Sunday, October 16, 2011


The statistics nerds over at the dating site Okcupid run a blog called Oktrends where they crunch their user data into really interesting posts. For example: Using a physical compliment in you frist message to someone brings down the odds that they will reply. In this case, using the word "pretty" as an adjective makes it a physical compliment, but as an adverb it works as just another word.

Graph shows reply rate of messsages containin the listed
keywords, plotted against the average reply rate of 32%
(This image belongs to OKcupid)

Pretty cool! Earlier in the year, Okcupid was updating Oktrends every few months, now however their last update has not been since April. (If by a miracle any of the Okcupid researchers out there read this, please start updating Oktrends again! Also I love you.)

Of course these results are only true for people who use Okcupid. Therefore we can only say that on average there are more twenty year old female Okcupid users that enjoy rough sex than any other age group. It is also important to not mix correlation and causation, but there's a whole book for that.

The fact that they do this makes me glad that Okcupid is somewhat transparent about what they are doing with user's personal information. Imagine the kind of correlations Facebook must discover (and not share with their users)? I really cannot suggest a single article, because they are all fascinating. If you haven't already, go and check it out. And let me know if you know of any other sites out there like this.


  1. Ok, it looks like i will have to start reading the links you posted, because i have no idea what OKcupid is. I'm guessing it's an online dating site.
    Also it's good to know that girls still love rough sex hahaha

  2. The guys over at OK Cupid are so awesome, i love statistics so I drool over all their reports.

  3. Hah, "your pretty". My pretty what?

  4. stadistics are great to optimize your webpages.

  5. Technology is really going to revolutionize some of the ways we view psychology. Should have saw these time of large user base "social data collections" coming. The web is rich with data. Science loves data.

    It reminds me of the World of Warcraft in game disease that the CDC got interested in because it was essentially a virtual model of the way a disease might spread in real life.