I've discussed this briefly here, but getting a congruent answer has always intrigued me. I've been reading loads of Daniel Kahneman's work recently & have finally got to an evidence-based model where theory predicts experience and practice: where the apparent dichotomy is explained.
Let's start from the top: are recumbents less likely to be hit by drivers? YES!
Now let's get into the why...
Kahneman writes about a 2 mind system, which is described elsewhere as a dual process account of reasoning. There's a fairly decent introduction here. The basics that you need to understand are the characteristics of the two systems. I'll use the terms that Stanovich and West coined that are used by Kahneman: system 1 & system 2. For more detail, there's plenty of research out there, but a basic overview would be that system 1 is the implicit, autonomic function (the stuff you do without thinking (like balancing a bike)) and system 2 is the explicit, effortful system. A simple example would be to ask you to complete the following sums:
(the answer to the second one is 408, in case you weren't sure (and the first one is 2, but you probably knew that)).
The other thing that's relevant here is how long it took you to solve the 2 equations. Even if you solved 2 using a calculator, it took seconds longer, right? Those seconds will mean a lot further down this piece. In addition, you probably considered different ways to solve the problem before jumping into one. You may even have evaluated it as you went through if you chose to do it in your head. This consideration is also quite critical further on.
OK, so far, so obvious: nothing shocking here is there? & it's all quite evidence-based and uncontroversial, but how does it relate to bikes....
Well, before we answer that, let's think about how driving works in this model. Well, just a thought before this:the brain has been around some 7 million years; how much of this has been spent at >30mph? Or more specifically: primates have been around for a few hundred thousand years; how much of this has been spent at >30mph? & finally: homo sapiens have been around for about 150,000 years; you know the question.....
There can have been no evolutionary advantage to being able to cope with these speeds & no reason to suggest that we might be able to deal with them competently. If we can add that dense urban environments are also very new, so 30mph in these environments is new condition in new environment & doubly (or doubly^2) odd....
(As a very limited biological aside, to my knowledge, animals who can manage these speeds get complete tunnel vision & fail to recognise all sorts of risks. I say "to my knowledge" deliberately since I know very little biology, so am happy to be corrected).
So since the brain is in an utterly alien environment, what does it do? What it always does when it can't answer questions: substitute and take shortcuts. So if I'm driving at 30 mph in a dense urban environment, what am I looking at? If you think you pick all of the relevant things, I have met you before. We pick things that are perceived as risks or threats and respond to them. There are those of you who have trained system 1 to look for more things and are more likely to pick important stuff, but you're unusual (remember), most of us don't.
So, if I'm riding a bike, can I rely on being seen? No. Brutal but simple. Even in primary position, you have no guarantees (in primary, you have an advantage in that all contact is seen as threat (socially or to insurance premiums or to delay in journey etc), but you're smaller, so less threatening. If you're not in primary, or, worse still, if you're in a weak secondary, don't expect anyone to see you. But you wear hi-viz, right, so it must be more visible, so I must be easier to see, right? (Ask yourself, how common is hi-viz? Can you expect something so common to crowd into the huge amount of potential things to be noticed by system 1?). Well, sorry, no. Hi-viz, is more visible in certain conditions, but visible and being seen are not equal. Being seen is all. Which means being seen as a threat or a risk, which means that system 1 will ask system 2 what to do. So, if you're on an upright, primary, looking like a Police officer or looking like an unconfident, blonde woman (thanks, Dr Ian Walker) is all that'll work.
So where do recumbents fit in all of this? What's the first reaction to a bent from most folk? WTF seems to summarise almost all of it for me. WTF is a fantastic reaction for a cyclist (or pedestrian) since it means that system 2 will be engaged. The first thing that will happen is slowing down whilst system 2 works out what to do. Slowly. Whilst accessing moral systems too.
We saw above that when system 2 is engaged then the outcomes tend to be more considered. Slowness, does not imply slow reactions, however; system 1 will be taking care of the basics whilst system 2 works out how to resolve this issue appropriately.
This model would predict then that under normal circumstances, system 1 will handle overtaking cyclists and will do this however it has been trained to undertake that manoeuvre. Where, however, something, threatening or unrecognised is seen, then system 2 will handle the manoeuvre. System 2 is more likely to undertake the manoeuvre with more care.
Furthermore, as familiarity increases and reduces the likelihood of system 2 engagement, system 1 will be being trained to pass 'bent cyclists with space.
As a caveat, there is no substitute for good road craft in staying safe on the road. If I have decent road craft (there is no substitute for good road craft: anyone cycling without a strong appreciation of this is daft; find a tutor if unsure, there's a bunch of them/us around), I'm prob fine. Especially if my road craft enables me to be aware of where I can and can't be seen & helps me to adjust for this... This model is looking more a relative probabilities in like for like situations. So I am certainly not implying that recumbent riders are safe by definition, but that they are less likely to be subject to ill-considered manoeuvres by drivers than if they were in the same position on an upright bike.
Which is something worth having.
Following feedback from a friend, I'll put another post up about the relative psychology of riders and how this impacts on safety; watch this space.
(I know there's a lot that's over simplified & under-explained, I was trying to make it readable: please ask if you want clarification. I'm confident in the theory & stake my life on it most days, so feel free to ask for explanation/clarification)
*Edit, there is a follow up post here*