- The di¤erence between the two sums of squares above,
SSB¼SSTSSW
¼
X
i;j
ðxixÞ^2
¼
X
i
niðxixÞ^2
is called thebetween sum of squares. SSB represents the variation or dif-
ferences between the sample means, a measure very similar to the nu-
merator of a sample variance; thenivalues serve asweights.
Corresponding to the partitioning of the total sum of squares SST, there is
partitioning of the associated degrees of freedom (df). We haveðn 1 Þdegrees
of freedom associated with SST, the denominator of the variance of the com-
bined sample. SSB hasðk 1 Þdegrees of freedom, representing the di¤erences
betweenkgroups; the remaining½nk¼
P
ðni 1 Þdegrees of freedom are
associated with SSW. These results lead to the usual presentation of the
ANOVA process:
- Thewithin mean square
MSW¼
SSW
nk
¼
P
iðni^1 Þs
2
P i
ðni 1 Þ
serves as an estimate of the common variances^2 as stipulated by the one-
way ANOVAmodel. In fact, it can be seen that MSW is a natural
extension of the pooled estimatesp^2 as used in the two-samplettest; It is a
measure of the average variation within theksamples.
- Thebetween mean square
MSB¼
SSB
k 1
represents theaveragevariation (or di¤erences) between theksample
means.
- The breakdowns of the total sum of squares and its associated degrees of
freedom aredisplayedin the form of ananalysis of variance table(Table
7.6). The test statisticFfor the one-way analysis of variance above com-
pares MSB (theaveragevariation—or di¤erences—between theksample
means) and MSE (the average variation within theksamples). A value
ONE-WAY ANALYSIS OF VARIANCE 265