Speed limit proposal based on bad statistics.

Winnipeg City Council is discussing lowering speed limits to 40 km/h in residential areas. The justification for this includes a study conducted in Edmonton that appears to show a great benefit to lowering the speed limit. It's even baked right in to the proposal before council.

 The proposal includes the following:

AND WHEREAS the City of Edmonton recently reduced speed limits to 40 km/h in several residential neighbourhoods with a 25% drop in severe collisions;

THEREFORE BE IT RESOLVED THAT Council ask the Winnipeg Public Service to consider lowering the residential speed limit in Winnipeg to 40 km/h

Here's the problem: THE RESULTS ARE INSIGNIFICANT.

From the study:

However, these reductions were not significant, as the 95% confidence interval included zero, implying no change or no effect. Generally, when a confidence interval is very wide like this one, it is an indication of an inadequate sample size (i.e. short “After” period) and implies poor precision. Consequently, the results of the collision analysis were inconclusive and additional research will be required to substantiate the impact of the pilot project on the number and frequency of collisions.

(emphasis mine)

That's a big problem with research related to public policy. Give a politician a study that says "implying reductions in predicted collision counts of 25% with a 95% confidence interval of -81%, 77%" and the politician will read "blah blah blah 25% reduction in collisions! blah blah blah."

I read "95% confidence interval of -81%, 77%" and see "these results are as useful as a tampon dispenser in my garage".


Some may also point out that the study showed a statistically significant decrease in operating speed in the study areas with the 40 km/h speed limit, but the decrease only averaged 3.95 km/h. Assisting with the decrease were the following factors:

i) a pre- and postcommunication plan;
ii) installation of new speed limit signage and setting up speed display boards, dynamic messaging signs and school dollies;
iii) implementing community speed programs (i.e. Speed Watch, Neighbourhood Pace Cars and Safe Speed Community vans) and
iv) using covert photo-radar trucks.

It was the result of a full-out blitz to increase awareness and reduce speeds, with a modest result of 4 km/h.


Look, I don't want kids to get run over by cars any more than you do, but this isn't the answer. Most people already drive a reasonable speed in residential areas. This isn't going to change that. What this will do is lower speed limits on collector streets where 50 km/h is a reasonable speed, but are deemed "residential", providing more opportunity for the police to set up radar traps to ticket drivers who are driving in a responsible manner.

What we need in this city is a common-sense approach to setting speed limits. It is insane that Kenaston Blvd has the same speed limit as Valour Rd. Set speed limits at levels that reasonable according to industry standards (ie. 85% percentile) and adjust as necessary for special cases like school zones. Let's not create misguided legislation based on inconclusive data and misinterpreted studies. That never turns out well.