is designed to parallel census data for the Unit- 

 ed States with respect to geographic divisions, 

 household income, population density and de- 

 gree of urbanization, and age of Panel 

 members. 



A sample of 1,500 households was selected 

 from the full Panel by stratified random samp- 

 ling. The smaller sample maintains corres- 

 pondence with the above mentioned census data 

 and provides a reasonable cross section of other 

 demographic characteristics, e.g. education, 

 occupation, race, religion, family size, and age 

 and sex composition of the children. 



The survey Panel may be considered closely 

 representative of the population of U.S. house- 

 holds with respect to the significant demograph- 



ic variables. Household surveys, however, are 

 particularly vulnerable to nonsampling errors 

 arising from unavoidable biases in the question- 

 naire and in the memories of the respondents. 

 To minimize the nonsampling errors, completed 

 questionnaires were monitored for obvious re- 

 porting errors. 



LITERATURE CITED 



LeBOVIT, C. 



1970. Foods eaten away from home. National food 

 situation. U.S. Department of Agriculture, Wash- 

 ington, D.C. 25-31. 

 PETERKIN, B., and B. EVANS. 



1965. Food purchasing guide for group feeding. 

 U.S. Department of Agriculture, Washington, D.C. 

 54 p. 



14 



