areas selected comprised the whole city, this was necessary 

 because the sample for subuniverse No, 3 (institutions) had to 

 be a 100 percent sample, in order to provide adequate basis 

 for analysis. 



Where a two-stage area sample was combined with a single- 

 stage list sample, the sampling fractions for the two stages 

 were so selected as to equal, in combination, the single samp- 

 ling fraction used for the list sample. In general, the sampling 

 fraction for the first stage in the two-stage area sample was 

 determined by the over -all sampling fraction for subuniverse 

 No. 3 (institutions). 



In some cases, supplementary sampling was done within certain 

 sampled areas for certain subuniverses to provide more 

 adequate samples. Thus, one subuniverse within one city was 

 in some cases sampled at more than one rate. These different 

 sampling rates, too, were offset by proper reweighting factors 

 that are reflected in the reweighted "Tabulating Cases" reported 

 above . 



D. Data Processing 



Each completed questionnaire was checked in against a "Call 

 Record Sheet" to insure that the proper sampling procedure 

 was followed. 



In addition each interview was reviewed by a trained editor for 

 completeness, accuracy, consistency, and quality. 



Codes for all open questions were developed from a represen- 

 tative subsample of interviews. In the coding operation, 

 standard quality controls were utilized to insure a high level of 

 accuracy. For instance, a record was kept of responses coded 

 into the miscellaneous categories with frequent review to deter- 

 mine whether or not a separate category should be added to the 

 code . 



All data were punched onto IBM cards ajid verification checks 

 were performed in accordance with standard tabulating proced- 

 ures. 



