Jordan Underwood
MMPO
Dr. Yao
October 14, 2009
Costly Errors of the Pollsters
No matter how hard a reporter or PR professional tries, there will always be error in what they report. The facts may be skewed, the questionnaire may be faulty or the data collection process may have gone wrong. Despite the numerous reasons error may occur, there are very few ways to combat them. Gawiser and Witt said it best when they noted that it is not eliminating all errors, which is near impossible, but it is minimizing them so the story, poll, or facts are represented in the clearest way possible. The main points that will be discussed are the summary of Chapter 12 of the required reading, my personal experience with the text, and finally a poll I discovered from the Washington Post.
Over the years one aspect of polling has remained the same: there are errors contained within it. There are several ways that this error can be accumulated through question order, non-response, background noise, interviewer error and data processing error. The first one discussed was question order. It is essential for the interviewer or the questionnaire to ask legitimate questions that their respondents can easily understand them so the responses are true and the results accurate. One way this can be affected is the question order. According to Gawiser and Witt, “Placement of questions can really determine what the results . . .will be” (Gawiser, Witt 91). In other words, with incorrect placement, the results will be skewed and this could throw off the entire poll story or report of the results. The next source of error comes from non-response. Non-response is defined as the failure of a respondent or a group of respondents, to respond in the same proportions as the other. This is one of the most crucial kinds of errors since the public opinion poll was created. This is true due to the fact that if one group reports more than the other, the poll will be biased because the results are shifted in one way or another. For example, in 1936, Literary Digest was conducting a poll about the upcoming election and they sent ballots to all registered voters so everyone who was registered had an equal chance of selection. The problem was not with the sampling error, but the response error because more Republicans turned in the ballots than Democrats so the poll showed that Landon was ahead by 12%, which was obviously untrue. This makes non-response the most difficult for a journalist to handle which is why they need to be on their toes and make sure they are accurately tabulating the data as well as the people need to be skeptical of the polls and not take them to seriously.
The next crucial error is background noise. What this boils down to is the fact that the public may express opinions without having any knowledge on the issue this can alter results because unless something is to be done to ensure the public has adequate information, the results may be skewed in some way. The following source of error is one of the biggest and that is interviewer error. The interviewers play a key role because they need to be able to choose correct respondents, administer the questionnaire, how to ask the questions and how to record the answers as well as the overall interaction between them and the respondents. Therefore several large errors can occur such as the interviewer is unprepared, they are biased towards one issue or another, or they have a lack of knowledge about the subject No matter what the error is, if it starts with the interviewer, many problems will follow. The final greatest cause of error is data processing error. These errors consist of human and computer errors, however, most errors come on from simple mistakes. Examples of these simple errors can be that the question is worded wrongly (different on paper than on the questionnaire), conversion of responses from open-ended questions to specific answers, lack of knowledge required to do data analysis. Overall, there are several errors that need to be taken into account when addressing polling results.
In all my years as a journalist, one thing has always remained the same: the interviewing process was the most difficult to master. As a young, spunky newspaper reporter in high school I always had hard-hitting ideas backed up with solid facts and reasoning, but that was only half of what was needed to make the story complete. The other half relied on representation from the student body through the use of interviews. The most difficult part of the interview process was meeting the people. It was not hard to meet teachers or faculty members because I was accustomed to seeing them either in class, the hallways or elsewhere on the school grounds, which made it simple to interview them. The hardest part was interviewing someone I did not know or had no previous contact with. Usually the best stories were the ones that dug deeper under the surface and wasn’t your typical “homecoming dance” story or a story about the football team. Instead, it was about something under the surface and that entailed interviewing someone who received less publicity in their everyday routine. This can make it difficult to read the person or ask them accurate questions because you may not know how they will react. Another difficult aspect of it is trying to get them to say something you need without coming out and directly saying, “Answer the question this way . . .” A key moment that I remember about interviewing came from a movie I watched in a beginning journalism class in high school. The movie was titled “Shattered Glass.” It came out in 2003 and portrayed a young writer who at the young age of 20 rose to journalistic fame as he became a top reporter for The New Republic, a prestigious New York Newspaper. It displays how the young writer not only fabricated his stories, but all his interviewees were made up. This led to a huge scandal and his world crashing down. Overall, in my experience with journalistic writing, interview errors cannot happen, otherwise the fate of the newspaper, poll, magazine article, etc, will be sealed.
In one of the supplemental readings (Asher chapter 2), a key point was discussed that directly correlates with the required reading from Gawiser and Witt. This was the concept of non-attitudes. Non-attitudes are crucial for the pollster to pay attention to because the poll results may not be completely accurate even though the pollster asked valid questions and gained results. They may not be accurate because the results may either be biased or not take into account the “I don’t know” or no response answers to the questions. Another key aspect of non-attitudes is the idea that the respondent may answer the question with a predisposition to the question and that may throw off the results. In other words, the respondent may answer the question with a mass amount of bias because they have prior knowledge concerning that issue and feel a strongly about it. A key example of this can be found in today’s issue of the Washington Post Online. A poll was recently released concerning the opinion of the U.S. population if an increase in troops in Afghanistan is necessary. This poll found that 47% are in favor of the increase, while 49% are opposed to it. In the 49% of those that are opposed to the increase, it would be accurate to assume that many of them ha pre-existing knowledge about the war or are involved personally, whether that be a family member or a friend who is serving overseas. The idea of pre-existing knowledge can also be found in the 47% approved side and this is revealed in the following paragraphs when the pollster notes that both sides “strongly support their side.” This poll story is key to the understanding of non-attitudes and errors in polling because when a person has a prior knowledge of the topic presented in the poll, the results can be skewed or wrongly represented because the respondent has strong emotions towards them. This is also made clear by the poll story writer because they note that there is a chance of a 3% error on either side.
Costly Errors of the Pollsters
October 22, 2009 by iowajournalism
Leave a comment