twitter, social science and mood
[link via David Lazer]
Twitter is getting lots of interest from social scientists. Here’s a piece from the current issue of Science about how “social scientists wade into the tweet stream” (the figure below is from this article). And, an NPR piece on a forthcoming Science article by Macy and Golder on affect and mood and twitter.
orgtheory.net 
The fact that the saddest day this year was the day Bin Laden was finally killed serves as an alarming reminder, for those who still needs it, about the sort of world we live in (I mean the world outside the US).
LikeLike
cynicman
September 30, 2011 at 1:03 am
If you look at the chart for hourly affect it rises in the morning and peaks toward the night. I can’t help but wonder how much of that is people saying “good morning” and “good night”. Similarly here how much of the new year’s peak is attributable to the phrase “happy new year”?
Still pretty cool.
LikeLike
RandomDude
September 30, 2011 at 3:42 am
What, if any, information do we have as to the demographics of twitter users? From my grumpy middle-aged perspective, calling this “average happiness” is a bit like sampling people on golf courses to ascertain “average income.”
LikeLike
andrewperrin
September 30, 2011 at 11:58 am
how is this not a fallacy of composition?
LikeLike
B
September 30, 2011 at 5:20 pm
RT @andrewperrin: What, if any, information do we have as to the demographics of twitter users? #nailedit
LikeLike
gabrielrossman
September 30, 2011 at 5:27 pm
First, before we completely dismiss this, the study is forthcoming. Secondly, it’s surprising to me to see how quickly people are dismissing this. Obviously, computation social science provides huge opportunities for studying large-scale behavior. Thirdly, “Dodds, Christopher Danforth of Vermont, and colleagues examined 4.6 billion tweets over nearly 3 years.” 4.6 billion observations from three years! No doubt this contains a large amount of retweets, but I am fairly confident that this sample is far more globally representative than the research produced by the vast majority of sociologists.
LikeLike
Trey
September 30, 2011 at 5:51 pm
It’s the average happinness of the Twitter population – one of the biggest countries in the world.
I fail to see the problem?
LikeLike
Anonymous
September 30, 2011 at 7:08 pm
[…] twitter, social science and mood « orgtheory.net – […]
LikeLike
Links for September 29th, 2011 through September 30th, 2011 | KevinBondelli.com: Youth Vote, Technology, Politics
September 30, 2011 at 9:02 pm
[…] twitter, social science and mood – […]
LikeLike
Daily Links for September 29th through September 30th | Akkam's Razor
October 1, 2011 at 12:01 am
My general reaction is that this type of stuff is empiricism gone awry. But – it sorta is what it is: new medium, lotsa data, descriptive, sex appeal. I’m sure there’s lots of un-programmatic, freakonomics-type work yet to come with twitter and related datasets. (Well, as well as good work – I haven’t read the most recent stuff to judge). But it gets attention, published in visible outlets – and, I think it is very interesting (for what it is). I think this type of work has its place, thus the varied outlets (some of the social “science” in journals like Science and Nature surprises me but I guess that is what they are looking for).
That Macy and Golder piece referenced above is now up on the Science site – “Diurnal and Seasonal Mood Vary with Work, Sleep, and Daylength Across Diverse Cultures.”
LikeLike
teppo
October 1, 2011 at 12:07 am
@Trey. “Computational social scientists” are remarkably cavalier about the representiveness of their “samples.” Many seem to believe that having lots of data scraped off the web absolves them of the need to worry about sampling weights, nonresponse bias, measurement error, and any number of related issues that people who collect survey data have fretted about for years. (I agree that many secondary *users* of survey data are likewise too cavalier, but the apples-to-apples comparison is of people who collect and first publish off the data.)
On your first argument, “forthcoming” doesn’t always mean “good,” even in a high-status journal like Science. Or, maybe especially in Science: I’ve read some real stinker social science articles in quasi-popular, general science journals, I suspect because the editorial staff ,may not know much about the social sciences and goes for quirky or controversial over convincing. (Note that I’m NOT referring to the Golder/Macy article in particular or making any kind of judgment about its quality, because I haven’t read it.)
BTW, what’s with the NPR article describing Golder as a “social and computer scientist”? He’s a graduate student in sociology at Cornell, which would seem to make him, well, a sociologist.
LikeLike
krippendorf
October 1, 2011 at 1:57 am
I didn’t mean to imply that “forthcoming” meant “good.” I meant that it hadn’t been released yet.
LikeLike
Trey
October 1, 2011 at 3:25 am
@Krippendorf: Have you read this particular study, or just making some general statements, and trying to (implicitly) dimiss this particular study?
Secondly: I’d much rather read a paper based on this kind of data, rather than a study that brilliantly has thought of all it’s methodological limitations, but has a fairly limited dataset and barely significant result that don’t explain much. Which is, I guess, +75% of all published research.
Furthermore, plenty of studies have examined the demographics of Twitter. Easy to put together.
LikeLike
Anonymous
October 1, 2011 at 6:56 am
@Anon: re measuring average happiness of Twitter users, sounds good to me; but that’s not how it’s published or marketed.
@Anon: plenty of studies have examined…: please cite, I’d be interested.
@Trey: sheer sample size doesn’t fix a biased sample. If the selection process is systematically skewed, more selections just means more skew.
LikeLike
andrewperrin
October 3, 2011 at 2:20 am
[…] the academy. The Golder and Macy piece in particular received sharp criticism in the comments on orgtheory, labeled variously “empiricism gone awry”, non-representative, and even fallacious (and […]
LikeLike
Seven-inch heels, natural language processing, and sociology « Code and Culture
November 21, 2011 at 4:46 pm
[…] Twitter is emerging as a popular source of data for scientists — see various twitter-related arXiv articles here. For example, here’s a piece validating the Dunbar number by looking at social interactions among 1.7 million people on Twitter (now published in PLoS ONE). At orgtheory.net I posted about a recently published Science piece attempting to measure aggregate mood by analyzing millions of tweets. […]
LikeLike
U got flu? Bio-surveillance, networks and twitter « StrategyProfs.net
November 28, 2011 at 6:31 am