Sampling errors for fertility surveys
Kish, L., Groves, R. M., & Krotki, K. P. (1976). Sampling errors for fertility surveys.
It is intended that as a result of this investigation the data gathered on sampling errors and relationships will be available to guide the designs of similar samples in other countries and will provide the basis to suggest methods for calculation, analysis, and presentation of sampling errors from future surveys. The immediate concern is with the World Fertility Survey. The investigation itself is based on 3 fertility surveys from the countries of South Korea, Taiwan, Malaysia, Peru, and the U.S. All of these surveys were conducted prior to the beginning of the World Fertility Survey. For each variable sampling errors were computed for means or proportions based on the entire sample, for about 24 diverse subclasses (domains of study), and for differences (comparisons) between about 12 pairs of those subclasses. Thus, for each survey a total of about 1000-1600 sampling errors were computed. Each of the calculations of sampling errors included "design effects" and intraclass correlations as well as the variances and standard errors. The aim is to mostly compute and present estimates of design parameters that can be used both simply and generally for diverse multipurpose designs. Simplicity and broad utility are the goals for which some compromises in precision have to be made. Within the methodology section, there is discussion of portability, the utility of portable measures of sampling variation, the use of roh and deft for imputation, calculation of roh values, formulas, variability of computed sampling errors and the need for averaging, and strategies for sampling error computation. The other major section of the report is a summary section which focuses on the results of 8 fertility surveys from the 5 c ountries previously indicated. Much of the discussion in the general summary and in the 5 separate reports concerns the relationships of sampling errors for entire samples to those for the averages (or marginals) over subclasses. The great range in values of roh for different variables in each of the surveys is the most important result in all these data.