PG851: GLM--ANOVA

This blog is for the use of students enrolled in either section of PG851. It is intended to be a forum for asking, discussing, clarifying, and helping.

Saturday, February 02, 2008

Factorial ANOVA Homework 1.C.

I understand from the notes why cell sizes being equal is important, and I understand why orthogonality (independence, right?) of predictors is important, but I don't know what balanced cells have directly to do with orthogonality. Any clues?
-Brandon

posted by brandon @ 1:08 AM  

3 Comments:

  • At 11:28 AM, Blogger Michelle.fukumoto@gmail.com said…

    Hi Brandon,

    I think the importance of the balanced cell sizes has to do with Section 5.6.5 in our book (p. 261) that talks about unequal sample sizes and the consequences of nonorthogonality. Just found it this morning.

    michelle

     

  • At 9:50 PM, Blogger Katie Byron said…

    I believe Michelle is right. If I remember correctly, equal cell sizes are part of orthogonality.

     

  • At 1:52 PM, Blogger Mari said…

    Consider the case in which you did not have equal cell sizes. Imagine, for instance, that you wanted to determine the effects of dog breed (Siberian Husky versus Bluetick Coonhound) and dog weight (under 25 kg or over 25 kg) on ratings of fearfulness.

    If 90% of the participants in the heavy dog group were also in the Bluetick Coonhound group, you would not be able to clearly separate the effects of dog weight from dog breed. Why not?

    Consult your text, and also think this through conceptually.

     

Post a Comment

<< Home