researcher then draws the implications of the results of these tests for
theory (see Chapter 1).Research that advances theory, by contrast, is
usually described ashaving an ii~dltctivequality. On the basis of new evi-
dence, the researcher developsa new theoretical concept or new relation-
ship or advances understanding of existing ones.
Not only does the researcher use data to illustrate the new concept,
he or she may also elucidate the relation of the new concept to existing
concepts. One researcher, forexample, developed the concept of edgework
based on his studies of people who skydive and fromrelated research on
people who seek out other dangerous situations (Lyng 1990).When de-
veloping a new concept, it is necessary to distinguish it from related con-
cepts and to explain its logical and causal connections to others (see also
Wieviorka 1988,1992).
Many theoretical advances come from detailed, in-depth examina-
tion of cases. Exploring diversity, for example, may lead to the discov-
ery of new social arrangements-and practices. The study of behavior of
the groupies who surround certain kinds of rock bands, for example,
might lead to new insights about the importance of rituals in contempo-
rary social life. The mere existence of novel phenomena also may chal-
lenge conventional thinking. Existing theories may argue that certain
ways of doing things or certain behaviors are incompatible, that it has
to be either one or the other. The discovery that "incompatible" ele-
ments can coexist calls such theories into question and may force re-
searchers to theorize about how such logically incompatible things can
coexist.
Research that gives voice also may lead to theoretical advances be-
e-.
cause such research often leaves existing'theories behind in its attempt
to see socialworlds through the eyes of their members. This openness to
the viewpoints of low-status and low-visibility people may expose the
inadequaciesof existingtheoretical perspectives.Finally,work that seeks
to interpret cultural or historical significance may also advance theory
because it too is based on detailed analyses of cases. For example, in-
depth research on the Iranian revolution could lead to new insights on
the importance of the interplay of religious ideology and political orga-
nization in large-scale political change.
Research that seeks to identify general patterns across many cases is
usually associated with the goal of testing theory (via hypotheses), and
less often with the goal of advancing theory, even though, as already
noted, testing theory does refine it. However, the analysis of broad pat-
terns can lead to theoretical advances (see, for example, Paige 1975;
Rokkan 1970,1975; Tilly 1984;Rueschemeyer et al. 1992).Sometimes hy-
potheses fail or are only partially supported, and researchers generally
want to know why. They may study additional patterns in their data to
find out why the theory they are testing does not fit the data well.
For example, using a generally accepted theory as a starting point, a
researcher might test the hypothesis that richer countries tend to have a
more equal distribution of income (that is, within their own borders)
than poorer countries. Analysis of relevant data might show that while
this pattern holds for most countries, among the richest fifteen or so it
does not-they might all have roughly the same degree of equality. This
finding might lead the researcher to speculate about the newly discov-
ered pattern: Why is it that greater national wealth does not lead to
greater equality once a certain level of economic development is
reached7 A variety of factors might be examined in the effort to account
for this pattern. This searchmight lead to the identificationof causal fac-
tors that suggest fundamental revision of the theory used to generate the
initialhypothesis about patterns of income inequality.
While the deduction-versus-induction distinctionis a simple and ap-
pealing way to differentiate kinds of social research, most research in-
cludes elements of both (see Stinchcornbe 1968). For this reason some
philosophers of science (for example, Hanson 1958) argue that all re-
, search involves rehoduction-the interplay of induction and deduction.
I It is impossibleto do research(withoutsome initialideas, even if the goal
is to give voice to research subjects. Thus, almost all research has at least
1 an element of deduction. S i a r l y , almost all research can be used to ad-
IIi, . vance theory in some way. After all, socialtheories are vague and impre-
:\ , cise. Every test of a theory refines it, whether or not the test is
supportive. Research involves retroduction because there is typically a
\ dialogue of ideas and evidencein socialresearch. The interactionof ideas
and evidenceculminates in theoretically based descriptions of sociallife
(that is, in social scientificrepresentations) and in evidence-basedelabo-
rations of social theory.
The Link between Goals and Strategies
It is clear that no researcher can tackle all seven goals at once, at least not
in the same study. A classic view of science says that it is a violation of
the scientific method to try to advance theory (goal 7) and test theory
(goal 2) in the same study. Data used to generate a new theory should
not also be used to test it. Most of the tensions between goals, however,
revolve around practicalissues.
Určeno pouze pro studijní účely
It is difficult, for example, to both examine innizy cases so that a gen-
eral pattern can be identified (goal 1)and study one case in depth so that
its specificcharactercan be understood (goal6). Even when it is possible
to do both, they don't alwaysmix well. What ifthe findingsfrom the in-
depth study of one or a small number of cases contradict the results of
the analysis of broad patterns across many cases?Which finding should
the social researcher trust? However, both kinds of research are impor-
tant because both help social researchers find order in complexity, order
that they can represent in their reports. The first type of research helps
socialresearchers identify what is general acrossmany cases-to discern
the underlying order that exists amid great variation; the other helps
them comprehend the complexityof specific situations directly.
Many different strategies of social research have emerged to accom-
modate its multiple and competing goals. As already noted, a research
strategy is best understood as a pairing of a general research objective
and a specific research method. Each strategy constitutes a way of link-
ing ideas and evidence to produce a representation of some aspect of so-_--___c----..-- ---cial life. ~ese~&sbate~@e~&ucturehow social researchers collect data
ahd-gike sense of what they collect. Even though some strategies are
clearlymore popular than others, there is no single "correct" way of con-
ducting socialresearch.
m the use of qualitativemethods to study commonalities
-"-....~,...,
These three strategies are discussed in detail in Part I1 of this book be-
cause they represent three very common but very different ways of car-
rying on a dialogue of ideas and evidence. The selection of these three
strategies does not imply that other strategies are not important or do
not exist. Indeed, there are plenty of qualitative researchers who study
diversity, and there are many researchers who use comparative methods
to study commonalities.The pairings emphasizedhere (qualitativemeth-
ods with commonalities,comparativemethods with diversity, and quan-
titative methods with covariation)have been selected because they offer
the best illustration of the core features of different methods. They also
provide strong testimony to the unity and diversity of socialresearch.
Qualitative researchers interested in commonalities examine-many__ - . . , . , . . .
aspe+s orfea.&res.of a relatively. . small number of cases in_.dpp.th.A. ...... -
study of how one becomes a marijuana user (Becker1953)is an example
of a qualitative study.
Comparative researchers interested in diversity study a moderate
L ~.
number o E s e s m a compreh~&sivemaniier, though& rGta2 much-kle-.."." ..-,,
tail-- as in_I__most qualitative.- __research:. .. A study of the checkered histoj of
democratic institutions in South American countries is an example of a
comparative study (E. Stephens1989).
Quantitative researchers interestep~inhow variables covary across_ _ . . , . _ ... . , , . . .. . . . ,.
cases--.. typically ex-: a;daeely sm&,.'i+er of features . . ~of cases. .
(that is, variables) across many, many cases. A study of the correspon-
dence between the lntenslty bf-p-ai.wc<@etition and the level of voter
turnout across all countiesin the United Statesis an example of a quanti-
tative study.
These three strategies can...~be plotted.... ,. .. in two dimensions~...... .~.......showingthe...,..~ ~
relati-z
of 'cases studTd...
tween studying cases and studying aspects of cases, or variables. Because
the energies and capacities of researchers are limited, they often must
i
F I G U R E 2.1
',
Cases, Aspects of Cases, and Research Strategies*
Comparative
research
Many
3 9u *
B:
4 0
Quantitative
research
Qualitative
research
Few 1
Few Many
Number of Cases
*Thethree research strategies are qualitativeresearch on commonalities,
comparativeresearch on diversity, and quantitativeresearch on relationships
between variables.
Určeno pouze pro studijní účely
50 Clrnpter2 The GonlsofSocinl Resenrclt 51
II choose between focusing on cases as wholes (qualitative research on
commonalities),focusing on variables (quantitativeresearch on relation-
ships among variables), or balancing the two in some way (comparative
research on diversity). It is possible to gain a detailed; in-depth howl-
edge of a small number of cases, to learn a moderate amount about an
intermediate number of cases, or to focus on limited informationfrom a
large number of cases.
The trade-off between number of cases and number of features does
not concern how much information social researchers can collect. After
all, socialresearchers can collectvolumes of informationon each of thou-
sands and thousands of cases (Davis and Smith 1988).The Internal Rev-
enue Senrice collects detailed information on millions of people every
year. The issue is how much information social researchers, or anyone
else for that matter, can shrdy; lhoru the information is studied (for ex-
ample, is each case examined individually?);and the releunilce of the in-
formation to a particular research question.
Imagine trying to grasp the nature of informal interpersonal net-
works in each of the top 500 U.S. corporations. It might take years to
unravel the informal networks of a single corporation. A social re-
searcher can gain this kind of intimateknowledge about only a relatively
small number of cases.
However, it might be possible to s w e y these same 500 corporations
and find out basic informationlike total assets, profitability, and number
of employees. The information from this s w e y would not add up to
intimate knowledge of each of the 500 corporations, but could be used to
examine relations amongvariables characterizingthese corporations.For
example, does large corporate sue pose an obstacle to profitability?An-
swering this question does not require in-depth knowledge of the work-
ings of nlty of the 500 corporations.Of course, such in-depth knowledge
would improve the analysis of the evidence on sue and profitability and
the representationof the results, but it is not essential to the study of the
general relationshipbetween these two variables.
It is important to note that Figure 2.1 represents the tendencies of
these three strategiesand does not establish absoluteboundaries around
these three strategies in any way. Some quantitative researchers, for ex-
ample, collect hunol-eds of variables on thousands of cases when they
conduct research, and they try to squeeze as much of this informationas
possible into the representations they construct. Of course, these repre-
sentations are still "big picture" representations of broad pattern of
covariationacross cases. Likewise, there are some qualitativeresearchers
who work in teams to increase the number of cases they study. Thus,
Figure 2.1 should be viewed as an attempt to depict the nature of the
typical representations that result from these three common strategies. /,..-.
Table 2.1 maps the relation between these three strategies and the
seven goals of social research discussed in this chapter. The column
headings of the table are the three general strategies; the rows are the
seven goals. The table shows the fit between goeealsddsEategies.
The three-different strategiesrange from intensive (qdifaTve study
of commonalities)to comprehensive (comparative study of diversity) to
extensive (quantitative study of the relationships among variables) in
their approach to cases. An intensive approach is best suited for goals
that involve close attention to specific cases; a comprehensive approach
is best suited for goals that involve examination ofrpatternof similari-
ties and differences across a moderate number of cases; an extensive ap-
proach is best suited for goals that involve knowledge of broad patterns
across many cases. It is important to remember, however, that the three
strategies examined here and in Part I1 are three among many different
strategies of socialresearch.
The goal of identifying general pattern (goalI), for example, is best
served by the quantitative approach, but it is also served by the com-
parative approach, though maybe not quite as well. (Thus, the p17'1lrn11~
strategy for identifying gener$ patterns is the quantitative approach; a
secoildnry strategy is the comp&rative approach.) A pattern is not general
if it does not embrace many cases. Also, most statements about general
The Goals and Strategies of Social Research*
Qunlifntiue Cortrpnmtiue Qrmatilnfiuc
Rcscnrclt Rescnrclr Rcsenrcl~
1. Identifying broad patterns secondary primary
2. Testing/re-g theory secondary secondary primary
3. Makingpredictions secondary primary
4. Interpretingsignificance P*~T secondary
5. Exploring diversity secondary primary secondary
6. Givingvoice primary
7. Advancing new theories P*~V primary secondary
'The thrrr rescarch stratcgirs are qunlil.%hvcresearch on commonalities, cornparalive
rescnrch on diversirv. and ~uanrilntiet.research un relationships bclwern \~ariablrs.
Prinrnnjindicates that the strategyis a very common way of achievinga goal;sccorldnn~
indicates that the strategyissometimesused to achieve a goal.
Určeno pouze pro studijní účely
52 Chnpter 2
patterns involve variables. Both of these features of general patterns
point to the quantitative approach as the primary strategy. The goal of
testing theory (goal 2) is served by all three strategies. Most theories,
however, are composed of abstract concepts that are linked to each other
and thus concern general relationships that'can be viewed across many
cases or across a range of cases. Sometimes a singlecase will offer a criti-
cal test of a theory, but this use of individual cases is relatively rare
(Eckstein 1975). Besides, from the perspective of most theories, single
cases are unique and therefore relatively unreliable as raw material for
testing theories. Likewise, the most appropriate strategy for making pre-
dictions is the quantitative approach. Most predictions involve extrapo-
lations based on many cases, the more the better, as long as they are
appropriate and relevant to the substance of the prediction.
The goals of interpreting significance and giving voice, by conttast,
are best served by a strategy that examinesa smallnumber of cases(often
a single historical episode or a single group) in depth-the qualitative
approach. Similarly, the best raw material for advancing theory is often
provided by strategies that focus on cases, which is the special forte of
qualitativeresearch and one of the strongpoints of comparative research.
However, all research, including quantitative research, can advance
theory. Finally, the goal of exploring diversity is best served by the com-
parative approach. However, because qualitative and quantitative re-
searchcontributeto knowledge of diversegroups, they too serve thisgoal.
The Social Nature of Social Research
Imagine a chart comparable to Table 2.1 constructed for a hard science
like chemistry or physics. Goals 4 and 6 would not exist, at least they
would not be considered major goals, and goal 5 would concern only a
handful of researchers. The remaining four goals (1,2, 3, and 7) are all
served by the quantitative approach-a strategy that addresses general
relationsbetween measurable aspects of the things socialscientistsstudy.
Goals 4,5, and 6 reflectthe socialnature of socialresearch. It is also these
goals that sometimesmake socialscientistsseem "unscientific," especially
to scientists,socialor otherwise, stronglycommitted to the other goals.
Consider again the goal of giving voice. Why should any particular
voice be privileged by social research? Why should a social researcher
try to enhance a particular group's visibility in society? Who cares
whether or not people who are not marginal can understand those who
are? Consider the goal of interpreting cultural or historicalsigniicance.
The Goals of Socinl Resenrcll 53
How do we know that the social researcher is not trying to whitewash
horrific events, or perhaps make the members of a truly destructive
group look We victims of oppression?Fmally, consider the goal of ex-
ploring diversity. By highlighting diversity, a social researcher may glo-
rify it. But too much diversity in society can tear it apart. Might it be
better to emphasize the things that we have in common, what most
members of society share?
These aspects of social research make it an easy target of criticism.
However, it is important to understand that no socialresearch exists in a
vacuum. Research on general patterns, for example, may simply privi-
lege what is nonnative. Allsocialresearch gives voice in one way or an-
other to some aspect of society. Similarly research that tests theories has
implicationsfor how we think about human nature, social organization,
and the differentkinds of socialworlds that are possible to construct. In
fact,because of the socialnature of socialresearch, all social research has
implications for the interpretation and understanding of anything that
people do or refuse to do together. Social research is inescapably social
in its implications. For thisreason, socialresearchers cannot escape bias,
regardless of which goals motivateresearch.
Určeno pouze pro studijní účely