Creo que es importante separar claramente la hipótesis y su prueba correspondiente. Para lo siguiente, supongo un diseño equilibrado entre sujetos CRF- (tamaños de celda iguales, notación de Kirk: diseño factorial completamente aleatorio).pq
es la observación i en el tratamiento j del factor A y el tratamiento k del factor B con 1 ≤ i ≤ n , 1 ≤ j ≤ p y 1 ≤ k ≤ q . El modelo es Y i j k = μ j k + ϵ i ( j k ) ,YijkijAkB1≤i≤n1≤j≤p1≤k≤qYijk=μjk+ϵi(jk),ϵi(jk)∼N(0,σ2ϵ)
Diseño:
B 1 ... B k ... B q A 1 μ 11 ... μ 1 k ... μ 1 q μ 1. ... ... ... ... ... ... ... A j μ j 1 ... μ j k ... μ j q μ j . ... ... ... ... ... ... ... A p μ p 1 ... μ A1…Aj…Ap B1μ11…μj1…μp1μ.1…………………Bkμ1k…μjk…μpkμ.k…………………Bqμ1q…μjq…μpqμ.q μ1.…μj.…μp.μ
es el valor esperado en la celda j k , ϵ i ( j k ) es el error asociado con la medición de la persona i en esa celda. Lanotación ( ) indica que los índices j k son fijos para cualquier persona i porque esa persona se observa en una sola condición. Algunas definiciones para los efectos:μjkjkϵi(jk)i()jki
μj.=1q∑qk=1μjk (average expected value for treatment j of factor A)
μ.k=1p∑pj=1μjk (average expected value for treatment k of factor B)
αj=μj.−μ (effect of treatment j of factor A, ∑pj=1αj=0)
βk=μ.k−μ (effect of treatment k of factor B, ∑qk=1βk=0)
(αβ)jk=μjk−(μ+αj+βk)=μjk−μj.−μ.k+μ
(interaction effect for the combination of treatment j of factor A with treatment k of factor B, ∑pj=1(αβ)jk=0∧∑qk=1(αβ)jk=0)
α(k)j=μjk−μ.k
(conditional main effect for treatment j of factor A within fixed treatment k of factor B, ∑pj=1α(k)j=0∧1q∑qk=1α(k)j=αj∀j,k)
β(j)k=μjk−μj.
(conditional main effect for treatment k of factor B within fixed treatment j of factor A, ∑qk=1β(j)k=0∧1p∑pj=1β(j)k=βk∀j,k)
With these definitions, the model can also be written as:
Yijk=μ+αj+βk+(αβ)jk+ϵi(jk)
This allows us to express the null hypothesis of no interaction in several equivalent ways:
H0I:∑j∑k(αβ)2jk=0
(all individual interaction terms are 0, such that μjk=μ+αj+βk∀j,k. This means that treatment effects of both factors - as defined above - are additive everywhere.)
H0I:α(k)j−α(k′)j=0∀j∧∀k,k′(k≠k′)
(all conditional main effects for any treatment j of factor A are the same, and therefore equal αj. This is essentially Dason's answer.)
H0I:β(j)k−β(j′)k=0∀j,j′∧∀k(j≠j′)
(all conditional main effects for any treatment k of factor B are the same, and therefore equal βk.)
H0I: In a diagramm which shows the expected values μjk with the levels of factor A on the x-axis and the levels of factor B drawn as separate lines, the q different lines are parallel.
H_0 = \mu_{A1}=\mu_{A2}
\mu_{A_1}