2. Preoccupation of the proprietor with the 

 Christmas rush, year-end inventories , or 

 clerical work. 



3. In the case of some schools, closing of the 

 schools for vacation during part of the 

 period of field work. 



4. Lack of interest in the study, on the part 

 of some proprietors. Some establishments 

 served only a minimum of hot solid food, 

 and seldom if ever used frozen processed 

 sea food. 



Nonresponse error is one of a number of factors affecting the 

 statistical significance of the findings. It is not taken into 

 account in the computation of tolerances, discussed in Section 

 III- A, Sampling Error. 



C. Response Errors 



While not measurable, response errors are likely to exist 

 where answers to questions are of a reasoning or quantitative 

 nature. Such errors may be defined as those introduced into 

 a study when respondents consciously or unconsciously give 

 superficial and/or incomplete answers to questions which 

 require reasons or numerical answers. 



Response errors in this study are more likely to occur in 

 numerical estimates since the only source generally available 

 is memory of detailed material by the respondents. 



For want of ajiy other guiding principle, one must assume that 

 these errors are randomly distributed. 



On the other hand, predominant errors may exist in quantitative 

 responses in the smaller establishments since their records 

 would tend to be less complete than those of larger establish- 

 ments. However, the effect of quantities erroneously reported 

 by small establishments would tend to be tempered by the 

 greater weight of quantities given by large establishments. 



16 



