Summary STA2020F - Week 2 (ANOVA)

, Week 2


-> SSFS2, =s
BUT


:. MS S2
=




BEFORE: We used an F-testfor HO: OP 62 =


B we used F* = S
NOW:We can use a for
F-cost Ho:M1 M2...
=

Mebecause
=


of


is
test-stat now;




*
Mstreat
I
S?
=




= treat

Ms resid sh resid



Can F*< I?
-Yes but
very unlikely
would imply that
var due to resid. error var. due to treatment

~
If F*<1, no need for F-test-> Ho
accept




Completely randomised block design



> An experimental design where the treatments are


allocated to the experimental units completely atrandom




-> One independentvariable treatment




-> each experimental is
unit allocated to
only one one


treatment
level/ group



-> experimental units are assumed to be identical except

for the applied
treatment