The Influence Of The Reliability Of The Dependent Variable On Statistical Power
PROBLEM: In the planning phase of developing quality educational research there are several critical decision points. Some of these decisions are made to en sure that the research will have adequate statistical power, that is, the capability to detect meaningful differences, if they do exist. In an...
| Main Author: | |
|---|---|
| Format: | Others |
| Published: |
Scholarly Commons
1977
|
| Subjects: | |
| Online Access: | https://scholarlycommons.pacific.edu/uop_etds/3492 https://scholarlycommons.pacific.edu/cgi/viewcontent.cgi?article=4488&context=uop_etds |
| Summary: | PROBLEM: In the planning phase of developing quality educational research there are several critical decision points. Some of these decisions are made to en sure that the research will have adequate statistical power, that is, the capability to detect meaningful differences, if they do exist. In an unpublished article, K. D. Hopkins from the Laboratory of Educational Research, University of Colorado and B. R. Hopkins from the University of the Pacific commented that all theoretical and empirical options for increasing statistical power had been investigated except for the reliability of the dependent variable. In this article, they derived a method for estimating the statistical power of the analysis of variance (ANOVA) and the analysis of covariance (ANCOVA) which factored out the influence of the reliability of the dependent variable. This method involved modifying the non-centrality parameters used in estimating the power of the ANOVA and ANCOVA tests. This modification required some untested mathematical assumptions and therefore needed to be empirically verified. |
|---|