Sec. 5.1 OBTAINING THE SAMPLE 119 



drawing numbers at random until a sample of the desired size was 

 obtained. Such a procedure would make it true that every possible 

 sample of the size n had been equally likely to be drawn at the outset 

 of the sampling, and this is necessary in random sampling. 



In many circumstances the procedure of sampling just outlined 

 would be unsatisfactory. A city might be made up of racial and 

 economic groups of such diverse opinions on the matter being studied 

 that it would not do to leave their representation in the sample to 

 pure chance, as in the random sampling just described. It would be 

 necessary to sample each group in accordance with its proportionate 

 part of the city's total registration of voters. 



It is noted that the sampling discussed above has supposed that the 

 sample will be taken by means of personal interviews. Any such 

 systems as calling persons on the telephone or mailing questionnaires, 

 which depend on voluntary and selective responses, or on their being 

 at a certain place at a certain time, are almost certain to produce 

 biased samples. The cause of their not responding, and hence not 

 being in the sample, may be associated with the type of response 

 they would have given. 



The theory and techniques of sampling in such a way that the 

 conclusions which can be drawn therefrom will be accurate and re- 

 liable are very extensive and cannot be covered here. The remarks 

 above merely point out a few of the more important and general 

 considerations. However, the reader can be warned to be critical 

 of any conclusions drawn from samples until he is satisfied that the 

 samples were taken in such a way that they should be representative 

 of the population about which conclusions have been drawn. If one 

 brand of cigarettes is said to be the favorite of a certain professional 

 group, we should at least wonder if that group was properly sampled. 

 Or, if someone returns from a foreign country and asserts that the 

 residents of that country hold certain points of view regarding a 

 matter of world-wide interest, we should wonder if he did an ade- 

 quate job of sampling public opinion in that country. Or, as a final 

 example, if someone seeks to obtain a sample by means of a mailed 

 questionnaire, we should wonder if those who do not respond have a 

 different opinion, say, from that generally expressed by those who 

 did return their questionnaires. If so, what population did those 

 who returned their questionnaires represent? 



