How Not To Teach Statistics

I had to teach statistics to my unmotivated grade 11 students. I thought I'd try to make things interesting. Textbook questions for a unit like this are typically boring and unrealistic. I decided that as a class we would create a survey that was interesting to teens, post the survey online, have everyone share the link then analyze the results. To me this was a good way to make the material interesting and relevant and was also a way to teach some digital media skills.

We created a Google form that would house the questions. The survey answers we're automatically dumped into a spreadsheet, where in theory they could easily be analyzed. I shared the link to the survey with my students through Moodle and they could share it using their preferred methods.

The trouble with data sets is that they are rarely very clean. Many of the questions generated by students in the class used a scale (poor, good, very good, excellent). This data was clean and would provide good graphing opportunities. I also wanted students to have numerical data. How else can you find measures of central tendency or measures of spread?  We did create some questions where the answer should have been numerical (How many hours of television do you watch per week?). The trouble is that many of the survey participants either didn't enter a number or entered numbers that were unrealistic. After the first day of releasing the survey we had 16 responses, six of which were unusable. I decided to scrap the project and we headed back to the textbook. We discussed the problems with surveys so that we could all learn from the situation.

I'll have to rethink this activity for next time. Ideally, we'd have a Wii in class and could generate a ton of data for a given game: scores, time played, levels reached, points/second, player biographics, etc. We could then analyze that data. Add a Wii to the wish list.