You can have an argument about what is the cause of most crashes, or of an individual crash. But it's unanswerable. Journalists usually choose the wrong question, so that it is unanswerable. So the issue seems difficult and complicated ... and interesting.
Normally there is an obvious boring answer.
In this case, you can just do statistics on the problem. Assume the cause is bad junction design. Then make an algorithm which answers "which features of a junction are correlated with which types of accident". Then make the assumption that the cause is bad cars, bad people, bad behaviour, etc. You'll find all the answers that way.
You'll find that it's almost impossible to crash a modern car. ABS, parts reliability, and modern wide tyres.
A certain generation remembers cars which just lost control or handled unpredictably or skidded in the wet. They believe in endlessly tweaking car design for safety. This is obsolete thinking. But the policy-makers of this world are (usually) the old.
For example some people say that speed kills. It's based on an observation that stopping distance is correlated with speed. This is wrong.
You need to gather statistics on how fast people travel versus how often they crash. You make a plot shownng speed versus rate. You control for different vehicle, road, and person types. This plot tells you the range of safe speeds.
So what's true in the elementary physics is irrelevant in sophisticated problems.
You probably find that the safe speed depends on the type of driver, and is generally much faster than the posted limit. Which is what everyone intuitively knows anyway.
You could approach it a different way. Correlate rates of accidents with rates of speeding tickets. If you could show that people who get speeding tickets have more accidents, you'd have a decent argument that speeding tickets are justifiable.
All this is still myopic. It focuses only on accident rates, forgetting things like utility and enjoyability. For example speed, throughput, flexibility, and other things are just as important as safety. Finding the safest possible solution is the easy part. Then you have to decide if you want it or not.
This journalist is complaining that others are looking at the issue wrong. Well he is wrong too!
If I can give a personal opinion, I think this is a weakness of schooling. People readily accept explanations like "stopping distance is proportional to speed squared, so reducing speed reduces crash severity", which is quite a difficult technical argument. But something like "to find the causes of accidently you find the correlation between each accident type and each factor", people can't grasp that. They ignore that type of insight.
That's because some percentage of people study physics or applied maths as teenagers, so physics thinking has penetrated society. Other types of maths like statistics economics and probability, are not understood. So society is deaf to the insights and benefits they could provide.