Every day, you hear or read about a poll or two about the primaries. Trouble is, the media do not tell you much about the polls.
Now, in the wake of the Iowa primary, observers are wondering why the polls were so far off on the Republican results. The margin of error was 9 percent. That is not due to chance.
Something is wrong with the surveys; but you cannot tell what, because all the news media tell you is the poll results.
They should be telling you the following:
1. When the survey was done. Most surveys take two to three days. Things change constantly in the primaries, so when a survey is done could affect results.
2. What is the sample size. With a sample of 1,000, a common number for surveys, the chance margin of error is 3.2 percent. If it is 500, then the chance margin of error is 4.5 percent. What this means is that differences less than the margin of error are not real, but are due to chance.
3. How respondents were chosen. This is a big problem in primary polls. A poll tells you they interviewed Republicans. What is a Republican? In Ohio when you register to vote, you do not indicate a party. You have to indicate party preference when you vote in a primary, but most people do not vote in primaries. It is is assumed that only a third of Ohio voters are registered by party. However, party registration is temporary. If you voted as a Democrat in last spring’s primary, there is nothing to keep you from voting in the Republican primary this year.
4. Of those chosen in the sample, with whom were interviews completed. Some surveys say likely voters were interviewed, which means an attempt was made to find out if the respondent really would vote. Others simply say Democrats and/or Republicans were interviewed which means a lot of respondents were people the survey takers knew would not vote. That will have some effect on the accuracy of results.
5. What exactly were respondents asked. What was the wording of the question? Was it “Which candidate do you like most?” or was it “Who would you vote for if the election were held today?” or something else?
In the absence of this information, there is no way to tell what went wrong with the surveys of Iowa Republicans.
Since media do not bother to disclose such information, you can expect more primary survey results that defy explanation. The polls are more entertaining than informative.
Guido H. Stempel III is a distinguished professor emeritus of journalism at Ohio University. Professor Stempel has a Ph.D. in mass communication from the University of Wisconsin and a master's in journalism from Indiana University. He has been on the Ohio University faculty since 1965 and served as director and graduate chairman of the journalism school, director of the Bush Research Endowment, and director of the Scripps Survey Research Center. He is a columnist for The Athens Messenger and The Highland County Press.