Start with the population and the intended comparison.
Define the target population, the way participants will be selected and the estimate you need. A survey of all members of a small organisation has a different design from a population survey with probability sampling.
Plan meaningful subgroups before setting a total. A study with hundreds of responses can still have too few in an important group to support a useful comparison.
Where the familiar 385 comes from.
For a simple random sample estimating a proportion in a large population, a common planning calculation uses 95% confidence, a margin of error of five percentage points and an assumed proportion of 0.5. The calculation is n = 1.96² × 0.5 × 0.5 / 0.05² ≈ 384.16, rounded up to 385.
This example has assumptions. Different precision, population size, sampling design, expected proportion or subgroup requirements change the answer. A finite-population correction may be relevant for a small known population.
An open link does not inherit those assumptions.
People responding to an open link or targeted advertisement are not automatically a simple random sample. Selection, access and willingness to participate can affect the results. Reporting 385 responses with a standard population margin of error would hide that distinction.
Record how the invitation reached people and who was likely to be missing. Weighting can address some known imbalances under assumptions; it does not make an unknown selection process disappear.
Turn the response target into a field plan.
- Estimate invitations and likely usable completions separately.
- Plan eligibility, refusals, partial responses and subgroup targets.
- Choose online, panel and fieldwork channels around access needs.
- Define when collection ends and which analyses the sample can support.