13 SMIDSY – looked, saw and turned anyway

SUMMARY – motorcyclists may believe that drivers don’t care about bikes or see them as a source of danger… they further believe the lack of empathy means they don’t look hard enough… this belief does not appear to be founded on any solid evidence, just speculation… gap acceptance studies show no right-of-way violations in thousands of passes… statistically, one study indicates that car drivers crash with other cars just as often as with motorcyclists and that they do not crash with motorcycles more often than motorcyclists with other motorcyclists… 

Whilst most investigations have focused on perception problems as explanation for why it is that drivers ‘look but fail to see’ motorcycles, or misjudge their speed and distance resulting in ROWVs, some studies have suggested that there are what might be called ’empathetic’ explanations. Motorcyclists are popularly viewed as a high risk group and ‘thrill seekers’, and studies have indicated that car drivers have negative attitudes towards motorcyclists. By extension, it has been suggested – quite seriously – that whilst some car drivers don’t care about motorcycles and their riders to ‘look hard enough’ for them at an intersection, others do not regard motorcycles as a potential danger. Having seen the approaching motorcycle, but cocooned in a car and not feeling threatened, the driver is then postulated to make a deliberate choice to turn in front of the PTW with an intention to force the motorcyclist to cede right-of-way, in a manoeuvre the driver would not have attempted in front of other vehicles.

The earliest expression of the ‘threat’ aspect of the SMIDSY collision seems to have come from Hurt and Dupont (1977). They wrote (and I’ve emphasised the ‘low threat’ part of the statement:

“If the approaching motorcycle and rider blend well with the background scene, and if the automobile driver has not developed improved visual search habits which include low-threat targets (such as motorcycles and bicycles, as contrasted with the high-threat targets presented by trucks and buses) the motorcycle will not be recognized as a vehicle and a traffic hazard exists.”

Not surprisingly, this theory was promptly given some prominence in the motorcycle press, and subsequently picked up by rider action groups who have issued ‘calls for action’ to address the assumed problem. Equally unsurprisingly, a few motorcyclists believe that they are ONLY put at risk because of a lack of care or though the deliberate actions of other drivers, and also that the perceptual problems previously discussed are just a convenient excuse for ‘looked but failed to see’ crashes.

How real is this empathetic explanation of the problem? Is there any factual basis for the belief that drivers deliberately target PTW? I don’t know, and I don’t think anyone really does. I’ve only been able to locate two later research papers which suggest something along these lines. In the earlier, Wulf et al (1989) looked into the issues surrounding collisions between cars and PTWs. Whilst suggesting several explanations you should now be familiar with:

  • the smaller physical size of the PTW compared with a car makes it less conspicuous
  • judging of distance and speed with a small frontal area or with only one headlight is difficult
  • because of the low numbers of PTW’s in traffic, car drivers do not expect to meet a PTW and are therefore less prepared to notice or recognise a PTW
  • because PTW’s seem to ‘lack relevance’ to most car drivers

Wulf also suggested that a car driver is inclined to ignore the presence of a PTW since the impact of a collision with a PTW is less threatening than with a car.

But understand this. Neither Hurt & Dupont nor Wulf ACTUALLY PROVIDED ANY EVIDENCE that for their theories that car drivers deliberately put motorcyclists at risk – they only advanced the POSSIBILITY that it COULD happen, not that it DID happen. I cannot find any study of junction collisions that has actually produced any kind of statistical figure for a deliberate ROWV as a proportion of the total number of junction collisions, only speculation that deliberate action could be an explanation.

Another popular belief is that drivers are able to tell the difference between a police motorcyclist and an ordinary rider and make decisions accordingly. The source for this is probably a paper by Caird and Hancock (1994) who suggested that compared with the average, some types of rider do not have as many collisions with cars:

“Within the motorcycle population, there is an under-representation of police motorcyclists and – somewhat paradoxically as they represent low contrast – Hell’s Angels.”

The implication is that having identified a police motorcycle – or a machine ridden by an Hell’s Angel – drivers deliberately hold back from a manoeuvre they would make in front of an ‘ordinary’ motorcyclist. It occurs to me that a plausible alternative explanation for the low incidence of collisions involving US police and Hell’s Angels is that the police ride in pairs and the Hells Angels frequently ride as a pack. Both a pair of riders and a pack are not only more visible, but only the leading riders in the pack are directly vulnerable.

I did find this statement [and as always, I would welcome new information] which comes from a US-based legal firm, HG.Org, who state:

“Deliberate hostile action by a motorist against a motorcycle rider is a rare accident cause. The most frequent accident configuration is the motorcycle proceeding straight then the automobile makes a left turn in front of the oncoming motorcycle.”

It’s unsupported – no reference to original material is provided. But considering the source – a US-based legal resources website – and the history of litigation in that country, I would not have expected the article to play down the possibility of ‘deliberate hostile action’. In fact, given the role of the business, I would have expected the reverse, that they would play up the ‘deliberate action’ angle.

But car drivers have more crashes with motorcyclists than cars, yes? That’s an article of faith amongst bikers. So I’ll take the opportunity to remind you about an observation from a recent paper from the Netherlands that I mentioned earlier:

“The majority of motorcycle crashes are crashes with a car. In these crashes, the police register the car driver as first offender more often than the motorcyclist. So in absolute numbers, many motorcycle crashes seem to be caused by car drivers. However, when adjusted for exposure, car drivers do not crash with motorcycles more often than motorcyclists with other motorcyclists. An analysis of different crash causes at intersections indicates that, relatively speaking, car drivers fail to give priority to a motorcycle as often as to a car.”

Let me highlight the two important statements. Firstly, De Craen and her team found that:

“relatively speaking, car drivers fail to give priority to a motorcycle as often as to a car.”

That makes it hard to see how drivers could be ‘singling out’ motorcycles at junctions with deliberately risky manoeuvres. They also found that:

“…when adjusted for exposure, car drivers do not crash with motorcycles more often than motorcyclists with other motorcyclists.”

For the vast majority of motorcyclists who believe that they are at risk from car drivers because “drivers don’t look properly for bikes” and that by contrast because they are bikers “we see other bikes” and don’t run into them, that will be a statement that is very hard to digest, but the study is based on police data. And just so you know that I have not misquoted or selectively quoted from the paper, here’s a screenshot of the section from relevant page:

motorcyclists crash with motorcyclists

Nevertheless, in the interests of balance I will point out that this is just one paper, by one team of researchers, using data from one country. It would be useful for the analysis to be repeated with fresh data from beyond the Netherlands.

de Craen et al (2011) went on to point out that motorcyclists are at greater risk ONLY in one scenario – when an oncoming car turns across a motorcycle’s path into a side road:

“In one situation motorcycles seem to be at a disadvantage compared to cars. This is when a car makes a left turn, and fails to give priority to an oncoming motorcycle. This specific type of crash occurs more often when the oncoming vehicle is a motorcycle than when it is a car.”

Their explanation was that the head-on view of the motorcycle which is “narrower than a car and has only one front light instead of two… gives less information about speed”.

But in the scenario below – the classic SMIDSY with the emerging car – it seems the driver is just as likely to make an error in front of another car as in front of the bike. That’s absolutely NOT what most of us would expect to hear. Of course, the consequences for the PTW operator are likely to be far more serious than for a driver of another car. That’s what loads the injury statistics, not actual crash rate.

Dealing with Traffic10

In any case, the fact is that most ROWVs in front of PTWs will NOT result in a crash. In the majority of incidents recognised by the rider as a ROWV, zero action – or only a trivial change of speed and direction – is required to avoid a collision. Emergency action is relatively rare.

So the key point then is not so much whether the rider THINKS that the driver should have waited (what we might term ‘bad behaviour’ because the motorcyclist considers the manoeuvre to be a violation of the traffic code) but whether the manoeuvre poses a GENUINE threat requiring a serious intervention by the rider to avoid a collision. What a rider needs to know is at what point does a ROWV change status. Is the manoeuvre:

  • irrelevant?
  • a mild annoyance?
  • a potential threat? (if the rider doesn’t take some mild defensive action)
  • a genuine threat? (requiring emergency evasive action)
  • unavoidable (potentially resulting in injury or death)

The answer is that although there have been a number of ‘gap acceptance’ studies, which look at the driver’s response to a motorcycle and how close to the car the motorcycle can get before the driver rejects the turning manoeuvre, my searches have turned up very little that investigates the SMIDSY from the rider’s perspective.

The only study I could find that actually asked motorcyclists what they thought was the early 1980 TRRL study I have mentioned before. They sent out a questionnaire to a sample of riders, asking:

“…whether or not they had experienced any accidents involving another vehicle or pedestrian in the last three months, and if so to declare:a) how many accidents.b) on how many occasions they were wearing their safety clothing at the time of the accidentc) what were the causes.”

They went on to analyse the responses:

“As accidents occur relatively infrequently, it was decided to investigate both accidents and near-misses where the motorcyclist considered that the incident was due to another motorist failing to see them. During the experimental period of one year 33 per cent of the sample suffered at least one such incident/accident or near miss. The distribution of the numbers of incidents reported by each respondent is given in Table 1 (reproduced below).”

Table 1 Frequency of accidents or near misses in last year / Number of incidents

No response4.7

In all, 150 responses were analysed, with 84% being reported as near-misses and 16% as accidents. It looks pretty bad, and in fact that authors assumed it was worse than it looked:

“It is quite probable that the situation is even worse than the figures above suggest as it is unlikely that the respondents were able to recall all such incidents which had occurred during the last year.”

A few things caught my eye.

Firstly, the survey asked participants to respond to a loaded question. It’s likely that riders who DIDN’T have an incident wouldn’t respond and you’ll note that there is no record of any motorcyclists who reported zero incidents. Whilst the sample isn’t big (just 150 motorcyclists), it does seem just a little unlikely that not one single rider had no crashes and zero near-misses. If every single respondent had at least one incident, that would explain the results, but the results are almost certainly skewed by a lack of reporting from riders who didn’t experience a crash or near-miss. I’m surprised that the zero data point is not recorded, all the same.

Secondly, only around 60% of the incidents reported occurred at junctions. The remainder involved overtaking collisions and rear-end shunts.

Thirdly, all those riders reporting eight or more incidents are lumped together. This gives a very strange shape to the frequency curve – a lot of riders having few incidents, then very few riders having four to seven incidents, then another large group having eight or more. To me, this seems an unlikely distribution. I would have expected the opposite – a ‘bell-shaped’ curve with relatively few riders at either extreme having either zero/few or many incidents, and the majority in the middle.

The authors didn’t seem to be concerned about this, but I’m doubtful that we can draw any valid conclusions from the near-miss reporting. It’s easy to tell the difference between a crash and no crash, but how near is a near-miss?

This is, of course, a highly-subjective question. At what point does a driver’s decision to pull out in front of a motorcycle change from a being ‘well-judged manoeuvre’ to what the motorcyclist perceives as a ‘near-miss’?

If you asked me, I’d class a near-miss as an incident where I had needed to brake hard or swerve to avoid a collision. Having spent many years in London as a motorcyclist and more lately a driver, I am used to heavy traffic, and tolerant of instances where drivers turn across the path of my vehicles even when I had right-of-way, even where the margin for error is limited. If you think about it, even if the margin for error was limited, it MUST have been adequate. If it had been inadequate, the manoeuvre could not have been completed without a collision, or at least some evasive action on my part.

We can turn to a study carried out by Ouellet and Kasantikul (2006) using a methodology similar to Hurt’s work in California but this time studying crashes in Los Angeles and Thailand. In both study areas, teams of investigators travelled to an accident scene immediately after a collision and conducted an analysis of the incident independent of the police investigation. In total, 969 collisions involving 1082 riders and 399 passengers in six different regions over a twenty month period were investigated (including motorcycle – motorcycle collisions – around one-quarter of the total).  In Los Angeles, 900 accidents were studied.

One of the key results was a determination of the time from ‘precipitating event’ (ie, the moment the collision partner began to turn into the rider’s path) to impact. In both Los Angeles and Thailand, the mean (average) time from precipitating event to impact was just under two seconds. Around one in five riders had more than three seconds from the precipitating event but still crashed, and between 1-5% had more than four seconds.

Given the entirely different driving conditions in the two cities, where there are very different levels of driver skill, traffic rule enforcement and deterrence, it’s interesting that the time from the precipitating event to the collision is almost exactly the same. The implication is that the ROWV is not dependent on ‘driver behaviour’ so much as human error.

What is alarming is that Hurt reported that the majority of riders in 900 crashes demonstrated poor collision avoidance skills. About one-third took no evasive action at all. Most of those who did take evasive action either chose the wrong action or executed their chosen action poorly, or both. Worryingly, rider training had no effect on collision avoidance performance. Similar results come from Germany. In a sample of 502 injury accidents in 2001-2002 (reported in ‘SafetyNet (2009) Powered Two Wheelers’), 279 of the motorcyclists had taken action to avoid a collision and 54 eventually lost control.

An important factor is the time it takes for the rider to respond. Most emergency stop measurements start with a rider who is expecting to stop. So all that’s measured is the rider’s physical reaction time (the time taken from detecting the signal to stop, to beginning to brake) plus the mechanical stopping distance (how hard the rider brakes, how good the brakes are etc). But it’s by no means guaranteed that in a genuine and unexpected emergency that the rider will PERCEIVE the need to take action instantaneously. In fact, this ‘recognition’ time can add anything from one to three seconds to the time it takes to stop. The riders in the studies just mentioned took longer to stop than the figures in the UK’s ‘Highway Code’. If you recall the gap acceptance infographic…

Gap acceptance

…you’ll note that I labelled five seconds as ‘safe’ and three seconds as ‘at risk’. This is the source of those figures.

For a motorcyclist less-experienced in heavy traffic (or one perhaps tolerant of what they see as ‘impatient drivers’), it’s easy to see how a driver accepting an ‘adequate but limited’ gap and the resultant proximity of the turning vehicle, could be perceived as a threat, even when the manoeuvre is successfully completed. From there, it’s no great leap to see how the motorcyclist could interpret this as either ‘careless’ or even a deliberate decision to put the motorcyclist at risk, even though that was neither the intent nor the belief of the driver.  But does any of this lend any credence to Wulf et al’s belief that the less-threatening nature of a collision with PTW means that drivers are likely to ignore the consequences of a collision with one?

Although we have to be careful assuming that laboratory studies accurately reflect real-life events, the evidence from at least one research study actually points the other way. Crundall et al (2012) tested three groups – 25 novice car drivers (mean age 20.6 years and a mean experience level of 1.6 years), 25 experienced car drivers (mean age 33.4 years / mean experience level of 14.8 years) and 24 ‘dual drivers’ with significant experience with both cars and motorcycles (mean age of 44.9 years / mean experience level of 25.7 years with cars and 20.0 years with a motorcycle). The study showed the subjects a number of videos which include ten scenarios with conflicting motorcycles, ten with conflicting cars, and ten with no conflicting vehicles. Another 42 clips were randomly interspersed. The study tracked response times reflecting when the participants thought it was safe to pull out were recorded, and the participants’ eye movements.

Although the results indicated that ‘dual drivers’ were more aware of approaching motorcycles and less likely to violate their right of way, Crundall et al also reported that “the most immediate finding from the analyses was the greater caution given to conflicting motorcycles than to conflicting cars. Both the percentage of safe responses and the [reaction times] reflect a greater safety margin in responding to motorcycles…” In other words, rather than taking greater risks, all drivers were more careful around motorcycles than when approaching other cars.

Let’s have another look at that “33 per cent of the sample suffered at least one such incident/accident or near miss” figure. It means that 67% of the respondents had ZERO incidents that were worth recalling. And what about those riders who reported incidents – how many miles did they manage to cover without incident?

The fact is that whilst ROWV incidents stick in our minds, the non-incidents vastly outnumber the collisions or even genuine near-misses. I don’t know of anyone who has calculated the ratio, but my guess is that it’s in the ratio of tens of thousands to one. Do I have any evidence for that claim? The early TRRL study also looked at what they termed ‘gap acceptance’ by sending a motorcycle around a large roundabout behind a car:

“An experimental motorcycle was introduced into the traffic flow and a study made of responses to controlled gaps which were presented… The 250 cc motorcycle followed a lead car, which was necessary to stop vehicles at the junctions and to enable the following motorcyclist to control the presented gap. The speed of the car and motorcycle was approximately 25 mile/h. Gaps between 1.5 and 5.0 seconds were used in the trials… In total 1,854 passes were recorded.”

What they found was that from a total of 1,854 opportunities, there were a total of 352 acceptances and 922 rejections with NOT ONE SINGLE ROWV during the study. (My emphasis.)

What was interesting was that their analysis of the data showed:

“…that the use of fluorescent clothing or a dipped headlight on the experimental motorcycle had no significant effect on the sizes of gaps accepted. The absence of any detectable changes in the gap acceptance behaviour of the motorists joining the traffic stream suggests that if the motorcycle is perceived at the junction when [typo for ‘then’?] the use of high visibility aids has no effect on this behaviour.”

I think there is a typo in that sentence, but the implication is that having seen the motorcycle, the conspicuity aids did not affect the driver’s subsequent decision-making in terms of accepting a tighter or wider gap.

Unfortunately, although rating none of the gap acceptances as a ROWV, the TRRL study doesn’t show which time gaps were accepted and which weren’t. The more recent study by Sager et al (2014) looked at motorcycle positioning and how that affected drivers’ decision-making. Working from the junction geometry and estimating the likely acceleration by the driver, they calculated that:

“…a three-second gap in a stream of oncoming traffic would not allow for the safe execution of a left turn, that a four-second gap would allow for the safe execution of a left turn, but leave very little safety margin, and that a five-second or more gap in the stream of traffic would allow for the execution of a left turn and leave a reasonable safety margin.”

These figures have to be treated with some care, as roads clearly vary in width and a safe gap will also vary with speed, but it’s a useful starting point, particularly as it seems to fit well with the work already mentioned by Ouellet on his ‘lane positioning hypothesis’ where he concluded that the ‘at risk’ zone for motorcyclists begins when they are approximately three seconds from the junction.

From what I can find in the research, in the UK at least, my belief is that collisions resulting from deliberate driver decisions to turn into an unsafe gap are far outnumbered by the perceptual errors that create ‘looked but failed to see’ problem.

Of course, when confronted with a ROWV that results in a genuine near-miss that requires emergency braking or swerve – and potentially a crash – the shock to the motorcyclist is significant and I’d suggest we are hyper-sensitive to ‘near-misses’. That being the case, it’s likely that as motorcyclists we are far more likely to recall these incidents, whilst at the same time forgetting just how many times the ‘driver looked, saw and did not turn’.


Since you’re here, I’ve a small favour to ask. If you feel able to make a small donation to the upkeep and continued development of SOBS, why not buy me a coffee? Each contribution is much appreciated. Each cuppa keeps me awake and writing! Thank you.


Continue your journey to the next page…

Last updated:

Friday 1 October 2021 – minor edit for clarity
Wednesday 30 June 2021 – minor edit for clarity and to correct spelling
Wednesday 1 May 2019 – minor edit for clarity
Saturday 27 April 2019 – emphasis added to the de Craen et al (2011) study results, screenshot of relevant section in the paper added Friday 14 December 2018 – added information about ‘deliberate hostile action’ and HG.Org reference, section rewritten for clarity
Saturday 15 December 2018 – added crash statistics comparing collisions between motorcycles and cars and cars and cars
Friday 23 November 2018 – updated with references to Wulf et al (1989), Crundall et al (2012),


de Craen, S., Doumen, M., Bos, N. & dr. van Norden, Y (2011) “The roles of motorcyclists and car drivers in conspicuity-related motorcycle crashes” SWOV R-2011-25

Crundall, D., Crundall, E., Clarke, D., Shahar, A. (2012) “Why do car drivers fail to give way to motorcycles at t-junctions?” Accid. Anal. Prev. 44 (1), 88–96.

HG.Org. (????) “Little known facts about motorcycle accidents” https://www.hg.org/legal-articles/little-known-facts-about-motorcycle-accidents-31124. Retrieved 14 December 2018

Hurt, H.H., DuPont, C.J. (1977), “Human Factors in Motorcycle Accidents”, SAE Technical Paper 770103, doi:10.4271/770103.

Ouellet, J.V., (1990) “Lane positioning for collision avoidance: an hypothesis”, Proceedings: The Human Element: 1990 International Motorcycle Safety Conference, vol. 2, Orlando, Florida, pp. 9.58-9.80.

SafetyNet (2009) “Powered Two Wheelers”, Project co-financed by the European Commission, Directorate-General Transport and Energy

Sager, B., Yanko, M. R., Spalek, T. M., Froc, D. J., Bernstein, D. M., Dastur, F. N. (2014) “Motorcyclist’s Lane Position as a Factor in Right-of-Way Violation Collisions: A Driving Simulator Study”, Accident Analysis and Prevention 72 (2014) 325-329.

Wulf, G., Hancock, P.A. & Rahimi, M. (1989) “Motorcycle conspicuity: an evaluation
and synthesis of influential factors”, Journal of Safety Research, Vol. 20/4


The material is free to all to access and use under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. That means you can share it with your family and friends, and re-use it for club magazines and websites, so long as you acknowledge the source and author and include the same Creative Commons license in the derived works.

Please note, this Creative Commons license excludes commercial use. If you wish to use any of my work for commercial purposes, including (but not limited to) articles in pay-for magazines or commercial websites, please contact me.

Creative Commons statement

Kevin Williams has asserted his right under the Copyright, Designs and Patents Act, 1988, to be identified as the author of this work.

Photo credit Paul Townsend https://www.flickr.com/photos/brizzlebornandbred/20001313491