the design of experiments game

I designed this recitation for the design of experiments module of IE 4516 Quality Assurance, and Prof. Hugh McManus reviewed it. It runs in one session, twenty slides, and it is the thing students remember.

• The Destruction of Reality Itself, the full deck

The premise

The Master wants to know which Doctor Who villains generate the most chaos, measured by a Chaotic Index of financial, life, and operational disruption.

Part one is a one-factor ANOVA across six villains in Excel. One response, one factor, six levels, alpha at 0.05. Part two escalates to simultaneous attacks, and students run a factorial design in Minitab, test interactions to three orders, check whether the residuals are normal, and recommend which villains the Master should hire and at what level of engagement.

Why it works

Nobody arrives with a prior about whether a Dalek outperforms a Cyberman. That is the point. In a quality course most examples come with an intuition attached, and students reach the right answer without the design. Here they have nothing to reason from except the experiment, so they have to argue from the design or not at all.

The deliverable is also split the way the learning is. Basic credit is the analysis. Extra credit is the recommendation with the trade-offs argued, which is the part that is actually hard.

I have been building games like this since my undergraduate final project, which was a multidisciplinary game built around a trebuchet. The other ones are on the games page.