Introduction
CORRECTION: In Applied Statistics for Engineers and Scientists by
Petruccelli, Nandram and Chen, the study you will undertake for this
lab is called a controlled experiment. However, it is really an
observational study. In your report, tell us why this is so.
Even though the study is not a controlled experiment, the point made
about blocking is still valid, and in this lab we will use the same
terminology used in discussing controlled experiments.
In Chapter 3 of the Petruccelli, Nandram and Chen book, the basic
motivations for blocking in experimental design are discussed.
In particular, the completely randomized design is contrasted with
the randomized complete block design.
Pairing is the simplest form of blocking. In fact, it is the special
case of the randomized complete block design in which there are two
treatments.
The design of this study will be what is called a paired comparison
design.
A paired comparison design is the simplest version of a randomized
complete block design.
To demonstrate the effect of blocking (or, in this case, pairing), the
data will be analyzed in two different ways: first, as paired
comparisons (paired by student), and then as if the data resulted from
a completely randomized design (which does not use student pairing).
The results should show you that "block what you can, randomize the
rest," is good advice.