Edgington, 1987 – Randomized Single-Subject Experiments and Statistical Tests

Question 1

Single-subject experiments important for studying…

Answer 1

• unusual types of people, e.g., multiple personality / DID
• studying several people over the long-term is difficult and expensive
• explain processes within individuals (studies with multiple participants lose individual data in favor of data averaged across the group)
• detect atypical responses
• needs and abilities can be precisely determined relative to an individual’s rights and freedom – diversity perspective
• how to adapt services to a particular client

Question 2

Characteristics of nonrandomized single-subject studies

Answer 2

• non-experimental; no variable manipulated (i.e., effects of divorce or bereavement) – manipulation could also be cost-prohibitive, impractical, or too artificial to simulate real life
• experiments should be used when feasible
• N=1 tend not to be randomized over time – adjustment of IV in behavior therapy from moment to moment; researchers’ ignorance of importance of randomization; concern for the welfare of the subject

Question 3

Advantages of Randomized single-subject designs

Answer 3

1. avoid danger of capitalizing on response trends that might have been there even with no tx
2. provides a basis for statistical test of tx effects
3. controls for bias
4. essential for being able to statistically test for tx effects in experiments and draw inferences

Question 4

Considerations for stopping tx in single-subject designs?

Answer 4

should be determined in advance or as determined by someone other than the research (so that a researcher does not stop tx due to knowledge of effects)