Once again, what you think is true is not true.
"To achieve that randomness, telephone polls generally make use of a random-digit dial system, in which a computer uses a pre-determined set of area codes and exchanges (the middle three digits, which identify the service provider and area), and randomly generates the last four digits. If trying to reach a particular population, like a racial or occupational group, researchers can purchase lists of numbers, then randomly pick from that list. Of course, since about half of American households now only use cell phones, pollsters have to include cell phones in the sample, despite the fact that federal regulations require cell numbers to be dialed manually, rather than with an autodialer, dramatically increasing the cost of calling. Pollsters have also had to increase the number of times that they call numbers back, trying to get a completion: seven or more attempts are now standard in the industry. Despite these efforts to achieve a random sample, though, response rates remain shockingly low, especially among younger people, Spanish speakers, Evangelicals, and African-Americans"
hbr.org
Polling companies want a random sample, the opposite of what you want. |
