Ettina wrote:
Teacher introduces us to SPSS - a statistical analysis program for social scientists.
NT reaction: Oh, I guess I need to learn this to pass the class. Crap. Well, I'll muddle through it.
My reaction: Where have you been all my life? (Proceeds to find every chunk of data I can possibly find to analyze, including the surveys I ran years ago and never properly analyzed.)
I like probability, too. When I was studying C++ about ten years ago, that's what really made the math dance, because you literally did not know the answer ahead of time. So, I'd loop through it multiple times using random numbers and keep track of the results. Then I'd write a program to solve the problem algebraically (this typically was harder). And if the two results were largely the same, I felt pretty good about it. A simple problem might be this: If a basketball player shoots 47% from the field, what's the chance of him having a relative dry streak where he makes 0, 1, or 2 out of the next ten shots?
I've thought about branching out to statistics involving medical research. Okay, we're keeping track of some outcome and we have an experiment group and a control group. Now, there's going to be some variation within the control group even if we don't do anything. So, we measure that variation within the control group, then we extend the tails of the graph, what it might look like if we had a larger sample for a control group. So, if it's a skewed bell graph, it's still a skewed bell graph just with the tails extended. And if it's a nice, smooth, 'normal' bell-shaped graph, it's still a nice, smooth, 'normal' bell-shaped graph just with the tails extended. Then we duplicate the experiment, Oh, say, 100,000 times using random numbers (technically, they're pseudorandom numbers) and we see how often we get the difference we observed between the experimental and control groups.