The interviewers employed on the survey were experienced 

 and capable members of the contractors field organization. They 

 were instructed to hold up interviewing after the first day's work 

 until their completed questionnaires were thoroughly edited to test 

 the quality of their work and their understanding of the instructions. 

 Thus they could be given further direction before proceeding with 

 the work on their assignment, or replaced if necessary. 



Coding and Tabulation Controls 



All coding was checked completely at the start, and sample 

 coding instituted only after difference© between coders had reached 

 an acceptably low level. Thereafter, a continuous check of coding 

 errors was maintained to make sure that the process was in control. 

 Open-end codes were established from a systematic sample of 500 cases 

 for any category with a frequency of greater than 1^ percent. 



A complete mechanical edit of the punched cards for consistency 

 and accuracy was carried out on the IBM 101 machine, and all errors 

 discovered in this process were corrected by reference to the questionnaire. 



SUMMARY OF MAJOR BINDINGS 



The Use of Canned Fish or Shellfish and its Relation 

 to Fresh and Frozen Fish or Shellfish 



Survey results indicate that canned fish or shellfish is 

 not directly competitive in any considerable degree with fresh or 

 frozen fish or shellfish . This is particularly so with respect to 

 canned fish. The following facts bear on this point: 



1. Almost all households served canned fish or shellfish 

 at one time or another as shown by the finding that 

 91% of the sample households had served these products 

 during the past twelve months. The use of canned 

 fish or shellfish was also quite stable throughout 

 each of the four regions studied. 



2. Those households that use fresh or frozen fish 

 actually also use more canned fish than average. 

 Calculations were made giving rough estimates of 

 the average number of servings in the four-we^k 

 period preceding the interview for these group sJ£/ 

 The comparison follows: 



2/ For this calculation (and others of a similar kind given in this 

 report) tabulations were used showing the number of times an item 

 was served in the past U weeks, from 1 time to 9 times or more . For 

 this last group, an average of 10 times was assumed. 



