ANOVA  ANalysis Of VAriance

Between Groups & Within-Groups

ANOVA Variance means “variation”
• Sum of Squares (SS) is the most common variation index
• BG & WG ANOVA • SS stands for, “Sum of squared deviations between each of a
– Partitioning Variation set of values and the mean of those values”
– “making” F
SS = ∑ (value – mean)2
– “making” effect sizes

So, Analysis Of Variance translates to “partitioning of SS”

In order to understand something about “how ANOVA works” we
need to understand how BG and WG ANOVAs partition the SS
differently and how F is constructed by each.

Variance partitioning for a BG design

Called “error”
Tx C because we can’t
account for why
20 30 the folks in a
condition -- who
10 30
were all treated
10 20 the same – have
different scores.
20 20

Mean 15 25

Variation among all

the participants – Variation among
represents variation Variation between the participants within
due to “treatment conditions – represents each condition –
effects” and “individual variation due to represents “individual
differences” “treatment effects” differences”
SSTotal = SSEffect + SSError
How a BG F is constructed How a BG r is constructed
Mean Square is the SS converted to a “mean”  dividing it
by “the number of things”
r2 = effect / (effect + error)  conceptual formula
SSTotal = SSEffect + SSError
= SSeffect / ( SSeffect + SSerror )  definitional formula
dfeffect = k - 1 = F / (F + dferror)  computational forumla
represents #
conditions in design
MSeffect SSeffect / dfeffect MSeffect SSeffect / dfeffect
F = = F = =
MSerror SSerror / dferror MSerror SSerror / dferror

dferror = ∑n - k
represents #
participants in study

An Example …

SStotal = SSeffect + SSerror

1757.574 = 605.574 + 1152.000

r2 = SSeffect / ( SSeffect + SSerror )

= 605.574 / ( 605.574 + 1152.000 ) = .34

r2 = F / (F + dferror)
= 9.462 / ( 9.462 + 18) = .34
Variance partitioning for a WG design
How a WG F is constructed
Called “error”
because we can’t Mean Square is the SS converted to a “mean”  dividing it
Tx C Sum Dif account for why folks by “the number of things”
who were in the same
20 30 50 10 two conditions -- who
were all treated the
same two ways – SSTotal = SSEffect + SSSubj + SSError
10 30 40 20
have different
10 20 30 10 difference scores.

20 20 40 0 dfeffect = k - 1
represents #
Mean 15 25 conditions in design
MSeffect SSeffect / dfeffect
Variation among Variation among F = =
participant’s difference SSerror / dferror
participants – estimable scores – represents MSerror
because “S” is a “individual differences”
composite score (sum) dferror = (k-1)*(n-1)
SSTotal = SSEffect + SSSubj + SSError represents # data
points in study

How a WG r is constructed

r2 = effect / (effect + error)  conceptual formula

= SSeffect / ( SSeffect + SSerror )  definitional formula
= F / (F + dferror)  computational forumla

MSeffect SSeffect / dfeffect

F = =
MSerror SSerror / dferror
An Example … This uses the data from
the BG design!!!!!! What happened????? Same data. Same means & Std.
Same total variance. Different F ???

BG ANOVA SSTotal = SSEffect + SSError

WG ANOVA SSTotal = SSEffect + SSSubj + SSError

SStotal = SSeffect + SSsubj + SSerror

1757.574 = 605.574 + 281.676 + 870.325 The variation that is called “error” for the BG ANOVA is divided
between “subject” and “error” variation in the WG ANOVA.
This is just an r2 = SSeffect / ( SSeffect + SSerror )
example -- = 605.574 / ( 605.574 + 281.676 ) = .68
don’t ever do Thus, the WG F is based on a smaller error term than the BG
this with real r2 = F / (F + dferror) F  and so, the WG F is generally larger than the BG F.
data !!!!!!
= 19.349 / ( 19.349 + 9) = .68

What happened????? Same data. Same means & Std.

Same total variance. Different r ???

r2 = effect / (effect + error)  conceptual formula

= SSeffect / ( SSeffect + SSerror )  definitional formula
= F / (F + dferror)  computational formula

The variation that is called “error” for the BG ANOVA is divided

between “subject” and “error” variation in the WG ANOVA.

Thus, the WG r is based on a smaller error term than the BG r

 and so, the WG r is generally larger than the BG r.