Using Comparative Methods
to Study Diversity
Introduction
Comparative researchers examinepatterns of similarities and differences
across a moderate number of cases. The typical comparative study has
anywhere from a handful to fifty or more cases. The number of cases is
limited because one of the concerns of comparative research is to estab-
lish familiarity with each case included in a study. Like qualitative re-
searchers, comparative researchers consider how the different parts of
each casethose aspects that are relevant to the investigation-fit to-
gether; they try to make sense of each case. Thus, knowledge of cases is
considered an important goal of comparative research, independent of
any other goal.
While there are many types of comparative research (see Frzeworski
and Teune 1970; Skocpol 1984; T i y 1984; Stinchcombe 1978; Lijphart
1971),the distinctivenessof the comparative approach is clearestin stud-
ies that focus on diversity (Ragin 1987). Recall that the qualitative ap-
proaches examined in Chapter 4 emphasize commonalities, and the
primary focus is on similarities across instances (for example, the fact
that hospital personnel assess the potential social loss of each dying pa-
tient). This concern for commonalities dovetails with an interest in clari-
fying categories and concepts (for example, the concept of potential
social loss and the situations in which it is assessed). In comparative re-
search on diversity, by contrast, the category of phenomena that the in-
vestigator is studying is usually specified at the outset, and the goal of
the investigation is to explain the diversity within a particular set of
cases (see, for example, Lijphart 1984; Rueschemeyer et al. 1992; Moore
1966;Nichols 1986).(Thistype of comparative research, which is the ma-
jor focus of this chapter, is examined in detail in Ragin 1987.)
Consider the followingexample of comparative research on diversity.
From the middle 1970sthrough the 1980smany less developed countries
experienced mass protest in response to austerity programs demanded
Určeno pouze pro studijní účely
by the InternationalMonetary Fund (IEVLF). These countrieshad accumu-
lated large public debts that they could not repay. In exchange for better
terms (for example, lower interest rates and longer repayment periods),
the governments of these countries agreed to IMF mandates that they
implement economic policies designed to facilitate debt repayment. For
instance, in some countries the IMFdemanded that the governmentstop
subsidizing the prices of basic commodities such as fuel and food. These
austerity measures saved governments money and made debt repay-
ment more feasible; they also provoked widespread protest among citi-
zens faced with new challenges to their economic well-being (Walton
and Ragin 1990).
A comparative researcher interested in these countriesmight contrast
the differentfonns of protest that occurred in response to these austerity
programs. In some countries, there were riots; in others, there were labor
strikes led by unions; in others, there were mass demonstrationsinvolv-
ing many different groups; in others, opposition political parties orga-
nized protests; and so on. Why did different kinds of protest erupt in
different countries? What causal conditions explain these different re-
sponses to austerity programs?And why did some countrieswith severe
austerity programs experiencevery mild mass protest?
To explain this diversity, a comparative researcher would first group
countries according to their different responses to austerity, placing all
those with riots in one category, those with demonstrations in another,
and so on. Next, the investigator would look for patterns of similarities
and differences. What are the similarities among the countrieswith riots
that distinguish them from all other countries? Perhaps, the countries
with riots also had repressive governments,widespread poverty, and se-
rious crowding in major urban areas, and none of the nomiot countries
had this specific combination of conditions. How did the countrieswith
mass demonstrations differ from all the other countries with austerity
programs? This search for systematic differences would continue until
the researcher could accountfor the diverse responses to austerityfound
in these countries.
Thus, in research that emphasizes diversity the focus is on the simi-
larities within a category of cases with the same outcome (for example,
countrieswith riots) that (1)distinguish that category hom other catego-
ries (countrieswith other forms of austerity protest) and (2)explain the
outcome manifested by that category. In other words, the study of diver-
sity is the study of patterns of similarities and differences within a given
set of cases (in thisexample, countrieswith austerity protests).
Contrasts with Other Research Strategies
As noted already, the main difference between comparativeresearch on
diversity and qualitative research on commonalities is that their basic
orientation toward cases differs. When qualitative researchers study
commonalities they usually view multiple cases as many instances of the
same thing.A qualitative researcher who interviews many taxicab driv-
ers uses these many instances to deepen the portrayal of this casethe
taxicab driver.
Comparative researchers who study diversity, by contrast, tend to
look for differencesamong their cases. Comparative researchers examine
pattems of similarities and differences across cases and try to come to
terms with their diversity. A comparative researcher might study the
settlement of Indochinese refugees in the United States in the 1970sand
1980s, contrasting the ways they were received in a variety of different
communities.It might be possible to distinguish four or five basic types
of receptionofrom hostile to indifferent to open to paternalistic and so
on-and then to pinpoint the factors (suchas size and wealth of commu-
nity) that determined these differentreceptions.
Another comparative researcher might study bars in a community
and contrast the different str:tegies they use to encourage and discour-
age drinking. Bars that cater to different customers (for example, bikers
versus business people versus lesbians) surely use different techniques.
In each of these examples, the research focuses on the diversity that ex-
ists within a specificset of cases.
Quantitative researchers (the focus of Chapter 6) also examine differ-
ences among cases, but with a different emphasis. In quantitative re-
search, the goal is to explain the covariation of one variable with another,
usually across many, many cases. A quantitative researcher might ex-
plain differentlevels of income across thousands of individuals included
in a survey by pointing to the covariation between income levels and
educational levels-people with more education tend to have more in-
come. In quantitative research, the focus is on differences in levels and
how different variables like income and education covary across cases.
In comparative research, by contrast, the focus is on diversity-pnttenenls
of similaritiesand differences.
Furthermore, the quantitative researcher typically has only broad fa-
miliarity with the cases included in a study.As the number of cases ex-
ceeds fifty or so, it becomes increasingly diEFicult to establish familiarity
with each case. Imagine a survey researcher trying to become familiar
with the lives of the thousands of people included in a telephone survey
Určeno pouze pro studijní účely
or a political scientist trying to keep up with major events in all coun-
tries. Neither task is feasible. There are practical limits to how many
cases a researcher can study closely.
The Goals of ComparativeResearch
The emphases of comparative research on diversity (especially, the dif-
ferentpatterns that may exist within a specificset of cases)and on famil-
iarity with each case make this approach especially well suited for the
goals of exploring diversity, interpreting cultural or historical s i w -
cance, and advancing theory.
Exploring Diversity
The comparative approach is better suited for addressing patterns of di-
versity than either of the other two strategies.Diversity is most often un-
derstood in terms of types of cases. The typical goal of a comparative
study is to unravel the different causal conditions connected to different
outcomes--causal patterns that separate cases into different subgroups.
This explicit focus on diversity distinguishes the comparative approach
from the qualitativeapproach. Recall that in qualitativeresearch the goal
is often to clarify categories with respect to the concepts they exemplify
by examiningsimilarities across the instances of a category (such as tafi
drivers).
One common outcome of comparative research is the finding that
cases that may have been defined as "the same" at the outset are differ-
entiated into two or more categories at the conclusion of the study. For
example, a researcher studying major U.S. cities that have elected Afri-
can-American mayors might conclude at the end of the study that there
are two major types of cities--those where interracial alliances elected
African-American mayors and those where black voters, who happened
also to constitute a majority of voters, elected African-American mayors.
The political dynamics and significanceof the electionscould differ con-
siderably across the two types.
The researcher studying governments that terronize citizens who op-
pose them might find that there are several main types, depending on
the international standing of the government in question. For instance,
when a government is supported by the United States and other major
powers, its terror may be overlooked.When a governmentlacks thissup-
port, its terror may be considered repugnant. Governments in the second
Using Coe~pnrntiueMetkods to Sttidy DiuersihJ 109
category would have to contend with the possibility that their actions
might provoke international sanctions or intervention and therefore
practice more covert forms of terror.
While comparative researchers often discern types in the course of
their examinations of patterns of diversity, they may also begin their re-
searchwith a tentative delineationof types. A common strategy is to cat-
egorize cases according to their different outcomes. The goal of the
research in thiscase is to unravel the causal conditions that generate dif-
ferent outcomes. If different causes can be matched to the different out-
comes, then the research confirms the investigator's understanding of
the factors that distinguish these cases. If not, then the frame for the re-
searchneeds to be reformulated.
For example, a researcher might examine the causes of differenttypes
of government repression. Some repressive governments, for example,
may simply harass their opponents-incarcerating them for short peri-
ods, subjectingthem to frequent questioning, opening their mail, and so
on. Other repressive governments may torture and kill their opponents.
Still other governmentsmay focus their repressive energies not on oppo-
nents, but on purging the less committed from their own ranks-mem-
bers of the ruling political party or clique. Stillothersmay attack random
members of society in order ;o maintain a general state of terror and obe-
dience (as Stalin did in the Soviet Union). It is important to understand
different types of repressiveness and the various conditions that explain
the emergence of each type.
The goal of exploring diversity is important because people, social
researchers included, sometimeshave trouble seeingthe trees for the for-
est. They tend to assume uniformity or generality when, in fact, there is
a great deal of diversity. A simple example: generally, governments that
are less democratictend to be more repressive.However, there are many
instances of repressionby democraticallyelected governments and many
instances of politically tolerant and lenient governments that are not
democratic. To understand government repression fully, it is necessary
to go beyond the simple identification of political repression with an ab-
sence of democracy and examine the different forms of government re-
pression that exist in all countries.
Ixtelpretiag Ct~ltc[vnlor Historicnl Sig7lificnnce
Comparative researchers focus explicitly on patterns of similarities and
differences across a range of cases. Relevant cases, in turn, are almost
always drawn from a specific and known set. Recall that in qualitative
Určeno pouze pro studijní účely
110 Clmnpter5
research (Chapter 4), much energy is often devoted to coming to terms
with the case. What is thiscase a case of?What concepts are exemplified
in this case?Into which larger social scientific categories, if any, does it
fit? In comparative research, by contrast, the researcher usually starts
with a good sense of the larger category that embraces the casesincluded
in the study because this category is usually specified beforehand (such
as "countries with austerity protests").
A researcher might focus on "military coups in Latin America since
1975" or "major cities in the United States that have elected African-
Americanmayors" or "recent U.S. federalcourt casesinvolvingthe rights
of AIDS patients" (Mushenoet al. 1991).In eachexample, the relevant set
of casesis defined in advance, and there is a finite,usuallymoderatenum-
ber of such cases. Typically, the category that establishes the boundary of
the set is historically and geographically delimited. In each of these ex-
amples time and place boundaries are either plainly stated (for example,
"Latin America since 1975")or implied ("recent U.S.").
This focus on circumscribed categories makes the comparative strat-
egy well suited for the goal of interpreting historically or culturally sig-
nificant phenomena, especially when there is a moderate number of
cases, as in the examples just mentioned. The category "major cities in
the United States that have elected African-Americanmayors" is histori-
cally significant in part simply because it is a relatively new and major
phenomenon. Prior to the expansion of civil rights in the 1960s, there
were no African-Americanmayors in major U.S. cities. It is culturallysig-
nificant because of the relevance of race and race relations to American
society. Likewise, the category "military coups in Latin America since
1975" is significant to those concerned with progress of democracy and
human rights in thisregion.
Because the comparative approach focuses on differences between
cases and the differentiationof types, it facilitates historical interpreta-
tion. Consider the category reoobitioi~.Some revolutions simply change
those who are in power or alter other political arrangements without
implementing any major changes in society. The revolutionaries that
overthrew Ferdinand Marcos in the Philippines did not attempt any fun-
damental changes in Philippine society. Other revolutions, by contrast,
bring with them regimes that seek to alter society fundamentally. Kings
are beheaded; property is coniscated; basic social patterns and relations
are changed forever. Revolutionary social changes of thisnature were at-
tempted after the French Revolution of 1789, the Russian Revolution of
1917, and the Chinese Revolution of 1949.
Using CoinpnrntiucMetlrods to Shldy Diversih~ 111
Revolutions that attempt fundamental social change are treated as a
distinct type by social scientists. These massive upheavals of society are
called social reool~itionsto distinguish them from revolutions that simply
change leaders or other political arrangements (Skocpol1979).By differ-
entiating social revolutionsfrom all other forms, researchers provide im-
portant tools for understanding and interpreting these massive social
transformations.When a major upheaval occurs, researchers can assess
whether or not it qualifies as a social revolution. If so, it can be com-
pared with other socialrevolutions.If not, then some other categorymay
be used (for example, conp d'etnt) to interpret the event and to specify
comparable cases. Generally, when a set of comparable cases can be
specified, these cases aid the interpretation and understanding of the
new case.
More generally, when socialscientistscategorizean event, they estab-
lish a primary analytic frame for its interpretation.Thus, the interpreta-
tion of historically or culturally significant eventsis often a struggle over
the proper classification of events into broad categories-a key concern
of the comparative approach.
Advancing T1reol-r~
Severalbasic features of the$omparative approach make it a good strat-
egy for advancing theory. These features include its use of flexible
frames, its explicit focuson the causes of diversity, and its emphasis on
the systematic analysis of similarities and differences in the effort to
specify how diversity is patterned.
In comparativeresearch, investigators usually initiate research with a
specific analytic frame, but these initial frames are open to revision. The
researcher interested in military coups in Latin American since 1975 al-
ready has a frame for the research-the frame for military coups. Recall
that the frames of qualitative research are fluid, and researchers may not
finish developing their frames until after all the work of collecting and
studying the evidenceis complete. In comparative research, by contrast,
frames are established at the outset of a research project, but they remain
flexible.Comparativeresearchersexpect their frames to be revised, and in
fact conduct researchin order to sharpen the ideas expressed in a frame.
A researcherinterested in welfare states in advanced countries might
start out with a frame that specifies two basic types of welfare states but
then conclude with a specification of three or four types (Esping-
Andersen 1990).Or, the researcher might conclude that there is only one
Určeno pouze pro studijní účely
! 112 Chnpter5
I
main type and that all deviations from this main type are best under-
stood as underdeveloped or incomplete expressions of the main type 0.
Stephens 1979). By altering initial frames in response to evidence, com-
parative researchers refine and elaborate existing ideas and theoretical
perspectives.
When conducting their research, comparative researchers are more
explicitlyconcerned with causation and causal complexity than are most
qualitativeresearchers. For example, when comparative researchers dif-
ferentiate types (such as types of government repression), they also try
to specify the combinations of causal conditionsconducive to each type.
What causes some regimes to concentrate their repressive efforts on re-
gime opponents? What causes others to focus their efforts on purging
troublesome members of the ruling party? And what causes still other
regimes to cultivate a general state of terror in the population at large?
This emphasis on causationis central to theory because most theories in
the socialsciences are concerned with explaininghow and why-that is,
with specifying the causes of socialphenomena.
To assess causation, comparative researchers study how diversity is
patterned. They compare cases with each other and highlight the con-
hasting effects of different causes. Comparative researchers view each
case as a combination of characteristics (for example, conditions relevant
to government repression) and examine similarities and differences in
combinationsof characteristicsacrosscasesin their effortto find patterns.
The Process of Comparative Research
The comparative study of diversity is neither as f l ~ das qualitative re-
search nor as fixed as quantitative research. Comparative researchers
typically start with a carefully specified category of phenomena that is
intrinsically interestingin some way (for example, countrieswith auster-
ity protests). They use analytic frames to help them make sense of their
categories, and they revise their frames based on their examination of
evidence.
In the course of their research they focus on patterns of similarities
and differences amongcases and assess patterns of diversity.This assess-
ment of diversity provides the foundation for improving or revising the
analyticframechosen at the outset of the study. Like qualitativeresearch,
the comparative approach stimulates a rich dialogue between ideas and
evidence. Researchers generate images from their data and adjust their
frames as they constructrepresentations of their research subjects.
Using ContpnrntiueMetlrods to Shidy Diversity 113
Selecting Cases
Comparative researchers usually initiatetheir research with a specificset
of cases in mind. Most often, this set has clear spatial and temporal
boundaries and embraces cases that are thought to be comparable with
each other, as in the examples already described. The degree to which
the cases that are selected actually belong to the same category (and
therefore are comparable)is assessed in the course of the research. While
conducting the investigation, the researcher may decide that some cases
don't belong in the same category as the others and can't be compared.
They also may reformulate the category as the research proceeds. Usu-
ally,however, such adjushents are modest.
Typically, the cases that comparative researchers select for study are
specific to their interests and to those of their intended audience, for ex-
ample, countries with mass protest against austerity programs. This cat-
egory of countries has clear spatial and temporal boundaries and
embraces a set of comparable cases. It is also an intrinsically interesting
set of cases. In short, it is just the kind of delimited empirical category
that is well suited for comparative investigation.
The comparative approach can be applied to many different kinds of
cases, not just countries. It is important, however, for the cases selected
to be comparable and to share membership in a meaningful, empirically
defined category. For example, the comparative approach can be applied
to the fraternities on a college campus, to ethnic and racial groups living
in a major urban area, to different religious congregations in a medium-
size town, to the truck stops alongInterstate55, or to the elections in the
congressional districts of a large state.The set of cases must be coherent.
Usually, they must also offer some potential for advancing social scien-
tific thinking.
Using Almlytic Frames
When researchers choose their cases, they also usually select their ana-
lytic frames. Essentially, a frame is chosen when the researcher specifies
what about the cases is of interest. The researcher interested in countries
with austerity protests may be interested in the different forms that the
protest took. This frame, which would be developed from the existing
social science literature on mass protest, would specify how people re-
spond to different conditions in different ways when they engage in po-
litical protest. In short, it would detail the different kinds of factors the
researcher should examine in a comparative study of mass protest.
Určeno pouze pro studijní účely
In some countries opposition groups may have many resources; in
others, they may have few. Groups with more resources are more likely
to engage in organized activities such as strikes and in other activities
that are relatively costly to participants. People on strike must give up
their wages. Thus, thisframe, which would be developed from the exist-
ing literature on social movements and collective action (for example,
Jenkins 1983),would direct the researcher to focus on resources, among
other things. Analytic frames help researchers see aspects of cases that
they might otherwise overlook, and direct their attention away from
other aspects.
Sometimesresearchers are interestedin many facetsof their cases and
don't select a frame until they are well along in their research. It might
take a while, for example, to determine what a comparison of countries
with austerity protests might best offer in the way of general social sci-
entific knowledge. Comparative researchers also may develop new
frames from their evidence, for example, a new frame for the study of
race and politics based on a study of cities where coalitions of white and
black voters have elected African-American mayors. This practice is less
common in comparative research than in qualitative research, however,
because comparative researchers start with a fairly good sense of their
cases and the empiricalcategory that embraces them (such as "countries
that experienced mass protest in response to IMT-mandatedausterity").
AimlyziizillgPnttenls of Diversify
In comparative research the examination of diversity-patterns of simi-
larities and differences-goes hand in hand with the study of causes.
Generally, researchers expect different causal conditions to be linked to
divergent outcomes in interpretable ways. Thus, the goal of the
researcher's examination of patterns of similarities and differences is to
identify causal links-how different configurations of causes produce
different outcomes across the range of cases included in a study. The
specification of different patterns of causation is the primary basis for
the differentiation of types.
In a study of how sororitiesgenerate a feelingof group solidarity,dif-
ferentways of generatingthisfeeling should affect the nature of the soli-
darity that is generated. The researcher might find that some sororities
generatesolidarityaround specialevents and rituals, while others gener-
ate it through routine activities that bring members of the sorority to-
gether on a daily basis. These different ways of generating this feeling
should have consequences for the nature of the solidarity observed. For
Using CoiiiynrntiueMethods to Shldy Diuersily 115
example, solidarity in sororitiesof the first type may be more visible but
also less durable, while in the second type, it may be more subtle but
more enduring.
If causes and outcomes cannot be linked in interpretable ways, then
researchers must reexamine their specification of causes and outcomes
and their differentiationof types. In many ways thisprocess of differen-
tiating types and specifyingcausal links specific to each type resembles
the "double fitting" of categories and images that constitutes the core of
qualitative methods (see Chapter 4). There is a dialogue between ideas
and evidence that culminates in a meaningful representation of the re-
searchsubject.The main differenceis that in qualitativeresearch the em-
phasis is on clarifying a category and enriching its representation,
whereas in comparative research the emphasis is on using contrasts be-
tween cases to further the researcher's understanding of their diversity.
Using Comparative Methods
Comparative methods are used to study configurations.A configuration
is a specific combination of attributes that is common to a number ofi
cases. For example, if all the countries that experiencedmass demonstra-
tions in response to IMF-mandated austerity were similar in having low
levels of economic development, high levels of urbanization, undemo-
craticgovernments,and poorly organizedoppositiongroups, thiswould
constitute a specific configuration of conditions associated with mass
demonstrations as a response to IMF-mandated austerity. The examina-
tion of patterns of diversity essentially involves a search for combina-
tions of conditions that distinguish categories of cases. Thus, researchers
look for uniformity within categories and contrasts between categories
in combinationsof conditions.
Data procedures appropriate for the study of configurations, formal-
ized by Drass and Ragin (1989), constitute the core of the comparative
approach to diversity. Comparative methods are used to examine com-
plex patterns of similarities and differences across a range of cases. Like
quantitative methods (see Chapter 6), comparative methods are used to
examine causes and effects, but the emphasis in comparativeresearch is
on the analysis of configurationsof causal conditions.
Before examining data procedures specific to comparative methods,
first consider an example that shows the main ideas behind the tech-
niques.
Určeno pouze pro studijní účely
A71 Overview of ColilparntiveMethods
An examplefrom the study of the repression of austerity protests is used
to illustrate general features of comparativemethods. Table 5.1presents
hypothetical data on sixteen countries that experienced austerityprotests
in the early 1980s.Eight of these countrieshad govemments that became
violently repressive in response to austerity protests; the governments of
the other eight did not.
The table shows differences and similarities among these sixteen
countrieswith respect to conditionsbelieved to be relevant to repression,
derived from an analytic frame for government repression. The condi-
tions include:
whether the country was politically aligned with the Soviet Union or
with the United States and Western Europe in the 1980s
m whether or not the country had undergone substantial industrializa-
tion prior to 1980
a whether or not the country had a democratic governmentprior to
the emergence of austerity protests
m whether or not the country had a strong military establishment prior
to the emergence of austerity protests- - -
The goal of comparative analysis is to determine the combinations of
causal conditionsthat Merentiate sets of cases. In this analysis, the goal
is to find combinations of casual conditions that distinguish the eight
countries with governments that became repressivefrom the other eight
countries. Careful examination of the similarities amonE the countries
Using Contpnrntive Methods to Stady DiversihJ 117
T A B L E 5.1
Simple Example of Comparative Methods*
Aligned Dfltrocrfltic Strong Voler~t
Cnse with USSR Indsstrinlized Gouen~tnolt Militnnj Repressiont
1 0 0 0 1 1
2 0 1 0 1 1
3 1 0 0 1 1
4 1 1 0 1 1
5 0 0 1 0 1
6 0 0 1 1 1
7 1 0 1 0 1
8 1 0 1 1 1
9 0 0 0 0 0
10 0 1 0 0 0
11 0 1 1 0 0
12 0 1 1 1 0
13 1 0 0 0 0
14 1 1 0 0 0
15 1 1 1 0 0
*Inthe columnswith causal or outcome conditions, the number1 indicates the presenceof
a conditionor "yes";0indicates its absence or "no."
'The two combinationsof conditionslinked to violent repressionare (1) absence of
democratic mvemment combined with a stron~military and (2)presence of a democratic- 0 - -~
with violently repressivegovernmentsshows that they do not share any combined with an absenceof indu;trializaiion.
single causal condition or any single combination of conditions. How-
ever, there are two combinations of conditions that are present in the set
of countries that had repressive governments that are both absent from
the set that did not. The sixteen cases are sorted in the table to highlight
these two combinations.
The first four cases share an absence of democratic governmentprior
to the emergence of austerity protests combined with a strong military
establishment. None of the cases in the lower half of the table (the eight
countries lacking violent repression) has this combination. The second
four countries with violent repression share two different conditions: a
presence of democratic governmentprior to austerity protests combined
with an absence of significant indushialization prior to the protests.
Again, none of the eight countries lacking violent repression has this
combination of conditions.
The results of the examination of similarities and differences thus
leads to the conclusionthat there are two different combinations of condi-
tions (or causal configurations) that explain the emergence of violent re-
pression in these cases. The first configuration (nondemocratic rule
combined with a strong military)suggests a situation where the military
establishmenthas gained the upper hand in part because of the absence
of checks (democratic government) on its power. The second configura-
tion (absence of signiEicant industrializationcombined with the presence
of a democratic governmentprior to the emergence of violent repression)
suggests a situation where a breakdown of democratic rule occurred in
Určeno pouze pro studijní účely
countriesthat lackedmany of the social structures associatedwith indus-
trialization (for example, urbanization, literacy, and so on). These social
structures are believed to facilitate stable democratic rule. Further re-
searchmight show important differences between these two sets of cases
with respect to the kind of repression that was inflicted on the protesters.
The cases are arranged in Table 5.1 so that the main patterns of simi-
larity among the countrieswith violent repression are easy to detect, and
the comparison of these cases with counhies lacking violent repression
is simplified. Specific procedures for assessing patterns of similarity and
difference are detailed in the next section. Before examiningthese proce-
dures, consider several general features of the comparative analysis just
presented.
1. Comparative analysis proceeds by comparing configurations of
causes-laws of the table-and not by comparing the presence or
absence of each causalcondition (that is, each of the first four
columns) with the presence or absence of the outcome (thelast
column--repression).
2. The comparative approach allows for the possibility that there may
be several combinations of conditionsthat generate the same general
outcome (governmentrepression in the example).
3. Comparativemethods can address complex and seemingly contra-
dictory patterns of causation. One causal condition (democratic
governmentprior to the emergence of violent repression)is impor-
tant in both its present and absent condition-it appears in both
causal configurations-but contributesin opposite ways.
4. The comparative approach can eliminateirrelevant causes. One
causal condition (whethernor not the country was aligned with the
Soviet Union) was eliminated as an important causal condition. Even
though it was considered a possible factor at the outset, examination
of similarities and differences among repressive and nomepressive
cases shows that this cause is not an essentialpart of either of the
key causal combinations.
The findings in Table 5.1 are easy to see. Usually, however, the pat-
terns are not so simple, and researchers must use more systematic com-
parative methods to help them analyze similarities and differences.
These techniques, explained in the next sections, make it possible for re-
searchers to find patterns that they would probably miss if they tried to
unravel differences simply by "eyeballing" their cases.
Using Conrpnrnh'veMetlrodsto Shrdy DinersihJ 119
Specgjiilg Cnuses nlzd Ol~tconzes
In the comparative approach each case is understood as a combination
of causal conditions linked to a particular outcome. Thus, the selection
of the outcome to be studied and the specification of causal conditions
relevant to that outcome are crucially important parts of a comparative
investigation.
Generally, in order to specify causes, the investigatormust be famil-
iar with the research literature on the outcome (for example, "govern-
ment repression") and with the cases included in the study. In this early
phase of the research, the investigator explores connectionsbetween so-
cial scientific thinking (for example, about government repression) and
the evidence. These early explorations lead to a clarification of the na-
ture of the outcome to be studied and a specification of the relevant
causes.
The comparative methods described in this chapter use what social
scientists call presence-absence dichotomies. This means that causal
conditions and outcomes are either present or absent in each case and
can be coded "yes" or "no," as in Table 5.1. Thus, instead of using a pre-
cise measurement of industrialization(forexample, tl~epercentage of the
work force employed in manufacturing)in the data analysis, an assess-
ment might be made of w h e v r or not substantial industrialization oc-
curred before a specific year (again, as in Table 5.1). The use of
presence-absence dichotomies simplifies the representation of cases as
configurations of causes. Research methods that focus explicitly on con-
ditions that vary by degree or level are discussed in Chapter 6.
In comparative analysis the number of causal conditions determines
the number of combinations of causal conditions that are possible. For
example, the specification of four causal conditions (as in Table 5.1) pro-
vides for 16 (that is, Z4) logically possible combinations of causal condi-
tions. Specification of five causal conditions provides for 32 (z5)
combinations, six causal conditions provides for 64 (Z6) combinations,
and so on. Causal conditions are not examined separately, as in studies
focusing on covariation(see Chapter 6), but in combinations.
Once causal conditionshave been selected, cases conforming to each
combination of causal conditions are examined to see if they agree on
the outcome. In Table 5.1, there is only one case for each combination of
causal conditions, so there is no possibility of disagreement. But what if
there were two cases in the first row (that is, two countries that com-
bined absence of alignmentwith the SovietUnion, absence of substantial
industrialization before 1980, absence of democratic government, and
Určeno pouze pro studijní účely
presence of a strong military), but in one countryprotesters suffered vio-
lent repression while in the other they did not? The researcher would
have to determinewhat additional factor (presentin one countrybut ab-
sent in the other) caused repression. This new causal condition would
then be added to the table for allcases.
If there are many causal combinations with cases that disagreeon the
outcome, then the investigator should take this as a sign that the specifi-
cation of causal conditionsis incorrect or incomplete. The close examina-
tion of cases that have the samepresence-absencevalues on all the causal
conditions yet have different outcomes is used as a basis for selecting
additional causal variables. Investigatorsmove back and forth between
specification of causal conditions (using social science theory and their
general substantive knowledge as guides) and examination of evidence
to resolve these differences.
Constrzlcti~zgthe Truth Tnble
Once a satisfactoryset of causal conditionsfor a particular outcome has
been idenwed, evidence on cases can be represented in truth tables. The
use of truth tables facilitates the analysis of patterns of similarities and
differences.
The first step in constructing a truth table is simply to list the evi-
dence on the cases in the form of a data table. Consider for example, the
data presented in Table 5.2. This table shows hypothetical evidence on
thirty suburban school districts surrounding a major metropolitan area.
The outcome of interesthere is whether or not the elementaryschools in
each district track students according to ability. When students are
tracked, they are grouped together into relatively homogeneous classes.
Students who learn things quickly are assigned to one class, while stu-
dents who learn things at an average speed are assigned to another, and
so on.
Having students of uniform ability together in the same room is
thought to simplifyteaching, making it more eicient. After all, it clearly
would be a mistake to put first graders and sixth graders in the same
room. Why not apply this same principle to students within grade lev-
els? The usual objection is that students who are assigned to the "slow"
group become branded low achievers and are rarely given the opportu-
nity to prove otherwise. Plus, being surrounded by "faster" students can
motivate a "slow" student to learn faster. Assigningstudents to the slow
group may seal their academicfate.
Using Cotnpnrntive Methods to Shldy Diversib 121
T A B L E 5 . 2
Hypothetical Data on Tracking in School Districts*
Scllool RRciol Clnss Cotnpetitine Unionized Abilihj
Dishict Diumihj Divenihj Elections Tencl~ers l'mcking
1 0 0 0 0
0
*Inthe columns with causal or outcome conditions, the number 1 indicates the presence of
a condition or "yes";0indicates its absence or "no!'
Určeno pouze pro studijní účely
The researcher in this example wanted to understand why some
school districts track elementary school students and others don't. The
table lists the causal conditions that the researcher, on the basis of an
examination of the relevant research literatures, thought might be im-
portant:
1. whether the school districtis racially diverse or predominantly
white
2. whether or not the school districthas a broad representation of
income groups (poor, working class, middle class, and upper middle
class)
3. whether or not the schoolboard elections in the district are open and
competitive, with good voter turnout
4. whether or not the teachers in the district are unionized
The first two factors (racial and class diversity) show the social com-
position of school districts. These factors are important because where
there is more diversity, members of dominant groups (for example,
whites in racially diverse districts) generally believe that hacking will
benefit their children most. The competitiveness of school board elec-
tions is important because the majority of voters usually disapprove of
tracking in elementaryschools. They believe this practice benefits only a
minority of students. In districts where school board elections are rou-
tine matters that attract little voter interest, however, the minority of
families that benefit from tracking might have more influence.Unioniza-
tion of teachers is included because the researcher believes that teacher
unions prefer tracking because it simplifies teaching.
The schooldistrictsare sorted in Table 5.2 according to the four causal
conditions, so that districts that are identical on these factors are next to
each pther. Inspection of the data shows that there are no districts that
have the same combination of scores on the causal conditionsbut differ-
ent outcomes. Districts 8-11, for example, all show the same pattern on
the four causal conditions; they also are identical on the outcome--none
of these districts tracks students according to ability. If the cases were
not consistent on the outcome, it would be necessary to examine them
closely to determine which other causal factors should be added to the
table.
Listing the data on the cases, as shown in Table 5.2, is a necessary
preliminary to the construction of the truth table. The idea behind a truth
table is simple. The focus is on causal combinations. Each logical combi-
nation of values on the causal conditions is represented as one row of the
Using Con~pnrntiveMethods to Study Diversity 123
truth table. Thus, truth tables have as many rows as there are logically
possible combinations of values on the causal conditions.If there are four
dichotomouscausalconditions,asin Table5.2, the truth tablewill contain
Z4 = 16 rows. Each row of the truth table is assigned an outcome score
(1 or 0, for presence-absence of the outcome)based on the cases in that
row. The first three cases in Table 5.2, for example, have the same combi-
nation of scores on the causal conditions (absent on each of the four con-
ditions) and the same outcome (absenceof tracking).They are combined
to form the first row of the truth table presented in Table 5.3. The number
of districts that make up each row of the truth table is also reported in
Table 5.3, so that the hanslation of Table 5.2 to Table 5.3is clear.
Sinlpl+jii tlle nut11Tnble
The truth table (Table5.3) summarizes the causal configurations that ex-
ist in a data table (Table 5.2). Listing configurations is not the same as
identifying patterns, however. Usually, comparative researchers want to
examine configurationsto see if they can be simplified. When investiga-
tors simplify configurations,they identify patterns.
A quick example of simplifi~ation:Look at rows 13and 14of the truth
table reported in Table 5.3. Row 13reports that schooldistricts that com-
bine the followingfour characteristics track students: (1)racial diversity,
(2) class diversity, (3) an absence of competitive school board elections,
and (4) an absence of teachers' unions. Row 14 reports that school dis-
tricts that djffered on only one of these four conditions-teachers'
unions-also tracked students. The comparison of these two rows shows
that when the first two causal conditions are present (race and class di-
versity) and the third is absent (competitive school board elections),it
does not matter whether teachers are unionized; tracking by ability still
takes place.
An easy way to represent this simplificationis to use uppercase let-
ters to indicate presence of a condition and lowercase letters to indicate
its absence. In this example, the word "RACE" indicates the presence of
racial diversity; "race" is used to indicate its absence. "CLASS"is used
to indicate the presence of class diversity, ''class" to indicate its absence.
"ELECTIONS" is used to indicate the presence of open, competitive
school board elections, "elections" to indicate the absence of this condi-
tion. "UNIONS'indicates the presence of teachers' unions, "unions" the
absence of this condition. Finally, "TRACKING indicates the presence
of tracking, and "tracking" its absence.
Určeno pouze pro studijní účely
T A B L E 5.3
Truth Table for Data on Trackingin School Districts*
Rncinl Clnss Conlpetitiue U~rioeized Ability N~mrberof
Row Diversity Diversih~ Electiot~s Teacl~ers Trnckitlg Dishictst
1 0 0 0 0 0 3
'In the columns with causal or outcome conditions,the number 1indicatesthe presence of
a conditionor "yes";0indicates its absenceor "no."
Vhe numberof districts is reported simply to remind the reader that each row of a truth
tablemay representmore than one case.
Thus, row 13can be represented as
TRACKING = RACE~CLASS.elections.unions
and row 14as:
TRACKING = RACE.CLASS.elections.UNIONS
where multiplication (.) is used to indicatethe combination of conditions.
These two rows can be simplified through combination because they
have the same outcome and differ on only one causal condition-pres-
ence-absence of teachers' unions. This simplificationstrategy follows the
logic of an experiment. Only one condition at a time is allowed to vary
(the "experimental" condition). If varying this conditionhas no discem-
Using ColltpnrativeMetllods to Shldy Diversify 125
ible impact on the outcome, it can be eliminated as a factor. Thus, the
comparison of rows 13and 14results in a simpler combination:
TRACKING = RACE.CLASS.elections
This rule for combiningrows of the truth table as a way of simplifying
them can be stated formally: If two rows of a truth table differ on only
one causalconditionyet result in the same outcome, then the causal con-
dition that distinguishes the two rows can be considerrd irrelevant and
can be removed to create a simpler combination of casual conditions (a
simpler term).
The process of combining rows to create simpler terms can be carried
on until no more simplification is possible. Table 5.4 shows all the sim-
plifications that are possible for the truth table in Table 5.3, using pres-
ence of ability tracking as the outcome of interest. In Table 5.4 the truth
table rows from Table 5.3 with outcomes of "1" (presence of tracking)
have been translated into the uppercase and lowercase names in the
manner just described. Panel A of this table simply lists the eight kinds
of districts that hack students accordingto ability. PanelB shows the first
round of simplification.Each of the terms from panel Acan be combined
with one or more other terms to create simpler terms. Whenever two
terms with four conditions are combined, the new term has three condi-
tions because one conditionhas been eliminated.
Panel C shows the second round of simplification. In this round,
terms with three conditions (from panel B) are combined to form
terms with two conditions. For example, the term labeled #17 in
panel B (race.class.UNIONS) can be combined with the term #21
(raceCLASS.UNIONS) to form a two-condition term (raceUNIONS).
All the terms from panel B combine with one or more terms from the
same panel to produce the three two-condition terms listed in panel C.
The three terms in panel C can be represented in a single statement
describingthe conditionsunder which tracking in these suburban school
districts occurs:
TRACKING = raceUNIONS +RACE.elections +elections.UNIONS
Tracking occurs when:
1. racial diversity is absent and teachers' unions are present,
2. racial diversity is present and competitiveschoolboard elections are
absent, or
3. competitiveschoolboard elections are absent and teachers' unions
are present.
Určeno pouze pro studijní účely
T A B L E 5.4
Simplification of Truth Table for Tracking (Table 5.3)
Pnncl A. Dishicts Tlmt Trock Shrdents
Row Cntrsnl Cottfigurntiorts
2 racedass~electiom~uMONS
Pnrtel B. First Rorntd of Sirnplifkxtion
-
Lnbelfor New Temt
Rows 2 + 4 -t race.class,UNIONS #17
Rows 2+ 6
Rows 2+10
Rows 4+ 8
Rows 6 + 8
Rows 6 + 14
Rows 9+10
Rows 9+13
Rows 10+14
Rows 13+14
Before acceptingthese tentativeresults, it is important to determineif
further simplificationis possible, as is often the case. Sometimes the pro-
cess of combiningrows to produce simpler terms (presentedin Table5.4)
Usiitg Contpnrntive Metllods to Study DiversihJ 127
generates "surplus" terms. A surplus term is redundant with other terms
and is not needed in the statement describingthe combinations of condi-
tions linked to an outcome. In short, some of the terms that are left after
the process of combiningrows may be superfluous. Recall that the goal
of comparative analysis is to describe diversity in a simple way. If the
results can be further simplified by eliminating surplus terms, as is the
case here, it is important to do so. The idea of a surplus term is best un-
derstood by examining the methods used to detect them.
The best way to check if there are surplus terms is to construct a
chart showing which of the original terms in panel A are covered by
which simplified terms in panel C. A simplified term covers a truth
table row if the row is a subset of the simplified term. For example,
RACE.CLASS.elections.UNIONS(row 14 of the truth table) is a subset
of the simplified term elections.UNIONS.
The chart showing the coverage of the simplified terms is presented
in Table 5.5. The simplifiedterm race.UNIONS covers the first four terms
from panel A of Table 5.4, while the term RACE.elections covers the
other four. The third simplified term (elections.UNIONS)does not cover
any of the rows uniquely; it covers two that are covered by the first sim-
plified term and two that are coveredby the second.Thus, the third sim-
plified term is surplus; it is redundant with the other terms.
By eliminating the thud simplified term the results of the analysis of
configurationscan be reduced to
TRACKING = raceUNIONS +RACE elections
This completes the procedure. The final statement says that tracking oc-
curs (1)when racial diversity is absent and teachers' unions are present,
or (2)when racial diversity is present and competitive schoolboard elec-
tions are absent. The first term indicates that in school districts that are
predominantly white, tracking is implemented if there are teachers'
unions. This finding supports the researcher's belief that teachers' unions
prefer tracking and specifies the conditions under which their interests
are realized-in districts where there is an absence of racial diversity. It
does not matter whether school board elections are open and competi-
tive or whether the district contains a broad range of income groups. The
second term indicates that in school districtswhere there is racial diver-
sity, tracking occurs when school board elections are not competitive.
They are routine matters that do not attracta lot of voter interest.In these
districts, it does not matter whether teachers' unions are present or
whether the districts contain a broad range of income groups. The sec-
ond term suggests that if voters become involved in school board elec-
tions, trackingwould be eliminated in racially diverse districts.
Určeno pouze pro studijní účely
Using Con~pnrntiueMethods to Sh~dyDiversity 129
The analysis of school districtspresented here shows the major steps
in using comparative techniques to unravel causalpattems.
1. Selectcausal and outcome conditions,using existing socialscience
literature and substantive knowledge to guide the selection.
2. Construct a sorted data table showing the scores of cases on these
causal and outcome conditions (Table5.2).
3. Construct a truth table from the data table, making sure that cases
with the same causal conditionsactuallyhave the same score on the
outcome (Table5.3).
4. Compare rows of the truth table and simplify them, eliminating one
condition at a time frompairs of rows (Table 5.4).
5. Examine the coverage of the simplified terms to see if there are any
surplus terms that can be eliminated (Table 5.5).
The terms that remain after step 5 show the simplest way to repre-
sent the pattems of diversityin the data. In the comparative analysispre-
sented in Tables 5.2 through 5.5, the goal is to explain why some school
districts track elementary students. The results show which types of
school districts track elementary students and distinguishes them from
those that do not. t
Conclusion
The brief overview of comparative methods presented in this chapter il-
lustrates some of the key features of the comparative approach.The most
important feature is its focus on diversity. Whenever a set of cases have
differentoutcomes (citieswith differentreactions to Indochineserefuges,
countries with different reactions to austerityprograms,bars with differ-
ent ways of encouraging patrons to drink or to not drink, and so on),
comparative methods can be used to find simple ways of representing
the patterns of diversity that exist among the cases. These methods iden-
tify similaritieswithin subsets of cases that distinguish them from other
subsets.
As in all forms of socialresearch, analyticframes and imagesplay an
important part in comparative research. Analytic frames provide pri-
mary leads for the construction of truth tables, especiallythe selectionof
causal conditions. The construction of the truth table itself is an impor-
tant part of the dialogue of ideas and evidence in comparative research
because the truth table must be free of inconsistencies before it can be
Určeno pouze pro studijní účely
simplified. Evidence-based images emerge from the simplification of
truth tables in the form of configurations of conditions that differentiak
subsets of cases.
In many ways the comparative approach lies halfway between the
qualitative approach and the quantitative approach. The qualitative ap-
proach seeks in-depth knowledge of a relatively small number of cases.
When the focusis on commonalities, it oftennarrows its scopeto smaller
sets of cases as it seeks to clarify their similarities. The comparative ap-
proach usually addresses more cases because of its emphasis on diver-
sity, and it is applied to sets of cases that are clearlybounded in time and
space. As Chapter 6 shows, the quantitative study of covariation seeks
broad familiarity with a large number of cases and most often views
them as generic, interchangeable observations.
Using Quantitative Methods
to Study Covariation
Introduction
The starting point of quantitative analysis is the idea that the best route
to understanding basic patterns and relationships is to examinephenom-
ena across many cases. Focusing on any single case or on a small num-
ber of cases might give a very distorted picture. Looking across many
cases makes it possible to average out the peculiarities of individual
cases and to construct a picture of social life that is purified of phenom-
ena that are specific to any case or to a small group of cases. Only the
general pattern remains.
Quantitativeresearchers construct imagesby showing the covariation
between two or more features of attributes (variables)acrossmany cases.
Suppose a researcher were to demonstrate in a study of the top 500 cor-
porations that those offeringbetter retirement benefits tend to pay lower
wages. The image that emerges is that corporations make trade-offs be-
tween retirement benefits and pay, with some corporations investing in
long-term commiiments to workers (retirement benefits) and some em-
phasizing short-term payoffs (wages and salaries). Evidence-based im-
ages such as these are general because they describe patterns across
many cases and they are pnrsiinoltiotis-only a few attributes or variables
are involved (pay and retirementbenefits).
Images that are constructed from broad patterns of covariation are
considered general because they condense evidence on many cases. The
greater the number of cases, the more general the pattern. A quantitative
researcher might construct a general image of political radicalism that
links degree of radicalism to some other individual-level attribute, such
as degree of insulation from popular culture, and use survey data on
thousands of people (includingpeople who are politically inert) to docu-
ment the connection.Qualitativeresearchersstudying this same question
would go about the task very differently. The images they construct are
detailed and specific, and they use methods that enhance rather than
condense evidence. Using a qualitative approach, a researcher might
Určeno pouze pro studijní účely