Because of the complex survey design of the NLAES, variance estimation procedures that assume a simple random sample cannot be employed. Research has shown that clustering and stratification of the NLAES sample may result in standard errors much larger than those that would be obtained with a simple random sample of equal size. To take into account the NLAES sample design, all standard errors of the prevalence estimates presented here were generated using SUDAAN (Research Triangle Institute 1994), a software program that uses appropriate statistical techniques to adjust for sample design characteristics.
