The Method to the Madness: How the Scientific Method Relates to Polling

Votor Polling Sheet: Image by RadioFan

Polling is essential to gathering scientific data with which to make political predictions or develop theories.  Since political scenarios affect large groups of people, physical experiments and case studies are not as economical in political science as they are in other forms of science. Sampling allows scientists to calculate data representative of the entire population.  Researchers who incorporate the scientific method in the way they collect, analyze and test data are more likely to reach accurate conclusions.

 Observing and Collecting Data

The scientific method is a set of procedures relying on the assumption that what is true is also observable.  When creating a questionnaire, political scientists must concern themselves with studying the “how” and “why” behind a variable, rather than relying on assumptions, previously held beliefs, or judgment.  Survey questions should be straightforward, closed-ended (yes/no, or choice of answers provided) and worded without bias. When selecting a survey sample, Jon R. Bond, President of the Southern Political Science Association, says the scientist needs to prepare for the study while considering reliability and replication. The questions, wording and sample demographics should be put out in the open so other researchers may replicate the study and prove or disprove resulting conclusions or theories.  Since political opinions among populations are constantly in flux, individual surveys can produce drastically different results when replicated. Solid theories are often derived from patterns observed in aggregate collections of poll data.  To make the poll truly scientific, the scientist should comply with the mathematical laws of probability, which require a random sample where each person in the population being studied has equal chance of being included in the sample.

Famous pollster George Gallup helped shape the process in which polls are conducted. Image courtesy of the Gallup Organization.

Analyzing Survey Data

Phenomena revealed in raw survey data can be used to create a hypothesis.  The hypothesis should propose a relationship between two variables, such as “women turn out to vote more often than men” or “voters age 18 to 25 are more likely to vote for the Democratic candidate.” With these hypotheses, women and voters age 18 to 25 were one set of variables, and voting was the other.  Multiple hypotheses may be derived from a single poll.  The variables should be as specific as possible, for example, “voters age 18 to 25” instead of “young voters” or “first-time voters.”  Sometimes additional hypotheses may be developed to test for correlation or causation among a related set of variables. For example, a pollster may observe that voters age 18 to 25 were, in fact, more likely to vote Democratic, but also that they were more likely to earn less than $60,000 per year. So, a new hypothesis may be formed stating “voters who earn less than $60,000 per year are more likely to vote Democratic.” Then the testing process would confirm whether one or both of the stated hypotheses are accurate. According to political scientist Duncan McRae, there is often an alternative explanation for what we think we have confirmed.

Testing Poll Result Assertions

A hypothesis derived from poll data can be tested based on the observable variables outlined in the hypothetical statement.  In polling, testing is usually conducted in the form of additional polls, some of which may replicate the original poll, and some of which may include changes, which should be noted in the poll analysis.  Changes in variables may prove whether the original and new variables have significant relation, and changes in the sample group may prove whether the hypothesis s consistent among different demographic groups.  Once hypotheses have been verified, they can be used to formulate theories. Scientific theories explain why and how phenomena occur, and in politics they can be used to predict voter turnout, election results and public opinion.

Poll Predictions: An Educated Guess

The most conclusive polls are those that occur on election day. Image by Man vyi

Even when a single poll is conducted in a manner that is consistent with the scientific method, political predictions are most valid when they are compiled using multiple polls conducted over time. However, this can be challenging in politics, since the landscape is constantly changing, and public opinion may have drastically shifted before an accurate result can be tabulated. Unscientific polls, such as straw polls, internet polls and television polls are frequently presented in the media, since they can be created quickly and many who view them do not understand that they are less accurate than scientific polls.

Sources

Bond, Jon R. The Scientification of the Study of Politics: Some Observations on the Behavioral Evolution in Political Science. Texas A&M University. Accessed October 17, 2011.

Wadsworth Media. Political Science and Scientific Methods in Studying Politics.  Accessed October 17, 2011.

© Copyright 2011 Marissa Selner, All rights Reserved. Written For: Decoded Science