You can't get much better analysis than on Radio 4's 'More or Less' (last Friday 1.30pm). The conclusion was that while there is some interesting research that shows helmets can both lower and increase risk there is no sound statistical evidence to really test these and show, on balance, which is the best strategy.
Which is what is evident in these hallowed forums where the proponents/detractors can't, when challenged, substantiate their claims with any scientific credibility.
Wear one if you want, don't if you don't but don't forget to always check the brakes, cables, headsets ... stuff that can really help you avoid crashes with no downside.
Bit of training and reading Cyclecraft will probably help. But I have no proof of that