of the Bureau of the Census at somewhat in excess of 48,000,000 in 1956.' If 

 this figure is used to approximate the number of households, on the average, 

 therefore, the 2,770 households included in the sample corresponds, roughly, 

 to a sampling rate of approximately one out of every 17,390 households. 



THE INTERVIEWING 



The survey was conducted by trained interviewers of the contractor. 

 Before surveying the individual hou.seholds in the sample, the questionnaire 

 on household preferences for canped fish and shellfish was pretested in the 

 New York metropolitan area, Duluth, Minnesota and New Hanover County, 

 North Carolina. Field interviewing for the survey began on June 11, 1956 and 

 continued for approximately three weeks. 



A good portion of the staff of W. R. Simmons and Associates Research, 

 Inc., participated in the revision of the questionnaire. This practice enabled 

 the questionnaire to reflect any local custom or local word usage which might 

 present problems and brought the knowledge of many experts in survey tech- 

 nique into the development of the questionnaire. 



The 185 interviewers who made the field contacts attended a number of 

 training sessions where they were given a set of written instructions, and were 

 required to pass a written examination demonstrating complete understanding 

 of the questionnaire and the written instructions. The interviewer filled out a 

 number of practice questionnaires by interviewing a number of strangers under 

 the supervision of trained supervisors. To assure fullest cooperation and maxi- 

 mum accuracy, the interviewer was instructed to read an opening statement 

 to the respondent which explained the purpose of the survey and guaranteed 



' Data in U. S. Census Bureau Current Population Reports, Series P-20, No. 68 

 adjusted for households not serving at least one main dish meal at home in 

 a week. 



