Recent developments in quality engineering methods have led to considerable interest in the analysis of variance, buiding a dispersion model, identifying important effects from replicated experiments and checking for significance by means of a half-normal plot. A methodology based on a chi-squared quantile plot is presented here for checking first the presence of heteroscedasticity, outliers and other data peculiarities, and after the estimation stage a new stepwise procedure tests for significant effects.
@article{urn:eudml:doc:40112,
title = {Quantile plots in the analysis of heteroscedastic models.},
journal = {Q\"uestii\'o},
volume = {16},
year = {1992},
pages = {135-145},
language = {en},
url = {http://dml.mathdoc.fr/item/urn:eudml:doc:40112}
}
Pepió Viñals, Montserrat; Polo Miranda, Carlos. Quantile plots in the analysis of heteroscedastic models.. Qüestiió, Tome 16 (1992) pp. 135-145. http://gdmltest.u-ga.fr/item/urn:eudml:doc:40112/