12/03/2007

If the Richmond Formula showed a 12% lead in an election, you'd be over the moon

For the uninitiated, the "Richmond Formula" is a formula used by the Lib Dems (and now copied by other parties) based on a system pioneered by Richmond Lib Dems in the 1980's which saw them take raw canvassing data from knocking on doors, and using the formula to extrapolate a potential result.

At the time that Richmond lib Dems developed the formula, they worked out which seats were "in the bag" from applying the formula, and then went to help in more marginal areas.

So what would Richmond Lib Dems have done with the 12% lead that Nick Clegg has over Chris Huhne (according to a Sky News poll) ? My guess is they would assume there is a clear winner.

I have to say, when i have run local elections I have assumed a 12% margin to be fairly accurate.

11 comments:

rob's uncle said...

Your clock is wrong: here I am commenting on a post timed at 21:13 at 20:31; if you can't be bothered to reset it from BST you should say so.

The Richmond formula is irrelevant here as you presumably know so I won't bother to explain why.

The sample used here is small, only c. 380 had voted, so the 95 % confidence interval is +/- 10 % = 46 - 66 % for Clegg so the probability he was truly ahead [> 50 %] is c. 80 % - not enough to bet the farm on, I suggest.

In any case the sample is certainly biased in some way, because You Gov is a biased sample of the electorate. However we cannot tell in what way until we get the result.

Nich Starling said...

Oh dear Rob's uncle. I never knew it was a "blog rule" that I had to say if I had changed from BST or not.

Actually, I have had a problem posting to my blog this evening, and had a funny error come up on screen. So I changed the time by one hour forward and posted it again, and that seemed to work. I can't see how that is against the law.

Anonymous said...

OMG How sad are this? Words fail me. As someone who has been involved in politics for thirty years I had never come across the "Richmond Formula" let alone other parties copying it. It just confirms the image of the Lib Dems as the anorak "how to win elections vol.26" party. A bar chart here...a two horse race there... what about policy...principles ... belief? So the self proclaimed "local campaigners" will move to another ward once the first one is in the bag? Pass the sick bag.

Chris K said...

Never believe the Richmond formula unsupported by other evidence.

This isn't a comment about the leadership election, just one about the difficulties of interpreting canvass figures.

The difficulties of interpreting opinion surveys conducted by professionals (which we're dealing with here) are different.

Duncan Borrowman said...

If I was the 2nd placed challenger in a two-horse-race byelection I would think it was all to play for and I could come up on the rails.
Of course the leading candidate may well relax with this lead...

Paul Walter said...

So you'd rely on applying the Richmond Formula to canvass results from just 1% of the constituency electorate, would you Nich?

Fascinating....

Anonymous said...

Anonymous - How patronising are you. Every party should have moveable activists who go in to "target" seats. The Richmond Formula (which if you haven't heard of it means your party use it by another name) allows you to move your moveable activists but leave the candidates and local campaign team to fight by themselves.

The Conservative Party did this in 2005 at the General Election using phone banks to gain samples from key seats so they knew which ones to fight whilst Labour did the same in 1997 and 2001. All parties in well organised seats do this in local elections.

Every party does this and to pretend otherwise shows that your are merely playing politics with your comment. Pathetic.

Nich Starling said...

Indeed, the Tories and Labour brag about using the same techniques as the Lib Dems. i have a tory friend who showed me a photocopy of the Lib Dem training handbook the Tories were giving away at a councillors conference once.

As for the stupid tosh about Lib Dems not caring. Remember, the Richmond formula only works because the Lib Dems work all year round, not just at election time.

Anonymous said...

The most obvious problem with the poll's sample is that it is significantly younger than the party's membership...

If you make a rough correction for that, the race is much closer.

Anonymous said...

Well anonymous, you must at least concede that this is a two horse race, and that only the Lib Dem can beat the other Lib Dem here


btw, I'd never trust the Richmond formula anyway, but it seems especially unsuited to this scenario.

Anonymous said...

If the Richmond Formula showed us 12% ahead I would be delighted.

However if the raw figures showed us 12% ahead of our main opponents with lots undecided, the Richmond Formula would not show us ahead at all!

Pages