Lets be clear: to really forecast September’s parliamentary election results with any degree of accuracy, you would need detailed, circuit-by-circuit polling in key marginal seats, the so-called "swing districts."
We don’t have that. We think, contrary to their claims, nobody does. And we’re not going to have that. So…what to do?
The next best thing, as far as I can tell, is to base your projections on previous elections’ results along with an educated guess about the government’s overall level of support. What you get from that exercise is an approximation – a rough approximation – that’s certainly prone to a lot of error, but is still – I think – better than nothing.
That’s the basic idea between the Original Swing-o-Meter: basically, an Excel-base elections projection tool that sets out to boil down all the complexities involved to a simple, easy to use model.
That Original Swing-o-Meter was kind of fun, but it was based on a fairly narrow range of data. Which is why this week we’ll be unveiling the Swing-o-Meter 2.0: a new, more sophisticated Swing-o-Meter based on an expanded set of data, and allowing finer grained control over the "Swings" to be forecast.
The new Swing-o-Meter will be based on data including both the 2007 Constitutional Reform referendum results and last year’s Constitutional Ammendment referendum results. By breaking down the circuits regionally, and on an Urban-Rural scale, it will allow a much clearer picture of the relative payoffs to concentrating in certain parts of the country, or in more rural or more urban parts of the country.
There’s a lot involved in making this new Swing-o-Meter, so I’m not promising it before Friday. But it is coming. And it is going to be wicked.