Saturday, October 20, 2007

Polls and Shoddy Reporting

The polls these days seem to be a mess and media coverage of them usually pretty bad. Three polls came up this week showing the Conservatives to have a very narrow lead over the Liberals nationally and the Liberals having a solid lead in Ontario. (h/t to Scott: here, here, and here).

Meanwhile, Ipsos comes along for the second week in a row showing the Tories in “majority territory” . I’ve got some serious questions about this though.

Why is it ALWAYS Ipsos that has the Tories polling higher than the other polling companies? Is there something we should know here? I mean the last Ipsos poll of the 2006 Election (1 day before Election Day) had it as Cons 38 – Libs 27, pretty far off the mark of what happened the next day, I’d say.

Second, where does this magic 40% national number called “majority territory” really come from? It’s total nonsense considering a party could have 40% nationally just because they’ve got 90% support in Alberta for instance. So the overall national number means nothing without reliable regional breakdowns, which none of these polling companies really ever provide because the MOE for each region is too large, so you end seeing like 10% regional swings in the span of week (which is really impossible).

Basically this kind of coverage is just really shoddy reporting trying to make a poll sound more interesting and pump up the election hysteria (“oh Harper’s in “majority territory” he must be itching to go to the polls now!).

Finally, I’d like to call one particular deceptive article to account today that really twists things: "Fears of Harper majority waning: poll

Take a look at the article and you see that: “A majority of 58 per cent said the best outcome of the next election would be a majority government,” and “Fifty-eight per cent of those in favour of a majority would rather have Harper as the prime minister in such a circumstance, compared to 28 per cent who preferred Liberal Leader Stephane Dion”

What’s 58% of 58% CanWest? Is that actually slightly less (33.6%) than the percentage of Canadians that voted for Harper in the last election (36%)? How can you say fears of a majority are “waning” when you’ve got no baseline to compare it to. 33.6% is still not that high given that I'm sure the Conservative support in this sample is grossly inflated (definitely at least in Ontario). I don't have any faith in these numbers given Ipsos' history.

I think reporting of polls in the media has always been pretty shoddy in general (ignoring MOE's and so on), but at least report the straight findings of them correctly, don’t give us misleading headlines and torqued spin.

And that goes for Bricker (the Ipsos guy) too because lines like “When you've got the other guy hiding under the bridge like a troll…” don’t really come off like you don’t have an agenda here do they?

What gives with Ipsos and CanWest media that report on them?

Recommend this Post