Well no, statistics can be relied upon if they're used honestly and people understand their limitations, but I've yet to see political poll results used honestly, so I agree with you there, polls are not to be relied on. That piece in the Globe & Mail this morning, for instance, had a banner at the top showing the results for the four largest parties, no mention of the Greens or any other party and no mention of the undecided vote, which when I've seen it mentioned tends to be around 18% if I remember correctly. Ipsos Reid showing the Tories at 43% clearly puts them into majority territory, but if there are 18% still undecided, that's really 43% of 82%, or around 35% of all voters, clearly still in minority territory. The way the results are presented assumes the undecided vote will be distributed on voting day the same way the decided vote is, which ain't necessarily so. These polls are also conducted by telephone, and anybody who doesn't have a land line--which is quite a few people these days, especially among younger folks--will never be part of one, so the sample is skewed. There's also no accounting in polling for people who won't vote, which as a percentage of all voters is generally larger than the level of support shown for any party. It's also known that publishing poll results affects people's voting behaviour in ways that can't be corrected for in the polling process.