Non-representative sample of Hunger Games responses

I had the idea for the Hunger Games survival analysis post Tuesday afternoon and published it about 24 hours later (and yes, in the meantime I did sleep, eat, and do a bit of real work as well). I thought it might hit a nerdy nerve by meshing pop culture and stats, and I was right. Three days later it's been read by over 12,000 people on my site alone, and the average time on page is long enough that I think folks are actually reading it and not just looking at the pretty pictures. It was picked up by Andrew Gelman and Jezebel (a Venn diagram with only this in the middle, I bet) and everyone from Stata to Discover Magazine shared it on Twitter.

All that to say, I think there's a market for explaining statistics and concepts from social science (I tried to work in some political science, economics, and psychology research) using pop culture tie-ins, so I may do some more of this.

For now I want to share some of the humorous reactions I've seen:

  • A classmates who is familiar with survival analysis but hasn't read the books saw the graphs and her immediate response was "Oh no, what happened on the first day? Those poor children!"
  • Richard Williams, commenting on the Stata listserv discussion: "If Stata can win over the Hunger Games crowd, SAS & SPSS are finished."
  • One of the comments on Metafilter kind of misses the point: "I love statistics, but come on: The major finding here is that Suzanne Collins did a good job creating a fictional dataset that shows some significant differences between groups. Yes, that's because statistics measures deviations from randomness, and Collins *made up the data* as part of her novel's plot." Shocking.
  • A friend to a friend of mine on Gchat: "he used Stata for a Good Thing. It was Interesting. That's im-[ahem]-Possible and he Did It.
  • Finally, the Teaching Assistant from last semester's generalized linear models class threatened to grade it as an assignment. Next time I'll use data where the assumption of the model (proportional hazards) aren't clearly violated...