Crunching numbers: A closer look at Portland’s road fatality rates

When you compute SD’s of deaths by year, you are assuming the counts are independent. Since car crashes and other events may involve multiple fatalities, this assumption is not warranted.

What if you adjust the numbers by miles travelled? I’ve often heard that bike fatalities are low overall but relatively high by miles travelled.

So sad.

I have a bad feeling that in a few years we’re going to look at this data and realize that the year Vision Zero was passed was also the year roadway deaths started trending up in a big way.

It’s kind of like the way Eugene passed a bike/ped plan and declared a desire to “go platinum” and then saw a huge downward trend in cycling (down 37% from 2009-14 from US Census ACS).

Sure, there’s external forces at play, mostly decreased fuel prices, but are our elected officials and city staff really not up to the task of overcoming those? I guess we’ll know in a couple of years.

http://www.kgw.com/mb/news/local/pedestrian-struck-on-southeast-hawthorne-boulevard-area-closed/303107910?c=n

Including the first three years, the overall downward slope is statistically significant for the overall fatalities. However, dis-aggregating by mode, only motorist fatalities have declined statistically significantly (p < 0.002). Fatalities in the other modes have not declined or increased statistically significantly.

Excluding the first three years (which I don't think is justifiable, those higher numbers of deaths in the late 90s certainly happened, and the number of deaths declined after that), none of the slopes are statistically distinguishable from no change. Motorist fatalities are close to having a statistically significant downward trend (p < 0.1), and motorcyclist fatalities have a similar p value for having an increasing number of fatalities. Pedestrians and bike riders show no significant trend at all (p < 0.8), with estimated rates of change of (the slope) of an increase of 0.03 bike rider deaths per year each year, and an estimated decrease of 0.01 pedestrian deaths per year each year.

An r^2 of 0.06 is not surprisingly strong, it is incredibly weak. An r^2 of 1 is perfect linear relationship, an r^2 of 0 is no relationship. The statistically significant downward slope for motorist deaths over the full time period has an r^2 of 0.4, which matches with noisy data.

Taking a five year running average, pedestrian fatalities are constant, motorcyclist fatalities seem to be rising slightly, bike rider fatalities had a good period in the mid-00's before rising again, and motorists fatalities declined early on before plateauing recently.

In terms of vision zero, it seems the city has improved motorist safety in the past 2 decades, but hasn't accomplished anything for people using any other mode.

Here in Greensboro, our county, Guilford, has a bit over 500,000 residents and averages about 45 fatalities over the past 5 years, with one bicycling death per year and 9 pedestrians (80% car, 2.2% bike, 17.8% ped.) Of course, this is the deep south and much more car-oriented here, so our rate per person ought to be higher, and it is, but not by as much as one might assume. Portland is only marginally less car-oriented than folks would like to think.

Note that VZ originated in Sweden as a methodology to reduce deaths of motorists. Tacitly, VZ in PDX has the same goal: successful implementation here overwhelmingly would affect motorists; we bikey types and peds would not have much to gain.

From a simple-minded statistics of infrequent events point of view there is 1 death each 11 days over the period of time analyzed. Trends may go up or down, but nothing is predictable.

At this rate, Vision Zero is nothing but a dream. The reality? A nightmare.