Kevin Rutan
19:169:SCA
Polling: A Science?
The word science alludes to many fields including chemistry, biology, geology and a plethora of other natural fields. It is harder to think of science within the fields of social studies, but the social and behavioral sciences are as true to the word as the natural science fields. They cannot be grasped in the same physical ways, but results from sources like polls are science. Particular phenomena like the “normal distribution” and the “nonresponse bias” are applied theories analogous to other fields and must be learned similarly (Gawiser and Witt 87).
For example, in a recent CBS poll covering the American people’s response to the economic bailout, 61 percent of Americans believe that the government should provide help to homeowners (CBS Inc). This information is more useful if you understand another scientific term applied to polls, “sampling error” (Gawiser and Witt 85). “Even a perfectly drawn and executed sample can provide a subset of the population that does not accurately reflect the population as a whole (Gawiser and Witt 85). Essentially, even if you survey 1000 people and they mirror the population’s proportions ethnically, socioeconomically, and politically the survey can still be wrong. In the CBS poll, the “sampling error” is plus or minus three percent, which means that the actual results range from 58-64 percent (CBS Inc).
Furthermore, in order to generalize these results, say from the CBS poll’s 1112 respondents to the country’s population, the researchers must use “Scientific probability sampling” (Asher 79). This ensures that the sample is random and representative, if “nonprobability sampling is used instead, than the results cannot be generalized (Asher 79). This would entail any sampling method that fails to pick subjects randomly. A few random street interviews, for instance, would not qualify as random samples despite the interesting nature of such interviews. In addition, any surveys in which the subjects seek the survey out, therefore selecting themselves to be polled, lacks a scientific basis as well.
Straw polls are an excellent example of this. In the 2008 caucuses, for instance, the race to pick a face for each respective political party was intense so of course every news medium wanted to outdo others in the industry by calling the winner first. On thegreenpapers.com, a collection of straw polls suggested that Mitt Romney would be the Republican Nominee in 2008. John McCain was projected as the nominee by one of the polls. As we know this was not the case, but the results of straw polls are inaccurate and this can be ascribed to the nature of nonscientific polling. This illustrates the importance of scientific polls and “sampling designs because they “tend to be more representative… because they generally avoid the selection biases inherent in nonprobability sampling” (Asher 80).
Overtime the internet has really allowed straw polls to flourish, but it has not been completely detrimental, as email has opened up an avenue for scientific samples as well. Email in particular has added an immense proliferation for polling that did not exist before. With telephones, for instance, there was always the problem of a large number of unlisted numbers. In 1996 “an estimated 29.6 percent of American households had unlisted numbers” (Asher 84). Furthering this, dialing randomly to add the possibility of reaching these people works, but often ends with confusion and the subject asking, “How did you get my number?” (Asher 85).
Overall, the science of polling is precisely that, a science. It requires the same patience and observation utilized in the other sciences and it requires the same use of the scientific approach to function properly. Scientific “sampling is the most problematic feature of public opinion polling,” but it is the heart as well (Asher 99).
http://www.cbsnews.com/stories/2009/02/23/opinion/polls/main4822664.shtml Poll: Public Wary of Bailouts. CBS Inc.
http://www.thegreenpapers.com/P08/StrawPolls.phtml. Election 2008 Straw Polls. Copyright 1999-2009. Richard A. Berg-Anderson.
Leave a comment