First, take a look at this chart – and, for ridership from January to April, note in particular the blue bars:
.
Second, compare that with The Province’s headline on the story with the above chart:
.
Here’s the writer’s explanation:
Total bike trips compiled by the city for the 12 months ending April 2013 — the most recent statistics available — show that ridership is down by 16,000 compared to the previous 12-month period from May 2011 to April 2012.
.
Third, try to explain how the article and headline passes the smell test.
you can see how much thought and energy went into this article when you look at the y-axis labels 75,000, 10,000, 20,000
One thing I see in that chart is that winter biking is up a fair amount since 2011. This may have been because of mild/dry winters in 2012 and 2013, but still that’s a nice thing to see.
Ridership is high because of casual summer riders. Young people who don’t have/can’t afford cars riding between beaches. More tourists ridiing on summer holidays.
And the city stats make large notice of Fireworks Night in their reports. Best view from Bridge.
The gloom of Winter riders are the true test. They are the regulars whom success will depend.
There is another explanation – the Province newspaper has a problem turning data into a chart. If you look at the story as it now appears you’ll see a quite different chart with data that reflects the numbers published on the city website. That does indeed show a slight drop on Burrard Bridge for the period they chose to report, although the Dunsmuir Viaduct added almost the same number of riders that the Burrard Bridge lost. If you roll the data forward two months (as it is now available) the drop on Burrard Bridge is less, (although it’s still a drop) and the increase on Dunsmuir Viaduct is far greater.
Overall the total trips on all four routes in the first six months of 2013 is quite a bit more than in either of the previous two years.
Well actually when you add May and June for this year stats get worse for the Burrard Bridge route.
When you plot the latest (monthly) numbers in Excel and add the trend lines they show that Burrard Bridge cycling is trending slightly down, Hornby and Dunsmuir Viaduct (the strongest growth) are trending up and Dunsmuir Street is trending up by a tiny bit…. We’ll have much better picture by the end of the yer when we have at least 3 full cycles of data for all bikeways.
So the Provice article is correct – Burrard Bridge numbers have stalled. They fail to mention the other bikeways of course, but those have nothing spectacular going for them…
See http://tinyurl.com/lg5tlso
… and the best way to increase cycling on Burrard bridge is to make better connections to it!
I would caution against using a Ordinary Least Squares-fitted regression against time with absolutely no explanatory variables to portray a trend. OLS assumes that the error terms are uncorrelated, which is generally not a reasonable assumption for time series datasets.
A few years ago some UBC students put together a paper on the effect of weather of bike lane use (available here: http://docs.trb.org/prp/12-2119.pdf) which provides good example of why OLS is inadequate. With an extra two years of data, it would probably be possible to fit it with a trend variable to see if there is any statistically significant increases/decreases emerging.
You could also test, for example, what effect opening additional lanes had on the original one (e.g. did Hornby actually increase use of Burrard Bridge?)
tl;dr: If people want to say “bike lane use was up/down for a year over year period from X to Y”, and the count data supports that, then it is a correct statement. However, this should not be seen as constituting a meaningful “trend”, either up or down,especially for a mode with as much statistical noise (i.e. weather effects) as cycling.
I don’t even think the Province got the numbers right. If anyone can make the above table using the City’s own data, please prove me wrong.
http://nwimby.blogspot.ca/2013/07/the-numbers-tell-story.html