As the saying goes, “There are lies, damned lies and
statistics.” Nowhere is this concept more prevalent than in the
polling industry, where analysts examine a wealth of information and
attempt to make some sense of it.
The “art” of polling — as some researchers refer to it — has been
mastered by some, lost by others, and is sought after by still more who
want to provide Americans with the most accurate “pulse of the nation”
in the polling research they do.
Nevertheless, just a few polling services are really regarded as
having mastered their trade, being considered reliable by both Americans
and those who use polling figures for “official” reasons.
One up-and-comer among that handful of reliable analytical polling
services is Rasmussen Research, developers of the
Portrait of America
“At Rasmussen Research, we are committed to unleashing the incredible power and truth of public opinion and consumer information in a pure, unfiltered form,” said Scott Rasmussen, director of the award-winning polling agency. With the advent of the Internet and the need for original content, “we plan to publicly release more polling data every year than the rest of the industry combined.”
Within a decade, he said, Rasmussen Research will have generated more than half of all polling data ever publicly released in U.S. history.
An independent assessment recently found that Rasmussen was the most accurate polling firm in the nation; Gallup finished second, with Zogby close behind.
And, Rasmussen said, his firm has developed a better way to take and measure telephone surveys at a cost far below that which other firms incur. That, he said, is one reason why his company has the ability to churn out much more data.
“The large number of interviews we conduct allows us to paint a richer picture of the data with details that are more accurate than those produced through traditional means,” he said. “Our strategy seeks to increase the value of our content by placing it in context through partnerships with leading industry sites.
“I hate it that the impression is that polls aren’t accurate,” said Rasmussen. “The fact is they are, and a lot of it depends, of course,
on the methodology.”
Rasmussen Research conducts public opinion surveys using a proprietary automated polling system. Participants in surveys hear a recorded voice ask a question and provide instructions on how to answer. Typically, the response instructions would be something like, “If yes, press 1; if no, press 2; if you are not sure, press 3.” Other times, the survey respondents are asked to respond in their own words.
And though the firm posts results of its surveys and polls online, Rasmussen said it does not gather polling data online.
“As with any opinion research, the most important step is designing the survey instrument. We take care in crafting questions designed to elicit the most important and useful information,” he said. Occasionally the firm will ask polling questions that have not been asked before. When that happens, “we always include other questions that can be compared to an existing body of survey research or actual data,” he said.
One of the most frequent complaints about polling firms is that they intentionally pick “partisan” or specific target audiences that are likely to provide answers sought by pollsters. While that may be true in some cases and with some non-professional survey personnel who work for a particular group, Rasmussen Research is non-partisan and interested first and foremost in accuracy.
“To place our calls, we randomly select phone numbers appropriate to the geographic market being surveyed,” Rasmussen said. “Throughout our company’s history, we have used both listed-number samples and other samples, depending upon the needs of the client. We have purchased samples from outside vendors and pulled our own samples from an in-house database. Currently, we purchase all of our samples from a single vendor — Survey Sampling, Inc.”
For election surveys, “we call a full sample of adults and then use a series of screening questions to determine which adults are likely voters. The screening questions include questions about their voting history, their intentions to vote in the upcoming election and other matters.
“Our system has been developed over several years and our work has frequently been validated by election results and other survey research,” he added.
For instance, for the first time Rasmussen has done some overseas polling, including exit polling in two London races. The firm managed to project within 1 percentage point the victorious candidate in each case.
Also, in the presidential primaries so far this year, the firm’s accuracy rate has been similarly repeated.
“In the New York primary, we declared Bush the winner an hour before CNN, and in Michigan, we beat CNN by 45 minutes in declaring McCain the winner,” Rasmussen said.
Progressive Review, which conducted the “Primary Pollster Run-Off” independent review of polling firms, ranked Rasmussen No. 1 in accuracy.
“The results suggest that, contrary to popular myth, polls do work, but that some work considerably better than others,” said Sam Smith, publisher of the magazine.