Learning from a STOP sign

I’ve been watching the behaviour of people driving through a junction in Edinburgh. In this article I’ll show you the junction, and I’ll demonstrate how most people break the rules here. That’s quite interesting to watch – but actually I’m not writing this article to criticise the rule-breaking. Later I’ll discuss why what we see here shows us that we need to be really careful when researching and analysing behaviours more generally – and how research about new infrastructure may sometimes end up being over-simplistic or misleading.

Here’s a photo of the junction….

…and here are links to the location on www.openstreetmap.org and on Google Streetview. The road layout looks like this (I’ve added arrows to indicate allowed traffic flow):

Most vehicles move through the junction as indicated here with red arrows.

The streets to the south (the bottom of these images) create a mini-gyratory system looping around a small green park.

The STOP signs are provided here because oncoming traffic (the longer red arrow above) can be on the ‘wrong’ side of the carriageway (as can be seen in the video below), and because visibility is limited.

The law is very clear about how people should drive when encountering a STOP sign, and I imagine lots of people know it well. You must physically stop your vehicle before proceeding, and then in addition you should ‘give way’ – not move on until you can do so without another driver having to change speed or direction. To quote:

“…every vehicle must stop before crossing the [wide white] line … and no vehicle must cross the … line… so as to be likely to endanger any person, or to cause the driver of another vehicle to change its speed or course…” (TSRGD 2016)

The Highway Code explains the same thing, but in easier language:

“You MUST stop behind the line at a junction with a ‘Stop’ sign and a solid white line across the road. Wait for a safe gap in the traffic before you move off.” (Highway code: Rule 171)

A video survey

The video below introduces the junction:

And this longer video shows an informal survey of behaviour here. I stood for 25 minutes (08:45-09:12, 16 Oct 2020) recording every vehicle which crossed the STOP line (I missed recording only 4 vehicles because of people walking across in front of the camera):

I count as follows:

  • In total 72 people drove (motor) vehicles over the stop line.
  • 64 of the 72 drove straight over the white line, some faster some slower.
  • 6 of the 72 crawled through it very slowly (definitely without actually stopping, but some might argue that they were effectively stopped)
  • 2 stopped fully, but on both occasions only because of a clear need to give way.


Our first observation must be that almost nobody driving obeys the law here, and that only a small percentage of people come close. We might find amusement in noticing that even the driver of the police van (at 2:14 on the video) makes no pretence that they are stopping.

But what more can we conclude? What do our observations teach us? And why does it matter that we think carefully about this?

So let’s think…. are any of the following fair conclusions, and if so which, based only on this survey and not on any other knowledge? And which form of words is the fairest to explain accurately?

  1. Drivers are irresponsible people.
  2. Most drivers are irresponsible people.
  3. All drivers break driving law.
  4. Almost all drivers break driving law.
  5. All drivers break some driving laws.
  6. All drivers ignore laws about STOP signs.
  7. Almost all drivers ignore some STOP signs.
  8. STOP signs don’t work.
  9. STOP signs don’t make people stop.
  10. Almost all drivers ignore these STOP signs.
  11. Almost all drivers break the law at these STOP signs.
  12. Most people who drive here break the law at these STOP signs.
  13. These STOP signs, at this location, don’t work.
  14. These STOP signs, at this location, don’t make people who are driving here stop.
  15. Some people drive through this junction dangerously.
  16. Some additional measures might be required here if we want to prompt people to obey the STOP signs.
  17. Some additional measures might be required here in order to prompt safer behaviours.
  18. Some additional measures might be required here in order to make the junction safe.
  19. Junctions with STOP signs are safe.
  20. Junctions with STOP signs are dangerous.
  21. This junction is safe.
  22. This junction is dangerous.
  23. This junction sometimes offers a rubbish experience for pedestrians.

But why does this matter?

I’m not asking this because I’m particularly interested in STOP signs.

I think that this matters because it teaches us a lot about how to do research about behaviour on our streets – and in particular it teaches us to be really careful at the moment when we look at research about new designs of infrastructure… things like bus stop bypasses (aka ‘floating’ bus stops) and continuous footway (if you’re not sure what a bus stop bypass/continuous footway is then please refer to the explanation at the foot of the article).

(I’m afraid that this is the section of this article where I get a bit more grumpy. Please know that I’m not commenting on any one piece of research, but some general tendencies I observe to exist – many of which aren’t the fault of actual researchers but of those who commission or interpret this research.)

Neither bus stop bypasses, nor continuous footway, are without issues, and they are both new designs to the UK. That makes people cautious. In theory that caution is a good thing.

IF we design these things badly they put people at risk. At bus stops there’s a risk of a collision between someone cycling and a pedestrian – with risks both from the collision and to people taking action to avoid one. If continuous footway is put in the wrong place, or if it’s badly designed, then it can create even bigger problems. At best poor designs create a form of ‘shared space’ where pedestrians have to negotiate with those driving by looking at them and waiting (hopefully) to be waved across. More often it creates something that looks a bit like footway, but which really works as a section of normal carriageway – increasing the chance that pedestrians walk (or wheel) straight out into the carriageway space when they really needed to behave more carefully. And bad examples of both can make things extremely difficult or impossible for someone with sight loss, someone who walks more slowly, or for those (like younger children) who are less able to negotiate a complex situation.

Inevitably, based on this caution, we end up wanting to study how people behave when faced with these designs in the UK.

But this is where we hit problems. What tends to happen is as follows:

  • We see a design which is proven in another country – typically the Netherlands – and because people don’t really understand what makes the design work, or because they have to compromise to get it installed, we create a poor copy.
  • Then we copy the poor copy, creating lots more poor copies.
  • Although we’ve done something new, we use the old words to describe it – so ‘contiunous footway’ becomes to UK designers what they see in the UK, even if it’s completely different to what’s found in the Netherlands. And all sorts of new rules are created for bus stop bypasses, which lead to designs which are very different to what we’d find at a Dutch or Danish bus stop.
  • Then – wanting to see if we’ve done a good job – someone commissions research, or someone who particularly likes or dislikes the design carries this out themselves. The research project looks at one of these designs, or maybe a few of them, probably with some cameras. The research asks really big bold questions like “is continuous footway safe?” – like I might have asked “do STOP signs make junctions safer?” before conducting my video survey.
  • The research comes up with some observations. These might be quite careful and accurate.
  • Either the researchers or people interpreting the research then try to answer the big bold questions – drawing conclusions which are far wider and which are presented as far stronger than is reasonable given the evidence.

In the last few years I’ve seen studies based on work at a tiny number of locations (often just one location), leading people (not necessarily the actual researchers) to make conclusions like:

  • ‘Cyclists’ don’t stop at zebra crossings at bus stop bypasses.
  • Bus stop bypasses are completely safe.
  • Bus stop bypasses are really dangerous.
  • In only 1 in 250 interactions between a pedestrian and someone driving on continuous footway does the pedestrian need to suddenly change their behaviour, so the risks involved with these designs are low.

My point, clearly, is that the behaviours of people cycling past bus stops will change radically, depending on countless DETAILS of the design of the bus stop, the cycle track, visibility, people’s behaviours, how busy the track is, how busy the bus stop is, how careful pedestrians are or are not being, what situations they’ve recently been dealing with, and so on. These design details are everything.

And the behaviours of people driving will change radically, depending on countless DETAILS of the design of whatever they are encountering as they turn into or out from a side road, how busy the roads are, how busy the footways are, whether the situation is a simple or a complex one, and on many many other factors.

Only fairly poor research will try to make conclusions about ALL bus stop bypasses, or ALL designs that continue a footway across the end of a side road based on a few observations of a few (often similar) designs.

Good research will make much more limited conclusions, and most importantly will focus lots of effort on trying to understand what design factors are relevant in each individual situation they study – looking far beyond minor details AND it will consider whether all the designs studied share the same faults.

The worst research will not just try to make conclusions about all examples of one type of intervention, but all ‘cyclists’ or all ‘drivers’ or all ‘pedestrians’ – almost as if each of these is made up of people not like the rest of us, following strange motivations, and exhibiting inexplicable behaviours.

Making research as intelligent as possible

So with that in mind let’s think very very carefully about what we can conclude from our video survey at the STOP signs. I think it’s probably fair to say that:

  • the people driving through this junction, at this time of day, do not stop their vehicles according to the law; and
  • a junction with STOP signs can sometimes fail to cause people driving to stop:

But then really, when we think about this, what we’ve observed is pretty much what we might expect. People who drive are just people, just as people who cycle, or pedestrians, are just people. And we all already know that in many situations people don’t stick to rules. In fact we know all of the following things don’t we:

People are just people

  • Most people don’t stick to rules if they think the rules are silly, or unnecessary, unless there’s likely to be a penalty for breaking them.
  • Many people stick more carefully to rules which other people think are important, particularly if that includes friends or colleagues.
  • If people break a rule which they don’t think is important, and there is no penalty, they are more likely to break it again.
  • If people see other people breaking the rules they are more likely to break the rules themselves.
  • If people feel a rule puts them at a disadvantage for no good reason, or if it puts them at more risk than if they ignore it, they are much more likely to ignore it.
  • People are more likely to ignore rules, despite them being sensible rules, if they can’t immediately see why the rule is sensible.
  • People are sometimes selfish, and power corrupts, so that those who are more likely to cause injury than to be injured are more careless about rules set up to protect the more vulnerable party.
  • People are impatient, imperfect, human, and therefore they sometimes do things which logically we’d expect them not to.

‘Drivers’ and ‘cyclists’ and ‘pedestrians’ and everyone else – we’re just people using a car, van, bicycle, or walking. There’s no mystery.

Unobvious detail matters

Which brings us back to the importance of much less obvious design details, and other factors. An example will illustrate:

People driving toward this ‘courtesy crossing’ in Kirkintilloch rarely stop to allow pedestrians to cross (there is no legal obligation to do so, but the crossing is intended to prompt this behaviour).

Why do you think that is? What can we conclude? Is grey a poor colour for this infrastructure? Are drivers bad people? Is it that courtesy crossings like this just don’t work well?

But hold on a moment – people driving toward this second courtesy crossing, also in Kirkintilloch, on the other hand do very often stop to allow pedestrians to cross.

What’s the difference? The second crossing is the same colour as the first, with the same design, in the same town, on the same road.

The truth is that – as far as I can tell – the colour isn’t the issue, nor the road category, the speed limit, the town, or many many many other factors. What’s different about the second location are other factors. People driving here have already encountered about four of these crossing points by the time they’ve reached this one – and they’ve had time to absorb the idea that they might need to behave differently. The carriageway doesn’t have a white dashed line dividing opposing lanes. There are other crossings like this one close together.

So we’ve learned, we think, a lot about courtesy crossings, but very little of it is about the materials, the colour, or ‘drivers’.

Returning to STOP signs one last time

Let’s look at the junction with the STOP signs one last time.

Based on the ideas above, and thinking more imaginatively, can we make some bigger guesses about what we see here:

  • Perhaps people driving through this junction at this time of day are mostly people who have done so before, and they can’t see why it’s important to stop at the STOP signs.
  • Based on that then it seems likely that truly dangerous situations rarely actually arise here, because if they were common more of the people who drive through the junction regularly would be being more careful.
  • The actual risk of an incident might seem to justify the signs, perhaps because it’s rare but because the consequences could be serious, but people here don’t trust that the signs are telling them something like this.

Some of those thoughts are really interesting I think – and they provide us with lots of interesting ideas about what we might need to change at this specific junction.

None of these ideas are too difficult to check either – checking might be difficult, but it’s surely not impossible – at least if I’m doing a decent piece of research. I can visit the same site on different days to try to collect just a little evidence on how many people are regulars here. I could probably speak to the driver of the waste lorry seen in my first video – there’s a good chance they could be traced, that they drive here regularly, and it would be interesting to get their observations about the manoeuvre they undertook. I could find a very different junction with different STOP signs, and evaluate differences in behaviour.

Or even better, what about this: I could find a location with STOP signs where people do regularly stop, and look at the differences in design. Wouldn’t that be interesting. I might need to look to other countries, but that would surely be a properly valuable exercise?


So here’s what I’m asking for in a nutshell:

If you’re studying behaviour on our roads, or if you’re interpreting research that someone else has carried out, or if you’re reading research or interpretations for your own interest, please remember that:

  • the people who are being observed are just people, whether they are walking, using a wheelchair, driving, cycling or jumping on a pogo stick;
  • given the above, it might help to be careful about your language, particularly if you’re studying something contentious (the word ‘cyclist’ is a particularly difficult one in this context, given that there are many existing stereotypes suggesting ‘cyclists’ are somehow different from other people);
  • behaviours will vary considerably between people, and according to all sorts of other complicated changeable factors like weather, time of year, and perhaps even day of the week, so record these, and study what this changes (if anything);
  • when making conclusions be very clear that what influences behaviours will be a vast array of complicated factors, many of which are difficult to observe, and some of which may be far more significant than those you first thought to study;
  • if the design idea that people want you to research is standard and common in other countries – particularly if it’s been well established for decades – then you really should include an evaluation of how the new designs you’re looking at in the UK compare to these, and note what behavioural differences you see.

So am I right? These are just my personal thoughts – ideas which arose watching this one junction, and looking at a limited number of pieces of research. Are you a researcher – let me know why I’m right, or wrong. Or are you someone with opinions about bus stops or continuous footway. Let me know if you think there are pieces of research which deal with the issues I’ve raised.

Explanation: Bus stop bypass / continuous footway

A quick explanation for those who don’t know…

If we are going to support cycling in the UK (or other countries) then we’re going to need to solve a few obvious problems. These include two very obvious ones.

Continuous footway

When people are cycling on a cycle track beside a big road then they need to be able to keep going, just as anyone driving does, as they pass junctions with much smaller roads – where there are no traffic lights to control traffic.

To enable this at side roads, the smaller road can be redesigned so that it carries very little traffic, which moves slowly, and which is only going to or coming from local destinations. As a result the pavement (i.e. footway) of the bigger road can continue unbroken alongside the bigger road, over the end of the smaller side road. People driving into or from the side road do so over the pavement, and take an appropriate amount of care because it’s obvious that this is what they are doing. If designed properly having this continuous footway, rather than having the footway broke, is invaluable to pedestrians. A cycle track can also be effectively part of this continuation of the pavement/footway (being very separate, but associated).

See “Design Details 1” for more, or “Design Details 2” if you just want a few images.

Bus stop bypass

When people are cycling along a cycle track, which is beside a big road, and where this road has bus stops along it, then the cycle track needs to keep those cycling on it safe even near the bus stops. We can’t just insist that people cycle on the carriageway in among the other traffic at those locations – not if we want the cycle track to feel safe.

One key solution to this is to take the cycle track around the back of the bus stop – to build the bus stop between the cycle track and the carriageway. Thus the cycle track bypasses the bus stop on the carriageway. There are other solutions too, such as having people alight from the bus directly onto the cycle track (with appropriate laws and designs to support this).

For some examples of different designs dealing with this issue in Copenhagen see “Copenhagen bus stops“.

See also…


  • I welcome comments, even if you want to tell me (politely and constructively) that you disagree, scroll below to view what others have said.
    If you landed here from a Twitter link on a mobile device you may need to press ‘Leave a comment’ below to see the comments on this article or you can reload the page to see them.


  1. Hi Robert, very interesting article and i’m glad i took the time to read it.

    I’m not sure i have ever encountered a junction with stop signs and a solid stop line (maybe i have and behaved in the same manner as those in your video). Perhaps the unfamiliar nature of the arrangement could contribute to the non conformance? i.e. you expect a give way situation at a junction like this so you treat it that way? I believe this arrangement is used frequently in the USA (I’ve seen it in movies and TV) and it would be interesting to see if it is obeyed there.

    That is my opinion on a lot of cycle infrastructure, the inconsistency of unfamiliar designs mean that people do not instinctively know how to react in certain situations (all users, pedestrians, cyclists and drivers).

    I hear a lot about highway code rule 170, 183 etc. as if this is enough to put peoples lives in danger with the comfort of having the law on your side, or the moniker of priority as if this is some kind of shield against danger. If people (perhaps myself included) cannot obey a very well signed rule then what chance is there that an average person knows individual rules of the highway code?

    I would agree that design is the key but there must be some kind of tipping point in terms of the introduction of new design? such as continuous footways? i.e. if you have different side road treatments in close proximity with different rules regarding priority how do we know that all users will clearly understand what to do and react accordingly?

    unfortunately, in my opinion this is a behavioral issue, as i think you have suggested (i may have picked that up wrong). It requires a lot of education and could require a generational change rather than the swift change many want. for example, dutch children would have experienced a lot of road safety issues as a pedestrian/as a cyclist before growing up to drive and therefore will be naturally more empathetic to the overall needs of all users and this may have contributed to the road conditions they now have?



    1. Thanks – actually the way I think of these things is that most of the behaviour we need has to be created by the design itself. For example, here a distinct and obvious narrowing of the carriageway, with a lovely solid looking bollard, meaning that people need to turn more sharply, would change how the junction feels. And that would be good IF we actually thought the best solution was to make people stop.

      There’s an element of what you call ‘education’ involved, yes, I agree… but I think that education has to come from experience not other places… so if you encounter lots of continuous footway based junctions then when you see one you know what to expect… but this only works to educate in the desired behaviour if most of the examples we encounter are set up properly to create the desired behaviour. The stop sign here is an interesting example – it feels as if it’s not necessary to stop here, and people clearly go through here all the time without stopping and without incident, so actually what happens is that experience of this junction teaches people that stop signs can be ignored more generally. Not good!

      On top of these things we can add some more traditional education – driving lessons and similar – to fine tune all of this. But that’s all it can do. Just fine tuning.

      That’s what I think anyway 🙂


    1. HI Simon – good to hear from you, and thank you.
      I tend to believe that the solutions to bus stop issues aren’t technical ones – they will be created by good design, and perhaps some changes in laws and standards. I keep meaning to write a blog post about this… maybe it’s time to do so?


  2. Robert,

    Such an excellently thought through and clearly written blog again. A wake up call to all researchers to develop appropriate and precise research questions, to recognise that the real world is one big natural experiment with many confounding factors, not the least of which is human behaviour which can vary between individuals, but also vary for the same individual depending on their circumstance on that day, to recognise the limitations there always are with research findings, and to then always try to specify what further research needs to be done to investigate those things on which conclusions could not be drawn. You could say that previous very long sentence of mine was the antidote to your beautifully clipped sentences!

    Despite the above, I would argue that we can begin to tease out quite a lot (NB quite a lot, not everything) with careful and full exploration of as many variables as are possible to collect within available budgets, and through statistical modelling that identifies the significance and effect size of the main effects of relevant factors, and the effect of the interactions between different factors relating, in our case of street use, either to a range of person, design or regulation related explanatory variables. Even within that process we still need to carefully acknowledge what we have not been able to measure. And we need some sort of corroborating qualitative evidence to support theories about causality (which you suggest of course with your interviews with local road users.)

    At the most atomised, all we can accurately say is that ‘this person, in this situation, behaves like this’, and that will be based on a record of one observation. Now that is fine, but it is what we might then call anecdotal and it cannot be applied to other situations, and it is not useful to help develop design rules, or approaches to regulation, or road user training. To do that for all roads and for all people, we need some sort of idea about where the central tendency lies, and where the extremes are, boundaries we would rather not cross with other than vanishingly small probabilities. What the analysis is trying to do that I outline above is to attempt to group in to aggregations of reasonable size these event chains that can be reasonably deemed to be similar, and from there general rules for policy application can then be developed.

    However, there is a strong element remaining in my mind always that sometimes we are kidding ourselves, and far to often we are revealing things that look appropriate statistically, or logical from some sort of causal chain analysis point of view. There is food for thought here: https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.0020124 (Admission: I have not read this paper in detail).


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 )

Google photo

You are commenting using your Google 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