bicycle safety in London (taken from an article in ASA's Significance Magazine): The City of London adopted a measure to improve safety of bicyclists. Before it was implemented, there were 6e-6 collisions per week per bicyclist, of which 5% were deadly. After implementation, the fatality of a collision dropped to 3%, but their frequency increased to 9e-6 collisions per week per bicyclist, of which 3% were deadly: riders started to believe in their immortality, because the measure was well advertised, and started doing silly things on the roads. Did the measure make a difference?


In [2]:
before = 6e-6
after  = 9e-6
fatalities_per_collision_before = 0.05
fatalities_per_collision_after = 0.03

Compute overall fatalities:


In [19]:
fatalities_before = before * fatalities_per_collision_before
fatalities_after = after * fatalities_per_collision_after
relative_change = fatalities_after / fatalities_before

print "fatalities before the measure:", fatalities_before, "; fatalities after the measure:", fatalities_after
if (relative_change < 1):
    print "riding a bike in London became safer by a factor of %3.2f" %(1.0/relative_change)
else:
    print "riding a bike in London became more dangerous by a factor of %3.2f" %relative_change


fatalities before the measure: 3e-07 ; fatalities after the measure: 2.7e-07
riding a bike in London became safer by a factor of 1.11