Snow Crash

I woke up at 12:30 PM today.

Now how did this happen? I'll explain a bit about yesterday. First, I woke up, got dressed, surfed the web a bit and then straight to school. For breakfast, I had the impossible combination of a Dunkin' Donuts sausage croissant, and Mountain Dew. This sort of "non-food" breakfast would keep me going through the test.

Took the exam. To say that the format of the exam was weird is almost a misnomer. Dr. S. tried to make this exam as much like exam P as possible. For example, you have an average of 4 1/2 minutes to complete each exam P question -- therefore the test had 10 questions in 45 minutes.

But he just---couldn't stop---tinkering with things. First, he decided he would penalize people for guessing on answers, which they don't do on Exam P. Your final score was:

(number of questions right) - 0.1*(number of questions wrong)^2

That ^2 means "squared". So if you got say, three questions wrong, that took almost a full point away from the questions you got right.

Then his instructions stated that all students must answer Questions 1 through 4, opening one's self wide open to that wrong question penalty. Which is not what they do on Exam P, not at all. During Exam P, you not only have the liberty to answer (or not answer) whatever of the 30 or so questions you get in 3 hours time, you have the liberty to guess answers as well.

Finally, you don't have to show your work on Exam P. On this exam...you did. Why, I have no idea, since in his system, a page with work shown and missed by a digit scores exactly the same as a blank page where one just guessed the wrong answer -- a penalty against the number of answers right.

So I did six out of the ten questions. One was a question on those first four that I couldn't get and just had to guess. I feel like I got two others. My time management wasn't that great -- Question 10 I might have had a stab at but I never got to it, obsessing over Question #2 which I was forced to answer. But the other five, I feel very good about. I also could have just guessed at another question, since two wrong squared times 0.1 would just be 0.4 taken off.

From gossip, one person answered seven questions and another answered eight. I don't know if they got those questions right, however -- if they did, I can rest assured that I'm not top student in the class.

I suspect that Dr. S's finally test distribution will be nothing approaching a normal distribution. It will probably be more like a uniform distribution.

For those non-actuaries reading, a "normal" distribution is the classic bell curve. There would be a lot of scores clustered around a certain number, and a few outliers on the high side and the low side. I don't know if a normal distribution of answers implies that the test is an appropriate tool, but the classic "A-B-C" distribution is basically a normal curve skewed to the right, with mostly A, B, and C grades an a few D and Fs on the left end.

A "uniform" distribution is essentially a distribution where any outcome is exactly as likely as another. The scores are all over the place, a "dartboard" distribution. If you divided the distribution into five equal pieces, you'd find the same amount of scores in each piece.

With those "crappy" test distributions, a bad teacher can just choose some arbitrary point, and claim that the people scoring above that point passed and those who didn't reach it failed.

So after that, more soda and Stats class.

Then home. We ate out. Lots more soda. Then some more soda before bed.

Then insomnia until 3:30.

The a carbohydrate/sugar/caffeine induced crash, which is why I'm writing to you at 1:15 PM having been up for only about an hour. I gotta stop doing this to myself.