I. Overview

II. Why we cannot get two identical groups: Selection

A. Self-assignment to group as a source of selection bias
B. Researcher assignment to group: An obvious source of selection bias
C. Arbitrary assignment to group as a source of selection bias
D. Matching: A valiant, but unsuccessful strategy for getting identical groups
1. The impossibility of perfectly matching subjects: Identical participants do not exist
2. The difficulty of matching groups on every variable: There are too many variables
3. Two difficulties with matching groups on every relevant variable
E. Problems with matching on pretest scores
1. Selection by maturation interactions: Participants growing in different ways or at different rates
2. The regression effect
3. Conclusions about matching on pretest scores
III. Problems with the pretest-post design:
A. Three reasons participants may change between pretest and posttest.
1. Maturation
2. History
3. Testing
B. How measurement changes may cause scores to change between pretest and posttest
1. Changes in how subjects are measured: Instrumentation
2. Changing the extent to which measurement is affected by random error: Regression
3. Changes in how many subjects are measured: Mortality
IV. Conclusions
A. Ruling out extraneous variables
1. Accounting for extraneous variables
2. Identifying extraneous variables
B. The relationship between internal and external validity

Key terms

Back to Chapter 9 Main Menu