O v.x.o mxo Mx,O M4x,O

This design is reguired whenever the nature of the experimental variables
is such that the effects are enduring and the different treatments must
be applied to nonidentical content. The sampling eguivalence of the two
sets of materials is essential -- M,Mcin sampling terms, egual to the sam-
ple MbMa).

Statistical tests deal with establishing the generalization across the
sample of lists or items and then computine an experimental effects score:
for a particular person (group) and employing this as a basis for generali«
zing across persons,

10. Non-eguivalent Control Group Design
(6) X O

O O

- The design involves an experimental group and a control group both given
a pretest and a posttest, but there is no preexperimental sampling eguival-
ence between the two groups. The more similar the experimental and the con-
CO tro groups are in their recuitment, and the more this similarity is coni
firmed by the scores on the pretest, the more effective the control over
the extraneous variables becomes. As a useful adjuncet to randomization,
matching of subjects in terms of pretest scores is effective, Due to the
non-eguivalent groups, application of ANCOVA is less plausible.

11. Counterbalanced Designs

Time 1 Time 2 Time 3 Time 4
Group A X0 X20 X30 X40
B X.0 X40 X40 X30
C X,0 Xx,0 x,0 X.0
DOO Xx,O x30 X,0 Xx410

These designs involve the case where experimental control is achieved or
precision enhanced by entering all respondents (or settings) into all treat
ments. The Latin-sguare arrangement is typically employed in which treat-
ments are applied in a restrictively randomized manner in turn to the
naturally assembled groups (or individuals). |

9 12. Separate-Sample Pretest-Posttest Design

R 0 (X)
R X 0