DESIGNS

In a true experiment, subjects are randomly assigned to treatments.

In a quasi-experiment, subjects are not randomly assigned, but are placed in groups by a naturally occurring phenomenon, e.g., self-selection.

Experiment allows causal conclusions that IV caused DV

Quasi-experiment allows limited causal conclusion, as it isn’t clear if IV caused DV or if other factors associated with group membership were the cause.

Pretesting is used as a stately to rule out initial differences on relevant variables.

If pretest differences exist, statistical methods are used for equating groups, but they are never certain.

If pretest differences don’t exist, we still can’t be certain other variables not assessed were the real cause.

Experiments might not be feasible.

Subjects cannot be randomly assigned

Cost

Ethics

Subjects won’t cooperate

Contamination of conditions

Design notation

X = Event or treatment O = Observation

 Group 1: X O

• One-Group Pretest-Posttest

 Group1: O X O

• Interrupted Time-Series

Group1: O O O O X O O O O

Group1: X O

Group2: O

• Pretest-Posttest Nonequivalent Control Group Design

Group1: O X O

Group2: O O

• Replicated Interrupted Time-Series Design

Group1: O O O O X O O O O

Group2: O O O O O X O O