Sampling for the 1947 Survey of Consumer Finances

Abstract

Abstract The designing of a sample for financial surveys requires special attention to families with higher incomes in order that adequate information may be obtained about what the “dollars will do” as well as what the heads of households will do. The chief problem posed here is the adaptation of area sampling to a procedure that is more efficient than the more usual stratified random sampling would be. The modifications introduced were essentially devices for disproportionate sampling of households classified by average rent values and estimated income levels. In determining sample size it must be recognized that the sampling errors of different items of information vary considerably from item to item. The findings from the 1947 Survey of Consumer Finances have been published in the June, July and August issues of the Federal Reserve Bulletin. The methods and procedures illustrate the application of area sampling to a practical survey problem. The sample consisted of small clusters of households within cities, towns and in the open country of 66 primary sampling areas.