Design of the Month

Cottage pie

The following quotation is taken from thelondonpaper of 11 July 2008.

At his Soho restaurant, [Aldo Zilli] cooked up six identically made cottage pies using mince, vegetables, wine and dairy from six different stores [the chains owning these shops were named].

These were then served up, identified only by the letters A to F, and taste-tested by six volunteers. ...

The first meal [named] gets a surprisingly positive reception, with testers praising the meat as "tender and lean".

But the second serving [named] is quickly rumbled as a no-frills budget meal, ...

The volunteers lavish praise on the next pie ...

The final dish ... trumps the lot with three eager (but rather bloated) testers labelling the "flavoursome meat" as their favourite.

After putting the forks down, our six brave volunteers mark the meals out of 10.

  1. What are the treatments, and how many are there?
  2. What are the experimental units, and how many are there?
  3. What are the observational units, and how many are there?
  4. Draw the Hasse diagram for the factors on the observational units, ignoring the treatments.
  5. Hence write down the skeleton analysis of variance, showing strata, sources and degrees of freedom.
  6. What is the halo effect, and how might it have operated in this experiment?
  7. Explain two other serious flaws with the design of this experiment.
  8. Design a better experiment to achieve the apparent aims of the reported one.

Other designs

Page maintained by R. A. Bailey