Prior to the November, 2000 presidential election George W. Bush had a comfortable lead in the polls. But on election day Bush ended up losing the popular vote by a small margin (although winning the electoral vote). Read the following article below written before the election:
Larry Kudlow (2000) The Weak-End of Polling, Retrived February 27, 2008, from http://www.nationalreview.com/kudlow/kudlow071800.html
After reading the article, please answer the following questions in length:
1. Why do you think different polls got such a wide rangeof estimates of Bush’s lead in the polls, ranging from a 12% lead to a 4% lead?
2. What samplingtechniques would you use to get the most accurate estimate of which candidate would win on election day?
1. Why do you think different polls got such a wide range of estimates of Bush’s lead in the polls, ranging from a 12% lead to a 4% lead?
In general, the differences between polls are due to biased samples of people chosen and biased questions based on their construction. Political polling can be an iffy prospect at the best of times. There are literally hundreds of pitfalls one can encounter when determining who to poll, what to ask, and how to ask it. One of the main factors in polling results that do not accurately reflect real outcomes is sampling errors. Within the realm of sampling errors are mistakes in timing, mode and method of polling, which, in turn lead to results that are not reflective of the actual numbers.
One mistake in sampling occurs as a result of timing of the survey. Phone surveys completed over weekends, for example, have a tendency to over-sample Democrats, and blue-collar workers. Additionally, the longer the survey, the more likely that a non-random sample will appear in responses. The average, random person does not have the inclination or time to participate in a lengthy survey. That being the case, the results often reflect the opinions of people with either unusual …
What are the best ways to sample individuals for opinion polls before election?