Hello everybody,
I am investigating the influence of a toxinon various clinical-chemical and haematological parameters in fish.
I have three treatments (control, low dose, high dose) in my experiment and 4 replicates (n=4). So a total of 12 tanks. Now, the fish from one tank are counting as dependent samples and therefore actually pseudo-replicates. Does it still make sense to sample more fish per tank? Do I have to pool the resulting data afterwards, as I statistically wouldnt be "allowed" to evaluate the fish from one tank on a per-fish basis? If we now assume that I sample 3 fish per tank, i.e. 12 per treatment (3x 4 replicates) and I have no significant tank effects, could I then increase my "n" to 12?
Regards
Finn
Designed as a two way analysis of variance (treatment x tank with three replicates per tank) and the analysis showed no significant tank effect, you are effectively getting the same experiment as 12 replicates/treatment in a one way AOV..
- Badges
- Science topic
- Similar topics
- Clinical Pharmacology
- Dosing