diet treatment model

[Rodrigo Badilla Cabrera]

Hi all
I have an experiment for growth with fishes and diet treatment´s in
differents tanks... I have differents diets (8) in 24 tanks, each diet in
triplicated, i want determinated the tanks and diet effects. I had use
genstat and spss for calculate them, but i have some comfussion how calculate
diet effect resting tank effect because i can not use tank as block for diet
treatment, i think that maybe diet.tank interaction???
I had calculate effect treatment: tank and diet, separatedly and have significants differences, but when put both effects join i don have signifficant differences in diet...
someone could me help please?
Regards

 

[Dason]

Only post a thread once please. I am deleting your other threads.

 

[Dason]

Yeah can you explain that again? It wasn't very clear to me what you're trying to say.

 

[Rodrigo Badilla Cabrera]

ok sorry, my english is not good...
I have to evaluate 8 differents diets replicated in 3 tanks each one (total= 24 tanks), i want determinated the significance of tanks and diet effects (as fixed effects), i want to know if i can subtract the tank effect to the treatment diet in a anova analysis,
i have tried with tanks nested in diet treatments, but i d like know if exist one way to substract the tank effect to diet treatament...
I had run the anova analysis with tank and diet, separatedly and have significants differences, but when run one anova with both effects in just one anova i dont found signifficant differences in diet treatments...

 

[Dason]

You can't really estimate tank effect unless you used the same tank more than once.

 

[Rodrigo Badilla Cabrera]

Hi Dason,
I did read that thank effect could give "noise" in some statistical-genetic analysis...variability of the model could be affected by tanks.
Anyway diets treatments and performance of fishes could be affected by tanks this is a fact, therefore diet treatments analysis is affected by tanks...i should use a nested or interaction analysis?
Thanks

 

[Dason]

Yes it could be affected by tanks. But there isn't enough data to estimate the tank effect. It just gets confounded with the random noise term in the anova.

 

[Rodrigo Badilla Cabrera]

ok, thank, in other way, is normal that when i calculate effect treatment: tank and diet like fixed effect, in separated anova, resulting that both effects have significants differences, but when run one anova analysis with both effects fixed in the same analysis i did not found signifficant differences in diet, its normal??..