See how your design choices move the sample size.

These are the exact models MethodVahti uses, running live in your browser. Move a parameter and watch three sample-size models, the stability of the synthesis, and a simulated saturation curve respond. The goal is not a number — it is understanding which of your choices the number actually depends on.

Design parameters

Heterogeneity0.35

0 = homogeneous sample · 1 = maximally varied. Usually your heterogeneity score.

Analytic depthexplanatory

Descriptive ≈ code saturation · theoretical ≈ meaning saturation.

Theme prevalence0.30

How common the target theme is. Rarer themes need more interviews.

Aim specificity0.65

A narrow, focused research aim lowers N (information power).

Theory strength0.50

A strong established framework lowers N.

Data quality0.75

Richer, higher-quality dialogue lowers N.

Evidential power0.80

How robust you want the evidence base to be.

Mixed methods (group contrast)

Three sample-size models → one synthesis

MethodVahti never gives one number alone. The shaded band is the stability range under ±0.05 perturbation of every input.

Optimal N (synthesis)

range 17–18 · information-power index 0.64

stable

model estimate synthesis stability band

Which choice does N depend on?

Each bar is how far the optimal N moves when that one parameter shifts ±0.10 — longest bar = the choice your sample size is most sensitive to.

Sensitivity sweep — N versus heterogeneity

Holding the other parameters fixed, how the synthesised N changes as the sample becomes more varied. The dot is where you are now.

Saturation simulation

A Monte-Carlo of theme discovery under your parameters: cumulative unique themes found as interviews accumulate (mean of 40 runs, shaded = spread). More heterogeneity means more latent themes, so the curve saturates later.

These are author-hypothesis models (Vahtian 2026), not externally validated. They are a transparent way to reason about trade-offs — the researcher decides N. See the open validation plan.