ABSTRACT
Government agencies in a country need to conduct surveys on a regular basis, monthly, quarterly, or annually, to monitor the economy, business, or human resources and to assess fhe future national or regional needs. The frequency of a sample survey would depend upon its utility to the society, resources available to conduct the survey, and the amount of burden to the respondent. The National Agricultural Statistics Service of the US Department of Agriculture conducts both quarterly and annual surveys that make use of rotation sample designs. The model-based approach has been considered to account for the correlation structure among survey estimates across time. Both the least squares method and the time-series technique are applied for statistical analysis and estimation. Considering the viewpoint of Hartley, the sample unit responses across time can themselves be modeled instead of modeling the periodic estimates obtained from the use of a rotation sampling design.
