household in the United States has a known chance of being included in the 

 sample. Thus a representative and unbiased group of households is selected 

 to reflect the responses of all the nation's households. This technique also 

 permits the calculation of the amount of sampling error involved in an estimate 

 of the response of the nation's households to a particular question derived 

 from the sample. 



Three simple steps have been followed in order to obtain the sample of 

 households deemed necessary to give reliable results for the limited funds 

 available to be spent on the survey. First, the approximately 3,000 metro- 

 politan and non-metropolitan counties in the United States were classified into 

 strata representative of various types of communities from which 100 localities 

 (metropolitan areas or counties) were selected. The second step was to select 

 440 clusters from among the 100 localities. A cluster is usually a Census 

 Enumeration District, or in large cities a block or group of blocks. The final 

 step consisted of selecting the households from among a listing of all dwelling 

 units situated within the boundaries of these clusters. The sample design per- 

 mitted the listing of newly constructed as well as other units during the course 

 of the survey which were not included in the pre-listing as the survey plans 

 were being developed. The procedures at each one of the steps mentioned, 

 with all their attendant statistical refinements which for the purpose of sim- 

 plicity are omitted here, gave an unbiased sample of United States households 

 in existence at the time of the survey and permitted gjroup totals to be esti- 

 mated from the sample. The localities, clusters and households were selected 

 by random procedures according to statistical formulas rather than by personal 

 choice. 



Households have been defined for the purpose of this report as a dwelling 

 unit in which one or more persons, related or not, have their main meal at 

 home at least once a week. This definition differs from the Bureau of Census 

 definition of a household although not drastically. The number of households 

 in the United States, as defined in this survey, has been estimated from data 



