle safety in a different way and not be so fearful that by encouraging increases in walking and cycling they would see an increase in traffic collisions and causalities”.

I doubt that there has been a “paradigm shift” except in terms of increased verbal “support” for walking and cycling, with an exception being in London – (mainly) characterised by segregating, rather than integrating cyclists with the general traffic flow. Public health professionals have been pointing out the benefits of walking and cycling for over twenty-five years, showing that the benefits of cycling in particular far outweighed the dangers even in then existing conditions (which were not to be tolerated anyway).

Christie and Pike are particularly concerned about criticism of helmet advocacy which draws attention to the adverse effects on cycling levels. Given the adverse effects of compulsion https://rdrf.org.uk/2013/12/17/the-effects-of-new-zealands-cycle-helmet-law/ , this is hardly surprising.(no comment is made by Christie and Pike about this evidence, nor the likely reasons for a lack of effect of mandation https://rdrf.org.uk/2013/12/27/the-effects-of-new-zealands-cycle-helmet-law-the-evidence-and-what-it-means/ .)


An exception

There is an alternative source of concern about SiN, namely from the new wave of cycle campaigners pushing for full y segregated cycle tracks as the key feature of a pro-cycling programme. Their argument is that such provision creates safe conditions to cycle, which then supports mass cycling. The consensus in RDRF is that while we’re happy to go along with the current push for “Space4Cycling” in the UK and London in particular, we have reservations about specific issues (driver behaviour at junctions, potential conflict with pedestrians at bus bypasses and on segregated cycle tracks), the lack of relevance on rural roads and more generally the possible adverse effects on the majority of streets where cyclists will travel in proximity to motor vehicles.

However, we are fully in agreement with the idea that cycling must be supported as an everyday form of basic transport carried out ordinary people wearing ordinary clothes, as opposed to a specialised athletic pursuit (although that can also be accommodated). Our main difference is a concern about “dangerising” existing cycling, and arguing for support for those people already cycling, as well as the full range of measures required for reducing danger at source for cyclists and others, whether highway engineering or other methods: vehicle technologies, law enforcement etc.

Our view is that high cycling rates and low cycling casualty rates per journey are achieved when cycling is regarded as a normal form of transport where special clothing and helmets are not seen as required. This appears to be at odds with Christie and Pike.


Why the panic?

There has traditionally been hostility towards increasing the numbers of people walking and cycling among “road safety” professionals, as “vulnerable road users” outside cars they are particularly prone to injury and death after collisions and not to be encouraged. (This opposition does not extend to a significant proportion of the “road safety” industry that advocates the far more hazardous form of transport that is motorcycling). Cycling, in particular, is seen as problematic.

One way this manifests itself is the way the problem is measured. For us this is critical, and requires a fundamental transformation in the way we conceptualise safety on the road. For traditional “road safety” professionals, it is the aggregate number of KSIs, or at least the aggregate number of a road user group KSIs. So, Christie and Pike mention: “…the relative high risk of injury among pedestrians and cyclists from deprived areas, where we know people walk more because there is less access to a car?” as a counter to SiN. But is there this “high risk of injury”?

What we know from such neighbourhoods is that there is indeed far more walking and street play. Taking into account this level of exposure, such pedestrians may in fact have quite a low level of risk – a low rate of injury per hour spent walking or playing in the street. But then an exposure based measure is not what the “road safety industry” is interested in.

Another indication of a scared reaction to prospects of increased cycling is: “… that 8 out of 10 injured cyclists result from single crashes not involving a collision with a motor vehicle”. Of course, as with all road crashes, there is the question of under-reporting. But there is no reason to think that there has been any change in the proportion of cyclist casualties that are under-reported: in terms of a trend, the picture is the same. Also, how serious are these collisions that might involve children playing (often off-road) on bicycles?

More importantly, is it really helpful to suggest that danger from the (ab)use of motor vehicles is not the problem? Is that what the “road safety” professionals want us to think?

The conceptualising of cycling as, we think, a key element in helmet advocacy.??? Sure enough, one of the authors mentions cycle helmet wearing as one of his two key points to consider when taking to the roads in this video . (The other is seat belt wearing which we would think of as grimly ironic in this context given the adverse effects of seat belt laws on cyclists and pedestrians) .And they ignore unreported pedestrian falls, which probably mean that figures for the relative danger of walking and cycling make cycling look worse than it is.




Christie and Pike do acknowledge that there is some sort of possible SiN effect, and call – as academics are so wont to – for further research. We would also agree with them that good quality highway engineering is required for cyclists and pedestrians.

But we need to point out the flaws in a belief system that has stood in the way of the more benign modes for far too long, and indeed been part of the problem for walkers and cyclists. If new research is indeed forthcoming, it has to take into account two features constantly highlighted by the road danger reduction movement. Firstly the hierarchy of danger in a car-centric society with discrimination against non-car modes, specifically, cycling as key. Secondly, that people constantly adapt to their perceptions of risk. Otherwise we will get nowhere.


NOTE* Linearity is the wrong word. The null hypothesis is that the risk does not change with cyclist numbers.



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s