there are certain areas in toronto that horwath ought to actually avoid, as an ndp bump is more likely to swing a red seat blue than orange.

i'm sure she has better riding data than i do.

so, i would put out a call to the ndp campaign to consult it very carefully - and to avoid taking modelling data seriously.

just briefly...

what the models do is they take data from across the province and try and squeeze it into specific ridings. so, if i'm talking to lawyers here, they're taking data that is overly broad and trying to force it into very specific scenarios that ignore the specifics, like incumbency and turnout and demographics.

personally, i'm going to argue that the whole idea is flawed - that the premise of modelling is overly broad and should be struck down as unconstitutional (the constitution being the predicate calculus, i suppose). but, that defence is untested.

riding polls ask people in the riding - it's directly empirical. much better. but, watch out for small sample sizes...