I have 2 to 5 subjects (different sea organisms) that will be exposed to several different treatments (varied doses of bacteria). Each group is done in triplicate due to the variability of the measuring instrument and organismal growth. For example, 3 controls of subject A, 3 of subject A with concentration X of bacterium 1, etc. The response measured will be relative fluorescence units by the treated and untreated organisms as a measure of viability and growth. These measurements will be taken at multiple time points for the same treatment, but again, in triplicate. In my mind this has the appearance of a repeated measures ANOVA, but the fact that each group/treatment is done in triplicate seems to violate what I've read about a repeated measures ANOVA.
$\begingroup$
$\endgroup$
2
-
$\begingroup$ I do not completely understand your design, but by any chance isn't mixed-design ANOVA what are you looking for? en.wikipedia.org/wiki/Mixed-design_analysis_of_variance $\endgroup$– rep_hoCommented Feb 29, 2016 at 10:32
-
1$\begingroup$ You have a mixed design. Perhaps it would be best to write this up as a linear mixed model rather than thinking about it in ANOVA terms. You have one source of variance which is not of substantive interest to you/which you regard as random, namely the differences between your replicates. You are interested in the effects of time (factor), type of organism (factor) and treatment (factor), your fixed effects. Hence a basic model: Y ~ random intercept for replicate + time x organism x treatment. You could add random slope, allowing treatment effects or the interaction to vary across replicates. $\endgroup$– BenCommented Feb 29, 2016 at 23:47
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
Partially answered in comments:
You have a mixed design. Perhaps it would be best to write this up as a linear mixed model rather than thinking about it in ANOVA terms. You have one source of variance which is not of substantive interest to you/which you regard as random, namely the differences between your replicates. You are interested in the effects of time (factor), type of organism (factor) and treatment (factor), your fixed effects. Hence a basic model: Y ~ random intercept for replicate + time x organism x treatment. You could add random slope, allowing treatment effects or the interaction to vary across replicates.
– Ben