

Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
An overview of two-way anova, a statistical method used to compare the mean of a response variable across different levels of two factor effects. The model equation, interpretation of coefficients, and the use of dummy coding, contrast coding, unweighted effect coding, and weighted effect coding. It also discusses the significance of interaction terms and the use of type i and type iii sums of squares.
Typology: Study notes
1 / 3
This page cannot be seen from the preview
Don't miss anything!


The main goal in two-way ANOVA is to compare the mean of a certain response variable across different levels of two factor effects.
The Model that we would like to fit is: Yi = B 0 + B 1 HISTi + B 2 BSEi + i (1) Where our variables for the ith^ individual: Yi = Perceived benefit of mammography. And Y¯ij = the average perceived benefit of mam- mography for history i and BSE j. Historyi =
{ 0 , N o History 1 , History BSEi =
{ 0 , N ot T aught BSE 1 , T aught BSE What does our model look like when Disease = 0? What does our model look like when Disease = 1? What does our B 0 estimate? What does our B 1 estimate? What does our B 2 estimate? What is the overall F-test testing in terms of our means? What will the individual H 0 : Bi = 0 test in terms of our means? When is it appropriate to use dummy coding, contrast coding, unweighted effect coding, and weighted effect coding?
2 Two-Way ANOVA Full Model w/ Dummy Coding Above we assumed that there was no interaction in the different effects. However as we sawin our first example that might not be a good assumption. Lets try refitting that model with an interaction term. We will be fitting the model: P Bi = B 0 + B 1 HISTi + B 2 BSEi + B 3 HISTi ∗ BSEi i (2) Using Dummy coding what will our: What does our B 0 estimate? What does our B 1 estimate? What does our B 2 estimate? What does our B 3 estimate? What will the individual H 0 : B 3 = 0 test? Lets try this in SAS: proc means;by HIST BSE; VAR PB; PROC GLM DATA=MAMMO;MODEL PB = HIST BSE HISTBSE/SOLUTION; OUTPUT OUT=OUT1 R=RY P=PY; PROC PLOT DATA=OUT1; PLOT RYPY;RUN; QUIT; Since our interaction term is insignificant I would refit this model as an additive model (nointeraction) and interpret the means separately.
How could we use the Type I and Type III sums of squares, for testing the significanceof two different research factors A (with four groups), B (with three groups) and A × B?
When would be the appropriate time to use Type I? Type III?