                                                 PANEL CLARIFICATION QUESTIONS 
June 25, 2013


This afternoon, Drs. Mary Christman and Dallas Johnson, asked EPA to clarify the 1-way ANOVA and the repeated measures 2-way ANOVAs as used in the Medaka multigeneration test (MMT) and Larval amphibian growth and development assay (LAGDA).

1) In the 1-way ANOVA corrected code, does the correction for MMT:
   a) Use the mean value for all individuals in a tank and the weights proportional to sample size as was done to calculate the mean for the tank in the LAGDA? Or
   b) Use individual observations as the experimental units and add a random effect for the tank within treatment and include denominator degrees of freedom adjusted according to the Kenward-Roger (KR) (or Satterthwaite) method? or
   c) Something else?
         
2) In the repeated measures ANOVA for MMT, the fixed effects are treatment and time, BUT that alone does not account for the repeated observations on the same experimental unit.
   a) How were repeated observations handled? That is what is the covariance among the error terms?
   b) How were repeated observations handled if a non-parametric ANOVA was performed because of lack of normality?

3) If the residuals show normality, but unequal variances among treatments, was a usual ANOVA and KR method or a nonparametric ANOVA done?
      

