The test to which cycle helmets are subjected - and which they often fail - are all basically similar. The helmet is attached to a headform (usually aluminium) which is attached to a rod with a fulcrum at one end. The headform is dropped, rotating around the fulcrum, and at some point impacts onto either a flat surface or a profiled anvil. The test mandates the amount of energy the helmet must absorb. It is not supposed to break.
The amount of energy is equivalent to hitting the ground at around 12mph.
That does not mean a 12mph impact, it means an impact with the ground after a fall from a stationary or slow-moving bike.
Impact energy rises with the square of speed. One helmet promoter claimed that a helmet would reduce the energy of a 30mph collision to 18mph, but this is wrong. It would reduce it to the equivalent of 27.5mph - 302 - 122 = 27.52.
It's pretty obvious that this is not a device designed to deal with collisions with cars, and I'm not aware of any manufacturer who sells them as such. However, most serious and fatal injuries are a result of collisions with cars, and many helmet promoters try to assert that helmets are highly effective in such cases. This is a hard sell for me; I can believe that there may be some protective effect in some crashes, but the idea of relying on something so obviously inadequate to the task and promoting it strenuously as an effective measure, is much more problematic.
As an aside, polystyrene foam (which is what helmets are made of) absorbs energy fairly well in crushing, but very poorly when it fractures. You can demonstrate this. See how much energy it takes to crush a section of polystyrene foam packaging, then see how much energy it takes to snap it in half. Helmets are supposed to absorb energy by crushing, but many "helmet saved my life" stories involve helmets that have failed, without extensive analysis it is not possible to say how much energy - if any at all - they absorbed.