Twitter's Real-Time Collective Mood Index

Recommended Sources

About Media Industry

BNET Media provides daily industry trends and news coverage with insights for managers and executives in publishing, print, broadcast, film, and online media. In addition to media company profiles, we bring you industry analysis on new partnerships, media products, mergers and acquisitions, labor and cost management, media buying, investments and a host of other important business issues.

We have a little bit of everything out here in California; once we even had a statewide “California Task Force to Promote Self Esteem…” and other values, which is what sprang to mind upon reading about those two University of Vermont statiticians who consider Twitter a “a hedonimeter,” a device that measures emotions like happiness.

Their goal is to establish a NYSE-type index of the national mood in real-time.

So far, in a related study of song lyrics, blogs and speeches, they have been able to determine that teenagers and the elderly are generally miserable, as measured by the frequency of the word “sick” and other negative terms. The pair say it seems people appear to be at their happiest in the 50s and 60s.

The researches say they hope their Twitter mood index will be useful, as the Google Flu Trends has proved to be, except of course in the case of swine flu in Mexico.

So, this is another example of how companies, academics and entrepreneurs are finding new uses for the real-time stream of information flowing over the micro-blogging service. In its monthly message announcing its latest data updates, Compete.com today noted that “the Twitter craze spells success for Twitter tools riding its coat tails.”

Twitter itself, however, is seeing its online traffic plateau again, after an eventful June news spike (Iran and MJ). According to Compete, Twitter’s web traffic grew only by a modest 1.25 percent in July. The top referral sites were Facebook, which drives 11.44 percent of Twitter’s web traffic, Google (9.94), Yahoo (5.08), MySpace (3.11), and Twitpic (2.54).

That means those five sites drive almost a third of Twitter’s online visits, thereby generating much of the flow of material the Vermont guys will be using for their mood index.

Related Bnet coverage of tech-based media hedonimeters:

The NY Times Seconds That Emotion

New Media Tools Go Beyond Multimedia

Tags: Twitter, Entrepreneurship, Management, David Weir

In addition to serving as a BNET Media analyst/blogger, David Weir is a veteran journalist and the author of several books. Weir is a co-founder and vice-president of the Center for Investigative Reporting, as well as an editorial board member of The Nation.

BNET User Analysis

Voice your 2cents

Web Buzz:

Auto Task Force Hiring. Obama Administration is Mum, though

Automotive - 243 days 14 hours 57 minutes ago

Need a job? Do you have some financial expertise? Do you know a little bit about the auto industry? Well, it seems that the Obama Administration's auto task force is hiring. It turns out that a new member of the task force, Harry J. Wilson, sent out an e-mail around Wall Street which was seeking applicants for up to four jobs on the task force....
VIDEO: ABL Targets Missile With Surrogate Laser

Aviation Week - 71 days 5 hours 48 minutes ago

This video from Boeing takes a little bit of interpretation, so I suggest you watch it, read the description below, then watch it again (once or twice). The surrogate high-energy laser referred to is a low-power laser that is fired through the ABL's beam-control optics to simulate the aircraft's actual megawatt-class high-energy laser....
Ball Aerospace to ship WISE spacecraft

Military & Aerospace Electronics - 104 days 16 hours 31 minutes ago

BOULDER, Colo., 12 Aug. 2009. Ball Aerospace & Technologies Corp. has completed a comprehensive pre-shipment review for NASA's Wide-field Infrared Survey Explorer (WISE) satellite. The pre-shipment review was completed on August 4, following assembly and performance testing. Ball Aerospace engineers have packed the satellite into a shipping...
Health Reform Summit: BIO's Invitiation Lost in the Mail

The In Vivo Blog - 263 days 11 hours 35 minutes ago

We've written plenty about the parallels and contrasts between health care reform vintage 2009 and the last big go 'round in 1993 with ClintonCare.Today's health care reform summit definitely fits in the "contrast" category. And not just because it was broadcast live on the web, in self-conscious contrast to the Clinton task force's closed door...
RIAA Still Filing Lawsuits...

TechDirt - 203 days 6 hours 16 minutes ago

Back in December, the RIAA claimed that it had discontinued its strategy of filing lawsuits against individual file sharers. Specifically, the RIAA's letter to Congress stated "we discontinued initiating new lawsuits in August." While we've pointed out in the past that this is wholly untrue , some Hollywood lawyers took us to task, claiming...

BNET TalkbackShare your ideas and expertise on this topic

Subscribe to this discussion via Email or RSS

1

TheNudger

08/12/09 | Report as spam

RE: Twitter's Real-Time Collective Mood Index

Being statisticians, the VT guys surely realize that the characteristics of a sample from a population are not necessarily the characteristics of the population. For that to be case you have to carefully select the sample - that's why pollsters and survey types get paid the big bucks.
The sentiments of addicted Tweeters, for example, are probably no more representative of the national mood than the sentiments of the "sample" of America that is showing up to rant and rave at the Town Hall meetings.
2

Nohohome

08/12/09 | Report as spam

RE: Twitter's Real-Time Collective Mood Index

Dear TheNudger: While what you say rings true to me in terms of any one cohort, it doesn't have much to do with the use of communication for measuring sentiment. We have to assume that the whole picture is never represented in organized measurement -- this certainly includes things being looked at such as national mood. Having never been to Twitter's web site, let alone use it, this doesn't render me at all uninterested in knowing what others get out of it, and in particular, put into it. These smaller cohorts are always helpful for piecing together the puzzle of what's going on. I think the real purpose (advertising/making money) driving what gets looked at in turn determines what's available to us on, say, the Internet. Given that, it becomes very important to glean what's going on and measurable at all -- including addicted Tweeters as well as ranters and ravers at Town Hall meetings. All of it is always swirling around us.
3

TheNudger

08/12/09 | Report as spam

RE: Twitter's Real-Time Collective Mood Index

@nohome. I agree completely on the value of small cohorts - I live (and die) by them in my business. The key to their utility in predicting anything from voter behavior to consumer preferences is to understand the selection biases involved in their creation.
For example, consumer reviews on the Web are empirically known to be biased toward the negative - it turns out that people who are satisfied rarely take the time to post comments, which those who are not are will do so with a vengeance. This bias is not so important for a site that just lists stories and solicits comments (e.g. www.consumerist.com), but become very important when you try and quantify consumer/citizen feedback by turning it into rankings or ratings (www.vanno.com, for example). It's sort of analogous to the systematic inaccuracies that pollsters found when polling was being done only on landlines, but the voter base (particularly younger people) was rapidly shifting to cell phones.
And I commend you for never having used Twitter - you (madam or sir) are possessed of an untrammeled soul.
4

TheNudger

08/12/09 | Report as spam

RE: Twitter's Real-Time Collective Mood Index

sorry - fingers got ahead of brain. the garbled snippet "...which those who are not are will do so with a vengeance." should read "...while those who are not will do so with a vengeance."
5

hotweir

08/13/09 | Report as spam

RE: Twitter's Real-Time Collective Mood Index

Thanks for the dialogue, guys. Since only about one in twenty Americans use Twitter, it is certainly representative only of a specific type of individual, which is a point I should have stressed, but didn't in the post. Also, something like ten percent of those on Twitter generate 90% of the Tweets -- an even more rarified circle. It's probably a similar demographic to bloggers, i.e., people who are pretty addicted to written communication forms. And we all know that writers are among the moodiest people on the planet!
6

emiltsch

08/13/09 | Report as spam

RE: Twitter's Real-Time Collective Mood Index

@TheNudger - good point about it not being representative...

A couple of add'l thoughts is the immediacy of the twitter user's thoughts/emotions - their "mood" could flip-flop on a topic in a matter of minutes, hours & days.

Also - with such a large portion of twitter user's being early adopter, Type-A personality types, I'd think the mood would be slanted towards the positive due to their general personality traits.

Not sure how the results of this sentiment measure would be applied...one thought: Trying to determine the best messages & content to tweet that may result in a higher click-thru rate (on links within messages sent)

IE: An overly negative mood may drive a marketer to try his hand at submitting motivational/self-improvement topics that touch on specific emotional hot buttons...

As for the traffic plateau, my guess is more people were out and about more in July & the mainstream attention twitter received earlier in the year grabbed a good portion of the curiosity seekers. I wouldn't be surprised to see another growth bump as the summer winds down, school/college begins and people start spending more time inside.

Just a couple of random thoughts w/absolutely zero research done to verify those ideas...besides, the study could also be a way for a couple of statisticians to burn through the grant money they received for the study.