Re: Gallup organization on polling methods

[dennis roberts]

random sampling has nothing to do with any particular person being selected ... but, rather, in the LONG run, making sure that sampleS will be representative of the target population ....whether i ever get into any random sample has no import

what is the target population here? all people who have telephones? all people who have telephones that are listed ... ? unlisted ? all people with and without telephones ... with those having telephones being listed? or, including UNlisted?

if gallup uses a method that uses ONLY random sampling of all telephone subscribers (residential) ... then in the LONG run ... samples would reflect that ... but, when they do sampling ... it is a SHORT RUN deal ... so, to insure that all of the important categories are included ... they would have to know aHEAD of time who fits into which category ...

thus ...

i will make the assumption that gallup CAN (if it desires) separate listed numbers from UNlisted numbers ... does anyone have any specific documentation that says otherwise?

that is ... i assume that gallup could (if they knew that about 2% of all #s were unlisted) ... ask the "computer" (or dictate to the group that does the selection) to stratify their sample so that about 2% are from the unlisted category ... thus guaranteeing some UNlisted representation in their sample ... on the spot ... in the SHORT RUN

does anyone know if that is true or not?

if it were true, which i assume is the case, then ... the rules are not the same for sampling ... you and i have no way of assuring that UNlisted numbers could be found if we wanted to do a poll ... but, gallup could

it's not the ability to contact ME specifically (if my # is unlisted), it is the ability to absolutely know that some of their sample WILL be from the UNlisted category ...

no one else who wants to do telephone sampling ... has any way to guarantee that

thus, they do know something about phone subscribers ... that, the average joe or mary does not



.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================