This means that the chances are, 2 out of 3, that if all eating 

 places in Denver had been interviewed with the same techiiiques 

 used in this survey this proportion would have fallen between 

 ,588 + .036 and .588 — .036, or between 62.4 percent and 

 55. 2 percent. 



The above procedure is followed when the "base" for the percent- 

 age whose standard error is desired is the same as the total 

 number of interviews according to the last column in the table 

 above. In other cases, the following slightly modified procedure 

 is followed. 



The proportion of all public and institutional eating places in 

 Denver, Colorado, with annual sales volume of less than $10,000 , 

 that bought frozen processed sea food in the preceding 12 months 

 is 39.6 percent (Denver Table 1, column 4, line 4); this percent- 

 age is based on a Denver subsample of 87 eating places (Ibidem, 

 line 1), which is not the total Denver sample of 216 places (see 

 table above, line 4). First we must adjust the percentage to 

 express it as a proportion of the total sample, as follows: 



39.6% X 87 (size of subsample) = 34.452 

 34.452 -1. 216 (size of total sample) = 16.0% 

 Applying the formula for simple random sampling: 



= / . 16 X .84 = /. 1 344 = / 



>/ 216 v/ 216 y 



dp = /. 16 X .84 = /. 1 344 - / 00062222 = .02494 



Applying the adjustment factor from the first column of the 

 preceding table above, line 4, 



dp = 1.05 X .02494 . .026187 (say .0262) 



13 



