Probit regression fits a dose-response relationship for a binary outcome and reports the effective-dose quantiles (LD50, ED50 and so on) with their CIs.
Probit is logistic's near-twin: same idea (link a binary outcome to a linear predictor), different link function (Φ⁻¹ vs logit). Historically used in dose-response bioassays where the latent tolerance distribution is assumed normal.
Effect sizes are interpretable as 'dose at which 50% (or other quantile) of subjects respond' — the LD50 / ED50 / EC50 framing standard in toxicology and pharmacology.
Dose column + binary response column.
Probit intercept + slope + dose at 5/10/50/90/95% response with CIs.
LD50 / ED50 + CI is the headline. Compare to logistic regression's similar quantile estimate; they differ by ≲5% at the median but diverge in the tails.