The survey was conducted during the month of June when 

 most families have not as yet commenced their summer vacations. 

 This period is also sufficiently removed from the influences of 

 Lent. 



For the first three stages, rigid control by Census data, 

 maps and other materials made it possible to state exactly what 

 chance every county, Metropolitan Area or final cluster had of 

 falling into the sample, whether or not the area actually happened 

 to be included. 



This is possible because every selection at every stage of 

 sampling was made from a complete list (or frame) with all units 

 specifically identified and the exact number of units known. The ex- 

 istence of such a list or frame is a basic req-rirement wherever selections 

 with known probability are to be made. For example, in the first stage 

 of sampling, the selection of the 100 localities was made from a com- 

 plete list of sach localities, which in total includes the entire 

 United States. In the same way the successive stages of sampling 

 followed the principle of probability selection from a list. 



Selection of the 100 Localities . All of the 3077 counties 

 in the United States were divided first oetween metropolitan areas 

 and non-metropolitan counties according to the 1950 Census classification. 

 Each metropolitan area was defined as a primary sampling unit, in- 

 cluding all counties within the area. These sampling units were 

 stratified into eight geographic regions following the Census regions, 

 except that the New England and Middle Atlantic regions were combined. 

 The metropolitan areas were further stratified by size, per capita 

 income, and the percent of increase or decrease in population from 

 19i|0 to 1950. For the South the percent of non-white population was 

 also used as a criterion for stratification. Such stratification 

 assured that all stratified elements of the population were repre- 

 sented in their proper proportion. One metropolitan area was selected 

 from each such stratum with probability proportionate to its 195>0 

 population. Each of the fourteen largest areas constituted a single 

 stratum and hence were automatically included. 



The non-metropolitan counties were divided likewise into 

 eight geographic regions and further stratified according to the 

 percent change in population since 19h0, the percent of population 

 which was urban and the rural level of living index developed by the 

 Department of Agriculture. One county was selected from each such 

 stratum with probability proportionate to the 19^0 population. 



300 



