Sample size calculator for two sample t test

Assuming two independent groups and a Normally distributed measurement, how many observations are needed to detect the difference that you specify, at the power that you specify, with the type I error rate that you specify?

The minimum required sample size per group is:
Histograms of how the data might look (if your measurement follows a Normal distribution):
The small superimposed curves show the sampling distribution of the group means, using the recommended sample size given above. The power and type I error rate dictates how much these curves should overlap, and the sample size controls how "peaked" or "flat" they are. If you increase the SD the curves are "flatter" and therefore overlap more; N increases to compensate If you increase the difference in means, the curves are further apart; N decreases to compensate.

Please e-mail me with issues, bugs, or feature requests

Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License