7.1 The shape of random variation

The starting concept was to roll one, two, and then three dice mapping the results onto the white board to show the distributions. This morphed minutes before class into passing out a single die to 12 student pairs and having the students tally data into a shared spreadsheet. 

The concept worked, with the complication that different individuals, and pairs, moved at vastly different paces through the data. With some working as pairs and others as individuals, there were only nine dice being rolled. Because the spreadsheet included the distribution for 12 dice, the data was padded using three columns containing the function =int(rand()*6+1). In retrospect this felt like a kludge. 

Karen and Kimilane started as a pair, and then were each given their own column to add another student generated data column. The Smartboard proved problematic with high latency - adding a column was difficult. Perhaps this was in part that nine groups were entering data to the same sheet. In part this was also a result of the memory limits of the board computer being exceeded. 

Lyviane and Sheral are working on the other side. 

Shardae started solo, Joseph arrived later. 

Around 9:20 the class stopped rolling. A histogram for one die was generated. Then the sum column was set up to sum two columns which led to a two dice histogram. Then the sum function was adjusted to sum three columns. Finally a 12 dice sum was generated, but the number of row for which there were 12 dice was only 83. That was too small to generate a clean normal distribution in a default chart.

Bear in mind that the chart was also subject to change because three columns were being dynamically generated by the random function. This instability of the shape was suboptimal.

In theory the activity is a good activity. More dice would allow every student to roll which would generate more data. Perhaps a pre-built set of frequency functions and charts could automate the generation of histograms. The concept is structurally workable and can demonstrate the desired outcome of a normal curve. A spreadsheet with some updated structures was prepared for possible future use.