By Kathy Frankovic, Ph.D., Former Director of Surveys at CBS News

When it comes to opinion research, it sometimes isn’t just a matter of meeting scientific standards, but about assuaging doubts about whether measuring should be done at all. This is especially true these days when questions have been raised about the accuracy of scientific polling in recent elections. Some of the criticisms made after the 2016 and 2020 elections were helpful, and survey researchers responded as scientists —reviewing their methods and making improvements where necessary and possible. Other criticisms are often the result of disappointment with what public opinion research reports. Like election denial, disappointment with a preferred outcome causes some people to question legitimate results. Of course, we all know that polls are a snapshot in time and opinions can change — and have changed — dramatically following major events. In 2022, pre-election polls gave Americans an accurate sense of who was favored in the elections and how public evaluations of them were changing as the election approached. They also underscored those elections where polls were just too close to say what might happen (and the days of vote counting after November 8 underscored the accuracy of many close pre-election polls).

“We all know that polls are a snapshot in time and opinions can change - and have changed - dramatically following major events.”

While election polls provide the most visible measures of public opinion research, we should not forget less visible but equally important measures such as the government’s estimates of unemployment produced by the Bureau of Labor Statistics, or basic information about the United States collected in between decennial censuses. These surveys are larger, more complex and even more rigorous than opinion polls can be.

Public opinion research has been with us for almost two centuries — scientific polling for less than half that time. Scientific polling is hard. Trying to replicate the population through random sampling — giving everyone a known probability of being included — faces serious problems, all recognized by practitioners: declining response rates, shifting modes of conducting research, and declining budgets. Despite this, the goal has always been to represent the public and even when those practitioners have been forced to move away from probability sampling there is always the recognition that polls require adequate representation and the opportunity for the public to be included. Otherwise, we may as well revert to counting crowd sizes, estimating editorial readership or trending Twitter posts — which would tell us something, just not what we are looking for.

Those unscientific methods have been used. Counting the number of newspaper editorials on one side of an issue or the other only measured editorial opinion among newspaper editors and — more likely — owners, even when quantifying the impact by adjusting for circulation. No matter how you quantified it, it did not have the value of scientific opinion polls.

Ignoring science when measuring public opinion will not tell us what we really need to know.

“Ignoring science when measuring public opinion will not tell us what we really need to know.”

Publicizing reports from good studies of opinions and preferences informs the public about many things that would otherwise be known only by those who can afford to commission their own studies — like government, political candidates and corporations. Few elites with their own access to commissioned polls admit to learning anything from published polls, but those polls democratize information, and are available to whomever wants to read.

There is really no question here. Social science is important for many reasons, and in a democracy, the voice of the people matters, and we need to measure it.

KATHY FRANKOVIC spent more than three decades at CBS News as the point person for the CBS News Poll and the CBS News polling collaboration with The New York Times. After the 2000 election, she was placed in charge of the network’s Election Night decision team, and successfully projected election results for elections from 2002 to 2008. She retired from full-time work at CBS News in 2009, but since then has been an Election and Polling consultant for YouGov, CBS News, Harvard University, Open Society Foundations and other survey research organizations.

She has served as President of both the World Association for Public Opinion and the American Association for Public Opinion Research, and sits on the Professional Standards Committee of ESOMAR. Kathy has won many national awards for her work conducting and explaining public opinion for the news media, including the 2008 AAPOR Award for Exceptionally Distinguished Achievement, that association’s highest honor, and the 2011 Roper Center Mitofsky Award for Excellence in Public Opinion Research.

Download as PDF