Tuesday, 28 July 2009

More evidence that Robert's model is wrong

by Irfan Ahmed

When I criticised Robert Smithson and his model that he has created to predict the outcome of the next general election there was an outcry by fellow PBers. I now have more evidence to add to my small evidence pile, arguing my corner that Robert’s model is wrong.

Roberts model, predicts that Burnley shall stay a Labour seat at the next general election and I think he and others need to reconsider their belief in the model. The prediction by Robert is as followed for Burnley:

CON 9217
LAB 11068

The problem with the model is that, it hasn’t looked at the political forecast for Burnley. First of all at the 2009 County Council elections the Lib Dems won 5 out of the six seats, moving on from that the reputation of the PPC is something that needs to be taken into consideration.

Gordon Birtwistle is running a mono causal campaign that has made him a celebrity in Burnley, the campaign “I want my Hospital Back” is really turning into a vote winner for Burnley something with neighbouring constituency Lib Dems are failing to benefit from.

Gordon has approximately 2000 or so poster sites, and this isn’t from what he and his team have told me but from what I have seen in Burnley. The Lib Dems are set to win in Burnley, yet Robert’s model predicts otherwise.

So what do fellow PBers think of the model, after having another piece of evidence put forward about the credibility of Robert Smithson’s model?

P.S. I don't have anything against Robert, if that's what people are starting to think after reading my comments on PB and on my own blog.


Hywel said...

The Burnley result at the General will be nothing like that prediction.

Irfan Ahmed said...

Is that in favour of my opinion or against?

MichaelK said...

It's in the nature of a model that it doesn't take account of specific local factors, so this doesn't have any bearing on whether Robert's model is better or worse than its rivals, such as UNS.

J said...

Are you going to analyse all 632 GB seats individually, or just ones you can shill for the Lib Dems about?

Ruaraidh said...

Irfan, I suspect you're missing the point a little. A model is never going to be able to take in every local factor - one seat's difference doesn't make it worthless.

Mike Smithson said...

Irfan - you simply have not understood what a model is and your criticism is invalid and inaccurate.

Go check with Baxter and Wells and you'll discover that they both also have Labour holding Burnley.

Irfan Ahmed said...


I have checked the election calculus and it points to a Labour hold and that's why I think these models are inaccurate, and should not be used as a way to predict the outcome of the next general election.

The best way in my opinion to work out the outcome is via a poll in every constituency but these are expensive and can be misleading, especially if the number of people polled is insufficient.

Irfan Ahmed said...


I can look at every seat in Parliament if you want and write about it on my blog. But for that I need people to fund my time and effort, which I don't think anyone will ever do!

J said...

So you're just going to shill for the Lib Dems.

That's fine; it means I know how seriously I should treat your contributions both here and at PB proper.

Irfan Ahmed said...


No I am not going to do that, I am looking at seats and have another seat that I want to look at and I shall write about it over at my blog and shall then share the link in PB Channel 1.

Philip said...

This is not "evidence" by any means.

If traditional models, eg Baxter, had Burnley going Lib-Dem but this R. Smithson's model had it staying Labour you could use this as an argument. But neither has them going Lib-Dem so it doesn't matter either way. This is not "evidence".

Morus said...

I think it was George Box who once said

"All models are wrong, but some are useful"

If you don't get that, you don't understand modelling.