> Qualitative Analysis

Trustworthiness, Rigor & Saturation

How to argue your analysis is sound — and how much data is "enough." · 11 min

Whichever path you took, you have to convince readers the analysis is trustworthy. The criteria differ by tradition — don’t borrow the wrong vocabulary.

Two vocabularies for rigor

• Codebook / positivist: reliability and validity — including IRR (Lesson 6) and a clear, auditable procedure.
• Interpretivist / reflexive: trustworthiness — credibility, transferability, dependability, confirmability (Lincoln & Guba, 1985) — plus reflexivity and quality criteria for the method (Braun & Clarke, 2021).

Common techniques (use the ones that fit your stance):

• Audit trail — dated memos and decisions, so others can follow your reasoning (Lincoln & Guba, 1985).
• Triangulation — multiple data sources, analysts, or methods (Denzin, 2012).
• Member checking — returning findings to participants. Useful, but not a simple “validation” stamp — treat it critically (Birt et al., 2016).

How much data is “enough”?

The traditional answer is saturation — collect until new data stops yielding new codes/themes. Guest et al., 2006 found saturation within the first ~12 interviews in their study. But the concept is slippery: people mean different things by it (Saunders et al., 2018).

Add interviews below and watch new codes taper to zero — the cumulative line flattens, and saturation is flagged once two interviews in a row add nothing new. (This demo uses a larger illustrative corpus, not the three-participant running example from earlier lessons.)

Two important refinements:

• Information power (Malterud et al., 2016): the more relevant information your sample holds (narrow aim, dense dialogue, strong theory), the fewer participants you need — a more defensible logic than a fixed count.
• In reflexive TA, saturation can be the wrong frame entirely, because meaning is generated by the analyst, not “found” and exhausted (Braun & Clarke, 2021b).