DDOT recently released crash data from 2014.
This isn't totally accurate becasue it's FARS data, which ignores some fatalities. In this case, Ian Wolfe's fatal crash in 2008 isn't counted because it was investigated by the Park Police and they, it inexplicably seems, do not report fatalities to FARS.
Injuries were up 26.7% from 2013, even while injuries nationally are down or flat (image below shows injuries per year - sorry it kind of sucks). Part of the increase is due to better reporting they state "In 2010, MPD and DDOT significantly improved recorded keeping, training MPD officers, and the crash and FEMS record management system; this resulted in an increase in the number of reported crashes"
It looks like Thursday and Friday are getting safer - good job pre-weekend!
So more cyclists are injured at times when cyclists are cycling. This isn't particularly useful. If we combined it with counts) then we could look at rates. Just to get an idea, I used CaBi data as a proxy for exposure, and then divided the number of injuries per month by the number of CaBi trips per month. There are three problems with this:
1. CaBi use may not perfectly mirror bicycle use
2. CaBi data covers DC, Arlington, Alexandria etc...and it might be different if I weren't so lazy and filtered out the non-DC data
3. The CaBi system isn't static, so some increase in use is due to increased numbers of system bikes, not overall increases in biking.
Also, I had to eyeball the injury number for each month. With all that in mind, what I found was that the month with the highest "rate" was October, followed by Septmeber, July and June. Lowest rate was February, followed by November and December. This is basically the opposite of what I expected. I expected the winter to have the highest rate because of less light, worse road conditions (ice and snow) and the inverse of the safety in numbers effect. So, I'll chalk it up to a methodology flaw (from the list above) until someone does it with better data.
Injuriess also tend to happen where cyclists are riding. Curious. Again, this could be more powerful it combined with counts to get rates. I think CaBi data would be less valid for where, since socio-economic factors start to come into play.
Once again, the kinds of people who we already know are more likely to ride bikes in DC, are also those who are more likely to be in crashes. Add counts, get rates.