PHD Discussions Logo

Ask, Learn and Accelerate in your PhD Research

Question Icon Post Your Answer

Question Icon

How can I determine if my observational or experimental sample is large enough to test a theory?

I'm planning an observational campaign to test a new model of galaxy morphology. Telescope time is highly competitive, so I need to justify my request for a specific number of targets. Beyond simple "the more, the better," what statistical framework should I use to calculate the minimum sample needed to reliably confirm or reject a predicted trend?

 

All Answers (1 Answers In All)

By Nitin Answered 3 months ago

This is a critical step in proposal writing. I always start with a power analysis. You need to estimate the expected effect size from your theory, choose a desired statistical power (typically 0.8 or 80%), and set your significance level (e.g., p < 0.05). Software like G*Power can then calculate the required N. Furthermore, I would recommend conducting a bootstrap analysis on any existing pilot data to see how your key metric stabilizes with increasing sample size. This dual approach theoretical power and empirical convergence makes a compelling case to time allocation committees that you're not asking for too much or, critically, too little.

 

Your Answer