Sort radiation harm into two camps first. Deterministic effects appear once a threshold dose is crossed and get worse with more dose. Stochastic effects are random hits, mainly cancer and inherited mutations.
For stochastic effects the rule is clear. More dose means a higher chance the effect occurs, but if it does occur, its severity is fixed and not set by the dose. So the true statement is that probability of effect is a function of dose, which is option B.
Knock out the rest. Saying severity scales with dose describes deterministic damage, so that is false for stochastic. Stochastic effects are assumed to have no safe threshold under the linear no threshold model, so claiming a threshold is wrong. Skin erythema and cataract are textbook deterministic effects with thresholds, not stochastic examples.
So the correct statement about stochastic effects is that the probability, not the severity, depends on dose.
\[\boxed{\text{Probability of effect is a function of dose}}\]