of the urbanized areas using the ordered 

 lists and 1950 census data on the number of 

 occupied dwelling units or households found 

 in each census tract, block or enumeration 

 district. These strata, seventy in number 

 for each urbanized area, vere constructed 

 to contain approximately the same number 

 of households In each. 



Each of the seventy strata vas then 

 divided into a number of small area seg- 

 ments having boundaries vhich could easily 

 be identified in the field by the inter- 

 viewers. Each such area segment contained 

 one or more clusters of households or sam- 

 pling units. The number of sampling vmits 

 or interviewer work loads assigned to each 

 area segment was based on data available on 

 the number of occupied dwelling units lo- 

 cated within these segnent boundaries. 

 These data were obtained from a variety of 

 sotorces including 1950 block statistics, 

 1950 enumeration district statistics, state 

 highway maps, etc. 



Tlie geographic strata In each city 

 were all constructed to contain the same 

 number of sampling units with the exception 

 of Detroit. In the central city portion of 

 Detroit, the area strata for the tracts in 

 the white zone (that is, the tracts with at 

 least 90 percent of their 1950 dwelling 

 units occupied by white families) were con- 

 structed to contain twice as many sampling 

 units as the remaining area strata estab- 

 lished for the Detroit urbanized «u:ea 

 sample. Initially, two sampling units were 

 selected with equal probability and without 

 replacement from each of the geographic 

 strata, yielding a sample total of ll»0 

 sanqpling units for each urbanized area. 



llie sample selection was accomplished 

 by choosing two random nxmibers for each 

 stratum between one emd the total number of 

 sampling units in the stratum. Thus, the 

 sampling rate was the same for all geo- 

 graphic strata within a city with the ex- 

 ception of those comprising the irtilte zone 

 in Detroit referred to above. Since these 

 strata contained twice as many sampling 

 units as the remaining geographic strata 

 in Detroit they were sampled at one-half 

 the rate of the remaining strata in that 

 urbanized area. The disproportionate sam- 

 pling in Detroit was deemed necessary to 

 yield sufficient Interviews with non-white 

 families for separate tabulation. 



The number of strata and sampling 

 units for the central cities and the re- 

 maining portions of the three xirbanized 

 areas are shown in the following table: 



Appendix Table 1 



NUMBER OF STRATA AND SAMPLING UNITS OF 



URBANIZED AREAS INCLUDED IN THE 



MOTIVATION SURVEY 



The decision to Include a sample of 

 non-white households in Orangeburg, South 

 Carolina was made after the sample for the 

 three principal urbanized areas was de- 

 signed and selected. The expected sample 

 size in each of the three urbanized areas 

 was then reduced from 8^ households to 

 725 households in order to shift a portion 

 of the field budget to the survey to be 

 conducted in Orangeburg Covinty. Rather 

 thsm design and select a new sample in 

 each of the three urbeuiized areas, twenty 

 sampling units in Birmingham, thirteen in 

 Boston, and twenty-eight in Detroit were 

 discarded at random with a condition that 

 no more than one sampling unit woxild be 

 discarded from any one stratimi. 



Strict field procedures were employed 

 to determine the eligible households 

 associated with the selected sampling units 

 in em unbiased manner. The interviewers 

 were required to list the occupied dwelling 

 units in each area segment containing a 

 selected sampling unit in advsmce of the 

 interviewing. The listings showed addresses 

 eind other necessary identification for all 

 dwelling units located within the botmdaries 

 of each Eu:ea segment. The enumerators were 

 provided with maps showing these boundaries, 

 as well as the steurting point and direction 



33 



