APPENDIX 



n>u 



he Sampling Plan and Selection Procedures 



The General Plan . The population under study in this 

 survey consisted of all private households in the United States. 

 To represent' this population, a multi-stage national probability 

 sample of those households was designed. This sample had to 

 meet the following requirements? 



1. Every household in the population must have a 

 known chance to be included in the sample. 



2. Estimates from the sample must be unbiased, 

 with calculable error margins. 



3. The tolerances of estimates produced by the 

 sample must be comparatively low so that the 

 estimates could be used with confidence. 



To meet these requirements, sample selections were made 

 with precisely known probabilities. The sampling selections were 

 in four successive stages as follows: 



1. One hundred localities (Metropolitan Areas 

 and non-Metropolitan counties! containing 

 189 counties were first selected. 



2. Within each locality cities, towns, villages 

 and rural townships were selected. 



3. Within these subdivisions, final clusters 

 (combinations of city blocks, Census Enumeration 

 Districts, etc.) were selected. 



U. In each cluster, all households were first 

 prelisted. Prom these lists, households 

 were selected for the sample with known prob- 

 ability* The individual in each household 

 primarily responsible for the meal planning 

 was designated for an interview. 



Interviews were conducted in 2,5>1j3 households selected 

 by the above procedure. 



299 



