In Canada, polls leading up to the election all the way to yesterday showed the Liberal and Conservative parties neck and neck. Yesterday's poll had them effectively tied in popular vote with anybody's guess as to who might form a government, presumably a minority one.

Now real results are in, and while not complete we see this:

Overall Election Results
Party	Elected	Leading	Total	Vote Share
LIB	127	11	138	37.43%
CON	85	8	93	29.36%
BQ	52	2	54	12.76%
NDP	17	5	22	15.15%
NA	1	0	1	.04%
OTH	0	0	0	5.26%
				308 seats

A remarkable difference and clear victory for the Liberals. As noted, we have seen this before. Surveys measure only "what people who bother to talk to pollsters want to tell pollsters." In spite of their claims of small margins of error, they can be very, very wrong.

It's important to not just learn when not to trust polls, but also to ask why, even when we see this sort of error time and time again, we continue to trust polls. We grasp at any information, even what we know to be unreliable.

It causes huge events. I remember in the 80s the provincial Liberal party seeing polls that showed a comfortable majority, so they -- based on the polls -- called an election. And were soundly trounced. (In part, in Heisenberg style, because people were annoyed they called an election for no other reason than their good poll numbers.)

So the idea to promote here: We often hear complaints from the "ordinary" folks that they don't like having to take all that Math in school because it will not be relevant to their life.

One course that everybody should take, and which is relevant, is a course on how to understand statistics and the misuse of statistics. Even if they came out of it not know a chi-square from a hole in the ground, they might be able to tell when stats can't be trusted. One hopes.