Suppose the intracluster correlation is small. Which of the following statements is correct about cluster sampling efficiency for a fixed total sample size? Select the best answer.

It is preferable to sample fewer clusters with more individuals per cluster because the design effect is close to $1$.

It is preferable to sample many clusters with few individuals because the design effect grows rapidly.

The number of clusters has no effect on variance for any value of intracluster correlation.

It is always optimal to sample one large cluster if intracluster correlation is small.