Technological and social change is forcing pollsters to rethink a lot of things about our trade as we move into 2014.
One of the biggest challenges is the move from landline phones to cellphones. Technically we know what to do, but the escalation in costs to do things right is creating headaches.
Acquisition of telephone numbers for cellphone samples is more costly — as much as eight times more per number in some circumstances. And it costs more to dial these numbers. Federal rules insist that we cannot use computerized predictive dialers to place calls to cell numbers. Rather, interviewers must dial each number by hand, individually, increasing the interviewer time wasted between interviews.
We also don’t want to interview respondents reached while they are driving or doing something else that might pose risks to either their safety or their concentration on our questions. So, more call-backs are required to reconnect with those voters who answer but cannot do the interview at that moment. And, most surprising, we find that cellphone interviews take longer to complete once started. This is a bit of a mystery. Perhaps some of this can be explained by “Can you hear me now?” syndrome, which afflicts cellphone users in places with marginal signal. But some of it might also be explained by the relationship between the age and educational attainment of cellphone users. Compared with landline voters, cellphone voters are generally younger and, in some states, less educated. So cellphone respondents might know less about political topics, and more frequently need to ask interviewers to re-read a question or take longer to form an answer.
But if these challenges were all we faced, no worry. These are manageable problems.
Unfortunately, the fixes are expensive. I have had some clients ask to skip the cellphone expense. The answer is “no.” The data would be worthless. Cellphone and landline people are different. One cannot substitute for the other.
Another interesting change is mandated by the rise of bloggers and other alternative media. At the beginning of our polls, we ask a “security screener” question, designed to terminate those who work for news media or other campaigns. There is nothing quite as embarrassing as having major portions of your campaign strategy — as tested in your benchmark poll — leaked because you interviewed someone in the news media or from the other party. Of course, respondents can lie and say they aren’t media or pols, but at least we should try.
Well, now we have bloggers. They are not “news media” or campaigners, but they love to capture polls in the wild. I had one astute blogger we interviewed unknowingly repost my questionnaire in full — almost perfectly rendered — in his blog. Impressive. Now I have added bloggers to my security screener.
Even questions as basic as race and ethnicity have to be reexamined these days. We have started following the Census Bureau example of offering “multi-racial” as a category. Guess what? There are a significant number of takers.
Hill is a pollster who has worked for Republican campaigns and causes since 1984.