Premium
Sums of squares and expected mean squares in SAS
Author(s) -
Driscoll Michael F.,
Borror Connie M.
Publication year - 2000
Publication title -
quality and reliability engineering international
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.913
H-Index - 62
eISSN - 1099-1638
pISSN - 0748-8017
DOI - 10.1002/1099-1638(200009/10)16:5<423::aid-qre351>3.0.co;2-w
Subject(s) - lack of fit sum of squares , least squares function approximation , explained sum of squares , variance (accounting) , generalized least squares , mathematics , non linear least squares , statistics , total least squares , linear least squares , residual sum of squares , total sum of squares , linear model , econometrics , regression analysis , accounting , estimator , business
The four different types of sums of squares available in SAS are considered, and a broad overview is given of how the similarities and dissimilarities between them depend upon the structure of the data being analyzed (for example, on the presence of empty cells). The fixed‐effect hypotheses tested by these sums of squares are discussed, as are the expected mean squares computed by SAS procedure GLM. Primary attention is given to linear models for the analysis of variance. Only two‐factor analysis of variance models are explicitly considered, since they are complex enough to illustrate the most important points. Numerical examples are included. Copyright © 2000 John Wiley & Sons, Ltd.