FOEEWORD 



This is the second of two volumes reporting on the 

 results of a survey of fish and shellfish consumption in public 

 eating places. Volume I, issued as Special Scientific Report — 

 Fisheries No. 218, gave an analysis of the responses to the 

 questionnaire on a national basis. Volume II gives an analysis 

 of the cross-tabulated responses on a regional, type-cf -establish- 

 ment, sales-size, asid city-size basis. 



This report contains cross -tabulations of the responses 

 to questions of only the weighted numbers of reporting establish- 

 ments. Put in anoliier way, the weighted numbers of non-reporters 

 are not included in these tables, primarily because appropriate 

 weights for these establishments were not available. Imputations 

 for these non-reporters on a national basis would have adjusted 

 the totals shown by roughly 10 percent for the United States as 

 a whole, 10 percent for restaurants, cafeterias, and drug stores, 

 13 percent for restaurants or cafeterias located in hotels, and 

 20 percent for drinking places, lunch counters, and refreshment 

 stands. Estimates of the number of establishments, adjusted to 

 "universe" totals, are given in tables 2 through 7 primarily to 

 indicate the magnitude of the adjustments. In preparing these 

 estimates, the assumption was made that the ■weighted numbers of 

 reporting establishiments were distributed in the same way as 

 those of the non-reporting establishments. Imputations of the 

 weighted numbers of non-reporters, however, are not included in 

 any "cell" for all other tables showing the percentage distri- 

 butions to questions included in the survey. 



Exact agreement of the percentages for similar ques- 

 tions in both reports was not possible, primarily because of 

 differences due to rounding. Consequently, the detailed per- 

 centages in some tables will not agree in total with those shown 

 in related tables. A review of these differences in agreement 

 of sub-total percentages with totals shown elsewhere indicated 

 that, in most instances, the differences were not large enough 

 to vjarrant adjustment. In the majority of cases examined the 

 percentages did not vary by more than 0.1 or 0,2 percent, and 

 rarely by more than 2 or 3 percent. The relative distribution 

 of the responses to individual questions are not seriously 

 affected by these deviations. 



