VARIATIONS ON EXPERIMENTAL DESIGN
    Preexperimental Research Designs
    Validity Issues in Experimental Research

    Social researchers sometimes speak of preexperimental designs in reference to study designs that do not meet the standards of the classic experimental model.

    The one-shot case study involves a single case that is subjected to an experimental stimulus and a posttest measurement of the dependent variable.  With no  pretest and no control group, however, it's not safe to attribute changes to the stimulus.

    Even in the one-group pretest-posttest design, the lack of a control group opens study design to misinterpretation as to the impact of the stimulus.

    The static-group comparison employs a control group but no pretest.  The risk here is that the two groups may not have been comparable at the outset of the experiment.  For example, the experimental group may be less prejudiced than the control group after the stimulus has been administered, but they may have been less prejudiced before.

    A major section of this chapter deals with problems of internal invalidity.  These are several ways in which the experimental stimulus may appear to affect the dependent variable, but the changes observed are really due to some other factor.  For example, a program of conjugal visits may appear to improve morale in a prison, when the improvement was really due to the prison getting a new chef.  You'll see how the classical experimental design helps us avoid this class of problem.

    Finally, there are problems of external invalidity, meaning that an experimental stimulus may really produce an effect within the experimental setting, but it wouldn't produce the same effect in "the real world."  We'll see how experimental designs relate to this problem.