Step 1: Re-state what snowball does well.
It is designed to locate rare/hidden populations efficiently by asking respondents to recruit peers (chain referral). This does increase the likelihood of finding the target traits ⇒ (c) is true.
A key use-case is estimating features of rare groups ⇒ (b) is true.
Step 2: Practical consequences.
Chains can grow large and geographically dispersed, raising fieldwork complexity, time, and cost—especially as waves expand ⇒ (a) is true.
Step 3: Why (d) is the exception (false).
Snowball is a nonprobability method with unknown selection probabilities; “sampling variance” in the probability-sampling sense is not well-defined, and realized variability can be high due to homophily/clustered networks. Costs are not necessarily “relatively low”; they may increase with recruitment waves and travel/logistics. Hence (d) is false.
\[
\boxed{(d)}
\]