Isn’t that just typical? After a weeks-long polling drought, the second survey of the night just popped into my inbox. Varianzas is showing a headline figure of 48% for the government, 47% for the opposition, 5% don’t know. Everyone else is getting far higher numbers for "don’t know/haven’t decided/won’t say", so I assume that 5% is after pushing hard for fence-sitters to say which side they’re leaning towards.
This poll is fresher – fieldwork September 10-16 – and done on the basis of 1300 interviews. The sample is pretty damn urban, though most pollsters try to compensate for this by adding extra lower-income respondents, who – the theory goes – vote more like rural people.
I’m a little leery about sharing it because, frankly, I think their projections are internally inconsistent: they first forecast 60-70 seats for the opposition, then share three National Vote Share scenarios, one of which has the opposition on 46.5% of the vote! I’m sorry, but if the opposition comes in at that level, they’re not looking at 60 seats, they’re looking at like 43-50 seats. Similarly, if the opposition gets the 51%+ of the polarized vote, they’re looking at 74-80 seats, not 70! So either their swingometer is screwy, or they pulled their seat-projections out of their bums.
Oh, plus they’re using a built-in PowerPoint template for their slides. Seriously, guys, make an effort!
Still, if you’re keeping score at home, their slides are here.