dennis roberts said on 12/4/02 8:56 AM: >So, we really have no idea of whether they are "typical" of the "population >of likely voters" or not. It was NOT a SRS of Americans who were "likely >voters".
while I appreciate the perspective you are bringing to this question, and certainly the process you describe is probably a closer approximation of what actually occured than is related in the article, we still have to come up with a rationale for deciding that the sample actually attained is not equivalent to a true SRS.... For it to not be equivalent we have to know that one of the following is true: 1) That the reason certain people are reached and others not reached is related to the questions being asked. If there is no relationship, then it is reasonable to assume the sample is adequate. 2) That people who 'slam' would respond differently to the questions being asked than those who do not 'slam.' If there is no relationship between slamming and the qeustion, then it is reasonable to assume the sample is adequate. 3) That households with multiple phones, or no phones, or who otherwise have higher or lower chances of being included in the sample may respond differently to the questions being asked. If there is no relationship, then it is reasonable to assume the sample is adequate. There may be other statements simlar to the above 3 that may be asked... but you get the picture. Show me some data that makes one believe that actual the sampling procedures are likely to get biases responses compared to a true SRS. Paul . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
