1.4: Political Science as a Social Science (2023)

The Scientific Method

Many social scientists employ the scientific method in the same ways that natural scientists do. Other social scientists are more in favor of rigorous interpretations, analyses, or in-depth case studies because historical events and contemporary social phenomena are too complex for simple causal models to address or because people are too self-aware to be measured and studied without distorting results. Nevertheless, all social scientists adhere to empirical, formalized methodologies. Science historian and philosopher Lee McIntyre argues that “the challenge in social science is to find a way to preserve our values without letting them interfere with empirical investigation. We need to understand the world before we can change it.” ⁽²⁾

You might have learned these scientific method steps in elementary school:

• Ask a question
• Research what others have learned about the question
• Formulate a hypothesis
• Conduct an experiment
• Collect and analyze data
• Communicate results

Regarding political science, the scientific method is a systematic, logically driven process to gather information and make conclusions about natural and social phenomena. Rather than focusing on an artificial step-by-step approach to understanding the scientific method, we can go into more detail on the features that distinguish science in political science from other ways of knowing.

The words empiricism, noun, and empirical, adjective, mean that scientists base their conclusions on careful verifiable observation and experience, rather than on intuition, revelation, prejudice, superstition, or anecdote. To westerners, empiricism is a cherished gift of the Renaissance and the Age of Enlightenment. For example, through his telescope, Galileo patiently observed four “stars” dancing around Jupiter, which led him to make the empirical statement that they were in fact moons orbiting around the planet. In addition, English physician Edward Jenner observed that farmhands who contracted cowpox earlier in life did not get smallpox, which led him to make the empirical statement that inoculating individuals with cowpox could protect them against smallpox. He tested this proposition on an 8-year-old boy named James Phipps. Phipps did not get smallpox. The result was the insight that inoculation made a person immune from the disease. These and many other examples illustrate empiricism’s power over other forms of knowing such as tradition or revelation.

Hypotheses, Concepts, and Variables

Aside from making careful and patient observations, the scientific method requires that we formulate hypotheses, conceptualize complex phenomena, and analyze constantly changing variables. Political scientists generate a hypothesis by asking a research question—an inquiry that asks how the political world operates or why it works the way it does.

The hypothesis posits an answer to the research question that you then test by conducting studies or experiments. The kinds of why or how questions that make good hypotheses are distinct from questions that elicit factual answers. For example, questions such as “What interests or organizations contribute the most money to political campaigns?” or “How many Supreme Court justices have been women?” are important. But they are the kinds of questions that typically elicit straightforward answers. Rather, here are some examples of large research questions in political science that make good hypotheses:

• Why does the United States—uniquely among advanced democracies—not have universal health coverage? A hypothesis might be that entrenched interests have been able to use the political system to block broad health coverage.
• Why do congressional incumbents have high reelection rates? A hypothesis might be that their financial advantage contributes greatly to their high reelection rate.
• How does the constitutional structure benefit some interests over others? A hypothesis might be that the constitutional structure privileges certain interests over others, particularly those who want to stop new policy over those who want to start it.
• How did conservatives go from spectacular defeat in 1964 to preeminence in all three government branches by 2000? A hypothesis might be that the conservative movement simply expanded to reflect real shifts in popular support on key issues that were favorable to the conservative point of view. In other words, shifts in public opinion caused the success of conservative politicians.

These kinds of questions are complex and require that scholars gather evidence from a variety of sources. Hypotheses must be supported systematically through a process of argumentation with political scientists who might disagree.

Not all hypotheses are the same. Here are the major categories of hypotheses:

Null hypothesis: There is no relationship between the two variables. On the question of money and incumbent reelection rates, the null hypothesis would be that there is no relationship between campaign budgets and chances of succeeding at the polls.

Correlative or correlational hypothesis: This simply suggests that two variables vary together. For example, a hypothesis may state that there is a relationship between religious fundamentalism and acts of terrorism. In doing so, the scientist is not speculating which variable is causing movement in the other.

Directional hypotheses: Since correlative hypotheses are not especially powerful, directional hypotheses posit a direction to the relationship in question. For example, a hypothesis could be that as religious fundamentalism increases, acts of terrorism increase. This is called a positive relationship—the value of one variable increases along with the value of another variable. A negative relationship involves the value of one variable decreases as the value of the other variable increases. For example, we might hypothesize that as personal income increases, willingness to support public transit decreases.

Causal hypothesis: Sometimes some of the variances in one variable is caused by the variance in the other variable. The two variables do connect. Causation is extremely difficult to establish. For example, let’s say that we could somehow measure the rise and fall of religious fundamentalism in the world and we have an accurate count of terrorist incidents over time. To establish causation, we would have to show a statistical relationship between the changing values for each variable and convince our readers of a valid link between the two variables. On top of this difficulty is the problem of social complexity. Rarely can a complex phenomenon such as terrorism be explained by one variable! Political scientists are much more likely to say that a certain percentage of the variance in terrorism can be explained by the variance in religious fundamentalism than they are to say that fundamentalism causes terrorism.

Hypotheses require the political scientist to conceptualize certain terms. A concept is a word or phrase that stands for something more complex or abstract. Political science is often concerned with big concepts such as liberty, democracy, power, justice, equality, war and peace, and representation. But there are many mid-level concepts in the discipline such as political development, political legitimacy, electoral realignment, and globalization. In addition, terms related to political ideologies—liberal, conservative, socialist, fascist, feminist, libertarian, and so forth—are also key concepts.

In turn, researchers need to define or operationalize concepts into measurable concrete variables. For example, earlier we hypothesized that as people’s income increases, their tendency to support public transit programs would decrease. How are we going to operationalize “income” as a variable that we can use in our analysis? We could ask a sample of people to tell us their income and then ask them questions about public transit. But, let’s say we wanted to rely on more concrete income records. We’ll still have questions to consider: gross income before taxes? Only wage income? Family income or individual income? As you can see, operationalizing concepts into measurable variables is not always easy.

A final comment about variables and testing hypotheses: the political scientist must control other relevant variables in the research design or methodology. For example, we might hypothesize that higher income causes people to tend to turn out to vote more, and indeed that’s what the data show. However, income also correlates well with higher formal education. How do we know whether we’re seeing the impact of income or education on voter turnout? We need to set a 'control' for education. One way to do that would be to sample only people with similar formal education levels and then break down the voting data by income within that educational stratum. Thus, we could look at only people with a bachelor’s degree but no graduate degree and see whether the tendency to vote within that group increases as personal income rises. Statisticians have developed mathematical techniques to control the effects of unwanted variables, but those techniques are beyond this textbook’s scope.

Experiments

Generally speaking, experiments come in two flavors: controlled and natural. A controlled experiment is one that is carefully set up by the scientist to control the variables that might affect the outcome, thereby isolating and evaluating the variable in which she is most interested. For example, we are interested in how conservatives and liberals respond to new information about health policy. We could survey 100 people from two groups: one conservative and one liberal. Then, we could develop an instrument to gauge the responses of conservatives and liberals. That instrument might be a knowledge questionnaire, a survey, or a behavior observation, depending on our hypothesis.

A natural experiment is an observational study in the real world where the scientist does not control the variables, but where natural processes or social events provide an opportunity for him to see the effect of a variable in action. Natural experiments are often compelling because they happen in the world around us rather than in a laboratory setting. For example, the Affordable Care Act—ACA or Obamacare—unintentionally created a natural experiment. The ACA required states to expand Medicaid to a larger percentage of poor people and funded them to do so. However, the Supreme Court struck down the mandate in 2012, thereby allowing states to choose whether or not to expand Medicaid. As it happens, states controlled by Republicans generally chose not to expand Medicaid, while states controlled by Democrats or that had a Democratic balance tended to expand Medicaid. Over a four-year period, researchers found that states that had expanded Medicaid reduced their mean annual mortality rate by 9.3 percent. Effectively, the 14 states that did not take advantage of the ACA had 15,600 people die who would not have died had the states expanded Medicaid. ⁽³⁾ This natural experiment allowed us to see the variable’s impact at the state level—was Medicaid expansion a net positive or negative on people’s health?

Falsifiability and Professional Responsibilities

Science’s emphasis on empiricism, conceptual clarity, variables, hypotheses, and experiments underscores another characteristic: falsifiability. Falsifiability—also known as testability—refers to the fact that scientific knowledge claims are subject to being proven wrong. Science philosopher Karl Popper argued that “A system is to be considered as scientific only if it makes assertions which may clash with observations: and a system is, in fact, tested by attempts to produce such clashes; that is to say, by attempts to refute it.” ⁽⁴⁾

Scientists make claims about the natural or social worlds and how they work. Those claims are so carefully documented that another scientist can either replicate the original study or marshal another set of observations with the explicit goal of testing whether or not the first scientist’s claim was correct. Systematically falsifying incorrect claims makes science progress toward greater understanding. We can empirically test a claim to see if it can either be refuted or confirmed.

What does an untestable claim look like? Namely, it is a theory that cannot be refuted. A couple of years before Charles Darwin published On the Origin of Species by Natural Selection in 1859, Gosse published a book called Omphalos: An Attempt to Untie the Geological Knot. Like Darwin, Gosse was trying to explain the increasing evidence that life had evolved over time. But Darwin used careful observations to explicate his theory of natural selection—a theory that was eminently falsifiable. Gosse, on the other hand, put forward a theory that God had created the currently existing plants and animals as well as fossils to look like evolution had taken place over a long period, but that in fact, God had created all life relatively recently. He reconciled his religious beliefs with empirical observations by developing a theory that could not be refuted.

When Darwin came along and wrote that finches on the Galapagos Islands had, through natural selection over time, modified their morphology to suit the kinds of things they ate on the various island ecosystems, Gosse’s adherents could simply say, “God just made the finches look that way.” Gosse’s claim is not falsifiable through any observation or experiment, whereas the theory of natural selection has passed literally thousands of tests for over 160 years. ⁽⁵⁾

Scientists of all stripes engage in common behaviors that support their work and to better understand each discipline’s study.

• professional conferences, scientists present their findings to their peers. There, they challenge each other, share new ideas and data sets, and develop common research interests around which they can collaborate.
• peer-reviewed journals publish peer-reviewed articles. If a political scientist sends a manuscript to International Studies Quarterly or any of dozens of political science journals, that manuscript will be farmed out to at least two other political scientists who have published in that field. They will review the manuscript and make comments on the methodology, the data, and the conclusions it offers. They will tell the editors of International Studies Quarterly whether the manuscript should be published, rejected, or sent back to the author for revisions. This is a blind process—the author of the manuscript does not know who is reviewing it, and the reviewers do not know who wrote the manuscript. The peer-review process is a robust way of ensuring credibility.

Political science is a member of the social sciences. While not all political scientists use the formal scientific method, they all adhere to empirical, falsifiable methods that are peer-reviewed. Political scientists at universities focus primarily on research and secondarily on teaching. Political scientists at community colleges focus primarily on teaching and secondarily on research.

What if . . . ?

What if we did a better job of developing scientific literacy among the American population? What impact would that have on our conversations about political issues that have scientific dimensions to them? How might those conversations be different? How would you go about promoting scientific literacy in America?

References

1. David J. Helfand, A Survival Guide to the Misinformation Age. Scientific Habits of Mind. New York: Columbia University Press, 2016. Page 22.
2. Lee McIntyre, The Scientific Attitude. Defending Science from Denial, Fraud, and Pseudoscience. Cambridge, MA: The MIT Press, 2019. Pages 193-194.
3. Sarah Miller, Sean Altekruse, Norman Johnson, Laura R. Wherry, “Medicaid and Mortality: New Evidence from Linked Survey and Administrative Data,” Working Paper No. 26081. The National Bureau of Economic Research. July 2019.
4. Karl Popper, Conjectures and Refutations: The Growth of Scientific Knowledge.London: Routledge, 2002. Page 345.
5. Donald R. Prothero, Evolution. What the Fossils Say and Why It Matters. New York: Columbia University Press, 2007, Page 9.