The 16 UK politicians most likely to be making up facts.

This article ranks UK politicians by how likely they are to back up their factual assertions with references. I use their Twitter streams as a starting point.  The method is somewhat noisey and it’s certainly not the sort of thing that I’d call science, but it illustrates some points nicely. As soon as someone comes up with a better method I’ll use that instead. This is a very rough first attempt –  There will be a proper full version around January, with plans to repeat on a monthly basis until the election in May.

I’m going to first give my motivation in terms of Wikipedia.  Then I’m going to describe my methodology such as it is (I’m including the source code), and then give the results.  This post clocks in at 1,500 words so it’s one of the longer ones…

Motivation: Wikipedia cites its sources, MPs should do the same

I like Wikipedia… I genuinely think it’s one of humankind’s greatest achievements…  Many politicians, however, do not…

Shabana Mahmood: …. Gentleman says that I had a career in the accounting industry, but I did not— I was a barrister specialising in professional indemnity litigation. I hope he did not get his information from my Wikipedia entry, which also has me down as two years younger than I am. http://www.theyworkforyou.com/pbc/2013-14/National_Insurance_Contributions_Bill/04-0_2013-11-21a.9.0?s=wikipedia#g9.10

Murdo Fraser: …with the duty, they must understand its meaning. During the evidence-taking sessions, we heard many different definitions of sustainable economic growth—somebody even suggested the one from Wikipedia, although I am not sure that that is helpful to the law-making process. http://www.theyworkforyou.com/sp/?id=2013-11-12.6.0&s=wikipedia#g6.12

Stewart Stevenson: …however one looks at it. Ken Macintosh referred to Fort Augustus and the first hydro power station that was built there. In 1896, the aluminium factory had what is described as—at least in Wikipedia, so it must be true— “the first large-scale commercial hydro-electric” generation. http://www.theyworkforyou.com/sp/?id=2013-06-18.3.0&s=wikipedia#g3.29

Patricia Ferguson: …in the form of a national tree. I sincerely hope that the Scottish Government will agree to formally recognise such an iconic image for our country. While doing some research for the debate, I noted Wikipedia’s bold assertion that the Scots pine is the “national tree of Scotland”. We know that that is not quite true, but it is interesting that such an assumption has been… http://www.theyworkforyou.com/sp/?id=2013-05-22.21.0&s=wikipedia#g21.4

Christopher Pincher: …on both Front Benches. It is also a pleasure to follow the hon. Member for Streatham (Mr Umunna), the shadow Secretary of State, who made a typically assured and polished speech—as I am sure his Wikipedia page will shortly remind us. http://www.theyworkforyou.com/debates/?id=2013-05-10a.259.3&s=wikipedia#g280.2

 

There’s also some small cases of people treating Wikipedia properly…

Paul Flynn: …of the English monarch as de facto Head of State. I want to clear up one point. There is a belief that the hon. Member for Bridgwater and West Somerset (Mr Liddell-Grainger) is 246th in line to the throne, and according to Wikipedia, the authority for that claim is the blog of “Mr Paul Flynn”. I advise anyone who wishes to repeat that claim to treat it with some caution, http://www.theyworkforyou.com/debates/?id=2013-01-28c.695.2&s=wikipedia#g712.5

However, people in glass houses shouldn’t throw stones.   Wikipedia is well known for requiring citations. It’s untrue that ‘every’ statement needs to be cited (although a sizeable proportion of Wikipedians would like it to be true) but in practice every statement that is likely to be challenged should be cited. In many articles this boils down to the same thing.

Politicians, on the other hand, have NO such restriction. When they quote a statistic, or fact, they are under NO obligation to provide a source.  And one of the things that depresses me about politics is that the electorate appear content with this situation.  Facts are important. Context is important. Both are especially important when you are making decisions that affect many many thousands of people.

I appreciate that during a speech, or particularly a debate, it’s a bit tricky to keep saying things like “If you look at goo.gl/8239 you’ll see” , but in the written word, there should be NO hiding.

 

But were can we find a large number of statements made by MPs in such a way that the prevailing culture is to provide links with statements?

 Methodology: Twitter as a proxy

In the UK our MPs are fond of political point scoring on the Microblogging site Twitter. Some of the points are sensible, some are rabble rousing of the worst sort.

Here’s what I did.  I took the set of MPs on Twitter (thank you to @tweetminister for providing the list) and for each member I pulled out all of their tweets over the last little while.  Checking every tweet by hand would be ridiculous, but we can put together some proxies and make the process more accurate.  First of all I want to pull out those tweets that are definitely from the MPs – so we remove the retweets. Next we want the tweets that are definitely statistics.  Examples might be:

Screen Shot 2014-11-23 at 12.25.05Which is definitely a statistic, but we have every reason to suspect that Brandon made it up.  If he’d given a reference, we’d know both the source of the statement and the context, we’d be able to have an informed debate on the subject (to keep this party neutral I also have a go at Rachal Reeves later on)

So what I’m going to do is just extract those tweets that contain the digits ‘0’ to ‘9’. I’m aware that I lose a lot of facts this way, but I certainly improve the fact:tweet ratio.  This is a weakness in the methodology but as I’m going to be comparing MP against MP we can cope with a little bit of noise in the signal.  This leaves us with a relatively small number of tweets (it varies by MP actually, some MPs tweet a lot without ever mentioning a number, something I find very odd).   Some have links in them and some do not. I want to know how many come with links and which are the politicians most likely to pick a fact out of the air.  So I write a little bit of Java code, and gather some results.

All the code is in this github repository.  The repositry contains both the code, and every tweet downloaded from every MP so you can have a look at how fair the approach is for any given MP (and have a sneaky look at how your MP did, every MPs file ends with a line giving their results).

 

Now, let’s be clear – I’m aware that this is rough mertic. I’m aware that a tweet like “If you want to see the 9 unicorns that the Tories are killing go to www.google.co.uk” counts positively.  I’m aware that having a link doesn’t make something true – as XKCD points out much better than I could:
https://i2.wp.com/imgs.xkcd.com/comics/citations.png?w=1000

But what I’m looking for here is the set of politicians that are willing to back up their assertions with something like this:

Screen Shot 2014-11-23 at 12.42.57

…compared to the set of politicians that are willing to just pull figures out of the air (Sorry Rachel, you get to be the good and the bad example).

Screen Shot 2014-11-23 at 12.42.46

 

Results: Cite your sources guys…

So… let’s see what the data tells us.  Here’s the top 20:

Rank TwitterId Name Tweets Tweets with figures Figures and links Percentage
1 @johnredwood John Redwood 890 78 78 100%
2 @michaelmeacher Michael Meacher 691 41 41 100%
3 @CliveBettsMP Clive Betts 233 25 25 100%
4 @MalcolmRifkind Sir Malcolm Rifkind 68 3 3 100%
5 @HazelBlearsMP Hazel Blears 81 2 2 100%
6 @glendajacksonmp Glenda Jackson 102 1 1 100%
7 @edvaizey Ed Vaizey 424 86 81 94%
8 @SMcPartlandMP Stephen McPartland 636 47 41 87%
9 @TonyBaldry Sir Tony Baldry MP 761 15 13 87%
10 @TogetherDarling Alistair Darling 130 7 6 86%
11 @RoryStewartUK Rory Stewart 798 201 170 85%
12 @GarethThomasMP Gareth Thomas MP 745 89 75 84%
13 @mike_weatherley Mike Weatherley MP 589 36 30 83%
14 @Owen_PatersonMP Owen Paterson MP 74 6 5 83%
15 @DanJarvisMP Dan Jarvis 575 53 44 83%
16 @ChrisRuaneMP Chris Ruane MP 360 35 29 83%
17 @nicolablackwood Nicola Blackwood 399 38 31 82%
18 @tcunninghammp1 Tony Cunningham MP 240 26 21 81%
19 @JonCruddasMP Jon Cruddas 602 39 31 79%
20 @JeremyLefroyMP Jeremy Lefroy 816 62 49 79%

Which is pretty impressive… from those 20 MPs… John Redwood storming out ahead and I’ll admit that Glenda Jackson appears to have been a touch lucky with happening to have a link in her single tweeted figure – I may have to introduce a cut off point in future iterations. I’ve limited it to 20 because these are the ones where I’ve a chance to look thought and check I’ve not done something crazy.

…and, more controversially, the bottom 16, and the reason it’s bottom 16, is because I found 16MPs who, in the time period I was looking at, didn’t cite a single figure…   (there were also about 15 who didn’t mention a single figure, which is alarming in it’s own right…)

@AlbertOwenMP Albert Owen MP 303 16 0 0%
@AnneMiltonMP Anne Milton MP 791 64 0 0%
@Craig_Whittaker Craig Whittaker 64 12 0 0%
@GordonBanksMP Gordon Banks MP 257 14 0 0%
@KeeleyMP Barbara Keeley 177 9 0 0%
@Keith_VazMP Keith Vaz MP 741 27 0 0%
@LindsayRoyMP Lindsay Roy MP 81 3 0 0%
@MPritchardMP Mark Pritchard 769 53 0 0%
@PhilipDaviesMP Philip Davies 320 16 0 0%
@RebeccaHarrisMP Rebecca Harris 21 1 0 0%
@SHammondMP Stephen Hammond MP 239 17 0 0%
@Valerie_VazMP Valerie Vaz MP 83 7 0 0%
@damiangreenmp Damian Green 424 29 0 0%
@jreedmp Jamie Reed 361 5 0 0%
@pamela_nash Pamela Nash MP 284 9 0 0%

I should give @AnneMiltonMP a bit of a pass on this.  It turns out that Anne keeps writing things like:

and about once a day:

which, of course, mean that she does undeservedly badly by this metric…

Pamela Nash is relatively blameless but one of her nine is this sort of one:

…all of these MPs might be exactly right with their statistics, but unless they back them up with the links, they are going to keep lowering the level of the debate…

 

 

 

 

One thought on “The 16 UK politicians most likely to be making up facts.

Leave a Reply