Appendix B 



Sample Design 



The saJi^ling operation for the nationwide survey of public 

 eating places was performed at two distinct levels, 



1, The selection of primaiy sampling units consisting 

 of metropolitan areas or certain combinations of 

 non-metropolitan counties from the more than 2,000 

 combinations of counties in the United States, There 

 were 230 of these primaiy sampling units selected 

 for this stu(^, 



2, The selection of small pieces of land (segments) 



from within the selected primaiy units for interviewing 

 all public eating places within the chosen segments. 



The probability of selection of any given eating place in the 

 sample is then the probability of selecting the land on which the es- 

 tablishment is located. 



Selection of Primary Sampling Units 



The 2,000 combined counties that serve as primaiy sampling units 

 were classified into one of four groups. 



Group I consisted of the 12 largest metropolitan areas. Group 

 H consisted of the primary vuiits \diich contained a city with 50,000 in- 

 habitants or more in 19^0. Group III consisted of the primaiy units having 

 less than 25 percent of the population residing on farras and certain others 

 having a very higli in-iid.gration rate between 1950 and 1953. Group IV 

 consisted of all the other primaiy units and contained more than three 

 quarters of the farras and farm population of the country in 1950. 



The primary \inits had been fui-ther classified into sub-groups 

 to form the primary strata. In group I, each metropolitan area occupied 

 its own separate primary stratum. In groups II and H primary strata had 

 been formed based on geographic and economic characteristics. In group IV, 

 the primary strata had been formed based on agricultural characteristics. 

 In total, 230 primary strata had been formed. 



From each primaiy stratum 1 primary sampling tmit was selected. 

 The distribution of sample primaiy units used in this survey is shown in 

 the map (see page 10), 



In addition to the area sample design the survey utilized the 

 Census Bureau's Current Mailing List of certain large establishments 

 outside of the sample areas. 



79 



