A friend of mine, Robin Hamman who looks after the BBC’s blog trials recently brought to my attention this study of the expressed political views of BBC Facebook users which was put out by conservativehome - an independent right wing UK website - back in October and was picked up by the Daily Mail.

The study (whose figures I have updated here using the same Facebook ad tool that conservativehome used) showed that of the 11,040 BBC staff registered on the site, 1420 staff put themselves in the “liberal” or “very liberal” category, compared with just 120 who labelled themselves “conservative” or “very conservative”. 420 regard themselves as “moderate” (the rest did not specify their political views). This compares to roughly 160k liberals and 56k conservative Facebookers in London and 847k liberals vs 233k conservatives in Facebook across the UK. (For the curious - there are < 20 self-confessed liberals working for Fox News in the US on Facebook compared with 40 conservatives and an equal number of moderates).

Of course this is somewhat embarrassing for the BBC as it provides further ammunition for those who would accuse it of liberal bias. The sample is a self-selecting sample from a self-selecting sample however, therefore no more than suggestive - and of course it includes large numbers of staff not involved in politically-sensitive work.

I find it interesting to note that the information provided probably included a large number of people who specified that their profiles (including their political allegiances) should be private. The privacy does not, however, protect users from being aggregated in order to be sold to - it is Facebook's ad sales tool that enables anyone to 'mine' Facebook to find out the expressed interests, ages and - yes - political affiliations of its users, grouped by organization. As this example shows, even aggregate data can be harmful to an organization when made widely available.

I also note that it is possible to attempt to advertise to Facebook users as young as 13 - and the ad sales tool says nothing about relevant regulations.

I just know there must be something out there in the HCI literature on how computer users are influenced by things like the size and position of buttons, or (of specific interest to me) by the size of text entry boxes. If you have a text entry box 6 lines long on a page and an otherwise identical page but with a 12 line long text entry box wouldn’t you expect the person writing in the bigger box to write more? (Even though in both cases the writer can in principle write as long as they like, with the text box developing scroll bars if they reach the end of the box)? I know that there is an interesting literature on the effect of defaults - I am hoping there is some similar research on this topic.

If you happen to know of evidence for this (or someone else at least asserting the same) either in the academic literature (preferably) or in the practitioner literature please comment…

The BBC has launched BBC Memoryshare
“A living archive of memories from 1900 to the present day.” They suggest that what is provided “may be used as a source of programme content for the BBC.”

It seems YouTube viewers watch more anti-immunisation videos than ones promoting immunisation, and public service videos on the subject were among the most low-rated. It seems Gresham’s Law can be applied to information on the Internet as well - bad (but interesting) information can drive out good.

Ironically, the JAMA study referenced is itself subscription only!