Physics307L:People/Koch/Ideas

Notes from mid-term survey

• 9 out of 16 people responded as of Sunday night, October 7.
• All of those are liking / loving the course, and expecting grades from B- to A+ (all are hoping for A or better).

Lectures

Overall, I was surprised at the positive feedback from lectures. Nobody admitted to hating them and there some good ideas:

1. People are eager to see more relevant data analysis techniques. I.e., we are going to slow, which is probably true, but I haven't been wasting too much time either. Specific things mentioned:
   • maximum likelihood
   • least squares
2. More hands-on examples (not just definitions) in lecture....student practice if possible (it occurs to me that lecture in front of computer terminals would be more appropriate! (I don't think we can squeeze that many people into the lab, can we?)
3. A common theme is that people are happy to learn things they haven't seen before (i.e., Physics307L is their only source of data analysis / error estimation, etc.)
4. Students like the computer usage during lecture (presumably compared to all chalk board).

Labs

1. A common theme is being confused / frustrated / uncomfortable w/ data analysis, due to lack of guidance and explanation.
2. A lot of positive feedback about using wiki for notebook
   • Makes it easy to share data w/ lab partner
   • Makes it easy to collaborate w/ other students (in addition to lab partners)
   • Overall, despite some drawbacks, seems like a huge success, compared with paper notebooks
3. Some negative wiki feedback:
   • Doesn't support all needed features...such as excel-type behavior.
4. Two weeks / lab so far seems to be appropriate, and students feel like they are able to focus more on taking quality data and learning how to use the instruments. There were some mentions that two weeks may be too long for some experiments, which of course is true. I should discuss options for dealing with this: figuring out further work to do; starting the next weeks' lab; taking a break.

September

1. When lecturing about error analysis, maybe use the Merck paper as a striking example?
2. Make sure to talk about Nelder-Mead (name?) simplex optimization and show animation, because it is cool.
3. Talk about statistical error versus systematic error
4. Excellent lab notebook example: Zane
5. Wikipedia standard error of the mean: correct, but not the easiest to read.
6. MIT OpenCourseWare Study Materials (Including lectures on error analysis)
7. I am thinking I should strongly recommend people purchase Bevington
   • Amazon used available, starting at about $20.