Applying Science’s Methods to Study Politics

Applying Science’s Methods to Study Politics
Political scientists use science’s methods to study questions as diverse as
the causes of war and the origins of public opinions. Studying political questions
in a scientific manner often involves the following:
• Specifying the question or problem with which the research is concerned
• Proposing a suitable explanation for the phenomena under study
• Identifying independent and dependent variables
• Specifying concepts thought to be useful in this explanation
• Formulating hypotheses and operationalizing concepts
• Clarifying measurement criteria
• Distinguishing between causation and correlation
• Thinking Logically
• Developing scientific theories.
Specifying the Research Question
One of the most important purposes of social scientific research is to
answer questions about social or political phenomena. A researcher might
attempt to answer questions about some important political behavior, such as:
Why are some political regimes more stable than others? Why do some
candidates for U.S. Congress receive more votes than other candidates do? Why
do public officials make the public policy decisions they do? And why do Supreme
Court justices take the positions they take on cases that come before them. In
every case, the researchers identified some political phenomenon that interested
them and tried to answer questions about the phenomenon.
The phenomena investigated by political scientists are diverse and are
limited only by whether they are significant (that is, would advance our
understanding of politics and government), observable, and political. Political
scientists attempt to answer questions about the political behavior of individuals
(voters, citizens, residents of a particular area, Supreme Court justices,
presidents), groups (political parties, interest groups, labor unions, ethnic
organizations), institutions (state legislatures, city governments, bureaucracies,
district courts), and nations.
The first task of a research is to specify the “why” question with which the
research is concerned. That will identify the phenomenon being investigated and
will point the research project in the direction of providing an explanation for that
phenomenon. Failure to specify the “why” question clearly can lead to confused
researchers as well as confused readers.
Students sometimes have difficulty formulating interesting and appropriate
research questions. Researchers also occasionally pose questions that are
simplistic, trivial, or impossible to answer. A research project will get off on the
wrong foot if the question that shapes it is inappropriate, unduly concerned with
discrete facts, or focused on reaching normative conclusions.
Political scientists seek knowledge about political phenomena. Although
the definition of political phenomena is vague, it does not include the study of all
human characteristics or behavior. For example, research studies guided by
questions such as “Why do some people drink coffee and others do not?” or “why
did dinosaurs become extinct?” might be interesting studies, but they would be
unlikely to yield fresh insights into political phenomena. Questions such as those
would be inappropriate for the study of politics and would be better addressed by
people in fields other than political science.
Research questions may also limit the significance of a research project if
they are unduly focused on discrete facts. Questions such as “Who is the
secretary of state?” or “Which interest groups contribute the most money to
political campaigns?” or “How many nuclear warheads does Russia possess?” may
yield important factual knowledge, but they will not sustain a research project,
While facts are important, they alone will not yield scientific knowledge—
knowledge about patterns, relationships, and explanations. Scientific knowledge
should be consistent with facts, but facts are not enough to yield scientific
explanations. Instead, political scientists are interested in advancing and testing
generalizations relating one phenomenon to another.
Another type of question that is inconsistent with political science research
methods is a question calling for a normative conclusion. Questions such as
“Should the United States deploy missiles in Europe?” or “Is the ‘actual malice
test’ too stringent a guideline for the solution of libel suits?” or “Was Bush Jr.’s
response to Katrina disaster racist and unfair to blacks?” are important and
suitable for the attention of political scientists (indeed, for any citizen), but they,
too, are inappropriate as presently framed. They ask for a normative or valuebased response, seeking an indication of what is good or of what should be done.
Proposing Explanations
Once a researcher has developed a suitable research question or topic, the
next step is to propose an explanation for the phenomenon the researcher is
interested in understanding. As noted in the previous section, these types of
questions are usually “why” questions in which the researcher is trying to explain
the variation in political phenomena—in other words, why political characteristics
and behavior occur at some times rather than others, or why they are likely to be
exhibited in some circumstances than in others.
To understand political behavior, political scientists propose explanations
for a phenomenon and then produce evidence bearing on these explanations.
Proposing an explanation involves identifying phenomena that we think will help
us account for the object of our research and then specifying how these two (or
more) phenomena are related. For example, political scientist Ted Gurr thought
political violence might be affected by a population’s sense of relative
deprivation. Garry Jacobson thought that congressional election outcomes might
be influenced by campaign spending. And Jeffrey Segal and Albert Cover tried to
find out if the personal attitudes of Supreme Court justices affect their judicial
Those phenomena that we think will help us explain the political
characteristics or behavior that interest us are called independent variables.
These types of variables are the measures of the phenomena that are thought to
influence, affect, or cause some other phenomena. Dependent variables are
thought to be caused, to depend upon, or to be a function of the independent
variables. Thus, if a researcher hypothesizes or posits that acquiring more formal
education will lead to increased income later on (in other words, that income may
be explained by education), then years of education would be the independent
variable and income would be the dependent variable. Likewise, if another
researcher posits that young people vote differently than older people (in other
words, that voting may be explained by differences in age), then age would be the
independent variable and voting the dependent variable.
Proposed explanations for political phenomena are often more complicated
that the simple identification of one independent variable that is thought to
explain a dependent variable. More than one phenomenon is usually needed to
account adequately for most political behavior. For example, suppose a
researcher proposes that an individual’s income and voting are related, with the
higher the income (independent variable), the more the individual is likely to vote
Republican (dependent variable). The insightful researcher would realize the
possibility that another phenomenon, such as the individual’s race, might also
affect his or voting choice. The proposed explanation for voting behavior, then,
would involve an alternative variable in addition to the original independent
variable, and the researcher would be interested in determining the relative
effect of each variable (income and race) on the dependent variable (voting
choice). This is done by “controlling for” or “holding constant” one of the
independent variables so that the effect of the other may be observed.
Sometimes researchers are also able to propose explanations for how the
independent variables are related to each other. In particular, we might want to
distinguish between which independent variables come before other
independent variables and indicate which have a more direct effect on the
phenomena we are trying to explain (the dependent variable). Variables that
occur prior to all other variables and that may affect other independent variables
are called antecedent variables. Variables that occur closer in time to the
dependent variable are themselves affected by other independent variables are
called intervening variables. The role of antecedent and intervening variables in
the explanation of the dependent variable differ significantly. Consider these
Suppose a researcher hypothesizes that a person who favors increased
military spending is more likely to vote for a Republican presidential candidate
than a person who does not favor military spending increases. In this case the
attitude toward military spending would be the independent variable and the
presidential vote the dependent variable. The researcher might wonder what
causes the attitude toward military spending and might propose that those
people who believe that Russia is an expansionist regime are more apt to favor
increased military spending. This new variable would then be an antecedent
variable, since it comes before and affects (we think) the independent variable.
Thinking about antecedent variables pushes our explanatory scheme further back
in time and, it is hoped, will lead to a more complete understanding of a political
phenomenon (in this case presidential voting). Notice how the independent
variable in the original hypothesis (favoring increased defense spending) becomes
the dependent variable in the hypothesis involving the antecedent variable (the
belief about Russia). Also, notice that in this example, beliefs about Russia are
thought to exert an indirect effect on the dependent variable (presidential voting)
via its impact on attitudes toward military spending.
Now let us consider a second example. Suppose a researcher hypothesizes
that the number of years of formal education affects one’s chances of turning out
to vote. In this case education would be the independent variable and voter
turnout the dependent variable. If the researcher then begins to think about what
it is about formal education that has this effect, he or she has begun to identify
the intervening variables between education and voter turnout. For example, the
researcher might hypothesize that formal education causes a sense of civic duty,
which in turn causes voter turnout, or that formal education causes an ability to
deal with bureaucratic details, which in turn causes voter turnout. Intervening
variables come between independent and dependent variables and help explain
the process by which one influences the other. The diagrams below present arrow
directions for the examples we just considered.
Beliefs about Russia Attitudes toward Military Spending Presidential Voting
(Antecedent variable) (Independent Variable) (Dependent Variable)