Chapter Summary

This chapter covered true experiments (sometimes called randomized experiments or randomized control trials), other randomized designs, and quasi-experiments. Pre-experiments were mentioned for comparison purposes only.

Five questions that companies should answer before beginning a business experiment are as follows: (1) Does the experiment have a clear purpose? (2) Have stakeholders committed to abide by the results? (3) Is the experiment doable? (4) How can we ensure reliable results? (5) Have we gotten the most value out of the experiment?

The necessary requirements for experiments to make claims of causality are (a) manipulation, (b) control, and (c) random assignment.

Four essential criteria for establishing causality in the experimental context are as follows: (1) the cause (IV) and effect (DV) are related; (2) the cause precedes the effect (i.e., a time order must be observed); (3) the cause and effect occur together consistently (i.e., cause and effect should covary); and (4) alternative explanations can be ruled out: the relationship between the IV and DV must not be due to a confounding extraneous “third” variable.

Threats to validity reduce drawing sound inferences. This chapter reviewed four general categories of validity, provided an internal validity exhibit to help you evaluate your design, and discussed how external validity affects the generalizability of findings.

True experiments with examples, the need for control groups, and when not to use true experiments were reviewed. Two other categories, other randomized designs (e.g., factorial and covariance analysis/blocking) and quasi-experiments, were also discussed.

Different matching procedures create equivalent or balanced groups for quasi-experiments. Some of the strategies that attempt to approximate the feature of random assignment were discussed.