Analyzing Labor Data Via Facebook Status
from the the-power-of-public-status dept
Sean over at the 463 blog has a cool post comparing mentions of “hired” vs. “laid off” on Facebook wall postings, and noticed a bit of a trend:
Who knows if it’s really indicative of anything in terms of whether the “worst” of the recession is over, but what I found worth noting was how cool it is that we can even analyze such data. Traditionally, trying to get a sense of whether or not more folks were getting laid off than hired, you were pretty much limited to various official stats that would be released. But thanks to the fact that people now share such things via Facebook or Twitter status, you suddenly can get at least a proxy bit of data. Now, there are obvious caveats, including the fact that the population on Facebook is clearly not a representative sample of the wider population (and, many of those most impacted by layoffs are probably least likely to be on Facebook), so I wouldn’t go tossing aside national labor stats just yet — but it is a sign of the new types of data that can increasingly be built from the fact that people are now sharing status information publicly.
Filed Under: labor data, recession, social networking, status
Comments on “Analyzing Labor Data Via Facebook Status”
what about trying Fired instead of Laid Off?
Re: Re:
Or got “Obamasized”.
Re: Re: Re:
>Or got “Obamasized”.
More like Bushwacked.
Another Source
I believe Google is analyzing a lot of this data using search queries, at least internally. Hal Varian has been repeatedly quoted as discussing how search queries provide valuable macro-economic data. See, e.g., http://feeds.wired.com/~r/wired/index/~3/M4yZ_3nVw28/nep_googlenomics Eric.
Lately I’ve been wondering if this sort of thing could be part of a business model for Twitter. That of course assumes that Twitter is in a better position to mine tweet data than anyone else.
A lot of companies invest quite a bit in market research type activities. Seems that Twitter mining could offer a unique view of.. well, everything.
Re: Re:
Misleading data, though. You would have to look at the overall data for twitter users, and then adjust it to match the full norms of society. Even then, if poorer people are less likely to be online (or less likely to use twitter) you end up with a dataset that doesn’t represent the full population.
Since the internet tends to skew towards slightly younger and slightly more affluent people, the numbers would tend to be off.
Re: Re: Re:
I’m not sure I’d call the data misleading; but you’re right it’s definitely data coming from a single segment of the population which may not be representative of the whole population.
I can envision situations where you’d want to mine this segment though. Apple might be one example. Any companies targeting programmers might be another (Portions of Adobe or Microsoft for example).
Re: Re: Re:
Not only that, but people use these feeds to disseminate news. So someone reading about more people laid off in the paper may put that in their status, and since that’s in the news a lot more right now…
you are missing the point, people
to the commenters nitpicking the validity of the data, you are totally missing the point of this chart.
of course, if you are a professional researcher or an economist you are going to weigh for statistical and demographic inconsistencies, etc. you also know that there are a lot of terms possible that will have different impacts.
one of the points of this is that the ability to measure the thoughts and impulses of many millions of people is available to anyone with an Internet connections. You only need about 45 seconds on your hands (that’s the time it took me). You won’t get perfect results, but what do you expect?
and, if Joe Blow can do this now, what could those mentioned professional researchers, advertisers and economists do once versions 2, 3, 4, etc come out?
and, btw, all this said, try to tell me that this chart doesn’t paint an interesting picture.
Re: you are missing the point, people
Not really – when we start to get closer to that point, people (like the Adbusters types) will purposely jam the system, signup up thousands of accounts and stuffing in fake data to skew the results as best they can. Everyone will write “I got fired” on their facebook on Monday, and then the next monday write “I got hired” just to screw the system over.
This sort of data is remarkably unreliable, because you are depending on people to tell the truth, the same people who are using screen names, VPNs, and all sorts of other stuff. About as trustworthy as asking inmates about how to be model citizens.
Re: Re: you are missing the point, people
All polling based systems are inherently flawed by requesting data from a (hopefully) statistically significant set of potential liars. This does not make it useless, only inherently flawed.
Re: Re: Re: you are missing the point, people
Like asking a bomb maker with no fingers how to make a bomb. Not useless, but inherently flawed.
Re: you are missing the point, people
You’re right it is very interesting that a certain level of ‘market research’ capability is being made available to the general public through this site. This type of trending is likely to expand to many other sites and blog searches because it has lots of ‘wow-factor’ and is a very easy online tool to build.
But yeah, the trend lines are not indicative of any specific trend in the economy, even though they are interesting to see. If you through in the S&P short term market rally across this graph you’d see it is actually rising as the hired & fired trends are falling (mid March the rally started and continued strong through May, now is rolling off). This could tell you many companies are getting their resources in line (manpower to needs), but it could also tell you nothing at all.
RE:
A lot of companies invest quite a bit in market research type activities. Seems that Twitter mining could offer a unique view of.. well, everything