Statistical Power Considerations When Testing Human Dynamic Response.
Abstract
The report investigates the use of the statistic (y-y ave)prime (V(hat, sup -1) (y-y ave) to test whether or not the observation y is from the same population as N normative observation vectors. Although the probability of type I error associated with the test can be set at any desired value, it is difficult to provide an assessment of the type II error probability, or equivalently of statistical power (the probability of correctly identifying dynamic response which is truly anomolous). In order to calculate power, the population from which y actually arises must be specified. Although that population will be unknown, power approximations may be obtained.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 03, 1974
- Accession Number
- ADA014710
Entities
People
- Dennis E. Smith