Warren Pearce and Brigitte Nerlich, from the University of Nottingham Institute for Science and Society, in collaboration with two researchers from Amsterdam, have published an interesting new paper looking at the twitter response to the launch of the IPCC WG1 Report (see also MSP blog post). They collected tweets sent between September 17th and October 8th 2013 (recall that the IPCC Summary report was published on Sept 27th). They included all tweets containing the term IPCC, whether or not they used the hashtag #IPCC, giving over 150,000 tweets. They then weeded this down (for example removing those that were about police complaints, and removing duplicates, retweets and ‘via’ tweets). They then cut down the number further, by focussing on ‘conversational’ tweets, i.e. those that included another @username, bringing the number down to about 60,000, involving about 11,000 twitter users.
There is a thoughtful discussion of ethical issues, where they say they decided not to mention any individuals in the paper, as doing so may “have unwanted side-effects” and “may bring them unwanted and disproportionate attention from those holding opposing views”. A lot of nonsense could have been avoided if such ethical considerations had been adopted more widely.
For the 239 most connected twitterers, a network analysis was carried out, giving colour-coded network diagrams. Some tweets were generated by a one-click campaign by an NGO called Avaaz – but apparently these don’t really influence the results. The network plots are colour coded and grouped. In one plot (fig 3), the colour indicates different types of communities; I don’t fully understand this (I guess the ‘communities’ are something the software does automatically), but they say that in the UK there are more connections between people with different viewpoints than in other countries. A second figure (fig 4) colour codes the network map according to whether the tweeter was supportive, neutral or critical of the IPCC. About half of the users were supportive, and a quarter critical. These results suggests that there is more ‘intermingling’ from the sceptic side: as they put it “attempts by unsupportive to connect with supportive were not always reciprocated”. In the conclusions there is a nice little dig at a very silly paper by Elsasser and Dunlap who claimed that climate sceptics live in an echo chamber.
From the maths point of view, there’s an interesting power law dependence in the graphs in Fig 1 and Fig 2 (the central sections of both log-log graphs form almost perfect straight lines).
Overall, I think this is quite interesting, and it studies the subject with a scientific, data-driven objectivity, which is more than can be said for some of the social science research in the field. As with Amelia Sharman’s paper on the sceptic blogosphere, some will say “we could’ve told you that”, but it’s good to see large data sets confirming general impressions.