A footnote on sample size

Reviewer asked why we stopped at 28 interviews. The honest answer is “funding,” but the defensible answer is saturation, so I went back and actually checked rather than asserting it.

Counting new codes per interview, in order:

• Interviews 1–8: 41 new codes
• Interviews 9–16: 12 new codes
• Interviews 17–24: 3 new codes
• Interviews 25–28: 0 new codes

That is a clean curve, and it does justify stopping. But it is worth being clear about what it justifies. Saturation says my coding scheme stopped growing. It does not say I have seen the full range of experiences. Those two things come apart badly whenever the sampling frame is narrow — and mine was, by design.

The reviewer is right to ask. The answer is: 28 is enough for the claim I am making, and would not be enough for the claim they might have expected me to make. I should have written that in the first place.