A survey organization uses cluster sampling, dividing a region into 400 geographic clusters of 30 households each. The intraclass correlation coefficient is ρ = 0.12. They sample 20 clusters and survey all households within each. What is the design effect compared to simple random sampling?

$DE = 1 + (m-1)\rho \quad \text{where } m = \text{cluster size}$