Wednesday, May 09, 2007

The Joys of Grading

Who knew that grading finals would be this much fun? And by 'this much fun' I mean really boring and frustrating.

My two final questions on the exam were meant to be free points. One asking about what statistics is, and the other about interpreting a simple confidence interval. They are messing these easy questions up too much for comfort, especially the confidence interval one. He is the question:

[6pts] Say Manhattan is debating a law about banning transfat in restaurants. People that are for the law say that in a study rats lived 0.42 years longer on average when they did not have transfat in their diet. Since you know a little bit more about statistics than the common folk, you take a look at the research and you find that the research does indeed say that rats lived 0.42 years longer when they didn't have transfat in their diet with a confidence interval of [-1.03, 2.52]. What would be your response to the people that want to ban transfat based on this study? (Explain why you say this.)

More than half the people say that the ban is a good idea because the average, 0.42, is inside the confidence interval. That is *so* not the right answer. No wonder policymakers are taking non-significant studies and turning them into ill conceived laws. Arg! After two semesters of statistics even my students still don't get the simple things.

(The answer I was looking for is that this study does not provide evidence that a transfat diet reduces the lifespan of mice since the confidence interval covers the number zero. Essentially the study was inconclusive.)


Braveharte said...

Are those real numbers? Just curious because that's a really wide confidence interval for an animal that lives 2-3 years. :) By the way I totally would have gotten it right. My students have trouble with confidence intervals too. I don't know why but they are harder for them to understand. I don't even have them interpret them. I just have them identify what they are and they usually fail that.

Ookami Snow said...

No those aren't real. I started writing the question about smoking bans but I changed it because I didn't want to be too topical.

Also the smoking data deals with odd ratios and the test already had a question about them.

So instead I made up a simple confidence interval to interpret. And geez if mice live 2-3 years then this study really doesn't tell us much.

Tuba Queen said...

I would have totaly agrivated you. Probably would have said something like.......I lost a total of 20 pounds since February!!!!!!