In the case of estimates by region, income, city-size, 

 family size, race, religion, age and employment status of the 

 homemaker, the sampling errors are, of course, larger, and the 

 results must be used with some caution. However, even for these 

 breaks, the estimates may be used with confidence wherever they 

 are based on all or nearly all sample households in the particular 

 region, income group, city- size group, etc. A full discussion of 

 the allowances to be made for sampling error in the case of esti- 

 mates for sub-groups is given in the Appendix. Comparisons of 

 classification data with independent data from Census sources, also 

 given in the Appendix, indicate that the sample reflects reasonably 

 well the current distribution of the population. 



Incomplete returns can be a serious source of bias in the 

 results of a survey if care is not used to reduce them to the mini- 

 mum feasible. In this survey the completion rate was 80 percent, 

 that is, 80 percent of the originally assigned 3*198 households 

 were interviewed. 



Refusals amounted to only U2 percent and thus were not 

 likely to produce any substantial bias. A second call was made on 

 a different evening in all cases where the respondent had not been 

 at home on the first call. Almost all of the potential bias from 

 the 15 percent not-at-home after two calls was eliminated by the 

 use of a special procedure under which each interview is weighted 

 according to the respondent's probability of being found at home 

 when the interviewer calls. This probability is determined by ask- 

 ing the respondent whether he or she was at home at a random time on 

 each of six nights. •=/ 



The Questionnaire and Field Work 



The questionnaire was developed in consultation with the 

 U. S. Fish and Wildlife Service and carefully pretested in three 

 different areas. Pretest results were analyzed in detail in conjunc- 

 tion with detailed reports frcm. the pretest interviewers, and this 

 analysis served as the basis for making the necessary revisions in 

 the questionnaire and in instructions to interviewers. A copy of 

 the questionnaire is included in the Appendix. 



1/ This technique is fully described in an article entitled,"A Plan 

 To Account For 'Not-At-Homes ' By Combining, Weights and Callbacks " 

 by Willard R. Simmons, The Journal of Marketing , July, 195U. 



