statrefs home‎ > ‎Main‎ > ‎Methods‎ > ‎

ANOVA


General Notes
 
ANOVA, or Analysis of Variance, is closely related to Linear Regression.  Both are related to General Linear Models (GLM) and Generalized Linear Models (sometimes abbreviated as GLZ).
 
The methods most often recognized as ANOVA are "Fixed Effects" ANOVA.  Other types of ANOVA include "Random Effects" and "Mixed Effects" studies. 
 
ANOVA methods can be used to analyze different forms of experiments, such as:

    Assumptions
     
    ANOVA (Fixed Effects) generally holds the same assumptions as Linear Regression.
     
    Residuals are assumed to be iid ~N(0,s2).  (The "~" tilde symbol means "distributed as".) 
    • iid := Independent and Identically Distributed
    • ~N(0, s2) := Normally distributed with mean = 0 and constant variance across the domain
    A discussion about the assumptions behind ANOVA can be found at StatSoft.

     

    References