In the Cold November Rain

This morning I finally matched up some of my FOI data with a hypothesis I’ve been rolling over in my mind for a while. I’m reproducing the resulting twitter-essay here, with annotations:

It’s November, which means police and other road safety folks are talking about pedestrian safety.

November typically sees more drivers hit pedestrians than any other month of the year. It seems to be a universal truth.

Police everywhere try to reduce the number of collisions in November by telling pedestrians to wear bright or reflective clothing.

Their reasoning is that in Nov. it gets dark earlier, and with the time change it’s even worse.

They say that while the weather is nice there are more people out walking in the dark, and therefore it’s hard for drivers to see them.

I’ve had a hypothesis for a while that this reasoning is crap, but I’ve had no way to test it, because how do you test that?

I FOI’d the Sudbury police for the dates and times of collisions with pedestrians over 5 years. That’s 487 data points.

I then got the sunrise and sunset times for the year specifically for Sudbury.

If the latest sunrise of the year is 08:08, and the earliest sunset is 16:39 what’s the share of collisions by period of day?

I thought, what if the same proportion of people get hit after 16:39 all year round? Then “wear bright clothes” is crap.

Here’s 5 yrs of crashes before 08:08, after 16:39, and in between for each month in Sudbury, Ontario:

The X axis is the month (Jan. = 1, Feb. = 2, etc.). The numbers in the bars are the actual number of collisions with pedestrians for that month according to time period. The colouring of the bars represent the percetage of collisions during that time period. A few things to keep in mind: “Daytime” for this purpose is defined as the hours between 8:08am and 4:39pm in every month (including the summer months). This seemed like the best way to control for daylight and commuting habits across the year.

You can see the highest % of crashes after 16:39 is in November. That seems to debunk my hypothesis doesn’t it?

But it’s not like November is the ONLY dark month of the year. And shouldn’t March, when we spring forward, have the least crashes? (that is, if November has the most because of falling back an hour, then march should have the least because of springing forward an hour)

Police tend to suggest that fewer people are out walking once it gets cold out. That hasn’t been my experience.

My theory is that in the winter, when there is snow, people drive more slowly and more carefully, particularly when turning.

And google “sneckdown“. It’s when the snow narrows the roadway, replicating the “neckdown”, a proven method of making streets safer.

Sudbury sneckdowns can be quite epic due to the amount of snow, and the excessively wide roads.

So maybe crashes go up in November because it’s dark, but after it stays dark drivers are more careful due to snow so crashes go down.

Then we shouldn’t be asking pedestrians to be more visible, we should be recreating, year-round, the conditions that make driving safer.

Look at my chart again: If darkness is the hazard, why is January relatively as safe as July?

In conclusion, my hypothesis is not proven, but by digging into these numbers and attempting to control for changes in daylight hours I’ve been led into some other interesting questions. And these aren’t questions that can be answered with aggregate stats on collisions per year or per month.

To answer these questions we need raw data on collisions — dates, times and locations — and it is irresponsible for the police to withhold this information unless compelled to release it through a freedom of information request.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s