Think tanks, newspapers and Twitter

Here are some fascinating graphs drawn from a statistical study of the political influence/orientation of British political think tanks carried out by economist Andrew Whitby - a study based on Twitter. 

Using the property of homophily - the tendency we human birds have to flock together with those who share our outlook and opinions - Andrew Whitby trailed all the Westminster MPs who use Twitter and noted down all the UK think tanks they chose to 'follow'.

The assumption here is that, in general terms, "you are likely to have similar views to those people that you follow or are followed by (or retweet or are retweeted by)." 

Measuring the difference between the proportions of Labour and Conservative MPs who follow each think tank could, therefore, give the world a strong indication of where that particular think tank stands on the political spectrum. 

The resultant graph rings true with me. See if it does with you too. 

There's only one possibly surprising result in that graph - namely the apparent left orientation of the Institute for Fiscal Studies, a body which is usually described (by the BBC) as 'respected', 'influential' and 'independent'. 

Andrew Whitby then conducted a similar study of whom MPs follow on Twitter, this time concentrating on the UK's newspapers, producing this graph:

That rings true too.

Such techniques remain useful for measuring bias as well, of course - and some have tried them already. 

The Centre for Policy Studies report into BBC bias, which many of you will remember, noted citations of think tanks by the BBC News website. It claimed to have found objective proof of BBC bias.

Its measure of left-right orientation, however, was somewhat unusual: It was based on how often those think tanks were cited by the (right-wing) Daily Telegraph and the (left-wing) Guardian - a technique which produced this rather bizarre chart:

Now, that doesn't ring true, by and large. The IPPR is NOT more right-wing than the Adam Smith Institute. Demos is left-leaning but most certainly not THE most left-wing think tank, or to the left of Compass. ResPublica isn't more left-wing than the Fabian Society - and nor is the Global Warming Policy Foundation. The Institute of Economic Affairs is most certainly NOT to the left of the Work Foundation either - etc, etc. 

For a statistical survey to really hit home it has to have very stable roots and not produce lots of obviously wrong results. 

The CPS went down a fruitful path by investigating BBC bias in this way, but undermined their own case because of this not-entirely-convincing way of defining left-leaning and right-leaning think tanks. They, thus, found themselves easily - and widely - dismissed. 

Maybe they should try again, using Andrew Whitby's Twitter results. Or someone else should try instead.