- Less time taken in class for data collection, more time for data analysis
- Everyone gets to make the measurement, instead of some students sitting on the sidelines
- More data possible, as we can bin the data sets for big group analysis
It's this last one that has the most impact on whether I do the data collection like this or not; if we need a lot of data, it's boring and cumbersome for each group to do it and not as easy to share as it should be within the classroom.
- The collection process has to be simple and cookbook enough for students to be able to do it independently, with essentially no equipment (simulations and videos are good)
- There's not any tinkering with the setup or designing the experiment
- It's not appropriate for introducing a completely new paradigm - not at all
- Students can't/don't easily ask questions, so you can get some junk data if you're not careful with the design and instructions
With those caveats, there are some cases in which I've found it quite useful:
Experiment 1: Period Measurement Techniques
Students did the experiment here.
The goal was to look at the effects of different measurement techniques on the period value - not just the mean, but the width of the distribution. I use this to introduce the concept of distribution width, too.
I put columns into the Google spreadsheet to calculate the period for each method and then paste the data into Excel, where I graph the normal distribution derived from each data set and graph them on common axes:
There's not much choice about this one - you need a big data set, and this is the easiest way to collect and analyze it that I've come up with.
This requires some manual cutting/pasting, etc., but the next one's nice and automatic:
Experiment 2: Resonance
I'm doing this one during the upcoming unit; I haven't done it before, but I think that it's relatively foolproof.
Students did the experiment here, which uses an applet dealing with the amplitude of a driven string. We're getting at the idea that a system will respond with a big amplitude only when it's driven at or near one of its natural frequencies (the frequencies of the allowable standing waves, in this case). This is a great intro to musical instruments: the buzzing of a mouthpiece or reed isn't really producing a single pitch, but a wide spectrum of noise, of which only the instrument's natural frequencies resonate and are heard.
This one is much cleaner with the data analysis. I used Python and GoogleCL to download the Google doc data, sort it, and graph it automatically.
The code looks like this:
And the output looks like this (fake data that I used to test it):
I do these infrequently (about once per term), but it can be a big help. I also wouldn't do them in any sort of context that required big paradigm-building, etc. They're straightforward cases where I need a lot of data, where that data can be collected over the internet or at home, and where the experiment's straightforward enough and involves an established setup or concept that I can trust them to give me accurate data.