Graves Being Graves

When Liberal pollster Frank Graves reported that the Tories were nearing majority territory, the party responded very unusually to otherwise good news saying that some pollsters will publish very high numbers for the Tories only to make a deal when they return to normal. Sure enough, a week later Frank Graves was doing exactly that. Nanos and Ipsos confirmed last week that the Tories had reached a high water mark, but EKOS is the first to report a sudden drop and make a big deal about it through their friends in the media. Graves is always delighted when he reports a decline in Tory support; "it appears that the slumbering electorate have come unhinged. Major oscillations suggest people are paying attention."

And why does EKOS have the Green Party at almost 12% while Nanos consistently has them in the 5% range? I'm curious, because if these are supposed to be truly random samples, they should average out to roughly the same. But EKOS repeatedly overrates the Green Party. What is it about their method that produces this irregularity? Is it a dependence on internet polling? It suggests that their method is not truly random and thus not statistically valid.

It should also be noted that on the Globe's politics page, they have a piece titled "Tory lead shrinks in face of Bev Oda furor" and another "Bev Oda affair fails to catch fire with voters". Shouldn't those be mutually exclusive? Harris Decima found that half the country "had heard nothing about the document-altering scandal", and yet we are supposed to believe that this "furor" is dragging down the Tories? This was headline news for a week and half the country knows nothing about it? My own site traffic has increased 30% in the last week, so there is something afoot. In a recent poll, 80% of my visitors voted that Bev Oda should NOT resign.