From 1992 until recently, Harold Andersen wrote a column for the Omaha World Herald (OWH); Andersen has recently given up the column in favor of a blog. He is also a former publisher for the OWH, and he and his wife Marian are the people that the UNL journalism building on campus here is named for. Andersen is very concerned about good practices in journalism. It's something he's written about repeatedly over the years. One of the things that he would write about is how journalists use results of polls. Here is a quote from a column he wrote in 2000:
When will journalists understand the limitations of public opinion polls and report the results of those polls with the proper qualifications and limitations clearly specified, keeping in mind, for just one example, that polls of a tiny percentage of a much larger mass of people don't reliably "show" or "find" the feelings of that large mass? Poll results allow projections (some of them very questionable), but they "show" or "find" only the answers of the tiny percentage represented by those who were polled.What he's saying, for example, is that if you run a national survey by sampling about 1000 of the 100 million voters in our country, you certainly know how those 1000 feel but that the sample is too small to tell you anything reliable about how the other 99,999,000 other voters feel.
Your writing assignment is to write a one page essay in reaction to this quote. Explain whether you agree or do not agree, and why. Why do you suppose that so many reputable organizations (such as the New York Times, the major networks, etc.) run such polls? What do you think this means about the business of polling? [Note that polling is big business. For example, the Gallup Organization is a major polling company whose operational headquarters are located in Omaha. Gallup also has a strong relationship with UNL via the
UNL Gallup Research Center.]
Below is the complete relevant part of the column from which this quote was taken. (Note that we now know that the 2000 presidential election was a dead heat, possibly more so than any other in our country's history, and that as luck would have it the poll in question in fact did end up very reliably indicating the feelings of the large mass of voters in that election!)
Excerpt from Harold Andersen's column in the Sunday, March 5, 2000, Omaha World Herald.
I decry again the careless use of poll results by too many journalists and the unqualified use of designed-to-alarm numbers by some government officials, by some in the news media and by a great many advocates trying to prove, by the numbers, that their cause is just. Examples of such use of numbers - sometimes simply careless, sometimes intentional - appear on the public scene with distressing consistency. Three recent examples: A front page New York Times story reporting the results of a Times/CBS News Poll said the poll showed that "voters' impressions of Texas Gov. George W. Bush nationally have diminished so substantially that now as many people dislike him as like him." In the first place, the poll doesn't show anything more than the answers given by 979 adults scattered across the country who said they are registered voters and then expressed their personal opinion. It may provide the basis for a national projection or estimate, but it doesn't "show" the national picture. The Times story moves on to a conclusion of fact that is downright silly: "If the general election were held today, the survey found, it would be a dead heat, with Mr. Bush drawing 45 percent and Mr. Gore 44 percent." (The poll in question is supposed to produce figures that allow the 979 responses to be projected as a reliable projection of how tens of millions of Americans feel, with a margin of error of only plus or minus 3 percent. Pollsters consider such a narrow projected margin - 45-44 percent in this case - a virtual dead heat.) In the first place, the survey of 979 possible voters didn't "find" any nationwide facts. When will journalists understand the limitations of public opinion polls and report the results of those polls with the proper qualifications and limitations clearly specified, keeping in mind, for just one example, that polls of a tiny percentage of a much larger mass of people don't reliably "show" or "find" the feelings of that large mass? Poll results allow projections (some of them very questionable), but they "show" or "find" only the answers of the tiny percentage represented by those who were polled.