Polling During A Crisis

By Kathy Frankovic, CBS News director of surveys

The pollsters in Greece have just confronted this issue. Over the last two weeks, forest fires ravaged that country, killing at least 60 people. People in Athens smelled the smoke. But the country was in the middle of an election campaign. Should the Greek pollsters have stopped polling?

The answer soon became clear: they should not. And they did not. The conduct of the government in fighting the fires - whether it responded quickly enough, and whether the response was competent enough - became an election issue. This was an opportunity to give the public its voice, and to learn how the fires might affect the September 16 election. That analysis could not be left only to pundits (who were likely to speak without supporting data) or to the politicians themselves.

In 2002, pollsters in the Netherlands faced a similar dilemma. Pim Fortuyn, the flamboyant Dutch politician and leader of an anti-immigrant political party, was gunned down in an Amsterdam street, just nine days before the election. The pollsters briefly halted polling (and politicians temporarily suspended campaigning), but soon began again. The assassination had changed Dutch politics, and it was important to understand why.

Polls are taken in times of crisis, or in times of great natural disasters, even when there may be problems in reaching some of the population. People who had been directly affected by Hurricane Katrina couldn't be interviewed in the days following the hurricane - many were away from their homes, and many telephone landlines were knocked down. But it was important to find out what people in the rest of the U.S. thought about the event and the government's response. Conducting polls made it possible to trace the decline in the public's assessment of the state and national response.

Similarly, many U.S. pollsters, including CBS News, conducted opinion polls the night of September 11, 2001. Although those terrorist attacks were aimed at New York and Washington, this was a national event, and Americans needed to have their voices heard.

Polling on 9/11 was difficult. Some organizations have interviewing rooms in places far away from the cities that were hit - in Utah or in suburban Philadelphia, for example - and interviewers there were able to arrive for work almost as easily as they normally would. But the interviewing room at CBS News is at the network's headquarters in the middle of New York City, and with the interruptions and police blockages of travel to and from Manhattan, many interviewers could not get to work at all.

Those who were able to come to work discovered that, in this crisis, many Americans wanted to talk, and that they appreciated the opportunity to express their views on the attacks. We designed the questionnaire to give respondents the opportunity to put their feelings in their own words, and to speak for the rest of the country. A very small number of respondents became extremely emotional, and interviewers generally ended those interviews as quickly as possible, so as not to add to the respondents' apparent pain.

But for many others, the interview may have been a cathartic experience.

In the days after 9/11, the CBS News Poll asked respondents to describe how they felt when they first heard about the attacks, a question which opened the door for respondents to give not just forced-choice answers, but to say exactly what they wanted to say. And they did. They talked about their feelings of shock and devastation on the day of the attack, and about how their feelings changed to anger as the scope of the damage became clear. They supported President Bush instinctively, and their approval of him only increased in the following days, rising from 50 percent in August before the attacks, to 72 percent on September 11 and 12, and up to 89 percent a few days later. They rallied around New York City and Mayor Rudy Giuliani. They expressed confidence in the government's ability to find and punish the terrorist leaders. They supported a military response, even if that might mean innocent civilians could be killed.

There were also a number of surprises in the process of polling - surprises which validated our decision to go ahead and poll in the wake of the attack. The process took respondents away from their television sets - if only for a few minutes - and in that sense it provided an outlet. Americans wanted to be taken away from the video replays of the attack and the talk about the collapse of the World Trade Center towers.

Pollsters worry about response rates and cooperation rates, which have been declining for a long time. But in crisis times, they rise! We can compute the cooperation rate by dividing the number of completed interviews by the number of completed interviews plus refusals. The average cooperation rate was 56 percent before 9/11, but it rose 17 points to 73 percent afterwards. The overall response rate (completed interviews divided by the total of refusals, respondents who are not available during the interview period and non-English speaking households) went from just about one-third to 44 percent after the attacks.

But like the president's approval rating, those high polling ratings also came back to earth. Within months of 9/11, the public's increased willingness to be polled slid back to pre-attack levels.
By Kathy Frankovic