1
$\begingroup$

My experiment has the response variable F the experimental unit is a cage the Treatment is rear (1/2 get standard rear & 1/2 get non-standard rear) and there are multiple measures of F over 12 days

making this more complicated is that there what I think are blocks? the experimental unit is a cage into which 12 groups (6 of each rearing treatment) are placed there are large cages and small cages stacked on top of each other on three different levels (top, middle, bottom) each tier has four cages in 2 of each treatment type (I think this means there are 2 replications) I think? it is a complete randomized block design but am not sure of even that...

Can you already tell I am at sea here on what the design is and how to analyze it... I have great software - and can make both R & Sigmaplot do various analyses... BUT I have no idea what to ask them to analyze!

I am imagining that I need to test for normality, but how do I do that when my DV has several columns for several days?

And once I have done normality tests, what do I look for?

Any help greatly appreciated - I have read multiple book chapters and after 12 hours of work nothing is any clearer and I am completely stumped and frustrated. If you can provide any help I would be eternally grateful!

$\endgroup$
1
  • 1
    $\begingroup$ Welcome to StackExchange ! You question would get clearer and more answer if you add a quick schema of your experiment. Beside, what did you had in mind when creating the design ? That would answer the questions about what to analyze. $\endgroup$
    – Emilie
    Oct 20, 2014 at 18:11

1 Answer 1

1
$\begingroup$

The normality assumption is tested on the level of a population to which the treatment is applied. Since no statistical test can mitigate the problem of uncertain design, it is important to narrow down what the question of interest was when the experiment was set up. Beyond the effect of the treatment, why was the repeated measures component of interest, ie why were cages sampled several times? This will help you decide whether you need to do repeated-measures analysis, but, as is, you do not provide sufficient information to discuss this part.

So, the design is blocked by level, and, assuming all assumptions are met (normality is not the only assumption to test for), you can do a one-way blocked ANOVA. And here's on rm blocked design Using lme to analyse a complete randomized block design with repeated measures: Is my model correct?

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.