3-level linear regression

AFNI Message Board
1

Hello,

Are there any suggestions about how to appropriately analyze our dataset? We have three conditions, which are three levels of a single factor, sentence complexity:

Condition 1: complexity = low
Condition 2: complexity = medium
Condition 3: complexity = high

What is the best way to analyze this at the group level? If we had four levels of the factor, we would use 3dANOVA2 using equally-space contrast weights (e.g., -3, -1, 1, 3). However, we have three levels, so using contrast weights of [-1, 0, 1] would not incorporate the medium condition. Suggestions?

Thanks!
William

2

William,

Try 3dLME by coding the three conditions with 3 equally-spaced numbers such as 1, 0, and -1 (or 1, 2, 3)and a random effect

-ranEff “~1”