Monthly Archives: November 2013

Analyzing Facebook Posts

The MIT Technology Review published an article about a Penn study analyzing Facebook posts to find correlations betweeen words/phrases and your demographic and personality profile. The actual study is available at the PLOS One Website.

There are lots of interesting correlations for posit, including ones for age predictors. Below are some keywords which are associated with some age groups.

“Words, phrases, and topics most distinguishing subjects aged 13 to 18, 19 to 22, 23 to 29, and 30 to 65. Ordered from top to bottom: 13 to 18 19 to 22 23 to 29, and 30 to 65. Words and phrases are in the center; topics, represented as the 15 most prevalent words, surround. (N~74,859; correlations adjusted for gender; Bonferroni-corrected pv0:001).

Words vs. Age Groups
13-18 19-22 23-29 30-65
  • school
  • tomorrrow
  • homework
  • English
  • bored
  • math
  • prom
  • hahaha
  • :D
  • <3
  • (:
  • semester
  • fuck/fucking
  • apartment
  • studying
  • campus
  • shit
  • roommate
  • at work
  • enjoying
  • office
  • beer
  • drinks
  • new job
  • company
  • apartment
  • daughter
  • my son
  • my kids
  • fb friends
  • husband
  • repost
  • blessed
  • children
  • prayer(s)

On the whole, I would say that the results do have a certain validity. If you’ve ever been on Facebook, I am sure you will have seen some of these words yourself for your age group. And while I don’t doubt the methodology at all, I would be handout the usual caveats for this kind of study.


Who’s in Facebook?

My first caveat is class assumption. The 19-22 word set is dominated by traditional collegiate life with the number one word being “semester” (followed by “fuck”). Other major collegiate-specific words include “campus, studying, classes” and minor words include “papers, exams, assignments, science, professor” and so forth. The list includes recreational words which could be collegiate or not (drunk, hangover), but many students also happen to drink (or talk about drinking) at college.

To me this means that this isn’t just a 19-22 year old sample, but middle class 19-22 year old sample. As many researchers such as danah boyd point out, it is important to note that not EVERYONE is in Facebook, and not everyone is in a particular social media environment.

Personality and Community?

The article also discusses correlations between personality type and word use. For instance, people who test as introvert are apparently interested in “computers” and “anime” (vs. “party” and “boys/girls” for extraverts), while those who are “neurotic” tend to use words like “depressed”, “sick of” and “fucking” (vs “success, basketball, lakers, success” for the emotionally stable).

Again, I don’t necessarily dispute the results, but I do wonder if the notion of “performance” has been taken into account. What I mean by performance is that people may write in a certain style and on certain topics in order to conform to some social norm such as what is expected of a particular gender.

To take a personal example, my Facebook network includes a lot of co-workers and family. I don’t necessarily share everything with everyone on Facebook. I watch a certain amount of manga, but I choose to not talk about it on Facebook since it’s not usually relevant to my circle. Instead, I tend to talk about the far more socially acceptable topic of pets and babies (I have a corgi and he is soooo cute!). I would be curious if my word cloud skewed towards extrovert or not.

Facebook may truly indicate personality preferences, but it is not the same thing as a personal journal.