Bimodal Twilight

I have not read, nor do I plan to read any of the Twilight books. I have not seen, nor do I plan to see any of the Twilight movies. If you love the books, Alex can provide you with all you could ever want on that topic. For the rest of us, I will not fan the flames by providing any snide remarks on the topic, nor poking fun at the story in any way.

My real purpose in bringing this up is to talk about normal distributions. Thrilling topic, isn't it? You're all quivering with anticipation I'm sure. (Or are you quivering because of my luminous marble-like pasty white complexion?)If we were to go measure something like . . . say . . . the mass of all your T-shirts, what would we expect to find? Logic would tell us that they'll weigh roughly the same, but there will be some variation. Some will be worn out and thin, and weigh less. Others will be a size to large, meaning more fabric (plus, still in mint condition because you don't want to wear a big ol' T-shirt) so they'll weigh more. If we were to graph the results, we would likely see a 'normal' distribution. We call this 'normal' because so much data comes out looking like this. The size of leaves on a tree. The height of 2 year-olds. The distance you can throw a baseball. Anything where there is an expected value, and then some events that fall a little beyond or a little short of that.
Now, many things are not 'normal'. If we were to expand our measurement to the weight of all your shirts, we might find it to be 'bimodal'. (Remember, the 'mode' of a data set is the single value that is most common. To be bimodal means there are two different values which are both significant maxiumums (though one will still likely be larger).) In the case of the shirts, we would expect that T-shirts are light, but that your vast collection of sweaters would introduce another graphical hump. Lots of things are also bimodal, typically things that depend on some factor with two possible outcomes that influnces the data. The weight of a group of people, for example. We would expect to see a peak where many of the women fall, and then a separate peak higher up for the men.Ok, that was all warm-up for what I really wanted to get at. What would we expect for a movie rating distribution? At first thought, maybe a normal distribution. Imagine seeing a movie with a very large group of friends and discussing it afterwards. You might expect that many people would rate it 'ok' with a few that liked it more, or a few that liked it less. If that's what actually happened, then maybe the movie you saw was 'I Am Legend'.

But when you think about it, this isn't really typical of peoples resonses. Usually a bunch of people think a movie is 'good' or 'ok' but there are a few who really, really love it. Such is the case with 'Ice Princess' which is nicely bimodal.
Or maybe you've seen a movie where many of the people thought it was ok, but some of them really liked it, while others really hated it. That might be 'Four Christmases'. (What is this? Trimodal?)But, I think it is really interesting (and likely unique) to find a movie with no middle ground. You either love it or hate it. And that, is Twilight. I think what we have here is the data from all the women who went to go see it, as well as the results from their fathers/boyfriends/husbands. ;)
All movie data sets came from imdb.com