James’s Musings

thoughts, photography, and geeky stuff
from an unrelentingly curious Silicon Valley entrepreneur

More Social Network Demographics

by James G. Beldock on September 9, 2008

By far the most pop­u­lar post on my blog (de­spite its length) was my post late last year about so­cial net­work de­mo­graph­ics (see “Age *Does* Matter: On the Demographics of Social Networks”). Among the phe­nom­e­na it dis­cussed was the clear­ly ex­po­nen­tial in­crease in Facebook user­ship across age groups (younger = faster). Here’s the chart:

In his post­ing A Profile of Online Profiles to­day to the ex­cel­lent bi-week­ly By the Numbers blog, Charles M. Blow, the New York Times’s Visual Op-Ed colum­nist, post­ed a chart that looks star­tling­ly sim­i­lar:

source:  A Profile of Online Profiles Charles M. Blow, New York Times

source: "A Profile of Online Profiles" Charles M. Blow, New York Times

How sim­i­lar are the­se curves? Pretty damn sim­i­lar! That’s re­mark­able, given that I was orig­i­nal­ly build­ing my data from stu­dents who grad­u­at­ed col­lege (or at least vol­un­tar­i­ly as­so­ci­at­ed them­selves with a col­lege class), and the Rapleaf data on which Blow’s chart is drawn comes from the broad­er pop­u­la­tion. [Thanks to my friend Michel Floyd for point­ing out that I had pre­vi­ous­ly glossed over this sign­f­i­cant method­olog­i­cal in­con­sis­ten­cy.] Nevertheless, the sim­i­lar­i­ty is striking—and thanks to a lit­tle quick-and-dirty over­lay­ment:

Comparing NY Times/Rapleaf Data to My Data

Comparing NY Times/Rapleaf Data to My Data

They’re both clear­ly ex­po­nen­tial. But there is a slight dif­fer­ence in the “mid­dle” of the curves. It’s pos­si­ble that the “re­al” data is a bit steep­er on the tails (lep­tokur­tic), that the de­mo­graph­ics have shift­ed in the past few months, but in all like­li­hood the rea­son for the slight dis­crep­an­cy in the mid­dle is that my data were from a dif­fer­ent sam­ple (col­lege grad­u­ates ver­sus gen­er­al pop­u­la­tion) and prone to cer­tain sys­tem­at­ic er­rors that RapLeaf’s aren’t. (For de­tails, see the method­olog­i­cal notes to my pri­or post.)

Blow’s data, by the way, are based on data pub­lished this sum­mer by my friend Auren Hoffman‘s com­pa­ny, RapLeaf, by the way. RapLeaf has a spec­tac­u­lar­ly rich data­base on so­cial net­work pro­files and has the unique abil­i­ty to track users from site to site.

Be Sociable, Share!
Michel Floyd September 10, 2008 at 8:36 am

Hey James, you’ve got a bit of an apples v oranges comparison here! Blow’s data is based on total users by age whereas yours is based on % of graduating class. So to compare effectively you at least need to know the distribution of the population by age (census has this). Secondly it’s not clear if your data is based only on college graduates, which are a small subsection of the population and which also have higher incomes and higher rates of internet connectivity.

Still, reversing Blow’s chart and overlaying it is pretty cute – I remember college profs doing this with viewfoils (which puts me in the demographic with only a 7% probability of being in a social network 😉

James Beldock September 10, 2008 at 12:51 pm

Michel is absolutely right: there is methodological inconsistency, and I’m going to make that clear in the post. However, I *did* correct for age (i.e., converted age of user to likely age at college graduation).

Social Bookmarks October 7, 2008 at 9:36 am

I don\’t normally leave comments… but I really enjoyed your post! I will be leaving a link back here in my blogroll! Thanks!

Comments on this entry are closed.

Previous post:

Next post: