| Conversion
of Model I Regression to Model II Regression and Application to ANCOVA
(Part I)
|
Summary
|
This paper defines model
I and model II regressions and distinguishes between them. The paper postulates
a convenient transformation from a model I design based gradient to a model
II design based gradient. Model I least square linear regression is more
commonly available in calculators and PC computer programs. The transformation
involves two auxilliary parameters. Evaluation of these parameters indicates
the degree of disparity between the two gradients. The paper
includes an example, two figures, and two tables.
|
| Conversion
of Model I Regression to Model II Regression and Application to ANCOVA
(Part II)
|
| Summary
|
This paper derives variance ratio tests
for Analyses of Covariances when the hypothesized regression equation results
from model II design. The paper considers single criterion and two criterion
cases. Two examples contrast the results with ANCOVA based on model
I regression equations. The examples illustrate need to consider predictor
normality and response-predictor variance ratios in the model II case.
The paper includes three figures and eight tables.
|