The following piece I penned for our weekly Digital Inspirational email at Blue Rubicon. Do sign up if you’d like such stories in your inbox every week.
The widespread and easy availability of data from Twitter makes it a popular source of insight into what the public is doing, thinking, eating, watching or buying. The Times’s Redbox newsletter for example now regularly features assessments of the different party leaders based on analysis of tweets.
But is this data reliable? Are people on Twitter representative of the rest of us? Much analysis – too much analysis, in fact – rests on the hope that it is.
Courtesy of pollsters GfK we now have some new data on this question. The pollster both polled likely voters as to their voting intention and then also looked at just those who are active on Twitter. The results were rather different – with Labour much more popular amongst Twitter user than amongst voters overall, and the Conservatives much less. Other parties were not much different between the two.
Full voting figures: Conservative 41%, Labour 28%. Twitter-only voting figures: Conservative 30%, Labour 39%.
A warning not to assume that Twitter = everyone.
UPDATE: Here’s further data which reinforces the point with a more general finding: self-selecting surveys online are not the same as public opinion.
UPDATE 2: Here’s some further data, including why it may not matter if online views are unrepresentative.