In order to reach a specific "error rate", such as +/- 4%, there is a specific number of individuals that you need to poll. Once they reach that number, it doesn't make sense to continue to call, unless they want to try for a lower error rate. And cost is definitely a factor.
I was called several months ago for the Rasmussen poll. It is totally automated, i.e. computer driven. I don't think that Time and Newsweek and others use this technique.
There are arguments pro and con. For instance, how many teenagers are answering the Rasmussen poll? With a live poll, the person taking the poll can make judgements that the computer can't.
Thanks, Just asking. I'm still suspicious about these techniques. I would think that in the information age more accuracy could be obtain, without incurring that much extra cost on the back end and the additional cost on the front end would be a worthwhile investment. But it's not my line of work so I guess I'll just have to accept it. LOL