6 BULLETIN 1443, U. S. DEPARTMENT OP AGRICULTURE 
The same questionnaire was used for all groups, so that variations 
noted among the groups in the replies to the same question were 
indicative of the differences arising through race, nationality, or dif- 
ferences in standards of living. 
In the conduct of this survey only those families that consumed 
meat were interviewed. Strict vegetarian families were not inter- 
viewed. Families using meat in small quantities were included. 
RESULTS OF THE SURVEY 
METHOD OF PRESENTATION 
The principal form of presentation of housewives' replies to the 
questionnaires is the summary table in which results for a single ques- 
tion from various cities were combined by population groups and 
group subdivisions. In compiling these summary tables no attempt 
was made to bring out the differences in consumer demand among 
cities, but to disclose these differences the replies to certain questions 
were arranged by cities.- In the city tables there has not been com- 
plete presentation of all data for all groups and classes for each city. 
Only the combined total number of replies from housewives of the 
middle and well-to-do classes of the American white group has been 
shown. It was believed that the two classes would more truly indi- 
cate the differences among cities in consumer preferences and buying 
habits, as they seemed to constitute the most representative portion 
of each city's population, when judged from the standpoints of race, 
nationality, and standard of living. 
The usual unit in the tables is the family or household. In nearly 
all of the tables the number of replies was based on the family or 
household, since most of the questions related to the household and 
not to the individual members. The few exceptions will be noted as 
the questions and tables are discussed. 
In presenting the analysis, the number of housewives giving the 
same answer to a question was expressed as a percentage of the total 
number of housewives replying to the question. This use of percent- 
ages permits direct comparisons among different population groups 
for which the number of replies varied greatly. Reference to the 
total number of replies to any question for any population group or 
class will indicate the extent of the data on which the conclusions 
were based. 
It should be made clear that the percentages were calculated to 
the first decimal place, not because the data justified such accuracy 
but because of the mathematical necessity of having the various per- 
centages relating to the several answers to a question total 100 per 
cent. Since some of the answers occurred in small numbers and con- 
stituted at times less than 1 per cent of the total replies, the use of 
tenths of 1 per cent was desirable in indicating the relative impor- 
tance of these answers. 
No relations were apparent among many of the phases of this study 
of consumer demand, and at this stage of investigations of this char- 
acter it was necessary to present the results for many questions 
separately without relating them to other questions. Each of the 
questions is therefore presented as a separate unit; but where rela- 
tions were thought to exist between various questions, they have 
been pointed out. 
