S.-60L
Prepared by George Bobinski, Jr,

Experimental and Ouasi-Experimental Designs for Research
by Donald T., Campbell and Julian C, Stanley, pp.l3-34

Abstract

The designs described in this section of the book are true experimental
designS which allow for the manipulation of a study"s independent variable
and the measurement of any impact on the dependent variable, The key is
that the experimenter has full control over the independent variable. These
are the most strongly recommended designs. They generally have high internal
validity and their weakness is external validity.
Summary

Three designs are proposed and explained, The first, (design 4), is

shown belows
Ro, X 0,
R O; Oy

This design is the most common of the three. It has good internal validity.
It controls for history (not intrasesšion, solutjon for this is to run simul-
taneously). Maturation and selection should be egual in each group. The
problem of instrumentation is handled by using a fixed test. If observers
are used, the same ones should be used for both experimental and control
sessions and they shouldn"'t be told which is which. NRegression should affect
both groupS egually. Selection effects are generally ruled out by random-
ization, matchine iS no sSubstitute for a randomization,

External validity of this design is weak, First there can be inter-
action effects involving X and another variable. Ex: effect may be specific
to groupS warmed up by the pretest, This is not really solvable, Inter-
action of selection and X, and Maturation can both be problems, These are
reduced by extending the boundaries in terms of settings and times. The

artificiality of the experimental setting (reactive arrangements) are also