It’s elections day in Vancouver. Go and read the party platform for the Non-Partisan Association (NPA) – a centre-right party – and you’ll find a graph on page 15. It’s part of a section on homelessness in Vancouver, and purports to show that it has increased since 2008 under the watch of the current mayor, Gregor Robertson. It is, as we’ll see, a very bad graph.
Pretty convincing, at first glance. Based on the relative lengths of the bars one would conclude that homelessness in Vancouver is indeed significantly “up since 2008.” But look at the values – 1575 homeless in 2008, 1600 homeless in 2011. The NPA has set the range of the chart from just 1560 to 1605, so that the difference between the two numbers is exaggerated. This is sneaky, because it gives a misleading impression of the true difference between the values: 1600 is only 1.016% bigger than 1575, but its bar on the chart is nearly 300% bigger. As a rule, bar charts should start from zero to avoid this type of confusion.
There is a more obvious reason why the NPA’s homelessness claim is wrong. It’s simply not possible to count something like the number of homeless people in a city with perfect accuracy. The committee in charge of the homeless census frankly acknowledges this in their report, and estimates the margin for error at up to 80 people (which actually seems conservative to me). Regardless, with this error margin there is no detectable difference between 1575 and 1600, because the two numbers fall comfortably within the range of measurement error. The 2011 media release from the census committee plainly says as much: “Statistically, the number of homeless [from 2008 to 2011] remains virtually the same” (p. 1).
If the NPA had included the appropriate range and error bars, the plot would have looked like this:
Well, probably a bit slicker than that, but you get the idea – it is absurd to claim that there is a significant difference between 2008 and 2011. It doesn’t take advanced statistics to figure this out, just common sense. Ironically, “Bringing Common Sense back to Vancouver” is the NPA’s campaign slogan.
We’re left to choose between two conclusions: either the authors of the NPA platform are stupid, or they are trying to bamboozle the voters. I’ll be charitable and assume it’s the former.
One more thing. If you look up the original data, which are available here, you’ll notice that total homelessness is only part of the story – the numbers are also broken down into sheltered homeless and unsheltered homeless. The number of unsheltered homeless people in Vancouver decreased from 815 to 145 between 2008 and 2011. That really is a significant difference. The correct conclusion to draw from these data is that while the total number of homeless in Vancouver has remained the same since 2008, the number of unsheltered homeless has significantly decreased.
The NPA’s misleading homelessness claim, and its equally misleading chart, might not seem like a big deal. They take up less than a page of the 24-page platform. But these issues matter. If it really is true that the current mayor’s policies are increasing the rates of homelessness, voters should be concerned. On the other hand, when political parties distort the facts to serve their own ends, they do real harm not only to the voters, but to those whom the policies ultimately affect: in this case, the homeless.
Disclaimer – I’m not voicing support for any particular political party here.