I have a question regarding an area that is often skimmed over in sample size calculations. How do sample units relate to replicates in the calculation of the sample size estimates.

For example, say I designed an experiment looking at how a particular treatment may impact fish. I have a control group with 5 independent tanks each containing 30 fish, and an equal amount of tanks and fish for the treatment group. Is my total sample size = 10 (number of tanks) or 300 (number of fish)? If the former, how do I incorporate the number of fish per tank into a sample size calculations?

  • $\begingroup$ Sorry, can you explain what you mean by dependent and independent here? As for sample units, are you talking in the sense of "primary sampling units" or clusters? $\endgroup$ – Marquis de Carabas Dec 23 '16 at 3:44
  • 1
    $\begingroup$ Hi, sorry I realised my question didn't make sense and edited. I think I now know the answer. Sample units (fish) are the number of observations to be used in sample size analysis, and dependency or clustering within replicates needs to be counted for using something like a mixed-effects model? $\endgroup$ – Tom Dec 24 '16 at 4:08

Your sample size is likely the number of tanks, not the number of fish. You can nest fish ID within tank in your statistical models to account for this.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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