How many responses do you need for a credible result — or, how much can you trust the responses you already have? For SoTL studies and unit evaluations. Nothing stored.
How many do I need?
What's my margin of error?
How many people could respond (e.g. students in the cohort)?
—
responses needed
Rule of thumb, not a rule. This uses the standard formula (50% response distribution = the most conservative case). It sizes for a *representative* sample — it doesn't fix a biased one. For sub-group comparisons you'll need more in each group. Pairs with the Survey & Likert Checker.
A free tool from NTLSN.com · Cochran's formula with finite-population correction · nothing leaves your browser.