Just a hunch

Apparently there are two polls coming out tonight. I'm not privy to either of them, but let me go out on a limb and predict they will show a healthy rebound in Liberal fortunes. Otherwise how to explain the persistent (and rising) election talk? UPDATE: In fact, the CTV poll (can't find the TVA/La Presse one) shows not much movement either way: Libs 38, Cons 27, NDP 15, Bloc 10. That's not good for the Grits, as they tend to drop during the campaign. Cons are up sharply in BC, Bloc as expected showing well in Quebec. The full data is are here. UPDATE: It was in fact a Leger/Journal de Montreal poll. Results (Quebec only): Bloc 46, Libs 35, Cons 11, NDP 7. Some narrowing of the gap, but not much. Laurent from Le blog de Polyscopique (the only fully bilingual blog I am aware of -- although this blog is conversant in Latin, as long as the conversation does not stray much beyond "the farmers are in the peninsula") notes that the Cons are up from 6% in the last poll, though it could all be within the margin d'erreur. UPDATE: The Election Prediction Project's current standings are: Liberals 105 Conservatives 58 NDP 11 Bloc Québécois 31 Too Close to Call 103 The Project attempts to divine how the vote will go riding-by-riding, overcoming one of the chief handicaps of national polls. On a related note, the folks who run the University of British Columbia's Election Stock Market are gearing up for another run, after their successful 2000 campaign. Like its forerunner, the Iowa Election Stock Market, it's a demonstration project in the power of markets as information-gathering devices. You buy and sell "stocks" tied to the electoral fortunes of various candidates/parties: share of the popular vote, number of seats, whatever. For example, the Iowa Presidential Vote Share market now shows Bush shares priced at 52 cents, to Kerry's 46 cents (Nader presumably accounts for the other two percentage points). So if Bush in fact collects 48% of the vote come November, you'd lose 4 cents per share if you bought in today. Whereas if you were smart enough to pick up Kerry at a discount, you'd earn the same amount. The theory is that people betting their own money will have the incentive to gather every scrap of information that could possibly affect the results: not just the polls, which as everyone knows are simply snapshots at a particular point in time, but everything, present, past or future, that might be germane. The IESM claims a superior track record to the pollsters in a dozen years or so it has been around, and it has the academic research to prove it. Similar markets are now up and running on the 'net, on every subject under the sun. Some well-known examples: the Foresight Exchange, and the Hollywood Stock Exchange. Of course, Ladbrokes works on essentially the same principle. ALSO: The UBC site also has a nifty election seat/vote predictor. You plug in your forecast of the average swing in vote from one party to another from the last election, and it spits out the implications if this trend applied across the country, or across individual provinces, using the updated riding boundaries. It's basically just a spreadsheet, but a useful one at that.