Monday, May 6, 2013

IPS - Day 57

Today the focus was on the bootstrapping method of resampling. I wanted students to understand fully how a bootstrap sample was created and used.

I first checked on the progress of data collection for the hand-washing study. Most groups had started to collect data and felt comfortable that they would have data to analyze for tomorrow.

I discussed the idea of bootstrapping one more time and used an applet to generate some data that we could use. The free-throw shooter applet allows you to specify how many free throw shots are taken. I set the number of shots to 10, as I wanted a small sample to illustrate bootstrapping.

The question of interest is whether or not the shooter can average 80% free-throw shooting. For our data, we had 7 makes and 3 misses. I numbered these from 1-10 and then generated two different samples drawn from these 10 shots, using replacement. In both instances, the bootstrap samples showed 80% made.

I then used the CPMP tools to generate 1,000 bootstrap samples. In this case, approximately 40% of the time, the results shows a completion rate of 80% or higher. The data suggests that an average free-throw completion rate of 80% would not be unusual if a shooter made 70% of the shots in 10 attempts.

I then asked the class to pick a data set we used (car data, chocolate data, or hand washing data) and generate three bootstrap samples from the data. There were some questions initially but students started to see the process of how a bootstrap sample is generated.

I asked them to consider the results of the samples they generated and what they mean and how they can be used. There were quite a few discussions about this. Again, students gradually started to see how these represented random outcomes that could be used to determine how likely specific results that we observe occur.

I closed with asking students to record in their notes the process of creating a bootstrap sample and how the samples are used as part of the statistical analysis process.

There was a little bit of time left, so I discussed the structure of the final exam that we will have next week. Students will be provided three scenarios and will be asked to complete an analysis and draw conclusions from the analysis.

Tomorrow the class will meet in the computer lab to crunch data they collected for the hand-washing study.

Visit the class summary for a student's perspective and to view the lesson slides.

No comments:

Post a Comment