Irene Hess

Stratification: A Practical Investigation

Abstract An investigation of stratification is reported for a universe with strong positive skewness. The choice of stratification variable, the method of constructing strata, the sample allocation, and the number of strata are considered. Comparisons are made among four types of allocation in combination with the corresponding optimum stratifications. For four simpler methods of stratification, equal allocation and allocation proportional to stratum aggregates are contrasted. Gains from stratification are examined for several estimation variables. Illustrations utilize data for the 1961 universe of United States nonfederal, general medical hospitals for short-term care. However, the findings apply to the stratification of other populations having similar distribution functions.

A “Replacement” Procedure for Reducing the Bias of Nonresponse

Abstract The effects of nonresponse can be reduced by simulating an increase in the number of recalls. The procedure consists in adding to the addresses of current surveys the nonresponse addresses from similar recent surveys.

On Noncoverage of Sample Dwellings

Abstract We present an investigation of the extent of noncoverage that can occur in area sample surveys of dwelling units in the United States, and of some relevant factors associated with noncoverage. We describe changes in field procedures which resulted in drastic improvements in dwelling unit coverage. Our noncoverage, formerly somewhere in the neighborhood of 10 per cent, was roughly the same magnitude as the total of all nonresponses; therefore, its reduction was important. In one year we brought our coverage rate to about 97 per cent of the number of occupied dwelling units reported by the Census Bureau. Thus noncoverage, as a source of errors of nonobservation in our surveys, has been greatly reduced in magnitude and importance. The potential bias is shown to be a function of the type of estimation used: large for simple expansions, less for means and usually even less for comparisons of subclass means. The amount of noncoverage is a function of the essential operating conditions used by a specifi...

SURVEY SAMPLING ACTIVITIES AT THE SURVEY RESEARCH CENTER

The Survey Research Centers primary concern is with the application of sample survey methods to a variety of social problems. The design, selection and servicing of a national sample of counties and households for personal interview surveys is a major sampling activity. The design of the county sample is discussed and its multipurpose uses illustrated with brief descriptions of several classes of household and nonhousehold sample surveys. Other types of sample designs developed at the Center are presented and some of the more unusual ones described.