SISL > 2012 Sustainability in Math Workshop > Activities > Solar panel statistical tests

Solar panel statistical tests

Summary

In this activity, students will use t-tests and Anova tests to determine whether there is a statistically significant difference in the number of watts of power produced on individual solar panels situated at Bryn Mawr College in Bryn Mawr, PA.

A data set is provided, though students will be encouraged to collect their own data at the Bryn Mawr website.

Learning Goals

It will build student skill in analyzing data in mechanisms designed to benefit society.

Students will gain practice in using statistics to demonstrate effectiveness.

Students will be encouraged to design an experiment to compare power from solar panel configurations.

Context for Use

This activity is appropriate for an individual or small group HW assignment in an introductory statistics class that covers t-tests and ANOVA tests.

Description and Teaching Materials

The downloadable Word document contains an introduction, a data set, a link to a web page where the students can find additional data, and a set of homework questions. Solar panel statistical tests (Microsoft Word 2007 (.docx) 21kB Mar16 13)

Teaching Notes and Tips

- The data set given here is fairly small. If the students are given time, they can go the Bryn Mawr website to collect their own larger set of data or augment the one given. - In order to use an ANOVA test, the underlying data should be somewhat normal. That will not happen if cloudy days, when little or no solar power is produced, are included. (This is why data from March 12 was not included and is why the times were chosen when power was at its maximum for the day.) Although the ANOVA test is fairly robust, these issues should be discussed. - For question 6, students should use a two-sample t-test. As before, a larger data set would allow a more powerful test. - Question 7 is not trivial. One possibility would be to collect data at, say, n=20 times for, say, k=5 different positions of the panels at Top of Pole. For each of the five positions, calculate the difference in the mean panel output between the 2*20=40 readings from the Top of Pole panels and the 12*20 = 240 Ground Mount panels. One will then have five samples of data, each with 20 differences. One could use an ANOVA test to determine whether the difference in means of those five samples is statistically significant. Of course, one would need to be on site to actually execute this experiment, which leads us to the next point. - It will be much more interesting to the students if they can use data from your school or community! Determine whether such data is available and use it instead of what is given here.