FEATURE15 April 2010

The future... in 140 characters or less


Tweetminster is running an online experiment to see whether it can predict the outcome of the UK general election based on what people are saying on Twitter. It will be one of the biggest tests to date of the predictive power of social media data.


It’s coming up to a year since we asked the question, ‘Can social media analysis be used to predict the outcome of elections?’ The question was posed in the wake of one company’s efforts to forecast the outcome of the American Idol contest, which – despite a run of success – fell down, quite literally, at the final hurdle.

At the time we posed the question the general consensus was ‘no’. But it was a theoretical ‘no’. No-one had tried to do it, you see – at least, not then or not that we were aware of. So thank heavens for Tweetminster: a project set up to help voters find and follow politicians on Twitter, which has built a model to predict the outcome of the UK’s upcoming general election based on the analysis of tweets.

The Tweetminster group credits as its inspiration an experiment conducted by Tokyo University graduates last year, ahead of the country’s general election in August, which analysed the correlation between online buzz and election results.

“The aim of the study was to assess if word-of-mouth mentions of candidates could help to predict which ones would be successful,” writes Tweetminster in an online paper. “The study found that in a majority of constituencies the most mentioned candidate won the seat.”

For its experiment, Tweetminster is identifying the most talked about parliamentary candidate in each constituency, from which it is forecasting the likely outcome of the election in terms of vote share. Its latest figures put the Conservative Party marginally ahead of Labour on 35% versus 32%, with the Liberal Democrats on 23%.

Compared to the more traditional pollsters, Tweetminster’s model seems to understate Tory support, while overstating that of its rivals (recalling Alison Macleod’s warning from last year that “social media tends to skew younger… it skews more left-wing”.)

Tweetminster says it is not looking to compete with polling methodologies – this is “an experimental study”, it says. Still, whatever lessons it learns will no doubt be of interest to the survey research establishment.

The work is, after all, at what the US Advertising Research Foundation (ARF) has described as the “cutting edge” of the listening revolution, a movement that demands that the research industry open its ears to what consumers are saying in public forums, rather than just asking questions all the time.

Speaking to Research earlier this year, Steve Rappaport, the author of the ARF’s Listening Playbook, explained that: “One of the interesting directions that listening is going is that as the research becomes better – as the quantification of listening data becomes better – the listening data as numerical data can be used in ways that traditional data is used: for forecasting, for plugging data into models, etc.

“I think the real potential of listening is in its ability to contribute not only to doing research in new ways, but in terms of developing new data that can be used in ways that support traditional modelling and forecasting. It can also provide new data that allow those disciplines to evolve further to become more predictive and more insightful.”

It is a school of thought seemingly embodied by the agency Conversition, which describes itself as a social media research company, as distinct from social media monitoring and social media analysis. “Social media research is the application of scientific marketing research principles to the collection and analysis of social media data such that valid and reliable results are produced,” says the agency’s website.

What this means in practice is that Conversition produces similar outputs to a traditional market research agency, only it sources its data from the social web rather than through surveys. There are some limitations to the approach. Brand or product awareness is impossible to measure, says chief research officer Annie Pettit, explaining that just because someone doesn’t mention a brand in an online conversation doesn’t mean they are unaware of it.

Otherwise, Conversition has most of the basic tracking measures covered. And, says Pettit, its wholly possible to use the company’s data to create working predictive models – and why wouldn’t you? “That is the whole purpose of market research,” she says.

Perhaps the best example to date of such a model is one constructed recently by social computing experts from Hewlett Packard. Using Twitter, researchers Sitaram Asur and Bernardo Huberman were able to predict the box office revenue of new cinema releases based on the rate at which tweets about a particular movie were generated.

In a paper detailing their experiment, the pair give a ringing endorsement of the predictive power of social media – one that should act as a rallying cry to all entrepreneurially-minded market researchers out there. “This work,” they say, “shows how social media expresses a collective wisdom which, when properly tapped, can yield an extremely powerful and accurate indicator of future outcomes.”


14 years ago

Thanks for the interesting discussion Brian. I wonder if you add anything in light of last night's live TV debates? I've also just written a blog comparing how the main parties are using social media in general which may be of interest - http://www.siliconbeachtraining.co.uk/blog/social-media-predict-general-election/ Natasha

Like Report

14 years ago

Hi Natasha, We have plenty of stuff about the leaders' debates appearing on the site soon. Keep an eye on the homepage.

Like Report