Doing Data Journalism Badly
from the flop dept
While there was a lot of rhetoric concerning the Instagram terms of service mess recently, most observers of these kinds of things noted that for all the claims from people that they were "quitting" the service, few people ever actually follow through on such threats. So, it surprised some this morning when the NY Post reported that Instagram users actually did flee in large droves, citing data from a company called AppData. They argued that there was a potential 25% decline in users, which is huge. The news apparently contributed to Facebook's stock dropping by nearly 3%... despite the claim being almost entirely bogus.
Even a quick look at the data that AppData presented should have raised some eyebrows:
As lots of folks pointed out, starting with Zachary Seward at Quartz (linked above), the drop shown does not occur until December 24th... a good week after the whole controversy went down. Furthermore, AppData does not measure all Instagram usage at all, but the small subset of people who have connected their Instagram accounts to their Facebook accounts. But most people just use Instagram as a mobile app, and AppData doesn't measure that.
Meanwhile, Robin Wauters over at TheNextWeb does a brilliant job of showing plenty of other services that AppsData's "data" also shows "plunged" during the same time period, including Yahoo! Social bar, which shot up... and then back down:
Or Pinterest, which dropped a bunch during that same time period:
Wauters lists a bunch of other examples as well, all suggesting that maybe the original claim by AppData, repeated by the NY Post (and then others) had little basis in reality.
Even a quick look at the data that AppData presented should have raised some eyebrows:
Meanwhile, Robin Wauters over at TheNextWeb does a brilliant job of showing plenty of other services that AppsData's "data" also shows "plunged" during the same time period, including Yahoo! Social bar, which shot up... and then back down:






Reader Comments (rss)
(Flattened / Threaded)
"Doing Data Journalism Badly" -- Yes, you are.
Yes, you can rely on dolts to remain dolts no matter how obviously exploited by greedy corporations. "If you're not paying for the product, you are the product", and "products" aren't very bright. Just at the moment, I think they should be made into Soylent Green, but perhaps that will pass.
Take a moment for Mike "Streisand Effect" Masnick and click:
http://en.wikipedia.org/wiki/Streisand_effect
Actual unsolicited testimonial: "Until I read Techdirt.com, I didn't know what shameless self-promotion was!"
[ reply to this | link to this | view in chronology ]
Re: "Doing Data Journalism Badly" -- Yes, you are.
[ reply to this | link to this | view in chronology ]
I was wondering why it opened down like that. Buying opportunity and rally back to $26. Great day for $FB daytraders.
[ reply to this | link to this | view in chronology ]
[ reply to this | link to this | view in chronology ]
[ reply to this | link to this | view in chronology ]
Say like trying to find a correlation between number of post saying "I quit", "I will leave", "I am cancelling" etc to drops in network activity, download of apps etc.
AppData may not be accurate enough to give exactly coordinates but it could be good enough to give a general direction.
Saying that I agree AppData may not be trustworthy enough to make claims about how many leave or to measure market share impact accurately, but it may be good enough to spot trends.
[ reply to this | link to this | view in chronology ]
Lies, Damn Lies, and Statistics
[ reply to this | link to this | view in chronology ]
Daily Active Users
[ reply to this | link to this | view in chronology ]
Add Your Comment