Sunday, 16 January 2011

Online Dating # 2

(Again, data shamelessly ripped from OK

Started with the Question:
Which of these options would you rather be true:
people either think you are beautiful or cute
people either think you are beautiful or ugly
people either think you are cute or unattractive
people either think you are beautiful or unattractive

We voted, with unsurprising results (R asking if he could choose "unattractive to everyone" as an option aside..)

Flashed over to some data from OK Trends:

Nice scatter, posed a couple of questions - and ended up identifying a couple of points with the same attractiveness but wildly different numbers of messages received.

Invented two sets of stats about these two women, and asked kids to give me some possible ratings they could have had to give them an average rating of "7" (for our purposes, the average was the mean):

Started talking about the differences between the two sets of ratings - pleasingly, words like "range" started coming up with only minor prompting.

They decided that by and large, the first girl had a larger range (the boys also decided by and large that they preferred the first girl!)

We then talked about what we could do to examine the "spread" of our data, and I gave them some data that we constructed histograms from (always surprised how difficult our little ones find histograms!)

One of the kids piped up that the first graph didn't make sense, that she should get more messages as basically no-one finds her unattractive.
We agreed, but then again, more people thought the second girl was super hot.

We compared a few other comparable women from the site:

and saw the pattern repeated.

We formed a hypothesis: you're more attractive if men argue about you.

again, from the data set given, the kids this time found the standard deviation of each woman of a similar level of attractiveness (i.e. all rated 7), and compared this to the number of messages received per week (the thick grey line is the average messages per week.
and came to the conclusion that if everyone thinks your hot, you get less messages, if there's disagreement, you get loads.
Hypothesis confirmed!

We didn't really get time to head on to the equation that OK Trends proposed, which explicitly shows that some types of ratings detract from your message possibility, which was a pity (as i say, histograms, damn!) but a good, eye opening lesson drawn (largely) from the real world that had kids arguing, which i generally take as a good sign!

The lesson PPT and Handout can be found at the links.