TABLE OF SAMPLING ERRORS 

 FOR NATIONAL SAMPLE (Continued) 



Percent Standard 



Having the Error in 



Ques. Character- Percentage 



No, Characteristics istic Points 



Characteristices in Section H - Percent 

 of households who own the particular pet 

 (dog or cat) that: 



H-lb Feed earned pet food containing fish to 



their dog 26.3 2.7 



H-lb Feed canned pet food containing fish to 



their cat 39.7 3»2 



Characteristics in Section H - Percent 

 of households who feed canned pet food 

 containing fish to their pet that: 



H-3 Usually buy five or more cans of pet 



food at one time U5.0 3»8 



Interpretation of the Table of Sampling Errors for National 

 Sample . The figures given in the final column of the table represent 

 one standard error. As an example, for the first characteristic 

 listed - the percent of all households that served canned tuna in 

 the past 12 months , which is given as 76.1 percent, the standard 

 error of sampling is 1.2 percentage points. This means that the chances 

 are about 2 in 3 that the sample percentage is within 1.2 percentage 

 points of the "true" figure, meaning by "true" here the figure that 

 a complete census would show using the same survey procedures. Two 

 standard errors correspond to a 19 in 20 probability. In this case, 

 the chances are 19 in 20 that the sample percentage lies within 2.1+ 

 percentage points of the "true" figure. This may be stated in terms 

 of confidence limits: the limits corresponding to the 19 in 20 

 probability are 76.1 minus 2.U and 76.1 plus 2.U or 73»7 to 78.5 



308 



