Analysis of Variance

Analysis of Variance - An ANOVA is a tabular presentation of the Sum of Squares (SS;variance) attributed to the model, the Sum of Squares attributed to error (Pure error and Lack of Fit error). and the Total Sum of Squares from the data. F-statistics on the significance of the regression model are included.

Two models are generally used:

1. Classification model, OR

2. Analysis model.

Classification models are generally restricted to a one- or two-way classification of complete and balanced data. A one-way classification may be based on a single factor or may be based on a main factor and a blocking factor, where the blocking factor is treated as a random factor. A two-way classification is based on two fixed effect main factors.

Blocking factors may be treated as random factors. An ANOVA for Blocked or Nested designs requires a different partitioning of sum of squares and degrees of freedom.

The descriptions "Factorial Analysis of Variance" or "Regression Variance" are sometimes used for the sum of squares attributed to a parameter in a regression. While the SUM OF SQUARES may be the same, the Degrees of freedom for an ANOVA are different from those associated with a regression parameter.

Learn more about the Regression tools in Six Sigma Demystified (2011, McGraw-Hill) by Paul Keller, in his online Regression short course (only $99), or his online Black Belt certification training course ($875).