2 THE EMPIRICAL APPROACH

TO POLITICAL SCIENCE

e
b ut
P olitical scientists Jeffrey Winters and Benjamin Page wonder CHAPTER OBJECTIVES

tri
if the United States, despite being a nominal democracy,
is not in fact governed by an oligarchy, a relatively small num- 2.1 Identify eight characteristics of

is
ber of very wealthy individuals and families.1 Their work leads empiricism.
them to conclude:

rd
2.2 Discuss the importance of
theory in empiricism.
We believe it is now appropriate to . . . think about the
possibility of extreme political inequality, involving great 2.3 Explain the five steps in the

,o
political influence by a very small number of wealthy
individuals. We argue that it is useful to think about the
2.4
empirical research process.

Describe practical obstacles

st
US political system in terms of oligarchy.2
that challenge the empirical
approach.
po

What are we to make of a (perhaps startling) claim such as this?

How do we know it’s true? Should we accept it? 2.5 Summarize competing
As the title of our book and this chapter suggest, we have perspectives.
confidence in a statement like Winters and Page’s if they arrive
y,

at their (tentative) conclusion through empiricism. This term

is perhaps best explained by reference to an old joke.
op

Three baseball umpires are discussing their philosophy of calling balls and strikes. The
first umpire says, “I call ’em as I see ’em.” The next one replies, “That’s nothing. I call ’em as
tc

they are.” Finally, the third chimes in, “Oh yeah! Well, they ain’t nothing until I call ’em.”
We put aside Umpires 1 and 3 until later in the chapter. For now, let’s concentrate on
the second one. We call him a strict or strong empiricist. He believes there are in fact things
no

like balls and strikes, and he can always tell the difference by merely looking at the pitches as
they are thrown. He believes no interpretation is necessary; the facts (the pitches) speak for
themselves, and the umpire simply reports on where the ball travels, nothing more, nothing
o

less. Importantly, this umpire believes that his observations are accurate and objective. The
teams, players, managers, and fans have no bearing, he believes, on his judgments.3
D

An empiricist, in other words, uses impartial observation to judge the tenability of arguments.
A political science “umpire” demands that data and measurements support whatever point is
being made. Statements can be believed and accepted to the extent that they are derived from
empirical or observational evidence. If, on the other hand, their “truthfulness” depends on belief,
authority, or faith instead of “hard data,” they are set aside for philosophers and others to evaluate.

25
Copyright ©2020 by SAGE Publications, Inc.
This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
26   Political Science Research Methods

Empiricism is an ideal. Most who adopt this methodology would admit that personal
judgment plays a part in their research—they are perhaps closer to the first umpire, who
calls the game as he “sees it.” But so important is empiricism that we need to take a detour
to clarify why many political scientists prefer this methodology to other ways of obtaining
knowledge. Although not everyone agrees, it does seem to have a “privileged” place in the
discipline, and we need to explore its philosophical basis. This leads us to a discussion of the

e
scientific method.4

ut
Although empiricism does have a dominant place in contemporary political science,
we stress that it has its share of critics, and we certainly don’t maintain that it is the only or

b
even the best way to study politics. There is plenty of room, we believe, for different research
stances. Proponents of alternatives work under many different labels, so we simply classify

tri
them as nonempiricists.5 Furthermore, there are substantial debates among empiricists over
appropriate methods and approaches, particularly over the advantages of quantitative versus

is
qualitative analysis.6 We’ll have more to say about this in chapters 7 and 9.

rd
ELEMENTS OF EMPIRICISM

,o
What, then, distinguishes the empirical or scientific approach? In our daily lives, we “know”
things in many different ways. We know, for example, that water boils at 212 degrees
st
Fahrenheit and that a virus causes Ebola. We also may know that democracy is “better”
than dictatorship. In some cases, we know something because we believe what we read in
po

the newspaper or heard on the radio or what a trusted authority told us. In other cases, we
know things based on personal experience or because they appear to be consistent with
common sense.
y,

Modern political science, though, relies heavily on one kind of knowledge: knowledge
obtained through objective observation, experimentation, and logical reasoning.7 This way
op

of knowing differs greatly from information derived from myth, intuition, faith, com-
mon sense, sacred texts, and the like. It has certain characteristics that these other types of
knowledge do not completely share. The ultimate goal of scientific research, which is not
tc

always attained, is to use its results to construct theories that explain political phenomena.8
Scientific knowledge exhibits several characteristics. Most important, scientific knowl-
edge depends on verification. That is, our acceptance or rejection of a statement regarding
no

something “known” must be influenced by observation.9 Thus, if we say that people in the
upper classes have more political power than members of the lower strata, we must be able to
provide tangible evidence to support this statement.
o

A contention cannot be accepted simply because someone said so or our instinct tells
D

us so. It must be supported by evidence. The empirical nature of scientific knowledge dis-
tinguishes it from mystical knowledge. In the latter case, only “true believers” are able to
observe the phenomena that support their beliefs, and observations that would disprove
their beliefs are impossible to specify. Knowledge derived from superstition and prejudice
is usually not subjected to accepted methods of empirical verification, either. Superstitious
or prejudiced persons are likely to note only phenomena that reinforce their beliefs, while

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   27

ignoring or dismissing those that do not. Thus, their knowledge is based on selective and
biased experience and observation.
On the flip side, some philosophers of science insist that a key characteristic of scientific
claims is falsifiability, meaning the statements or hypotheses can in principle be rejected
in the face of contravening empirical evidence. A claim not refutable by any conceivable
observation or experiment is nonscientific. In this sense, the findings of science are usually

e
considered tentative, because they are “champions” only so long as competing ideas do not

ut
upend them. Indeed, the philosopher Karl Popper argued that scientists should think solely
in terms of invalidating or falsifying theories, not proving them.10

b
In view of the importance of verification and falsification, researchers must always
remain open to alterations and improvements of their research. To say that scientific knowl-

tri
edge is provisional does not mean that the evidence accumulated to date can be ignored or
is worthless. It does mean, however, that future research could significantly alter what we

is
currently believe. In a word, scientific knowledge is tentative, and because of this property,
empirical research is thought to be self-corrective.

rd
Scientific knowledge is supposedly “value-free.” Empiricism addresses what is, what
might be in the future, and why. It does not typically address whether or not the existence

,o
of something is good or bad, although it may be useful in making these types of determina-
tions. Political scientists use the words normative and nonnormative to express the distinc-
tion. Knowledge that is evaluative, value-laden, and concerned with prescribing what ought
st
to be is known as normative knowledge. Knowledge that is concerned not with evaluation
or prescription but with factual or objective determinations is known as nonnormative
po

knowledge. Most scientists would agree that science is (or should attempt to be) a nonnor-
mative enterprise.
This is not to say that empirical research operates in a valueless vacuum. A researcher’s
y,

values and interests, which are indeed subjective, affect the selection of research topics, peri-
ods, populations, and the like. A criminologist, for example, may feel that crime is a serious
op

problem and that long prison sentences deter would-be criminals. He or she may therefore
advocate stiff mandatory sentences as a way to reduce crime. But the researcher should test
that proposition in such a way that personal values and predilections do not bias the results
tc

of the study. And it is the responsibility of other social scientists to evaluate whether or not
the research meets the criteria of empirical verification. Scientific principles and methods of
observation thus help both researchers and those who must evaluate and use their findings.
no

Note, however, that within the discipline of political science, as well as in other disciplines,
the relationship between values and scientific research is frequently debated. We have more
to say about this subject later in the chapter.
o

An additional characteristic of scientific knowledge helps to identify and weed out prej-
D

udices (inadvertent or otherwise) that may creep into research activities.11 Scientific knowl-
edge must be transmissible—that is, the methods used in making scientific discoveries
must be made transparent so that others can analyze and replicate findings. The transmis-
sibility of scientific knowledge suggests “science is a social activity in that it takes several
scientists, analyzing and criticizing each other, to produce more reliable knowledge.”12
To accept results, people must know what data were collected and how they were analyzed.

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
28   Political Science Research Methods

A clear description of research procedures allows this independent evaluation. It also per-
mits other scientists in some cases to collect or obtain access to the same data and test the
original propositions themselves. If researchers use the same procedures but do not replicate
the original results, something is amiss, and the reasons for the discrepancy must be found.
Until then, both sets of results are suspect.
In an effort to improve the transparency in political science research, in 2012 the

e
American Political Science Association adopted a set of data accessibility and research trans-

ut
parency (commonly referred to as DA-RT) principles as part of its Guide to Professional
Ethics.13 The guidelines declared that “[r]esearchers have an ethical obligation to facilitate

b
the evaluation of their evidence-based knowledge claims . . . so that their work can be tested
or replicated.” Researchers are expected to meet this obligation in three aspects:

tri
1. Data access: Researchers making evidence-based knowledge claims should reference

is
the data they used to make those claims. If these are data they themselves generated
or collected, researchers should provide access to those data or explain why they

rd
cannot.
2. Production transparency: Researchers providing access to data they themselves

,o
generated or collected should offer a full account of the procedures used to collect or
generate the data.
st
3. Analytic transparency: Researchers making evidence-based knowledge claims
should provide a full account of how they draw their analytic conclusions from the
po

data (i.e., clearly explicate the links connecting data to conclusions).

Adherence to and implementation of these principles is not without controversy and debate
y,

among political scientists.14 In later chapters in which we address data collection and analysis,
we will discuss how these principles can be met. In some cases, this may pose substantial chal-
op

lenges and raise other research ethics issues.

tc

HELPFUL HINTS
no

TYPES OF ASSERTIONS

It is sometimes tricky to tell an empirical state- Empirical arguments, by contrast, often use varia-
o

ment from a normative one. The key is to infer tions of to be or direct verbs to convey the idea that
the author’s intention: Is he or she asserting that “this is the way it really is in the world.” Naturally,
D

something is simply the way it is, no matter what people occasionally believe that their values are
anyone’s preference may be? Or is the person stat- matters of fact, but scientists must be careful to
ing a preference or desire? Sometimes normative keep the types of claims separate. Finally, people
arguments contain auxiliary verbs, such as should often state opinions (beliefs) as if they were a mat-
or ought, which express an obligation or a desire. ter of fact in rhetorical sentences, as in “No tax

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   29

hike ever created a job.” Without verification, this is — Empirical: A verifiable assertion of
not an empirical statement. “what is”
When reading research reports or (even more
—— Normative: An assertion of “what should be”
important) when following political discussions in
the media, on the internet, or on the campaign trail, —— Rhetorical: A statement to the effect that
try to keep in mind that statements that seem to be “my belief is a fact”

e
of the same type can be surprisingly different:

b ut
This idea of transmissibility leads to another characteristic of empirical knowledge: it

tri
is cumulative, in that both substantive findings and research techniques are built upon
those of prior studies. As Isaac Newton famously observed of his own accomplishments, “I

is
have stood on the shoulders of giants.” He meant that the attainment of his revolutionary
insights depended in part on the knowledge other scientists had generated in the previous

rd
decades and centuries. The process of constantly testing and refining prior research pro-
duces an accumulated body of knowledge. (You’ll see examples of this fact in chapter 3,
which explains literature reviews.)

,o
Another important characteristic of scientific knowledge is that it is general, or appli-
cable to many rather than just a few cases. Advocates of the scientific method argue that
st
knowledge that describes, explains, and predicts many phenomena or a set of similar occur-
rences is more valuable than knowledge that addresses a single phenomenon or case.15 For
po

example, the knowledge that states with easier voter registration systems have higher elec-
tion turnout rates than do states with more difficult systems is preferable to the knowledge
that Minnesota has a higher turnout rate than does Alabama, for example. Knowing that
party affiliation strongly influences many voters’ choices among candidates is more useful to
y,

someone seeking to understand elections than is the simple fact that John Doe, a Democrat,
voted for a Democratic candidate for Congress in 2018. The empirical approach thus strives
op

for empirical generalizations, statements that describe relationships between particular sets
of facts.16 For example, the assertion that positive campaigns lead to higher voter turnout
tc

than do those that are characterized by mudslinging and name-calling is intended to sum-
marize a relationship that holds in different places and at different times. Furthermore,
many political scientists would assert that insofar as it is possible, it is important to quantify
no

the relationships—for example, by how much do more difficult voter registration systems
depress voter turnout, or how many fatalities might be avoided on average by states adopting
stringent vehicle safety inspection programs? Another characteristic of scientific knowledge
o

is that it is explanatory; that is, it provides a systematic, empirically verified understand-

ing of why a phenomenon occurs. In scientific discourse, the term explanation has various
D

meanings, but when we say that knowledge is explanatory, we are saying that a conclusion
can be derived (logically) from a set of general propositions and specific initial conditions.
The general propositions assert that when things of type X occur, they will be followed by
things of type Y. An initial condition might specify that X has in fact occurred. The observa-
tion of Y is then explained by the conjunction of the condition and the proposition. The goal

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
30   Political Science Research Methods

of explanation is, sometimes, to account for a particular event—the emergence of terrorism,

for example—but more often it is to explain general classes of phenomena such as wars or
revolutions or voting behavior.
Explanation, then, answers “why” and “how” questions. The questions may be specific
(e.g., “Why did a particular event take place at a particular time?”) or more general (e.g.,
“Why do upper-class people vote more regularly than, say, blue-collar workers?”). Observing

e
and describing facts are, of course, important. But most political scientists want more than

ut
mere facts. They are usually interested in identifying the factors that account for or explain
human behavior. Studies of turnout are valuable because they do more than simply describe

b
particular election results; they offer an explanation of political behavior in general.
An especially important kind of explanation for science is that which asserts causality

tri
between two events or trends. A causal relationship means that in some sense, the emer-
gence or presence of one condition or event will always (or with high probability) bring

is
about another. Causation implies more than that one thing is connected to or associated
with another. Instead, it means one necessarily follows the other. Chapter 1 touched on the

rd
issue of why economic inequality appears to be increasing in the United States. Some polit-
ical scientists, for instance, believe that “de-unionization” (the weakening of organized

,o
labor) has led to (caused) an increase in inequality in the United States. But is there, in fact,
a causal connection, or is the relationship merely fortuitous? Statements asserting cause and
effect are generally considered more informative and perhaps more useful than ones simply
st
stating that an unexplained connection exists. But they are difficult to establish. The issue
of how to design a research strategy in which an investigator is intending to demonstrate
po

causality is discussed in chapter 6.

Explanatory knowledge is also important because, by offering systematic, reasoned
anticipation of future events, it can be predictive. Note that prediction based on explanation
y,

is not the same as forecasting or soothsaying or astrology, which does not rest on empiri-
cally verified explanations. An explanation gives scientific reasons or justifications for why
op

a certain outcome is to be expected. In fact, many scientists consider the ultimate test of
an explanation to be its usefulness in prediction. Prediction is an extremely valuable type
of knowledge, since it may be used to avoid undesirable and costly events and to achieve
tc

desired outcomes. Of course, whether or not a prediction is “beneficial” is a normative ques-

tion. Consider, for example, a government that uses scientific research to predict the out-
break of popular unrest but uses the knowledge not to alleviate the underlying conditions
no

but to suppress the discontented with force.

In political science, explanations rarely account for all the variation observed in attri-
butes or behavior. So exactly how accurate, then, do scientific explanations have to be? Do
o

they have to account for or predict phenomena 100 percent of the time? Most political sci-
D

entists, like scientists in other disciplines, accept probabilistic explanation, in which it is not
necessary to explain or predict a phenomenon with 100 percent accuracy.
Scientists also recognize another characteristic of scientific knowledge: parsimony, or
simplicity. Suppose, for instance, two researchers have developed explanations for why some
people trust and follow authoritarian leaders. The first account mentions only the immediate
personal, social, and economic situation of the individuals, whereas the second account accepts

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   31

these factors but adds deep-seated psychological states stemming from traumatic childhood
experiences. And imagine that both provide equally compelling accounts and predictions of
behavior. Yet, since the first relies on fewer explanatory factors than does the second, it will
generally be the preferred explanation, all other things being equal. This is the principle of
Ockham’s razor, which might be summed up as “keep explanations as simple as possible.”

e
ut
THE IMPORTANCE OF THEORY

Theory plays an important role in research. The accumulation of observed relationships

b
sometimes leads to the creation of a theory—that is, a body of statements that system-

tri
atize knowledge of and explain relationships between phenomena. The process of reasoning
going from specific observations to a general explanation or theory is known as induction.

is
Two crucial aspects of empirical theory are that (1) it leads to specific, testable predictions,
and (2) the more observations there are to support these predictions, the more the theory is

rd
confirmed. You will notice that many, if not most, published examples of empirical research
are theory-driven; that is, the predictions or propositions about the relationships under
investigation are based on what a theory or theories would lead us to expect. The process of

,o
deduction entails reasoning from a general theory to a specific expectation.

An Example: Proximity Theory of Voting

st
To clarify some of these matters, let us take a quick look at an example. The “proximity
po

theory of electoral choice” provides a concise explanation for why voters choose parties and
candidates.17 Superficially, the theory may seem simplistic. Its simplicity can be deceiving,
however, for it rests on many years of multidisciplinary research18 and involves considerable
y,

sophisticated thinking.19 But essentially the theory boils down to the assertion that people
support parties and candidates who are “closest” to them on policy issues. Furthermore, this
op

theory would predict that candidates will try to position themselves so that they are closer to
more voters than are their opponents.
Take a particularly simple case. Suppose we consider the immigration debate. Positions
tc

on this issue might be arrayed along a single continuum running from, say, “All undocu-
mented immigrants should have a path to citizenship” to “All undocumented immigrants
should be deported” (see figure 2-1). Proximity theorists believe that both voters and candi-
no

dates (or parties) can be placed or located on this scale and, consequently, that the distances
or proximities between them (voters and candidates) can be compared. The theory’s predic-
tion is straightforward: an individual votes for the candidate to whom he or she lies closest
o

on the continuum.20
D

To expand a bit, theorists in this camp argue that (1) analysts using proper measurement
techniques can position both issues and candidates on scales that show how “close” they are
to each other and to other objects, and (2) voters vote for candidates who are closest (most
proximate) to themselves on such scales. People choose nearby candidates out of their desire
to maximize utility, or the value that results from one choice over another. Knowing this
fact, candidates adjust their behavior to maximize the votes they receive. Adjusting behavior

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
32   Political Science Research Methods

FIGURE 2-1 ■ Proximities on Immigration Issues

Proximities on Immigration Issues

Candidates

A B C

e
b ut
Amnesty Keep as is Deport

tri
is
Voter 1 Voter 2

Immigration Scale

rd
means not only taking or moving to positions as close as possible to those of the average or

,o
typical voter (the so-called median voter) but also, if and when necessary, obscuring one’s
true position (that is, following a strategy of ambiguity).21 Figure 2-1, for instance, shows
st
that Voter 1’s position is closest to Candidate B’s; therefore, Voter 1 would presumably vote
for that person. Similarly, Voter 2 would prefer Candidate C. Note also that Candidate A
po

could attract Voter 1’s support by moving closer to the middle, perhaps by campaigning on
an “amnesty-only-for-children-of-illegal-immigrants” platform.
The proximity theory has many of the characteristics of an empirical theory. Note that it
does not take a stance for or against one side or the other in the immigration debate. Rather,
y,

it explains why things happen as they do, and it offers specific and testable predictions. It
is also an implicitly causal theory in that it hypothesizes that the desire to maximize utility
op

“causes” voters to vote for specific candidates. It is general since it claims to apply to any
election in any place at any time. As such, it provides a much more sweeping explanation
tc

of voting than a theory that uses time- and place-bounded terms such as “the 2014 guber-
natorial election in Pennsylvania.” In addition, it provides a parsimonious or relatively sim-
ple account of candidate choice. It does not invoke additional explanatory factors such as
no

psychological or mental states, social class membership, current economic conditions, or

even partisanship to describe the voting act. Most important, although the proximity the-
ory rests on considerable formal (and abstract) economic and decision-making reasoning,
o

it puts itself on the line by making specific empirical predictions, which can be checked by
asking voters (1) their positions on immigration and (2) how they voted.
D

As a theory, it incorporates or uses numerous primitive or undefined terms such as issue,

candidate, and utility. These words and concepts may have well-accepted dictionary mean-
ings, but the theory itself takes their common understanding for granted. When a theory
is challenged, part of the dispute might involve slightly divergent interpretations of these
terms. At the same time, the theory makes explicit various other assumptions. It assumes,

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   33

among other things, that a researcher can place individuals on issue dimensions, that people
occupy these positions for reasonably long time periods, that voters are rational in that they
maximize utility, and that candidates have objective positions on these issues.22 Moreover,
by assumption, certain possibilities are not considered. The theory does not delve into the
question of whether or not a person holds a “correct” position on the scale, given his or her
objective interests. Finally, to test the proximity or spatial idea, researchers assume that one

e
can assign individuals meaningful spatial positions by asking certain kinds of questions on

ut
surveys or polls.23 This may be a perfectly reasonable assumption (we touch on that matter
in chapter 10), but it is an assumption nevertheless. Still, spatial modelers, as those who use

b
proximity theory are called, go to great lengths to define and explain key concepts. How
distance is defined is a serious matter because different definitions can lead to different sub-

tri
stantive conclusions.24

is
The Explanatory Range of Theories

rd
Theories are sometimes described by their explanatory range, or the breadth of the phenom-
ena they purport to explain. Usually, one does not have a theory of “why Donald Trump
won the election in 2016.” (It is, of course, possible to find several theories that account for

,o
this particular outcome. But note that the 2016 election results are an instance, or “token,”
of the kind of event with which these theories deal.) Instead, a good theory of electoral
outcomes presumably pertains not only to a specific presidential contest but also to other
st
presidential elections or other types of elections in other times and places.
In the social sciences, so-called narrow-gauge or middle-range theories pertain to lim-
po

ited classes of events or behaviors, such as a theory of voting behavior or a theory about the
role of revolution in political development.25 Thus, a theory of voting may explain voter
turnout by proposing factors that affect people’s perceptions of the costs and benefits of
y,

voting: socioeconomic class, degree of partisanship, the ease of registration and voting laws,
choices among candidates, availability of election news in the media, and so forth.26 Global
op

or broad-range theories, by contrast, claim to describe and account for an entire body of
human behavior. A theory of federalism might explain why subnational governments do
not adopt redistributive programs (those that redistribute wealth from wealthier residents
tc

to poorer ones) as much as one might expect based on need.27 A really general theory, for
example, might attempt to account for increases or decreases in economic inequality in
no

any society at any time.28 In short, theories play a prominent role in natural and social sci-
ences because they provide general accounts of phenomena.29 Other things being equal, the
broader the range of the things to be explained, the more valuable the theory.
o

A BRIEF OVERVIEW OF
D

THE EMPIRICAL RESEARCH PROCESS

So what exactly is the empirical or scientific research process? In reality, no scientist in
the field or laboratory adheres to a prescribed set of steps like someone following a script.
Scientists rely not just on formal procedures but also on intuition, imagination, and even

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
34   Political Science Research Methods

luck at times. Nevertheless, we may conceptualize what they do by identifying the underly-
ing logic of their activities. Here is an idealization of a scientific research program.

Development of an Idea to Investigate or a Problem to Solve

A scientist gets topics from any number of sources, including literature about a subject,
a general observation, an intuition (or hunch), the existence of conflicts or anomalies in

e
reported research findings, and the implications of an established theory. For example, a

ut
report on income inequality may indicate that it varies considerably from country to coun-
try or that it is increasing. A logical response would be to ask why. As another instance,

b
consider newspaper accounts that suggest that evangelical Christians tend to support con-

tri
servative candidates because of “moral values.” Several research questions are raised by these
accounts: Do evangelicals base their choices of candidates on their proximity to candidates’

is
positions on moral issues while other voters base their choices on other types of issues such
as economic issues? Is turnout among evangelicals higher in elections where there are dis-

rd
tinct differences between candidates on moral issues than in elections where the differences
are small?

Hypothesis Formation
,o
After selecting a topic, an investigator tries to translate the idea or problem into a series
st
of specific hypotheses. As we see in chapter 4, hypotheses are tentative statements that, if
confirmed, show how and why one thing is related to another or why a condition comes
po

into existence. These statements have to be worded unambiguously and in a way that their
specific claims can be evaluated by commonly accepted procedures. After all, one of the
requirements of science is for others to be able to independently corroborate a discovery. If
assertions are not completely transparent, how can someone else verify them? In the preced-
y,

ing example, we might hypothesize that evangelical Christians are more likely than others
to base their vote on candidates’ positions on moral issues.
op

“Data” Collection
tc

This is where the rubber meets the road: the essence of science comes in the empirical testing
of hypotheses through the collection and analysis of data. Consider the case of religion and
voting just mentioned. We need to define clearly the concepts of moral values and evangelical
no

Christian. We might, for instance, tentatively identify evangelicals as people who adhere to
certain Christian denominations and moral values as attitudes toward abortion and same-
sex marriage. A researcher could write a series of questions to be administered in a survey or
o

a poll to elicit this information. Only when concepts are defined and decisions made about
D

how to measure them can data collection and analysis begin.

Interpretation and Decision

At some point, the investigator has to determine whether or not the observed results are
consistent with the hypotheses. Though simple in principle, judging how well data support

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   35

scientific hypotheses is usually not an easy matter. Suppose, for example, we find that 75
percent of evangelical Christians opposed same-sex marriage and 90 percent of these indi-
viduals voted for a House candidate in 2014 who opposed same-sex marriage. So far, so
good. But suppose, in addition, that 70 percent of nonevangelicals also opposed same-sex
marriage and that more than 90 percent of these people also voted for House candidates
opposed to same-sex marriage in the same election. It appears that attitudes might be affect-

e
ing voting, but the data do not necessarily establish a connection between religious prefer-

ut
ence and whether or not votes are based on moral issues. Weighing quantitative or statistical
evidence requires expertise, practice, and knowledge of the subject matter, plus good judg-

b
ment (and this skill is often difficult to teach). Still, chapters in this book are devoted to
showing ways to make valid inferences about tenability of empirical hypotheses.

tri
Modification and Extension

is
Depending on the outcome of the test, one can tentatively accept, abandon, or modify the

rd
hypothesis. If the results are favorable, it might be possible to derive new predictions to
investigate. If, however, the data do not or only very weakly support the hypothesis, it will
be necessary to modify or discard it. Let us stress here that negative results—that is, those

,o
that do not support a particular hypothesis—can still be both interesting and beneficial.30
As we suggested earlier, some scholars, such as Popper, believe that science advances by
disproving claims, not by accepting them. Consequently, a valuable contribution to science
st
can come from disconfirming widely held beliefs, and the only way to do that is to replicate
or reinvestigate the research upon which the beliefs rest. The key is not so much the result of
po

a hypothesis test but how substantively important the hypothesis is to begin with.

REACTIONS TO THE EMPIRICAL

y,

APPROACH: PRACTICAL OBJECTIONS

op

Empirical research problems arise because many important concepts are abstract or have
many meanings or are value-laden. Chapter 1 showed that a concept such as “talking like
tc

a man” needs to be defined carefully and clearly, and that finding an adequate definition
of “economic inequality” can be difficult. Should we be looking at individuals or house-
holds? Should we use annual income—calculated before taxes, after taxes, or after adding
no

to individual or household income publicly supplied in-kind benefits such as health care or
job retraining? Or should we try to measure net wealth (assets minus debts)? The following
chapters take up some of these questions.
o

Furthermore, political scientists must face the fact that human behavior is complex,
perhaps even more complex than the subject matter of other sciences (genes, subatomic
D

particles, insects, and so on). Complexity has been a significant obstacle to the discovery of
general theories that accurately explain and predict almost every kind of behavior. After all,
developing a theory with broad applicability requires the identification and specification of
innumerable variables and the linkages among them. Consequently, when a broad theory is
proposed, it can be attacked on the grounds that it is too simple or that too many exceptions

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
36   Political Science Research Methods

to it exist. Certainly, to date no empirically verified generalizations in political science

match the simplicity and explanatory power of Einstein’s famous equation E = mc 2.31
There are still other obstacles. The data needed to test explanations and theories may
be extremely hard to obtain. Indeed, often the potentially most informative data are totally
unavailable. People with the needed information, for example, may not want to release it for
political or personal reasons. Pollsters, for instance, find refusal to answer certain questions,

e
such as those designed to measure attitudes toward ethnic groups, to be a major problem in

ut
gauging public opinion. Similarly, some experiments require manipulation of people. But
since humans are the subjects, the researchers must contend with ethical considerations

b
that might preclude them from obtaining all the information they want. Asking certain
questions can interfere with privacy rights, and exposing subjects to certain stimuli might

tri
put the participants at physical or emotional risk. Tempting someone to commit a crime, to
take an obvious case, might tell a social scientist a lot about adherence to the law but would

is
be unacceptable nevertheless.

rd
Self-Reflection and Individuality
Like any other organisms, humans are aware of their surroundings. They have the addi-

,o
tional ability to empathize with others and frequently attempt to read others’ minds. As John
Medearis put it, “human beings—individually, but especially jointly—are self-interpreting
and reflective, capable of assigning meanings to their actions and revising these meanings
st
recursively.”32 Observations of this sort led many social scientists and philosophers to question
whether or not the scientific method can be applied to the study of something as intrinsically
po

language based as politics. This doubt appears later in the chapter, when we discuss interpre-
tation versus explanation. In the meantime, let us point to a practical problem. Since humans
are self-reflective and empathetic creatures, they often anticipate a researcher’s goals and
y,

adjust their actions accordingly (e.g., “The investigator seems to favor immigration reform,
so I will too”).
op

When it comes to studying political behavior such as voting or decision making, another
difficulty arises. Many experiments in science assume that the entities under investigation are
for all intents and purposes identical and, hence, can be interchanged without fear of compro-
tc

mising the conclusion. An iron ion (Fe+) from one source is as good as another from somewhere
else (no matter where in the universe) when it comes to studying iron’s reaction with oxygen.
no

But can the same be said of humans? Consider a political scientist who wants to investigate the
effects of negative campaign advertising on attitudes. Suppose Jane and Mary are subjects in a
study. We cannot assume that they will react to the experimental stimuli exactly the same way,
even though they are the same age, gender, political persuasion, and so forth.
o

Social scientists have to get around this problem by using groups or samples of individu-
D

als and then examining the average effect of the stimulus. Any generalization that results has
the form: given subjects with characteristics A, B, . . . , X (the stimulus) on average affects Y
(the response) by approximately N units. In other words, sometimes the basic units under the
scientist’s microscope can be considered pure, even if they are complex molecules, but not
so in political science. The objects political scientists study are multifaceted and conscious

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   37

beings with volition of their own who often change opinions and behaviors; thus, state-
ments about them must necessarily be tentative, general, and time bound.
Finally, there is the inescapable subjectivity of politics. We provide an example that
bedevils research into the studies of power. Most political scientists would agree that, if an
oligarchy exists in the United States, it should at a minimum make or heavily influence key
policy decisions. The problem is, how does one objectively identify “key” policies? Should

e
the choice be left to the judgment of the researcher or knowledgeable/informed experts?

ut
Or are there concrete indicators or measures of importance? Suppose we want to classify
decision A as “important.” On what grounds do we make the assignment? The number of

b
people A affects? Its cost? The number of times it is mentioned in the press? Its length in
legal codes? The number of times it is litigated? Any or all of these might be useful. But for

tri
a variety of reasons, none of these may capture the significance (or lack of significance) of a
decision. Importance often comes from how people interpret or understand policy A, and

is
understanding of this sort, many assert, lies beyond the scope of empirical sciences.33
All of these claims about the difficulty of studying political behavior scientifically may

rd
have merit. Yet they can be overstated. Consider, for example, that scientists studying natural
phenomena encounter many of the same problems. Physicists cannot directly observe ele-

,o
mentary particles such as quarks. Nor can astronomers and geologists carry out experiments
on most of the phenomena of greatest interest to them. Indeed, they cannot even visit many
of the places they study most intensively, like other planets or the center of the Earth. And
st
what can be more complex than biological organisms and their components, which consist of
thousands of compounds and chemical interactions? Stated quite simply, it is in no way clear
po

that severe practical problems distinguish political science from any of the other sciences.

Is Political Science Trivial or Irrelevant?

y,

The empirical approach in political science, with its advent in earnest in the 1960s, seemed
to bring with it all the accoutrements of rigorous natural sciences: equations and math-
op

ematical models, statistical analysis, instrumentation and quantification, computers and

electronic databases, esoteric concepts (e.g., “multidimensional issue spaces”). Yet practi-
cally from the moment the empirical or scientific perspective arrived on the scene, doubters
tc

and skeptics appeared. In the late 1960s and later in 2000, well-publicized “revolts” against
hard-core empiricism took place. Among other complaints, critics pointed to the trivial
no

nature of some of the “scientific” findings and applications. Common sense would have
told us the same thing, they argued. Of course, as we explained earlier, there is a difference
between intuition or common sense and scientific knowledge. To build a solid base for fur-
ther research and accumulation of scientific knowledge in politics, commonsense knowl-
o

edge must be verified empirically and, as is frequently the case, discarded when wrong. Still,
D

“scientism” left many political scientists dismayed.

A more serious criticism of the scientific study of politics is that it leads to a failure to
focus enough scholarly research attention on important social issues and problems. Some
critics contend that, in the effort to be scientific and precise, political science overlooks the
moral and policy issues that make the discipline relevant to the real world. Studies rarely

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
38   Political Science Research Methods

address the implications of research findings for important public policy choices or political
reform. In other words, the quest for a scientific knowledge of politics has led to a focus on
topics that are quantifiable and relatively easy to verify empirically but that are not related to
significant, practical, and relevant societal concerns.34 A related criticism is that researchers
are using increasingly sophisticated statistical methods to investigate politics. Understanding
these methods, and hence being able to discuss results, is beyond the reach of many political

e
scientists, not to mention students and the general public. This has deleterious effects on

ut
the level of discourse about what is known about important political and social issues.35 These
considerations take us back to our umpires. Can researchers really emulate Umpire 2 (the

b
strict empiricist) who claims to “call ’em as they are”? Many think not. Political scientists,
having been exposed to decades of philosophizing about limitations and problems of the

tri
“scientific method,” probably now admit to being like Umpire 1 and call balls and strikes as
they see them. This doesn’t mean their research is totally subjective or a matter of opinion; but

is
it is, they realize, so contingent on time, place, language, and culture that finding scientific
laws and truths of politics is problematic. Instead of calling them hard-nosed empiricists, we

rd
might better call today’s political scientists modest or constrained empiricists.

COMPETING POINTS OF VIEW

,o
As widely accepted and useful as science has become in modern times, serious philosophers
st
and social scientists have challenged these premises. We cannot explain all of their objec-
tions here, but the essence of their argument is that certain aspects of human life are simply
po

not amenable to systematic and objective analysis. More important, an uncritical faith in
realism, objectivity, and material causality is unwarranted. We concentrate on two points:
y,

1. Human actions cannot be explained scientifically but must be interpreted from the
point of view of the actors. Meaning and understanding are the proper goals.
op

2. Social scientists have to realize that the world, far from having an independent
existence that they observe directly, is partly constructed by observers themselves.
tc

To oversimplify, we shall say these two viewpoints constitute “nonempiricism.”

no

HELPFUL HINTS
ASSUMPTIONS OF EMPIRICAL RESEARCH
o
D

— An empiricist (“I-call-’em-as-they-are” umpire) —— Materialism: Only concrete and observable

makes assumptions about methodology. (if only indirectly) entities have causal
efficacy.
—— Realism: There is a real world that exists
independently of observers. (It’s there even if —— Denial of supernatural causes: Explanations
we aren’t there to see it.) of phenomena based on mysterious,

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   39

unknowable, unobservable, “hidden” forces experience or data. (Don’t take anything on

are unacceptable. faith alone.)
—— Regularity: Natural phenomena (human —— Irrelevance of preferences: To the maximum extent,
behavior and institutions) exhibit regularities one’s values and biases should not affect the
and patterns that can be revealed by reason decision to reject or accept an empirical claim.
and observation.

e
—— Theory and causal explanation: The goal
—— Verification and falsification: Statements about of science is to create general, verified

ut
the world must be verified or falsified by explanatory theories (even laws).

b
tri
Interpretation

is
Some people question the empirical strategy because the subject matter, human institutions

rd
and activities, differs from the behavior of material objects such as atoms or stars, and these
differences raise all sorts of complexities. One indicator of the inapplicability is that prog-
ress in developing and testing contingent causal laws has been agonizingly slow.36 Moreover,

,o
both the methods and the content of the discipline have not come close to the exactitude and
elegant sophistication of sciences such as biology or physics, and, consequently, nowhere
can we find empirical generalizations with the level of precision and confirmation enjoyed
st
by, say, the theories of relativity and evolution.
Skeptics argue that there are good reasons for this outcome. Since politics inescapably
po

involves actions—that is, behavior that is done for reasons—and not mere physical move-
ment, analyzing it brings up challenges not encountered in the natural sciences. Opponents
of the empirical approach claim that scientific methods do not explain nearly as much about
y,

behavior as their practitioners think. The problem is that to understand human behavior,
one must try to see the world the way individuals do. And doing so requires empathy, or the
op

ability to identify and in some sense experience the subjective moods or feelings or thoughts
of those being studied. Instead of acting as outside, objective observers, we need to “see” how
individuals themselves view their actions. Only by reaching this level of understanding can
tc

we hope to answer “why” questions such as “Why did John still vote in the last election even
though he was bombarded by countless attack ads on television, the internet, radio . . . every-
where he turned?” The answers require the interpretation of behavior, not its scientific expla-
no

nation in terms of general laws. In short, interpretation means decoding verbal and physical
actions, which is a much different task than proposing and testing hypotheses.
Given this way of looking at the research task, some social scientists advocated stressing
o

the interpretation or empathetic understanding of actions and institutions. One of the ear-
D

liest and best-known proponents of this methodology was Clifford Geertz, an anthropolo-
gist, who felt that “man is an animal suspended in webs of significance he himself has spun.
I take culture to be those webs, and the analysis of it to be therefore not an experimental
science in search of law but an interpretive one in search of meaning.”37 As a simple example
of the difference between empirical and interpretative approaches, take journalist James
O’Toole’s analysis of a close Pennsylvania U.S. Senate election in 2010: “it’s now a pretty

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
40   Political Science Research Methods

close race, according to the polls and the body language of the campaigns.”38 Here he relies
on both an empirical tool (polling) and intuition (the “body language of the campaigns”).
Those who closely follow electoral politics would perhaps agree that a minimum of interpre-
tation and subjective analysis is always helpful.
Another way of looking at interpretation is to consider the concept of social facts. What
exactly are things like political parties, elections, laws, and administrative regulations? In

e
what sense are they real? They do not have the same kind of material existence as atoms,

ut
bacteria, and mountains, but have an entirely subjective existence only in the minds of peo-
ple living in a particular culture. One philosopher remarks that “minds create institutions.

b
There would be no money or marriage or private property without human minds to create
these institutions.”39 How, then, should they be studied? The sociologist Émile Durkheim

tri
told his students to take them seriously: “the first and most basic rule [of social inquiry] is to
consider social facts as things.”40 And many political scientists almost instinctively adhere to

is
that principle. Nonetheless, the notion that much of what is studied is socially constructed
raises some thorny epistemological issues.

rd
Constructionism and Critical Theory

,o
Most political scientists take reality pretty much as a given. That is, they posit that the
objects they study—elections, wars, constitutions, government agencies—have an existence
independent of observers and can be studied more or less objectively. But an alternative per-
st
spective, called the social construction of reality or constructionism,41 casts doubt on this
uncritical, perhaps blasé attitude. According to constructionism, humans do not simply
po

discover knowledge of the real world through neutral processes, such as experimentation or
unbiased observation; rather, they create the reality they analyze. This position is perhaps
another way of saying, “Facts do not speak for themselves but are always interpreted or con-
y,

structed by humans in specific historical times and settings.” This stance may be likened to
Umpire 3, who you may recall says the phenomena under investigation “ain’t nothing until I
op

call ’em” as though the very act of umpiring creates its own reality.
One version of this position admits that entities (for example, molecules, planets) exist
separately from anyone’s thoughts about them, but it also insists that much of what people
tc

take for granted as being “real” or “true” of the world is built from learning and interaction
with others and does not have an existence apart from human thought.42 Consider the term
Democratic Party. Instead of having an independent, material existence like an electron or a
no

strand of DNA, a political party exists only because citizens behave as if it exists. This means
that two individuals who come from different social, historical, and cultural backgrounds
may not comprehend and respond to the term in the same way. What is important in study-
o

ing, say, individuals’ responses to Democratic candidates is fathoming their personal beliefs
D

and attitudes about the party.

Constructionist thinking now plays a strong role in international relations theory, where
a concept such as anarchy is not considered a “given and immutable” cause of the behavior
of states (for example, their desire for security through power politics). Rather, concepts like
this one have to be understood in terms of what actors (individuals, states) make of them.43

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   41

The constructionist viewpoint, which comes in innumerable varieties, challenges the

idea of an objective epistemology, or theory of knowledge. Such ideas, however, are of a
deeply methodological nature and raise deep philosophical issues that go well beyond the
task of describing the empirical methods used in the discipline.44 We thus acknowledge that
the scientific study of politics is controversial but nevertheless maintain that the procedures
we describe in the chapters that follow are widely accepted and can in many circumstances

e
lead to valuable understandings of political processes and behaviors. Moreover, they have

ut
greatly shaped the research agenda and teaching of the discipline, as can be seen by looking
at the evolution of the field in the twentieth century.

b
The emergence and domination of the empirical perspective have also brought about
renewed interest in normative philosophical questions of “what ought to be” rather than

tri
“what is.”45 Part of the discipline has become receptive to variations of critical theory, or
the belief that a proper goal of social science is to critique and improve society (by mak-

is
ing it more just and humane) rather than merely understand or explain what is going on.
Critical theorists feel, in other words, that simply analyzing a polity as it is amounts to a

rd
tacit endorsement of its institutions and the distribution of power. Contrary to the idea that
science should be value-free, critical theorists argue that proposing and working for reforms

,o
are legitimate activities for the social sciences. They therefore analyze institutions, practices,
ideologies, and beliefs not only for their surface characteristics but also for their “hidden
meanings” and implications for behavior.
st
Take, for example, the statement “I’m just not interested in politics.”46 An empiri-
cal political scientist might take this simply as a cut-and-dried case of apathy. He or she
po

might then look for variables (e.g., age, gender, ethnicity) associated with “not interested”
responses on questionnaires. A critical theorist, by contrast, might ask, “Does this person
really have no interest in current events? After all, isn’t everyone affected by most political
y,

outcomes, like decisions about taxes, war and peace, and the environment, and thus in fact
have an interest in politics? So perhaps we have a case of, say, ‘false consciousness,’ and it
op

is crucial to uncover the reasons for lack of awareness of one’s ‘real’ stake in politics. Is the
indifference a matter of choice, or does it stem from the (adverse) effects of the educational
system, the mass media, modern campaigning, or some other source?”
tc

Here is another case. An important challenge to research in political science (as well
as in other social science disciplines, such as sociology) has come from feminist scholars.
Among the criticisms raised is that “the nature of political action and the scope of political
no

research have been defined in ways that, in particular, exclude women as women [emphasis
added] from politics.”47 Accordingly, “what a feminist political science must do is develop
a new vocabulary of politics so that it can express the specific and different ways in which
o

women have wielded power, been in authority, practiced citizenship, and understood free-
D

dom.”48 Even short of arguing that political science concepts and theories have been devel-
oped from a male-only perspective, it is all too easy to point to examples of gender bias in
political science research. Examples of such bias include failing to focus on policy issues of
importance to women, assuming that findings apply to everyone when the population stud-
ied was predominantly male, and using biased wording in survey questions.49

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
42   Political Science Research Methods

A related complaint is that political science in the past ignored the needs, interests, and
views of the poor, the lower class, and the powerless and served mainly to reinforce the belief
that existing institutions were as good as they could be. Those who agree with this com-
plaint are called “critical theorists.” Concerns about the proper scope and direction of polit-
ical science have not abated, although nearly all researchers and teachers accept the need to
balance the scientific approach with consideration of practical problems and moral issues.50

e
Let’s wrap up our discussion so far before returning to the all-important question: What

ut
difference does all this philosophizing make? Table 2-1 lists some of the key differences
between what we have been calling the empirical and nonempirical schools.

b
tri
is
TABLE 2-1 ■ Methodological Perspectives in Political Science

rd
Nonempirical Empirical

Goals To understand behavior Causal explanations and predictions of individual

and institutional behaviors

,o
To interpret actions
General theory and laws
Information of practical use
st
“Value-free” knowledge
po

Assumptions Social facts (at least) are Realism (appearance and reality are the same).
“constructed.”
Independent, objective observation is possible.
Institutions are social
Behavior and, implicitly, institutions exhibit
creations.
regularities.
y,

Objective observation is not

Claims about the real world must be verified.
generally possible because
op

our very senses are affected Attitudes (values, biases, beliefs) must not affect
by culturally defined and observation and analysis.
imposed prior beliefs. There are no causeless effects.
tc

Totally value-free research

is impossible.
no

Basic tool kit Qualitative Primarily quantitative

Methods Qualitative analysis (e.g., Field studies and observation, content and
ethnography, content and document analysis
document analysis, study of
Case studies and comparisons
o

discourse)
Experiments and field experiments
D

Case studies and

comparisons Mathematical models
Surveys
Statistical analysis of data
Simulations

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   43

Nonempirical Empirical

Objections Observation is Takes “politics out of political science.”

impressionistic, subjective,
Concentration on formalism, quantitative
and nonsystematic.
measurement, and mathematical analysis leads to
Knowledge is trivial and practically meaningless results.

e
“nontransmissible.”
Findings are tainted by the

ut
investigator’s values and
biases.

b
Alleged biases Conclusions are affected Inherently favors the status quo and existing

tri
by political and social power structures.
ideologies.

is
Source: This table is based partly on tables in Colin Hay, Political Analysis: A Critical Introduction (New York: Palgrave, 2002),

rd
chap. 1.

CONCLUSION
,o
In this chapter, we described the characteristics of scientific knowledge and the scientific
st
method. We presented reasons why political scientists are attempting to become more scien-
tific in their research and discussed some of the difficulties associated with empirical polit-
po

ical science. We also touched on questions about the value of the scientific approach to
the study of politics. Despite these difficulties and uncertainties, the empirical approach is
widely embraced, and students of politics need to be familiar with it. In chapter 3, we begin
y,

to examine how to develop a strategy for investigating a general topic or question about
some political phenomenon scientifically.
op

TERMS INTRODUCED
tc

Actions. Human behavior done for a reason. 39 Cumulative. Characteristic of scientific knowledge;
no

new substantive findings and research techniques are

Constructionism. An approach to knowledge that
built upon those of previous studies. 29
asserts humans actually construct—through their
social interactions and cultural and historical prac- Deduction. The process of reasoning from general
o

tices—many of the facts they take for granted as hav- theory to making predictions about events or behav-
D

ing an independent, objective, or material reality. 40 ior in specific situations. 31

Critical theory. The philosophical stance that disci- Empiricism. Relying on observation to verify
plines such as political science should assess society propositions. 25
critically and seek to improve it, not merely study it Explanatory. Characteristic of scientific knowledge;
objectively. 41 signifying that a conclusion can be derived from a set of

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
44   Political Science Research Methods

general propositions and specific initial considerations; Normative knowledge. Knowledge that is evalu-
providing a systematic, empirically verified under- ative, value-laden, and concerned with prescribing
standing of why a phenomenon occurs as it does. 29 what ought to be. 27
Falsifiability. A property of a statement or hypothesis
Parsimony. The principle that among explanations
such that it can (in principle, at least) be rejected in
or theories with equal degrees of confirmation, the
the face of contravening evidence. 27

e
simplest—the one based on the fewest assumptions
General. A characteristic of scientific knowledge is that and explanatory factors—is to be preferred; sometimes

ut
it be applicable to many rather than just a few cases. 29 known as Ockham’s razor. 30

b
Induction. The process of reasoning from specific Social facts. Values and institutions that have a sub-

tri
observations to theories about behaviors or events in jective existence in the minds of people living in a par-
general. 31 ticular culture. 40

is
Interpretation. Philosophical approach to the study Theory. A statement or series of related statements
of human behavior that claims that one must under- that organize, explain, and predict phenomena. 31

rd
stand the way individuals see their world in order to
truly understand their behavior or actions; philosoph- Transmissible. Characteristic of scientific knowl-
ical objection to the empirical approach to political edge; indicates that the methods used in making sci-
science. 39

Nonnormative knowledge. Knowledge concerned ,o

entific discoveries are made explicit so that others can
analyze and replicate findings. 27
st
not with evaluation or prescription but with factual or Verification. The process of confirming or establish-
objective determinations. 27 ing a statement with evidence. 26
po

SUGGESTED READINGS
y,

Box-Steffensmeier, Janet, Henry Brady, and David Collier. King, Gary, Robert O. Keohane, and Sidney Verba. Designing
The Oxford Handbook of Political Methodology. New Social Inquiry: Scientific Inference in Qualitative Research.
op

York: Oxford University Press, 2008. Princeton, NJ: Princeton University Press, 1994.
Brady, Henry E., and David Collier, eds. Rethinking Social Kuhn, Thomas. The Structure of Scientific Revolutions. 2nd
Inquiry: Diverse Tools, Shared Standards. Lanham, ed. Chicago: University of Chicago Press, 1971.
tc

MD: Rowman and Littlefield, 2004. Nielsen, Joyce McCarl, ed. Feminist Research Methods:
Elster, Jon. Nuts and Bolts for the Social Sciences. Cambridge, Exemplary Readings in the Social Sciences. Boulder,
UK: Cambridge University Press, 1990. CO: Westview, 1990.
no

Hay, Colin. Political Analysis: A Critical Introduction. New Rosenberg, Alexander. The Philosophy of Social Science. 3rd
York: Palgrave, 2002. ed. Boulder, CO: Westview, 2007.
Hindmoor, Andrew. Rational Choice. New York: Palgrave Silver, Brian L. “I Believe.” In The Ascent of Science, 11–28.
o

Macmillan, 2006. New York: Oxford University Press, 1998.

D

NOTES

1. Jeffrey A. Winters and Benjamin I. Page, “Oligarchy 2. Ibid., 744 (emphasis in original). Also see Jeffrey A.
in the United States?,” Perspective on Politics 7 Winters, Oligarch (New York: Cambridge University
(December 2009): 731–51. Press, 2014).

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   45

3. During his Senate confirmation hearing, Chief refer to themselves as “formal modelers” or “rational
Justice John Roberts came close to capturing the choice theorists.”
essence of the empirical viewpoint when he told 8. Whether or not political science or any social science
the committee, “Judges and justices are servants of can find causal laws is very much a contentious issue
the law, not the other way around. Judges are like in philosophy. See, for instance, Alexander Rosenberg,
umpires. Umpires don’t make the rules; they apply The Philosophy of Social Science, 3rd ed. (Boulder, CO:

e
them.” He added, “My job is to call balls and strikes Westview, 2007).
and not to pitch or bat.” CNN.com, September 12, 9. Ibid., 107.

ut
2005. Accessed June 3, 2015. Available at http:// 10. The most ardent proponent of the idea that science
www.cnn.com/2005/POLITICS/09/12/roberts.state- really amounts to an effort to falsify (not prove)

b
ment. In other words, judges “see” the law and the hypotheses and theories is Karl Popper. See, for exam-
facts of a case as they are. Judiciary Committee chair ple, The Logic of Scientific Discovery (New York: Basic

tri
Joe Biden, however, challenged Justice Roberts on his Books, 1959).
umpire analogy: “So, as much as I respect your met- 11. Alan C. Isaak, Scope and Methods of Political Science,

is
aphor, it’s not very apt, because you get to determine 4th ed. (Homewood, IL: Dorsey, 1985), 30.
the strike zone. . . . Your strike zone . . . may be very 12. Ibid., 31.

rd
different than another judge’s view.” Washington Post, 13. APSA Committee on Professional Ethics, Rights
“Transcript: Day Two of the Roberts Confirmation and Freedoms, A Guide to Professional Ethics in
Hearings,” September 13, 2005. Accessed January Political Science, 2nd ed., rev. (Washington, DC:
10, 2015. Available at http://www.washingtonpost
.com/wp-dyn/content/article/2005/09/13/AR200
,o
5091300979.html. In other words, the senator believes
American Political Science Association, 2012).
Accessed February 11, 2019. Available at https://
www.apsanet.org/portals/54/Files/Publications/
st
judges may act like Umpire 3, who in a sense “con- APSAEthicsGuide2012.pdf.
structs” reality in his own way. 14. For example, see Jeffrey C. Isaac, “For a More Public
po

4. It might be more accurate to use the words “scien- Political Science,” Perspectives on Politics 13, no. 2
tific methods,” since to define what is and what is not (2015): 269–83; Kristen Renwick Monroe, “The Rush
science is a notoriously tricky task, and not everyone to Transparency: DA-RT and the Potential Dangers for
agrees on an exact definition. Qualitative Research,” Perspectives on Politics 16, no. 1
y,

5. Those who follow the philosophy of social science, (2018): 141–48. The January 2014 (vol. 47, no. 1)
or epistemology, know that naming the sides in issue of PS: Political Science & Politics contains a
op

these methodological debates is virtually impossible. symposium on openness in political science.

Someone we might label a nonempiricist might very 15. It may be tempting to think that historians are inter-
well foreswear the tag. We are just attempting to sort ested in describing and explaining only unique, one-
tc

out tendencies. time events, such as the outbreak of a particular war.

6. See James Mahoney, “After KKV: The New This is not the case, however. Many historians search
no

Methodology of Qualitative Research,” World Politics for generalizations that account for several specific
62, no. 1 (January 2010): 120–47; Henry E. Brady events. Some even claim to have discovered the “laws
and David Collier, Rethinking Social Inquiry: Diverse of history.”
Tools, Shared Standards, 2nd ed. (Lanham, MD: 16. Isaak, Scope and Methods, 103.
o

Rowman and Littlefield, 2010). 17. Many varieties of this theory exist, but they share the
components presented here.
D

7. Careful readers will note that we are combining all

sorts of activities under one label. Specialists in one 18. Anthony Downs, an economist, provided one of the
method or another often call themselves different first explications of the theory in An Economic Theory
things to emphasize the kind of research they do. For of Democracy (New York: Harper & Row, 1957). His
instance, those who rely on deductive reasoning and ideas in turn flowed from earlier economic analyses.
do not spend much time observing the world often See, for example, Harold Hotelling, “Stability in

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
46   Political Science Research Methods

Competition,” Economic Journal 39, no. 153 (1929): 27. See Paul E. Peterson, The Price of Federalism
41–57. Accessed February 11, 2019. Available at (Washington, DC: Brookings Institution Press, 1995).
https://www.jstor.org/stable/2224214. 28. A good example is Thomas Picketty, Capitalism in
19. See James Enelow and Melvin Hinch, The Spatial the Twenty-First Century (Boston: Belknap/Harvard,
Theory of Voting: An Introduction (New York: 2014).
Cambridge University Press, 1984). 29. Isaak, Scope and Methods, 167.

e
20. This expectation assumes that immigration is impor­ 30. An often-remarked-on characteristic of scholarly jour-
tant to the voter—that there is not some other issue nals is that they tend to report mostly positive find-

ut
that is more important that may cause the voter to ings. An article that shows “X is related to Y” may be
prefer another candidate. more likely to be accepted for publication than one

b
21. Kenneth Shepsle, “The Strategy of Ambiguity: that asserts “X is not related to Y.” Whether or not a
Uncertainty and Electoral Competition,” American “negative result” makes a significant contribution to

tri
Political Science Review 66, no. 2 (1972): 555–68. knowledge depends on the importance of the origi-
22. As an example, see Anders Westholm, “Distance ver- nal claim. Suppose that a team of psychologists found

is
sus Direction: The Illusory Defeat of the Proximity that “love and marriage” really do not “go together.”
Theory of Electoral Choice,” American Political That would be worth publishing.

rd
Science Review 91, no. 4 (1997): 870. 31. For further discussion of complete and partial expla-
23. Here is an example: “Please look at . . . the book- nations, see Isaak, Scope and Methods, 143.
let. Some people believe that we should spend 32. John Medearis, “Review of Perestroika! The Raucous
much less money for defense. Suppose these peo-
ple are at one end of a scale, at point 1. Others feel
that defense spending should be greatly increased. ,o Rebellion in Political Science by Kristen Renwick
Monroe,” Perspectives on Politics 4, no. 3 (2006): 577.
33. For an effort to objectively measure policy impor-
st
Suppose these people are at the other end, at point tance, see David Mayhew, Divided We Govern: Party
7. And, of course, some other people have opinions Control, Lawmaking, and Investigations, 1946–2002,
po

somewhere in between, at points 2, 3, 4, 5 or 6 . . . 2nd ed. (New Haven, CT: Yale University Press,
Where would you place YOURSELF on this scale, 2005).
or haven’t you thought much about it?” See vari- 34. See Charles A. McCoy and John Playford, eds.,
able v3142 in the 2004 American National Election Apolitical Politics: A Critique of Behavioralism (New
y,

Study. Accessed February 11, 2019. Available at the York: Thomas Y. Crowell, 1967).
Survey Documentation and Analysis, University of 35. See Isaac, “For a More Public Political Science.”
op

California–Berkeley, website: http://sda.berkeley.edu/ 36. See Alexander Rosenberg, The Philosophy of Social
D3/NES2004public/Doc/nes0.htm. Science, 3rd ed. (Boulder, CO: Westview, 2007).
24. The conceptualization of distance and other mat- 37. Clifford Geertz, The Interpretation of Cultures (New
tc

ters related to the proximity theory are debated in York: Basic Books, 1973), 5; see also following discus-
Westholm, “Distance versus Direction,” 865–73; and sion, pp. 6–7.
no

Stuart Elaine Macdonald, George Rabinowitz, and 38. “Federal Spending Front and Center in Pa., Wash.
Ola Listhaug, “On Attempting to Rehabilitate the Senate Races,” PBS NewsHour, October 26, 2010.
Proximity Model: Sometimes the Patient Just Can’t Be Accessed February 11, 2019. Available at https://
Helped,” Journal of Politics 60, no. 3 (1998): 653–90. www.pbs.org/newshour/show/federal-spending-
o

25. A good example is Theda Skocpol, States and Social front-and-center-in-pa-wash-senate-races.

39. Colin McGinn, “Is Just Thinking Enough?,” review
D

Revolutions: A Comparative Analysis of France, Russia,

and China (New York: Cambridge University Press, of Making the Social World: The Structure of Human
1979). Civilization, by John R. Searle, New York Review of
26. See Raymond E. Wolfinger and Steven J. Rosenstone, Books, November 11, 2010. Accessed February 11,
Who Votes? (New Haven, CT: Yale University Press, 2019. Available at http://www.nybooks.com/articles/
1980). archives/2010/nov/11/just-thinking-enough/.

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 Chapter 2 • The Empirical Approach to Political Science   47

40. Émile Durkheim, The Rules of Sociological Method and 44. For an excellent collection of articles about the pros
Selected Texts on Sociology and Method, ed. Steven Lukes and cons of studying human behavior scientifically, see
(New York: Free Press, 1982), 60 (emphasis in original). Michael Martin and Lee C. Anderson, eds., Readings
41. The term constructionism encompasses an enormous in the Philosophy of Social Science (Cambridge, MA:
variety of philosophical perspectives, the descrip- MIT Press, 1996).
tion of which goes far beyond the purposes of this 45. Isaak, Scope and Methods, 45.

e
book. The seminal work that brought the ideas into 46. This example is based on an article by Isaac D. Balbus,
sociology and from there into political science is “The Concept of Interest in Pluralist and Marxian

ut
Peter L. Berger and Thomas Luckmann, The Social Analysis,” Politics & Society 1, no. 2 (1971): 151–77.
Construction of Reality (New York: Doubleday, 1966). 47. Kathleen B. Jones and Anna G. Jonasdottir,

b
An excellent but challenging analysis of construction- “Introduction: Gender as an Analytic Category in
ism is Ian Hacking, The Social Construction of What? Political Science,” in The Political Interests of Gender,

tri
(Cambridge, MA: Harvard University Press, 1999). ed. Kathleen B. Jones and Anna G. Jonasdottir
Equally important, members of this school have (Beverly Hills, CA: Sage, 1988), 2.

is
widely varying opinions about the place of empiricism 48. Kathleen B. Jones, “Towards the Revision of Politics,”
in social research. Many constructivists feel their posi- in The Political Interests of Gender, ed. Kathleen B.

rd
tion is perfectly consistent with the scientific study of Jones and Anna G. Jonasdottir (Beverly Hills, CA:
politics; others do not. Sage, 1988), 25.
42. See John R. Searle, The Construction of Social Reality 49. Margrit Eichler, Nonsexist Research Methods: A
(New York: Free Press, 1995).
43. Alexander Wendt, “Anarchy Is What States Make
,o
of It,” International Organization 46, no. 2 (1992):
Practical Guide (Boston: Allen and Unwin, 1987).
50. See the articles comprising “Political Science and
Political Philosophy: A Symposium,” PS: Political
st
391–425. Science and Politics 33, no. 2 (2000): 189–97.
po

STUDENT STUDY SITE

y,
op
tc

Give your students the SAGE edge!

SAGE edge offers a robust online environment featuring an impressive array of free tools and resources
no

for review, study, and further exploration, keeping both instructors and students on the cutting edge of
teaching and learning. Learn more at edge.sagepub.com/johnson9e.
o
D

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.
 e
b ut
tri
is
rd
,o
st
po
y,
op
tc
no
o
D

Copyright ©2020 by SAGE Publications, Inc.

This work may not be reproduced or distributed in any form or by any means without express written permission of the publisher.