I had to post something to move that horrible Stephen Harper head further down the page.

I'm trying to understand the polls that have been coming out recently over the past few days. First we have the SES rolling polls which for the past week have been giving the Conservatives a growing lead over the Liberals, then yesterday and today a Leger poll and an Ipsos Reid poll showed the Liberals still in the lead.
All three polls show the Liberals at 32 and 33 percent among decided voters, but while Leger and Ipsos Reid show the Conservatives at 30 and 31 percent, the SES rolling poll has them at 37%. That's quite a difference, and as the Sun has been more than pleased to point out it makes it appear as if the Conservatives have the makings of a good lead.

The Leger Poll was taken over a period of 4 days, with a total of 1500 respondents.
The Ipsos-Reid Poll was also taken over a 4-day period with a total of 2000 respondents.
The SES poll is a little more complicated. Simply put, it is a poll of 600 people (a very small sample) taken over 3 days. SES polls 200 people per day, and each day adds the new data to the poll and drops the old data from 3 days ago. Get it? Got it? Good.

Anyway, I can't believe how little of this stuff I remember from university. I'm supposed to have a degree in political science, and I can remember taking an entire year of this polling crap in second year, but do you think I can remember it now? Suffice to say that a poll of 1000 people is necessary for a poll to have an acceptable margin of error. The SES polls fall far short of that mark.
These SES polls are so out of whack with Ipsos-Reid, which sets the standard for polling in this country, that the folks at SES are going to have to spend some time reworking their model before the next election. Somebody said, and I've just done a bit of unsuccessful googling to try to get their name, that election polls can be weapons of mass destruction. With so many newspapers picking up these SES polls and concluding that the Conservatives are leading, it's entirely possible that this could have an affect on how some people vote. Conservatives, feeling secure that their party has a good lead, may find it less urgent to get out and vote. Some NDPers may choose to vote Liberal. Whatever.

Conclusion. There is something wrong with the polling model that SES is using in this election. It's nice, it's gratifying, to have day by day polling results, but it might be smarter to ignore SES poll results this election and wait a few days for more thorough poll results.