BBC BLOGS - dot.Rory
« Previous | Main | Next »

Online 'sentiment' around the prime-ministerial debates

Rory Cellan-Jones | 12:50 UK time, Friday, 30 April 2010

The prime-ministerial debates have been a fascinating experiment, not just in the way that television can influence an election campaign but in a new way of combining the web and TV to make both a richer experience. One example - to blow the BBC's own trumpet - is this page which allows you to review all the key moments of last night's debate and see a transcript of what was said.

Another innovation has been the way a growing number of people have followed these major events via two screens - a laptop perched on their knees as they sit in front of the television, connecting with social networks to comment live on what they're seeing. This has in turn made possible an industry of what's being called "sentiment analysis", which attempts to understand what has been said online.

We've been asking a firm called Lexalytics to give us instant readouts of Twitter sentiment - last night's chart showed David Cameron and Nick Clegg both attracting roughly equal numbers of positive comments, with Gordon Brown in negative territory.

Lexalytics chart showing sentiment towards leaders

Other firms have been using similar techniques. Linguamatics, based in Cambridge, says it has used sophisticated text-mining software and linguistic analysis to understand how Twitterers reacted. This chart maps positive tweets for each leader, in relation to each question asked during the debate.

Chart showing

The analysis offers a perhaps puzzling result. Counting positive tweets during the debate, it says Nick Clegg had 37% of them, followed by Gordon Brown with 32% and David Cameron in 31%. That's not a result reflected in polls and other sentiment measurements. My hunch is that the volume of tweets may have been higher for Gordon Brown than David Cameron - and for all those positive ones, there were actually more that were negative.

Another company, Meltwater Buzz, tells me it does a much wider sample of online opinion, looking not just at social-media updates but also at blog comments, newspaper articles and just about all readable web content. It has tracked sentiment throughout the election campaign, and produces results very different from those I've seen elsewhere. For instance, the Meltwater tracker shows Nick Clegg achieving lift-off during the second TV debate, not the first. The explanation offered is that it was then that the satirical #nickcleggsfault craze took off, and this was measured as extremely positive for Mr Clegg. The firm's latest trends show David Cameron forging ahead, following Gordon Brown's "bigoted woman" gaffe and then last night's debate.

Chart showing

Once again, let me stress that the technology of "sentiment analysis" is untested - at least in the political arena - so all these charts and conclusions should be taken with a massive pinch of salt. But some of the practitioners are gearing up for the ultimate test, by trying to predict the election result. A week from now, we will know just how seriously to take this new technique.

Comments

  • Comment number 1.

    I don't know who these polling and stats companies have been following, but the #leadersdebate hashtag was completely full of vitriol for David Cameron all night long for all three nights.

    Maybe its because they use computers, but from what I saw and what I read, the idea that there is any nice sentiment out there on Twitter to Cameron from the general populous is incredibly misguided.

    I'll call it now. Hung Parliament with Labour being the largest party.

  • Comment number 2.

    There is an online timeline of predicted outcome here https://livetalkback.com/blog/?p=87 - it's generated from a timeline of live opinion votes here https://m.livetalkback.com/polls/poll/665 - interesting to watch this change as the campaign evolves.

    It's going to be interesting to see if these online methods are more, or less accurate than the traditional pollsters.

  • Comment number 3.

    I concur. The polls being run by TwtPoll bears this out as well. https://twtpoll.com/r/p5wt6a

    Hence my disbelief on the 10 o'clock news with the polls stating Cameron had won it...

  • Comment number 4.

    **embarrassed blush** er Rory, you've just described my approach to all the PM debates. Me with Laptop (on the knee), TV, Bottle of Beer and following the tweets/Facebook comments and "the worm".

    It is a highly addictive way of getting almost real time feedback on a live debate. Every nuance, gesture and point is virtually analysed to death. However in amongst the dross, there are some excellent points raised. Points that later appear in certain political analysts blogs and articles! ;)

    After the event, it is always interesting to compare my "on-line experience" to the debate with the often jaundiced political media interpretations. Quite often I find myself asking, "were they watching the same debate as me?"

    Oh well! the vagaries of perception I suppose!

  • Comment number 5.

    A fascinating case study in the evolution of new uses of technology. Sentiment analysis is a new field, and is going to have teething problems - I'd imagine that Twitter users skew more Liberal Democrat than the general population simply by virtue of demographic trends, but that will probably be decreasingly true as time goes by.

    The future? Maybe. Fascinating in the here and now? Unquestionably.

  • Comment number 6.

    This comment was removed because the moderators found it broke the house rules. Explain.

  • Comment number 7.

    Heres a proposition; it is younger voters that are more likely to be tweeting, blogging, on facebook, watching youtube debates (https://www.youtube.com/ukelection). And online it seems that the LibDems and Cleggmania is alive and well. This then begs the question on how relevant online opinion is, after all young people are among the least likely to vote. Except that this election for whatever reason seem's to have motivated unprecedented numbers of new voter registrations and a disproportionate number of these are in the 18-24 age bracket (https://tinyurl.com/2e8682o%29 This could all mean that the carefully calculated weightings of all those opinion polls are wrong and this election still really is wide open.

  • Comment number 8.

    Whilst this was all very interesting to start with, after basically writing the same blog after every political debate, it has become a bit dull. Surely the Apple/Adobe war warrants a blog post? It is far more of a technology story than this. Twitter is not at all what I would consider as technology, More of a use (and limited at that) of technology. It would be far better for Twitter and all the other online services simply be seen as just another form or analysing the political spectrum. Essentially Twitter is simply people stating their mind in a much more public way. Technology hasn't changed anything really here.

  • Comment number 9.

    "The Triumph of Elfin Lord Clegg" by our correspondent from Sirius Five.

    In the mighty play-off last night, Elfin Lord Clegg was the mighty victor against Data Cameron and the doughty Gordon Rumplebrownfairmanlethorpe. In an epic battle an inspired Elfin Lord Clegg channelled the spirit of Al Gore to become the "mighty communicator". Data struggled robotically - even sacrificing his eldest child to the STATE SCHOOL SYSTEM ! (Old Etonians loaded their trebuchets). Rumplebrownfairmanlethorpe - fuelled by his commitment to end child poverty - hailed the child trust fund, and the child tax credits, designed to to give ordinary Britons a life chance. Data imploded, Nick's ears glowed. I'll still vote Labour, but I might be persuaded to have a Babycham with the Elfin Lord...

  • Comment number 10.

    Hi Rory and everyone else who commentated for the links to sentiment analysis.

    I wondered if you cared to share your opinion on the tv debate analysis of the comments on the BBC have your say forum I have carried out:

    https://electiontrends.blogspot.com/

    There are different forms of analysis, including sentiment analysis so please have all look at all our different posts.

    We are in the process of analysing the entire bbc have your say comments on the election and the forum blogs. So please follow me on on twitter if you wish https://twitter.com/electiontrends

    kind regards and best wishes
    Jayne Coulthard

  • Comment number 11.

    "I don't know who these polling and stats companies have been following, but the #leadersdebate hashtag was completely full of vitriol for David Cameron all night long for all three nights."

    I'd hazard a guess that this is because most people using twitter are likely to be quite young and have only known a Labour government. They probably think the arrogance of Brown is what a politician is supposed to be and don't like anything different.

  • Comment number 12.

    This comment was removed because the moderators found it broke the house rules. Explain.

  • Comment number 13.

    Democracy UK on Facebook conducted a poll regarding the performance of the leaders in the last debate which put Nick Clegg as the winner in 46%.

  • Comment number 14.

    This comment was removed because the moderators found it broke the house rules. Explain.

  • Comment number 15.

    Rory, I think the sentiment analysis is great - definitely gives an 'on the ground' slant to the complement the usual. I'm also curious to know what the twitter volumes are as mentioned above - I was under the impression that the voter apathy was quite high this time around.

  • Comment number 16.

    When's the Technology blog coming back?

  • Comment number 17.


    Judging by the number of comments on this blog, compared with say Nick Robinson's blog (several hundred per post), who cares what people using Twitter think about anything, the debate included? It is a very small proportion of voters.


  • Comment number 18.

    The proof of the pudding for these technological gadgets (like Twitter), social sites (like FaceBook) and video feeds (like YouTube) will become self-evident AFTER the elcections.
    Did voting numbers jump? i.e. Did voter apathy drop?
    Did the polling results off these technological tools hold up (i.e. Clegg, Cameron and Brown dragging up the rear)?
    I can't wait for the election results. I want to see how the actual results reflect all this twittering, in your face, tubal nonsence that has gone on over the last couple of months.
    My prediction: There will be little correlation between these technological "things" and the actual vote results, but I've been wrong before (Mind you, my error margin is less than 0.0001%).

  • Comment number 19.

    As a not so young Twitter user (per one of the comments below about Twitter users being young), I am greatly suspicious about many of the tweets. Likewise comments on blogs such as this. Sadly I think that social media networking has been highjacked during these pre-election weeks by party activists of all colours. What could be a great way of galvanising grassroots democracy, has been taken over by people who care little for democracy, debate or free speech. Getting their party's view over and trashing the others seems to be their aim and at any cost - even to the extent of the Labour party's Twitter "czar" (note to David Cameron and Nick Clegg - sack all tzars!), breaking the law. A sad use of an innovative and accessible form of communication (yes even for the not so young!)

  • Comment number 20.

    This comment was removed because the moderators found it broke the house rules. Explain.

  • Comment number 21.

    How to make decisions in Modern Britain. I need to get the brakes on my car fixed. Should I stick with the mechanic who seems to know a lot about cars or go with the new guy who would look better on a Mechanics Head2Head TV debate.

  • Comment number 22.

    The UK election race is pretty interesting.......... but it all is the same old shit, I watched all the three live debates and there was hardly any difference between the 3 accept few point scoring. I mean how is it that 3 different parties with different concepts have almost the same views on all the big issues. Don't you guys think its about time to come up with a new way of politics or maybe more so A different way of government. Like something new, cool, mordern, better and more interesting where everybody can easily take part and is more willingly interested.

    For this election my vote is for Nick Clegg because it seems like he can make the biggest difference in the same old crap system.

  • Comment number 23.

    @Sam - think you're stuck between a a rock, a hard place and a steel wall unfortunately - none of the candidates in my opinion have sufficient credibility to run the country. Given the lame Gordon Brown responses and David Cameron's constant Brown-bashing without offering substantial alternative policies, I think you're right - Nick Clegg is probably the best of the three.

    Rory, any chance of the sentiment analysis continuing a short while after the election please - I would be curious to see what the reaction of the British public is to the outcome - particularly if it ends up as a hung parliament. I reckon it would be sort of like the winner's curse phenomenon when it comes to purchasing things in an auction!

  • Comment number 24.

    We did a pretty similar analysis as well and what we found was that in relative terms, Nick Clegg received the most negative comments before the debates. After the success in the debates, the sentiment regarding him improved significantly. We analysed social networking sites like Twitter and Facebook but also other sites like forums, blogs and online news comments.
    An article of the study can be found here: https://www.whitevector.com/blog/2010/05/10/2010-general-election-the-role-of-social-media/
    However, as said, sentiment analysis is definitely interesting but mostly insufficient. In many cases, negative comments could even be more desirable than no comments at all. To go further in the analysis, you need to look for absolute and relative discussion shares, discussion spread to different sites, most activated consumer groups etc.
    Cheers,
    Essi, Whitevector

  • Comment number 25.

    How to make decisions in Modern Britain. who cares what people using Twitter think about anything, the debate included? It is a very small proportion of voters.Should I stick with the mechanic who seems to know a lot about cars or go with the new guy who would look better on a Mechanics Head2Head TV debate.

 

BBC © 2014 The BBC is not responsible for the content of external sites. Read more.

This page is best viewed in an up-to-date web browser with style sheets (CSS) enabled. While you will be able to view the content of this page in your current browser, you will not be able to get the full visual experience. Please consider upgrading your browser software or enabling style sheets (CSS) if you are able to do so.