Response variance in double sampling for inclusion probabilities. – Godwin A. Udofia
The mathematical model developed by Hansen, Hurwitz and Bershad, for response errors under simple random sample design is extended to double sampling for inclusion probabilities. Under the double sample design and a linear response model, the total response variance is decomposed into components that reflect the different sources of response error. The effect of variation of the number of interviewers on the contribution of the simple response variances due to the various sources of response error is highlighted. A justification for proper identification of factors that involve correlated response deviations before the start of the survey and the need to determine the most efficient combination of the number of interviewers and the numbers of respondents are also given.
Keywords: Components of response variance, double sampling, inclusion probabilities, response bias, sample estimates.