One way is to use a preselected panel to select interviewees from — rather than just randomly dialing the population at large. A panel comprises a large number of preselected volunteers, who agree to be surveyed on occasion (often in exchange for money or prizes). Since the location and demographics of a panel are known in advance, it’s not difficult to chose a “representative” sample for each poll. The panel would include a proportionate number of cell-phone users.
Another advantage of panels is that response rate is usually far better than it is for random dialing to the population at large. Response-rate bias is one of the most serious sources of error in polls & is not easy to quantify. Panels have a big edge in reducing response rate bias.
Only to those who give a crap about what a poll says.
Or those who change their stance on a subject to fit a poll, whether they believe the new stance or not.
Which group do you fit into, that makes you believe it is a 'serious issue'?
Or are you just responding to a poll?