OkTrends - The Math of Dating

OkTrends is a blog run by the folks at the matchmaking website OkCupid. Their website generates huge volumes of data on personalities, attraction, and sex and they've been crunching those numbers to see what interesting relationships pop out.

For example, in this graph Christian Rudder graphs how attractive people appear in their photos based on the time of day the photo was taken:

From OkTrends - Don't Be Ugly By Accident

In the graph, the blue line is the average attractiveness of photos grouped by time of day. Superimposed on that trend is another graph in yellow and brown of the height of the sun during the day and night. When the sun crosses the horizon that's sunrise and sunset. Mr Rudder is able to show evidence of Golden Hour; the hour after sunrise and before sunset when vivid colors fill the sky and make pictures more interesting.

There's tons of SL Math in this blog. Mr Rudder fits a sinusoidal curve to the attractiveness data in an attempt to show that photos taken in the afternoon and late at night are often most appealing.

From OkTrends - Don't Be Ugly By Accident

I'm planning to add this graph to my unit on sine and cosine. I imagine the discussion will be memorable

• What does the trend suggest about good photos? 
• Does the trend appear to be a good fit of the data?
• What is the period of the graph? Would it have been reasonable to expect the period to be 24 hours?
• If you were to perform the same study from scratch, how would you do it?

Finally, here's a graph comparing the number of (self-reported) sexual partners straight men and women have had to gay men and women:

From OkTrends: Gay Sex vs. Straight Sex

After thinking, "Wow, those trends are very similar," my next thought was, "Is that an Ogive?!" It seems like the only place I ever see Cumulative Frequency Charts is in IB exams but here's one in the wild!

Some questions I might ask students are:

• Where do the two trends differ the most? What does that mean about the similarities and differences between the sexual habits of the gay and straight communities?
• How many partners has the median straight/gay person had?
• If this graph were to be turned into a frequency distribution, what shape would it have?
• Followed up with: All of this data is self-reported. If respondents were to lie, what kinds of lies might they tell and how would that effect the shape of the expected distribution?

Some of the topics in the blog are a bit too racy for my classroom, but there are still a lot of useful gems. Honestly the whole thing is just fascinating.