Thoughts on politics and life from a liberal perspective

Wednesday 20 May 2009

I'm going to be on Radio 4's "More or Less" this Friday!

I have just been called by Radio 4's "More or Less" presented by Tim Harford and they have asked me to be a guest on this week's programme! They want to talk to me about the story I broke recently about the apparent correlation between how safe an MP's seat is and the likelihood of them being implicated in the expenses scandal that I posted about here and here.

As the programme is about numbers, they want to focus mainly on the numbers and maths behind what I discovered. It is being recorded tomorrow and will be broadcast on Radio 4 at 1:30pm in the afternoon this Friday 22nd May.

So if you want to hear what I sound like then please tune in to hear me or listen again/podcast it.

Right, I'm off to phone my mum!

UPDATE: I've just discovered that it is repeated on Sunday at 8:00pm too.


Costigan Quist said...

Congratulations - I'm horribly jealous of course.

Alix said...

That's marvellous! I mean Polly Toynbee and Guido are one thing, but a nerdy programme on Radio 4 - get in! That's our core audience listening there :-D

Jack Gibson said...

Be a bit careful what you say about your findings when you appear. They'll almost certainly get a statistician or social scientist to comment on them. They might well point out that:

1) Your data are incomplete, and very likely to be subject to selection bias (the DT concentrating on more senior MPs in safer seats).

2) There is likely to be effect modification by the holding of ministerial or shadow-ministerial posts by those with larger majorities. Holding a ministerial post brings with it a slightly different set of rules about designation of main residences (and other stuff like virtually compulsory claims for official use of grace and favour residences).

3) Members in safe seats tend to have been in parliament longer, so there's likely to be confounding by knowledge of what can be got away with (although you could argue that that's on the causal pathway).

4) T-tests and M-W U tests aren't really appropriate to these sort of data. Chi squared tests for trend would be a better bet, or mabye a logistic regression against size of majority...

5) And probably some other stuff I haven't thought of ;)

Mark Thompson said...

Thanks Jack.

I understand 1, 2 and 3 but you have lost me with 4.

I don't think they need a statistician or social scientist as Tim Harford is a renowned economist who knows his stuff.

They know that I am not a statistician and I think that they are just interested in how someone out walking their dog who has a bit of a maths background can run a bit of data through a quick analysis and end up with something that has spread around the internet like this.

If they do come heavy on the stats then I will invite them to contribute to the debate on the blog and contribute their own thoughts as all the other commenters have been doing! I don't think they will though. they are trying to make a radio programme which is accessible to their listeners, most of whom would not know what a chi-squared test is, as indeed I don't!

I am not going to claim that my findings are anything other than what they are. A first cut at analysing the data currently in the public domain by an enthusiastic amateur.

Sunny H said...

heh! nice one!

Anonymous said...

Well done Mark - I hope you put on your posh R4 voice.