ND OTHER EXPERIMENTAL DESIGNS

lean sguares for a wo-
one factor, and assum-
fixed, are as follows:

ce components as de-
ace component which is

s that the correct error
nate. The correct error
fects is the (C x S)/R

) compute the reguired
2C observations will be
totals by T,. and T,.,
lumn summed over sub-
ummed over columns is
ed over columns is T,...
as follows:

ES

IA

49.9 ILLUSTRATIVE EXAMPLE OF A TWO-FACTOR EXPERIMENT WITH REPEATED MEASUREMENTS ON ONE FACTOR 317

ROWS

[19.2] ralb) To —-TRČ
SIR
[19.3] Val x X Ti, SE x T;.

WITHIN SUBJECTS

R €
[19.4] vy

COLUMNS
| LA 8. T?
[199 RŽ T.. nRC
RxC
l RO € , | R , | € h T:
[s] PPD ac! TIR et TRE
SCIR

ROC n
Hh4) x x NXu- 5X xET.-I

TOTAL.

Te?

ROC n
[19.8] ba ba bi X2,; — RE

In practical computation a number of these terms can be obtained by
simple subtraction.

19.9 ILLUSTRATIVE EXAMPLE OF A TWO-FACTOR
EXPERIMENT WITH REPEATED MEASUREMENTS
ON ONE FACTOR

Table 19.5 shows illustrative data for a two-factor experiment with re-
peated measurements on one factor. Two groups of four subjects were
used. Bach subject was measured under five experimental conditions.
The totals reguired for computational purposes are also shown in this table.