2

What Is a Case Study?

The Problem of Definition

Gerring, J. (2007). “Chapter 2: What Is a Case Study?”. En Case

Study Research, Cambridge, Cambridge University Press: pp.17-36.

The key term of this book is, admittedly, a definitional morass. To refer

tative, small-N, 1 (b) that the research is holistic, thick (a more or less
to a work as a "case study" might mean: (a) that its n1ethod is quali­

comprehensive examination of a phenomcnon),2 (c) that it utilizes a

particular type of evidence (e.g., ethnographic, clinical, nonexperimen­
tal, non-survey-based, participant-ohservation, process-tracing, histori­
cal, textual, or field research),3 (d) that its method of evidence gathering
is naturalistic (a "real-life context"),4 (e) that the topic is diffuse (case
and context are difficult to distinguish), 1 (f) that it employs triangulation

properties of a single observation,7 or (h) that the research invescigates

("rnultiple sources of evidence")/; (g) that the research investigates the

the properties of a single phenomenon, instance, or example. 8

Evidently, researchers have many things in mind when they talk ahout
case study research. Confusion is compounded by the existence of a

I Eckstein (1975); Ge�rgc and Bennett (2005); Lijphart (1975); Onun, Feagin, and Sjoberg
(199 1: 2); Van Evcra (1997: 50); Yin (1994).
2. Goode and Hart (1952: 331; quoted in JVlitchell 1983: 191); Queen (1928: 226); Ragin
( 1 987, 1997); Stoecker ( 1991: 97); Verschuren (2003).
l (;eorge and Bennett (2005); fiamel (1.993); Hamn1ersley and Go1nm (2000); Yin (1994).
·I Yin (2003: 13) .
., Yin (1994: 123).
(> !hid.

· ( :;nt1plwll and SL1nky ( 1963: 7}; Eckstein (1975: 85).

H · 11,· i� i� 111·< >h;1hly 1 ht' 1rnJsl C(llllllHin 11ndcrst:inding of the rern1. George and Bennett (2005:
I'/), 1 1>1· 1·x;11nplc, dl"li1H· :l l,l�t· .1� "':111 i11-�1;111n· Df ;1 cbss of events." (Note that else\vhere
111 tl11· •..11111· d1.1)'1l'1· ilw1· 1111.-i 1li.11 iii,· .111.tly�ic; of th;11 in!-.r;111cc will he sn1:tll-N, i.e.,
q11.1l 1 1.111H·. I 'w(· .i\·," t l,l,·11 I 'I Ill I I 1, ') .111d l 111,··. ( _)()()/: \ <; l )
.
I '
18 [. Thinhing about Case Studies
large number of near-synonyms - single unit, s ingle subj ect, single case,
N = 1, case-based, case-co ntr ol, case history, case 1nethod, case record,
case w ork, within-case, clinical research, and so forth. 9 As a result of this
profusion of terms and 1neanings, pro ponents and opponents of the case
study marshal a wide range of argum ents but do not seem any clo ser to
agreement than when this de bate was first broac��d sev eral decades ago.
Jennifer Platt notes that "111uch case study tbeor1z1ng has been conce�[U­
ally confused, because to o many different themes have been packed into
,,,
th e 1"dea 'case s tud y. 10
_ .
H ow then should the case s tudy be understood? The first six options
enumer:tcd �bove (a-f) see1n inappr opriate as general definitions of
the topic, since each implies a substantial shift in m eani�g �elative to
es tabli shed usage. One canno t subs titute case stud for qualita tve, eth o­
! � �
graphic, process-tracing, holistic, naturalisti�, diffuse, or triangulation
_ of nation-states (across son1e te1nporal frame); in a study that atternpts
WI" thout feeling that some thing has been los t 111 translatio n. These ter1ns
are perhaps better unders too d as describing certain kinds o f case studt. �s,
not the topic at large. A seventh o ption, (g), equates the case study with
the study of a single o bservation, the N = 1 re search design. This is
_ Fo r s tudents of political science, the arch e typal case is the do1ninant
l ogically impossible, as I will argue. The eighth opt10n, (h), centenng
on phenomenon, instance, or example as the key term, 1s c orrec as far
�
as it goes but also ambigu ous. Imagine asking sorneone, "What 1s your
-instance?" or "What is your phenomenon?" A case study presupposes a
relatively bounded phenomenon, an implication that none of these terms
captures.
Can this concept be reconstructed in a clearer, more productive
fashi on? I begin this chapter by stipulating a series of definitions. I then
present a typology of researcb designs, understood according to the pat­
terns of spatial and temporal e vidence that they draw upon. A final sectJOn _
addresses a c entral definitio nal questi on, na1nely, whe ther case studies
should be unders tood as exclusively "small-N" analyse s.
9 Davidson and Costello (1969); Franklin, Allison, and Gorman (1997); Hersen and
Barlow (1976); Kazdin (1982); Kratochwill (1978).
10 Platt (1992: 48). Else\\'here in this perceptive article, Platt (1992: 37) con1ments: " the
_
diversity of the themes \\'hich have been associated with the term, and the vague1�ess of
50111e of the discussion, causes sorne difficulty.... In practice, 'case study method ,n tts
heyday [in tht intenvar years] see,ns to have n1eant some pern1utat1on of the fol\o\v1ng
components: life history data collected by any means, personal documents, unstructured
or not any attempt was 01;-1cle to generaI17.e from the111 , aO)' atre1npt at holt�tic study, ::ind
intervie\v data of any kind, the close study of one or a s1nall number of c�ses \vhether
. _
.'
n·i·.111;1r c1npi ri 1 ;11 1111wv1 itlll wil h 1·:1,·h 11l \wr."
· ·
1Hlll·qtt;llHILt11Vl" lI,l t;t -,11,
,·,lv, -
. . ,s
.
. . These L·o11111rnH::nts have nc1tlwr :1 nL'Cl'S�:1rv 1(1!',II .1\ 1\01. ,1
\'<
What ls a Case Study? 19
Definitions
}' or purposes of m ethodological di scu ssi on, it is essential to develop a
v ocabulary that is consistent and clear. In arriving at definitions for key
terms, I rely on ordinary usage (within th e language regi on of social
science) as much as possible. However, becaus e o rdinary usage is often
an1biguous, enco1npassing a range of 111eanings for a given term (as we
have seen above for "case"), some concept reconstructi on i s unav oidable.
At the end of this discussion, I hope it will be cl ear why this particular way
of defining terms might be useful, at leas r f or m eth odol ogical purpos es.
11
Case conno tes a spatially d elimited phen o1nenon (a unit) observed at a
single p oint in tirne or o ver some period of tin1 e . It cornprises th e type of
phenomenon that an inference attempts to explain. Thus, iu a s[udy that
attempts t o elucidate certain f eatures of nati on-states , cases are comprised
to explain the behavi or of individuals, cases are comprised of individuals,
and so f orth. Each case may provide a single observati on or multiple
(within-case) observations.
political unit of our time, the nati o n-state. However, this is a matter o f
convention. The study of smaller social and political units (regi ons, cities,
villages, communities, social groups, fan1ilies) or specific institution s
(political parties, interest groups, businesses) i s equally commo n in 1nany
social science disciplincs. 12 In psych o logy, 111edicine, and social work the
notion of a case study is usually linked to clinical research, wh ere individ­
uals are the preferred units of analysis.13 Whatever on e's ch osen unit, the
1ne tho dological issues attach ed to the cas e study have nothing to do with
th e size of the cases. A case may be created o ut of any phen o1nenon so long
,.1 s it has identifiable bo undaries and co1nprises the primary object of an
inference.
Note that the spatial boundari es of a case are often m or e apparent
i-han its ten1poral boundaries. We know, n1 or e or less, where a country
hcgins and ends, while we may have difficulty explaining when a country
11 In the following analysis, 1 take a "rninimal" approach to definition (Gerring 2001:
_ _
( :haprcr 4; Gerring ,1nd Barresi 2003). Scholars e,nbedded in a particuh1.r rescc1rch st'tring
_
1n:1y c.:huosc sorncwhat different terms and 1neanings.
1 ·' hlr di.,n1,sion of .\nhnational studies in political science, see Snyder (2001).
11 l :or\i11i (21l04); l),1vidso11 and Costello (1969); Hcrsen and Barlo\v (1976); Franklin,
. .. t\ll1,,l11, .11111 (;,H·111;111 ( I 'Jl)7); Rohin:-,on (200 I). For discussion of the 1neaning of the
1,·1111 "'1.1�c \t1HI\'," �t"(' lk11h1,;ll, ( ;u]d:-,tt"ill, :ind J'dl'ad ( 1987: 37 1 ); C:unningharn (1997);
r-.i.-111.1111 ( 1,iiiiq 1 .11,.i v'"1·., 111111·11 (.)on n.
 I. Thinking about Case Studies
20
begins and en<ls. Yet some te1npora l bounda ries 111ust be assun1ed. This
is particula rly importa nt when ca ses consist of discre te events - cri ses,
revo lut ions, leg is lative a ct s, and so forth - within a single unit. Occasion­
a lly, the ten1pora l boundar ies of a case are 111ore obvious than its spatial
bounda ries. This i s true when the phenomena under s tudy are eventful but
the unit undergoing the event is amorphous. For exan1ple, if one is study­
i ng rerror ist attacks it may not be cle ar how the sp atial unit of analysis
should be un derstood, but the events themse lves may be wel l bounded.
A case study may be understood as the intensive study of a single case
where the purpose of that s tudy is - at least in part - to shed light on a
la rger c la ss of cases (a population). Case study research n1ay incorporate
sever a l cases, that is, n1ultiple ca se s tudies. Ho\vever, at a certa in point it
wi ll no longer be poss i ble to investig ate those cases intensively. At the point
wh er e the emphas is of a study shifts from the individua l case to a samp le
of c ases, we sha l l say that a study is cross-case. Evidently, the distinction
between case study a nd cross-case study is a m atter of degree. The f ewer
ca ses there are , and the more intens ive ly they are studied, the more a work
merits the a ppella tion "case study." Even s o, th is proves to be a u seful
distinct ion, and much follows fron1 it. Indeed, the entire book rests upon
it. Al l emp ir ica l work may be cl assified a s either c a se study (compris ing
one or a few cases ) or cross-case study (comprising m any cases).
An addition a l i1nplicatio n of the term "ca se study" i s that the unit(s)
under specia 1 focus is not perfectly representative of the popul ation, or is
at least questionab le. Un it ho1nogeneity a cross the sample and the popnla­
tion is not assured. If, for ex a1nple, one is studying a sing le H2 0 n1olecule,
it n1ay be r eason a ble to assume tha t the behavior of that molecule is iden­
tica l to that of all other H2 0 molecu les. Under the circurnsta nces, one
would not refer to such an investigation a s a "case s tudy, " reg ardless of
how intensi ve the investiga t ion of that s ing le molecu le might be. In social
science settings one r are ly f aces phenomena of such consist ency, so this
is not an issue of gre at practical sign ificance. Nonetheless, intrinsic to the
concept is an element of doubt about the bi as that n1ay be conta ined in a
sa mple of one or severa l.
A few addition a l terms may now be formally defined.
An observation is the 1nost basic e lement of any empirical endea vor.
Convention a lly, the number of observations in an ana lysis is referred to
with the letter N. (Confnsingly, N may a lso be o sed to design ate the num­
ber of czi ses in a study, a usage that is usu a lly cle ar fro 1n context.) A sing le
observation n1ay he understood as containing several din1cnsions, each
nl ,..v h il · h 111;1y he 111c;1s11rcd (;1cross disp;1r;1tc ohsl'rv;itio 1 1 ·, ) .1·, . 1 l'ill'i, ,/1/('.
What Is a Case Study? 21
Where the propos ition is causa l , these may be subdivided into depen­
dent (Y) and independent (X) variables . The dependent va riab le refers to
the outcome of an inves tigation. The independent variable refers to the
explanatory (causa l) factor, that wh ich the outcome is s upposedl y depen­
dent on.
A case n1a y consis t of a si ngle observa tion (N = 1). This would be true,
for example , in a cross-section al a na l ysis of 1nultiple ca ses. In a ca se study,
ho,vever, the case under study a hvays provides n1ore th an one observarion.
These may be constructed di a chronica l ly (by observing the case or some
subset of \Vithin-case units over time) or synchronicall y (by observin ba
within-case varia tion at a single point in time), as discussed bcln\v.
This is a cl ue to the fact that case studies a nd cross-ca se studies usuall y
oper ate at different level s of analysis . l'he case study is typically focused
on vvithin-case variation (if there i s a cross-case co1nponent, it is prob a ­
bly secondary in i1nporra nce to the within -ca se evidence). The cross-case
s tudy, as the n a n1e suggests, is typically focused on cross-ca se variation ( if
there is also vvithin-case variation, it is prob a b l y seconda ry in in1portance
to the cross-case evidence). They have the san1e object in view - the expla­
nation of a population of cases - but they go about this task differentl y.
A samtJ!e consists of whatever cases are s ubjected to forn1a l analysis;
they are the in1n1ediate subject of a study or case study. (Confu singly,
the term "sa n1ple" n1ay a lso refer to the observations under s tudy. But
at present, we treat the san1pl e as consisting of c ases.) In a ca se study,
the sample is sn1 a ll, hy definition, cons isting of the single case or ha ndfu l
of cases th at the rese archer h a s under her lens. Usua lly, hovvever, vvhen
one uses the tern1 "sa1nple" one is i1nplying th a t the nun1ber of cases
is large. Thus, " sample-b ased work" v.rill be understood as referring to
large-N cross-case methods - the opposite of case study work. To reiterate,
the feature distinguishing the case study format from a sample-b ased (or
"cross-case") research design is the number of ca ses fa l ling within the
s;.1 n1ple - one or a few versus n1a ny - a nd the corresponding thoroughness
with \vhich each case is studied. C ase studies, l ike large-N sampl es, seek to
represent, in a ll ways relevant to the propos ition at hand, a population of
L\1ses. A series of case studies might therefore be referred to as a sa1 np le if
they are r elativ ely brief and re lat i vely numerous; it is a matter of emphasis
;111d of degree. 'J'h e 1nore case s tudies one has, the less intensivel y e ach one
i'i sn1dicd 1 and the n1orc confident one is in their representativeness (of
sonH' hro;1dcr popuL1tion), the 1nore likely one is to describe the1n as a
'>.11npl(' r:U ht-r tli:1 1 1 ;1s :1 SLTics of c;1sc studies. For practica l re asons -
111ilc',',, 1 l 1 .1t 1'>, .1 '.!11tly i'> ('7'tr;1ordi11;1ril� long · 1 h c c:1sc study research
1
22 I. Thinking about Case Studies TABLE 2.1. Case study dataset with tu;o cases

format is usually limited to a dozen cases or fewer. A single case is not

unusual.
Granted, in so1ne circu111srances a single study n1ay con1bine the t\VO y
Obs 1 . 1
elen1ents - an intensive case study and a 111ore superficial analysis con­
Ohs 1.2
ducted on a larger sample. These additional c,1ses are often brought into Obs 1.3
the analysis in a peripheral \Vay - typically, in an introductory or con­ Obs 1.4
cluding section of the paper or the book. Often, these peripheral cases arc Obs 1.5
surveyed through a quick reading of the secondary literature or through Obs 1 . 6
Obs 1 . 7
a statistical analysis. Son1etimes, the status of these informal cases is left Obs 1 . 8
in1plicit (they are not theorized as part of the for111al research design). (Jhs l.9
This 111ay be v..rarranted in circun1stances ,vherc the relevant compari­ Case 1 Obs I.JO
son or contrast between the forn1al case(s) under intensive study and the Obs l . J J
Obs l.12
peripheral cases is obvious. Thus, studies of American exceptionalis1n, in Obs 1.13
enun1erating features of the A1nerican experience, often assume that the Ohs 1.14
United States is different fron1 European countries in relevant respccts. 1 4 Ohs 1.15
In this situation, the additional cases - the UK, France, Germany, and (Jbs 1.16
so on - provide the necessary background for whatever arguments are Obs / . 1 7
Obs l . / 8
being n1ade about America. They are present, in the sense that they carry Obs l.19
an irnportant burden in the analysis, hut perhaps they are not forn1ally Popufation Sample Obs 1.20
accounted for in the author's research design. For our purposes, what is Obs 2. 1
Obs 2.2 ...... ... ... ...
significant is that 1nost vvorks combine case study and cross-case study . '
Obs 2.3
components, \vhether or not the latter are explicit. Methodologically, Obs 2.4
these approaches are distinct, even though they n1ay be integrated into Ohs 2.5
a single work. (Indeed, this is a good way of approaching many subjects.) Obs 2.6
Continuing with our review of key ter1ns, the sample of cases (large (Jhs 2. 7
Obs 2.8
or sn1all) rests within a population of cases to which a given proposition Obs 2.9
refers. rfhe population of an inference is thus equivalent to the breadth Case 2 Ohs 2. 10
or scope of a proposition. (I use the terms proposition, hypothesis, infer­ Obs 2. 1 1
ence, and argu1nent interchangeably.) Note that most san1ples are not Ohs 2. l2 , ...... ..... ,......
Obs 2.13
exhaustive; hence the use of the term "sa1nple," referring to samfJ!ing
Obs 2.14
from a larger population. Occasionally, however, the sample equals the Obs 2.15
population of an inference; all potential cases are studied. Obs 2.16
For those familiar with tbe rectangular form of a dataset, it may be help­ Obs 2. 1 7
Obs 2.18
ful to conceptualize observations as rows, variables as colu1nns, and cases
Obs 2. 19
...
as either groups of observations or individual observations. Several pos­ Obs 2.20
sibilities are illustrated in the tables presented here: two cases (Table 2.1),
multiple cross-sectional cases (Table 2.2), and tin1e-series cross-sectional
cases (Table 2.3).

H A l l l ('III.I ( j (J <) l ).
I', i p 1 i l . 1 t i 1 1 1 1 l ; S.1ntpk I ; ( :.1:-.('" .1.; ( }h..,crv.1! i()ns ( N J : ..:..:. 40; Vari;1hks = .(

' I
TABLE 2.2. Cross-case cross-sectional dataset with /Ort)' cases TABLE 2.3. Time-series cross-sectional dataset

' '
Y, y
Case I (T I )
X, X. Y y

{
Obs I. I
('asc :! (T 2)
Obs I
Ohs 2 Obs 1.2
Case J Obs 1.3 cr,J
(T 4)
Case 3 Obs 3
Obs 1.4
( 'ase5 (Ts)
('asc 4 Ohs 4
Obs 1.5
('ase () (T 1 )
Ohs 5

'"" {
Obs 2. 1
er,)
Ohs 6
C)bs 2.2
(T ,)
Case 7 ()hs 7
()bs 2.3
(T ,)
Case 8 Ohs 8
Obs 2.4
(T 5)
Case 9 Ohs 9
()bs 2.5
(T , )
Case JO Ohs 1 0

c,. , {
Obs 3 1
(T 2)
Case J J ()hs J 1
Obs 3.2
(T ;)
C'ase 12 Ohs 12
Ohs I 3 Obs 3.3 ..
Case 13
Obs 3.4 cr4J
,

(Ts)
Ca.se 14 Ohs 14
Obs 3.5
(T 1)
Obs 15

,,. ,{
Case 15
Ohs 4. 1
(T 2)
Case 16 ()bs 16 , .. ..... . ........
Ohs 4.2
(T ;)
Case 1 7 Ohs 1 7
Ohs 4.3
(T 4)
Case 18 Obs I 8
Obs 4.4
(T s)
Case 19 Obs !Y Population Sample Obs 4.5
(T 1 )
Population Obs 20

'"'" '{
Sample Case 20
Obs 5.1
(T 2)
Case 21 Ohs 2 J
Obs 5.2
Case 2) Obs 22
Obs 5.3
(T 4)
Case 23 Ohs 23 (Ts)
Obs 5.4
(Ts)
Case 24 Obs 24
Ohs 5.5
(T 1)
Obs 26 ,...... ...... ........
C'ase 25 Obs 25

{
(T 2)
Obs 6. l
Case 26 Obs 6.2
(T ,)
C'ase 27 Obs 27 Case 6 Obs 6.3
(T4)
Case 28 Ohs 28 ()bs 6.4
(Ts)
Case 29 Obs 29 Obs 6.5
(T ,)

'""' ' {
Obs 3 0
()bs 7.2 (T 2 )
<;'ase 30 Obs 7. 1
Case 3 1 Obs 3 1
Case 32 Obs 32 Obs 7.3 (Ts)
Case 33 Ohs 33 Obs 7.4 (T,)
Case 34 Obs 34 Obs 7.5 cu
(T 1)
C'ase 35 Obs 35

'"" {
Ohs R. l
Cl,)
Case 36 Obs 36 Obs R.2
(Ts)
C'ase 37 Obs 3 7 Obs 8.3
(T,)
,... .
Case 38 Obs 38 Obs 8.4 ...... · · · ··· · ·
Case 39 Ohs 39 .. ,. ....
Obs 8.5
,
(Ts)
Case 40 Obs 40

l '1 1 1 i 1 1 l . 1 1 i ( 1 1 1 I ; S;1111pk I ; < ::1scs = 8; ()bscrv,nions (N) = 40; Time (T) = l-5;
Popt1Lition = l ; San1plc = 1 ; Co.scs = 40; Observations (N) = 40; Variables = 3 . \'.11 !. , 1 , 1 ,·� l.

.' \
26
It must be appreciated that all these terms are definable only by ref­
erence to a particular proposition and a corresponding research design.
A country 1nay function as a case, an observation, or a population. It all
depends upon what one .is arguing:. In a typical cross-country tin1e-series
years. 15 Hov,rever, shifts i n the level of analysis of a proposition necessarily
regression analysis, cases are countries and ohservations are country­
change the referential rneaning of all terms in the semantic field. If one
111ovcs dov,rn one level of analysis, the new population lies within the old
population, the ne\V sample within the old san1ple, and so forth. Popula­
tion, case, and observation are nested within each other. Since most social
science research occurs at several levels of analysis, these terms are gener­
ally in flux. Nonetheless, they have distinct meanings ,vithin the context
of a single proposition and its associated research design.
Consider a survey-based analysis of respondents "rithin a single coun­
try, under several scenarios. Under the first scenario, the proposition of
interest pertains to individual-level behavior. It is ahout hov,r individuals
behave. As such, cases are defined as individuals, and this is properly clas­
sified as a cross-case study. Now, let us suppose that the researcher wishes
to use this san1e survey-level data drawn from a single country to eluci­
date an inference pertaining to countries, rather than to individuals. Under
this scenario, each poll respondent constitutes a ,vi thin-case ohservation.
If there is only one country, or a few countries, under investigation -
and the inference, as he.fore, pertains to n1ultiple countries - then this
study is properly clas�ified as a case study. If 111any countries are under
study (with or with out individual-level data), then it is properly classified
as a cross-case study. Again, the key qnestions are (a) how 111any cases are
studied and (b) hoVir intensjvely a.re they studjed - \vith the understanding
that a " case" e1nbodics the unit of concern in the central infcrence.
To co1nplicate n1atters further, the status of a work 111ay change as it is
digested and appropriated by a com1nunity of scholars. A meta-analysis
is a systen1atic atten1pt to integrate the results of individual studies into a
quantitative analysis, pooling individual cases drawn from each study into
a single dataset (with various v..reightings and restrictions). The ubiquitous
literature review or case study survey a i1ns at the same objective in a less
synoptic fashion. Both statistical n1eta-analyses and narrative literature
reviews assirnilate a series of studies, treating them as case studies in
authors. 1 6
some larger project - v,rhether or not this was the intention of the original
I ', \( "(', L)'. . , l ' 1 1t'\v.,r•,l,i ('t .d. (.li)IH)).
I ,. \ 1 1 1·., · 1 , I I " I \\' 1 I•,," I I ' 1 1!) I ) ; 1 1 1, .1·. ( I ' I '·11.
I. Thinking about Case Studies What Is a Case Study? 27
A Typology o f Covariational Research Designs
'In order to better understand what a case study is, one n1ust con1prehend
wha : 1t_ '._ s n t. 'l'he distinctiveness of the case study 1nay be clarified by
� _
plac1 �g 1t vv1th1n a broader set of 1nethodological options. Here , I shall
classify research designs according to (a) the number of cases that they
encori:pass ( ne, several, or 1nany ) , (b) the kind of X/Y variation that they
�
exploit (s � at 1al or temporal), and (c) the location of that variation (cross­
.
case or vv1th1n-case). This produces a typology with ten possible cells, as
depicted in Table 2.4.
Variations on the case study forn1at occupy five of these ten cells, desig­
nated by the shaded regions in Table 2.4. Type 2 represents variation in a
single case over time (diachronic analysis). Type 3 represents \Vithin-case
variation at a single point in tin1e (synchronic analysis). Type 4 combines
synchronic and diachronic analysis, and is perhaps the most comnion
appro::lch in case study work. "fhus, Robert Putna,n's classic study of
Italy, Mal<.ing Democracy Work, exploits variation across reo-ions .and
over ti1ne in order to test the causal role of social capital. 1 7 °
It is con1n1011 to combine several cases in a single study. lf the cases are
con1prised of large territorial units, then this co1nhination 1nay he referred
to as the "co1nparative" 1nethod (if the variation of interest is prin1arily
_ or the "co1nparative-historical" method (if the variation of
interest is both synchronic and diachronic). 1 8 It should be pointed out
synchron1c)
tha� these tern1s are used pri1narily \vithin the subfield of con1parative
.
politics. Other tern1s, such as "n1ost-sin1ilar" and "most-different," 111ay
be used as v,rell. 'Thus, \vhile a case is alv\.rays singnlar, a case study v,rork
or research design often refers to a study that inclndcs several cases.
The larger point is that the evidentiary basis upon which case stud­
ies re�y �s plural, not singular. Indeed, there are five possible styles of
C( J �ar1at1onal evidence in a case study. Usually, they arc intermingled _
.
d, tferent sorts of analysis will be employed at different stages of tbe anal­
ys1 � - so that it is often difficult to categorize a study as falling neatly into
a smgle cell in Table 2.4.
' !"he botto1n half of Table 2.4 lays out various cross-case research
� lcsig11s, ,vhcrc the most important elen1ent of the empirical analysis
111volvcs con1pansons _ across many cases (1nore than a handful). Cross-case
I 1 ' 1 1 1 1 1 . 1 1 1 1 ( ] 'J'l.l).
11' t l 1 1 1 I H " ,·1,111p.11'.1 tin· 11H·1hod .\( (' c :oll itT ( 1 99.1 ) · I i · 1h·1rt ( 1 97'1 1 971) P · ) · :l!l d
' , · 11 , , . ; I Ze\VOfS{l
.
l . - i 1 1 w ( I 'I ' I I ) ; R 1 , l 1 1 1 · 1 i l 'Jr,')); .111d \11wkcr ( I ')7(,). ( ) 1 1 1 l w co111p.1r:11ivl".. historicil method
,\ l . 1 l 1, >1u·1· ,11 1 d 1{1 1 , •,, l 1 , · 1 1w1 ,·1 ( . ' 1 11 1 \ ) , ( h1 i l w 111�1<11'\' , i i 1 l w n11np:1r:lli 1·1· llH"th ! id, ,l
1 , · 1 1 1 1 1 l i . 1 1 1 1 . 1 1 1 ,·11·. 1, ... I 1 , , l 'o 1 , , , · 1 l ' l ' J 1, ,., , I . 1 • • •. , , , · l l I I . , 1 1 \
 28 I. Thinking about Case Studies What Is a Case Study? 29

TABLE 2.4. I<esearch designs: A covariatiunal t)'pology evidence. The classification of a research design always depends upon
the particular proposition that a researcher intends to prove. Potentially,
each of the foregoing cross-case 1ncthods 111ight also be employed in the
Spatial Variation Tcn1poral Variation
capacity of a case study. (That is, a case study 1nay enlist cross-sectional,
No Yes ti1ne-serics cross-sectional, hierarchical, or hierarchical time-series tech­
1Yone [Logically impossible]
2. Singlc�case study niques.) It all depends upon the proposition in question (i.e., what sort
(diachronic)
One { of phenomena it is about, and hence what sort of phenomena constitutes
3. Single-case study 4. Single-case study
Within-case (synchronic) (synchronic + diachronic) "cases") and on the degree of analytic focus devoted to the individual
cases.
Cross- case &
Several -C 5. Comparatl\· e method
ivithin-case
The N Question
Cross- case
A1any { Traditionally, the case study has been identified \.Vith qualitative n1ethods
Cross-case & and cross-case analysis \Vith quantitative methods. This is how Franklin
v.1ithin-case 9. H10rarchical
Giddings put the 111atter in his 1924 textbook, in \Vhich he contrasted t\VO

,Vote: Shaded cells are case study research designs.

analysis vvithout any explicit temporal con1poncnt (type 7) is �1su � lly clas­
sified as cross-sectional, even though a ten1poral..component 1s s1111ulatcd
with independent variables that are assun1ed to precede the dependent
variable. An exan1ple was illustrated in 1'able 2.2. When an explicit ten1-
poral coinponent is included, we often refer to the analysis. as tin1e-scries
cross-sectional (TSCS) pr pooled time-series (type 8). This format was
illustrated in "fable 2.3. When one exa1nines across-case and within-case
variation in the same research design, one is said to be e1nploying a hier­
archical model (type 9). Finally, when all forms of covariation are enlisted
in a single research design, the resulting method n1ay be described as a
hierarchical time-series design (type 10 ). 1 9
I t bears repeating that I have listed the n1ethods most con11nonly iden­
tified vvith these research designs not with the intention of distinguishing
labels but rather with the intention of illustrating various types of causal
fundan1entally different procedures:
In the one \Ve follo\.v the distribution of a particular trait, quality, habit or other
phcnon1enon as far as \Ne can. In the other we ascertain as con1pletely as v·.: e
can the nun1ber and variety of traits, qualities, habits, or ,vhat not, co1nbincd in
a particular in::.tance. The first of these procedures h.:is long been known as the
statistical method. . . . The second procedure has ahnost as long been known as
the case n1ethod.20
In the intervening decades, this disjunction has become ever more en-
'
sconced: a contrast betwccn " stats " anll " cases, " " guant " and " qua1. "
Those who work with numbers are apt to distrust case study methods,
while those who work vvith narratives are likely to be favorably disposed.
I believe that this distinction is not intrinsic, that is, definitional. What
distinguishes the case study n1ethod fron1 all other methods is its reliance
on evidence drawn from a single case and its attempt, at the sa1ne time,
to illun1inate features of a broader set of cases. It follows from this that
the number of observations (N) e111ployed by a case study may be either
s111all or large, and consequently may be evaluated in a qualitative or
qu�1ntitative fashion.2 1

19 It will be noted that, like n1ost case studies, hierarchical models involve a 1nove1ncnt across
levels of analysis. Ho,,vever, while a case study moves dou.ni fron1 tbc_ primary level of
analysis (to within-case cases), a hierarchica� model moves �P· Th �1s, if classroorns are
the pri tnary unit of analysis in a study, one might en1ploy a h1erarch1cal model to control
for the effects of brgcr cnscs - schools, districts, regions, and so forth . But one \vould 1:ot
cinploy individu;i\ studt·nt· s ;1s cases in such an analysis (not, that 1s, w 1 t l 1 11111 ch:1ng1ng
ilw 1 1 1 1 i 1 ()( :11t.d:· ,i, f()r ilH· l'lltil'\' ,1 11dy).
·' 0 c ;idd111i-:., ( 1 924: 94). Sec al�o IVft"f"hl ( 19 5 4); Rice (192 8: C:h,1pter 1); anJ Stouffer (1941:
\4l)).
·' 1 l'lii, ,t·L·1 i1rn l'xpL1i11:-. :111d cL1horatcs on a therne rirst articulated by Lundberg (1941),
lull1 >INt·d liy ( ::11 1 1plicll ( I '>7.'i/1 ')HH) - irself :1 revision of Campbell's earlier perspective
({ . 1 11q1lwll .l1lll S1. 1 1 il1·v 1 ()/,.l). I li�corit ·. 11 h:1ll:is1 for t·his vit.:w 1n:1y be g;irncrcd froin
i ll! ' 111 ' 1,I < 1 1 1·\111· 1·1 1 1 1 1·1 1 1 . i l 1 1··,1·, 1 1·, 11 1 1 1 p,r, l 1 01l1J)', }', 1·1> 1 1 t 1 1H11 1 h· d;Hl'd 1 0 i-lw puhlicirion of
( , 1 1 ·. 1 . 1 1 I l 1 n , d , , 1 I n l 1 1 w r '-. I f,·111,·11/,· ,/,·1 / ' ·· \'• f,,.,,/,,. .. tl,, i 1 1 I .'i<,0. 1 1 1 1 l 1 i, \V ( ) i I. , I k 1·,l' 1 1 .1 1 u l
30 1 . Thinhing about Case Studies
In order to see v.rhy this 1night be so, let us consider how a case study
of a single event - say, the French Revolution - \Vorks. Intuitively, such
a study provides an N of 1 (France). If one ,vere to broaden the analysis
to include a second revolution (e.g., the An1erican Revolution), it would
be conunon to describe the study as cotnprising two observations. Yet
this is a gross distortion of what is really going on. The event known as
the French Revolution provides at least tv·.ro observations� for it will be
observed over time to see -...vhat changed and what rc1nained the same.
These patterns of covariation offer essential en1pirical clues. l'hey also
construct 111ultiplc observations from an individual case. So N = 2, at the
very least (e.g., before and after a revolution), in a case study of type 2
(in Table 2.4).
If, instead, there is 110 te1nporal variation - if, for exan1ple, the French
Revolution is exa1nined at a single point in time - then the investigation is
likely to focus on cross-sectional covariational patterns within that case,
a case study of type 3 (in Table 2.4). If the primary unit of analysis is the
nation-state, then ,..vithin-case cases 111ight he constructed fron1 provinces,
localities, groups, or individuals. The possibilities for within-case analysis
are, in principle, infinite. In their pathbreaking study of the International
Typographers Union, Lipset, Tro\V, an<l Colen1an note the variety of
within-case evidence, which included union locals, union shops (vvithin
each local), and individual men1bers of the union.22 It is not hard to
see why within-case N often swa1nps cross-case N. 'fhis is bound to be
true \Vherever individuals comprise vvithin-case ohservations. A single
national survey \Viii produce a much larger sample than any conceivable
cross-country analysis. Thus, in 1nany circumstances case studies of
type 3 con1prise a larger N than cross-sectional analyses or time-series
Barlo\v (1976: 2-3) report, Fechner developed "1neasures of sensation through several
psychophysical n1cthods. With these 1nethods, Fechner \Vas able to detennine �ensory
thresholds, just noticeable differences (JNDs) in various sense tnocblities. What 1s com­
mon to these methuJs is the repeated measurement of a response at Jifferent intensities or
different locations of a given stimulus in an individual subject . . . It is interesting to note
that Fechner was one of the first to apply statistical n1ethods to psychological problems.
Fechner noticed that judg1nents of fJNDs] in the sensory tnodalities varied son1e,.vhat
from trial to trial. To quantify this variation, or 'error' in judgrnent, he borrowed the
nonnal law of error, de1nonstrated that these 'errors' were normally distributed around
a mean, which then became the 'true' sensory threshold. This use of descriptive statis·
tics anticipated the application of these procedures to groups of individuals at the turn
of the century when traits of capabilities were also found to be normally distributed
around a n1ean." Hersen and Barlow note that Fechner, the pioneer, " was concerneJ
with v;1ri,1hility 1l'ithi11 the subject." See also Queen (1928).
'·' l ip�t" I , ' l 'ruw, .11HI ( :olt-1na11 ( l 9.'ifr 422).
What Is a Case Study? 31
cross-sectional analyses. For example, a recent review o f natural resource
management studies found that the N of a stndy varies inversely with its
geographic scope. Specifically, case studies focused on single communities
tend to have large san1ples, since they often en1ploy individual-level
observations; cross-case studies are 1nore likely to treat comn1unities as
comprising observations, and he11ce have a smaller N.23 This is a con11non
pattern.
Evidently, if a case study combines ternporal and within-Ca/Jc variation,
as in case studies of type 4, then its potential N increases accordingly. And
if cross-case analysis is added to this, as in the con1parativc method or
the comparative-historical method (types 5 and 6 in Table 2.4), then one
realizes a further enlargement in pote11tial observations.
These facts hold true regardless of whether the method is experimen­
tal or nonexperi1nental. It is also true of counterfactual reasoning, which
typically consists of four observations - the actual (as it happened) before
and after observations, a11d the before and after observations as recon­
structed through counterfactual reasoning (i.e., with an i magined inter­
vention). ln short, the case study docs not preclude a large N. It sin1ply
precludes a large cross-case N, b y definition. Indeed, many reno\vned case
studies are data-rich and include extensive, and occasionally quite sophis­
ticated, quantitative analysis. Frederic Le Play's work on working-class
families incorporated hundreds of case studies. 24 Robert and Helen Lynd's
study of Muncie, India11a, featured surveys of hundreds of respondents
in "Middletdwn." 25 Yankee City, another pioneering community stu<ly,
included interviews with 17,000 people.26
What, then, of the infamous N = 1 research design that haunts the
i1naginations of social scientists every\vhcrc?27 This hypothetical research
design occupies the empty cell in Table 2.4. The cell is empty because it
represents a research design that is not logically feasible. A single case
()hserved at a single point in tin1e without the addition of within-case
observations offers no evidence whatsoever of a causal proposition. In
1 rying to intuit a causal relationship from this snapshot one would be
t·ngaging in a truly random operation, since an infinite number of lines
llligbt he drawn through that one data point. l do not think there are any
'1 l't)lt"t"1(" ;111d ( )s1ron1 (2005: 1 1).
'·I l ', l' ,11lk 1 · ( I tJ70).
I n1.I , 11HI l .ynd ( 1929/ I <J '16).
... \\'.1111("1" , I I H I ! 111\t ( J 1).-1 ] ).
:\, lw11 . 111d '.11i 1 l . i l ( I ' J .� ' l ) ; ( ;,·.Ide\ ( I 'l 1JO); ( ;(lldd11>1·1w
{ 1 997); King, l( l'o h,111e, :111d Vcrh,1
1 1 •1•1,1 I . I wln-1 ,,, ,11 i I '1.'J ·, : 1 1 ) . I 'i; I 1 , 1 ' J ' ! . J ) .
'l'l.
32 I. Thinking about Case Studies
exa111ples of this sort of investiga.tion i n social science research. Thus, I
regard it as a myth rather than a inethod.28
The point becon1es even clearer if v.re consider the case stu<ly in rela­
tion to a time-series cross-section (1'SCS) research design, as illustrate<l in
Table 2.3. Let us imagine that cases are con1prised of countries and that
te1nporal units are years; hence, the unit of analysis is the country-year. In
Table 2.3, each case has five observations and thus represents a single
country observed over five years Cr 1_5 ). Now, consider the possibility
of constructing a case study fro1n just one of these observations - a single
country at a single point in time. This seems an unlikely prospect, unless of
course there is significant within-case variation during that year. Perhaps
this country, during those tvl'elve 111onths, offers a critical juncture in vvhich
the variables of theoretical interest undergo a significant change. Whether
the temporal era is short or long (and we can i111agine much shorter and
much longer ten1poral periods), the significant feature of most case stud­
ies is that they look at periods of change, and these periods of change
produce (or are regarded as producing) distinct observations - classically
"before" (pre-) and "after" (post-) observations. Alternatively, it may be
possible to exploit spatial (cross-sectional) evidence in that country at
that particular tin1e - for example, with extensive docun1entary records
or a systen1atic survey. In these circumstances, one can easily i1nagine
a case study being constructed from a single observation in a time-series
cross-section research design. But this can be accomplished only by subdi­
viding the original observation into multiple observations. N is no longer
equal to 1 .
The skeptical reader 1nay regard this conclusion a s a sen1antic quibble,
of little import to the real world of research. If so, she might consider
the following quite co1nn1on research scenario. An ethnographic study
provides a thick description, in prose, of a particular setting which is
intended to uncover certain features of other settings (not studied). The
prose stretches for five hundred pages in a draft manuscript and is rather
28 The one possible exception is tbe deviant case that disproves a deterministic proposition.
However, the utility of the deviant case rests upon a broader population 0£ cases that
lies in the background 0£ a case study focused on a single case. Thus, the N o f such a
study, I would argue, is greater than one - even if no within-case evidence is gathered.
The rnore in1portant point is perhaps the follo\ving. No one has ever conducted a case
study analysis that consists of only a single observation. If the point of the case study is
to dcrnonstratc that a single case of such-and-such a type exists (perhaps with the goal
of falsifying a deterministic proposition), then it is likely to take a good dc;1I of \Vork to
csr;1hlish rhl' focrs of that case. This work consi�ts uf 1nulti pk wi1lii11 l".l',1' "l,�crv.11·ions.
1\ 1 ;,1i11, I IH' N ,� l \ l lll'h l1i )'. IHT 1!t;111 lltll'.
What Is a Case Study? 33
repetitive; certain patterns are repeated again and again. In an effort to
reduce the sheer volume of descriptive material, as well as to attain a n1ore
synthetic analysis, the researcher begins to code the results of her labors
into standardized categories: she counts. Has she, by com1nitting the act
of nun1eracy, now converted a case study into so1ne other type of study?
(If so, what shall we call it?) Note that the object of her study does not
vary, even though the prose is now co1nbined with some form of quanti­
tative analysis, which 1nay be sin1ple or sophisticated. The introduction
of statistical analysis does not - should not - disqualify a study as a "case
study.)'
The Style of Analysis
To be sure, non-case study work is by definition quantitative ("statisti­
cal") in nature. Tbis is so because whenever one is atten1pting to incorpo­
rate a large number of cases into a single analysis, it will be necessary to
reduce the evidence to a small nurnber of dimensions. One cannot explore
1 ,000 cases on their ov,rn terms (i.e., in detail). (One might sin1ply accu­
n1ulate case study after case study in a compcn<lious mnltivolun1e vvork.
However, in order to reach any 1neaningful conclusions about this pile of
data it will be necessary to reduce the infor1national overload, which is
why God gave us statistics.)
With case study evidence, the situation is evi<lcntly more con1plicated.
(:ase studies ofay e1nploy a greJt variety of techniques - both quantitative
and qualitative - for the gathering and analysis of evidence. This is one of
the intriguing qualities of case-study research and lends that research its
characteristic flexibility. Thus, it see1ns fair to say that there is an elective
;tfhnity between the case study forn1at and qualitative, small-N \Vork,
,·vcn though the latter is not definitionally entailed. Let us explore why
this might be so.
(:ase study research, by definition, is focused on a single, relatively
hounded unit. That single unit may, or n1ay not, afford opportunities
lor large-N within-case analysis. Within-case evidence is so1netimes quite
t''<l l'nsivc, as \vhen individual-level variation bears upon a group-level
i11fcrl'ncc. But not always.
{ :onsidLr the following classic studies, each of which focuses on the
. 1 1 1 i1udcs and characteristics of An1crican citizens. The American Voter, a
I ollahor;1 t ivc effort hy Angus C:a111phcl!, Philip Converse, Warren Miller,
.111d I >on. l i d Sinkcs, t'\;11ni11cs p11hlic opinion on a 1,.vidc range of topics
1 1 1 . 1 ! .nt · 1 ho1q', lll ! o in!lt1('lll'(' ('l('L'l ( )r.il h('li;1vior tl1ro11gh the i11st-run1cnt
 r·'-"·'-·
34 I. Thinking about Case Studies What Is a Case Study? 35

of a nationwide survey of the general public.2 9 The People's Choice, by TABLE 2.s. Case study and cross-case study research designs co,npared
Paul Lazarsfcld, Bernard Berelson, and Hazel Gaudet, is a longitudinal
Sh1dy
Largest
panel stu<ly focusing on 600 citizens living in Erie County, Ohio, who Subjects Cases San1ple Analysis Population

1,000+
were polled at monthly intervals during the 1940 presidential campaign
' (Campbelt ei:).I., tlw United (ind�viUual:.:} Quant
Citizens of
to determine \Vhat influences the ca111paign may have had on their choice 1,000t Ap1crica:ns
of candidates.30 Middletown, by Robert and Helen Lynd, examines life Cross-: L9GO) SnJ.tcs.

600
case
in a midsized city, including such topics as earning a living, making a
Erie County; Qti.ailt
(Lazarsfelrl 1948) (indivitlu�lsJ
study , Citi?.ens of
Thr: People 's Cho"if:-'e
home, training the young, using leisure, taking part in religious prac­ 2,()00 A111cricans-
tices, an<l taking part in con1n1unity activities (these are the sections into OH

Cilizl!ns of
(Lynd and Lyi1d.
which the book is divided). The Lynds and their accomplices rely on a
(,dties)
{ M�'••••
300+
Quttht A!nJet,tllil
great variety of evidence, including in-depth interviews, surveys, direct i.,rL1:ncie, u,r &Qujt '
1929/1956)
observation, secondary accounts, registers of books checked out of the Case

Working men 15
studv
library, and so forth. 3 1 Folitical Ideology, by Robert Lane, attempts to
([,wuo 1 962) of"Eas:rpo1t" (indivii;tuals)
; Political ideofrJgy
uncover the sources of political values in a subsection of the An1erican IS 0.ill.
public, represented by fifteen subjects who are interviewed intensively by
the author. These subjects are 1nale, white, 1narricd, fathers, between the 1\II categories (subjects, cases, anJ.lysis, population) refer to tbe primary inferences prodnccd by
ages of t\venty-five and fifty-four, \Vorking-class and vvhite-collar, native­ illl' study in question.

born, of varying religions, and living in an (unnamed) city on the eastern

seaboard. 32
A sun1n1ary of so1ne of the n1ethodological features of these four stud­
ies is contained in Table 2.5. Note that the first t\VO studies (The Atnerican
Voter and The People's Choice) are classified as cross-case and the second
pair (Middletown and Political Ideology) as case studies. What is it that
drives this distinction? Clearly, it is not the type of subjects under study
(all focus prin1arily on individuals), the nu111ber of observations (which
range from small-N to large-NJ, or the breadth of the popubtion (all pur­
port to describe features of the same country). The style of analysis differs
in one respect: only in the case stud\es does qualitative ana.lysis co111prise a
significant portion of the research. This, in turn, is a product of the num­
ber of cases under investigation. Where hundreds of individuals are being
studied at once, there is no opportunity to evaluate cases in a qualitative
29
Ca1npbell ct al. (19fi0).
JO Lazarsfdd, Berelson, ;:in<l Gaudet (1 948). A larger roll, ,.,,rith 2,000 respondents, \Vas
taken initially, as a way of est;:iblishing a baseline for the chosen panel of 60 0. Tn addition,
special attention \V3S paid to those whose vote choice ch;:inged during the course of the
panel. These might be looked upon as :1 series of case studies nested \Vithin the larger
panel study. However, because this sort of analysis pbys only a secondary role in the
overall ;1n:1\ysis, it seen1s foir to char::icterize this rese,1rch design ::is " cro�s-c1se."
11 l .y11d :ind Lynd ( 1 929/1956).
I .' I ,1111· ( ] '!(,/).
rnanner. Ry contrast, \vhere a single case (as in Middletoiun) or a s1nall
n umber of cases (as in Political Ideology) is under study, qualitative anal­
ysis is usually de rigueur - though it may be combined with quantitative
analysis (as in Middletown).
The reader will notice that subtle differences in the research objective
of a study �'ln shift it fron1 one category to another. If, for exan1ple,
Robert and Helen Lynd decided to treat their surveys as representative of
illdividua!s in the general public (across the United States), rather than as
representative of cities in the United States, then Middletown would take
on the methodological features of The People's Choice: it would becon1e
:1 cross-case study. Indeed, this is a plausible reading of some portions of
t h a t studrs
linportantly, the technique of analysis e1nployed in a case study is
11cil' sirnply a function of the sheer nun1ber of within-case observations
.1v:1iLd-,lc in that unit. It is, more precisely, a function of the nun1ber
()f ,·0111/hira/Jle observations available within that unit. Consider Robert
I .:nH·\; intensive interviews. Clearly, lots of "data" was recovered from
t hese lL"ngthy discussions. However, the respondents' answers \\'ere not
i I itlt'd su as to confonn to standardized variahles. Hence, they cannot
[11 " li:111dh·d \vi1 hi11 :i d;l LlSL't forn1Jt, usually referred to as a "san1ple"
( . i l 1 h o 1 11•.h \.Vt' li.1\'l' OlT:1sio11:, lly t•tnploycd this tcrn1 in a broader sense).
( l! ( 1 1111·,(·, [ .111c ( ( H i l d h.1\' i' l · l i 1 l'>l'll lo rt'l· odc tlicst' in l"l'rvit·,vs to :11lov·./
36 I. Thinking about Case Studies

for a quantitative analysis, reducing the diversity of the original informa­

tion in order to conforn1 to uniform parameters. It is not clear that much
,vould have been gained by doing so. In the Lvcnt, his study is li1nited to
qualitative forms of analysis.
l'his issue is treated at length in a later chapter. For the 111oment, note
the fact that case study research often provides a piece of evidence pertain­
ing to A, another piece of evidence pertaining to B, and a third pertaining
to C. There may be many observations (in total), and they may all be
relevant to a central causal argu1nent, even though they are not directly
co111parable to one another. These are referred to in Chapter Seven as
nonco1nparable observations.
fn sumn1ary, large-N cross-case research is quantitative, by definition.
This 111uch confonns to usual perceptions. Hov,,.rever, case study research
may be either qualitative or quantitative, or both, depending upon the
sort of within-case evidence that is available and relevant to the question
at hand. Consequently, the traditional association of case study work with
qualitative n1ethods is correctly regarded as a n1ethodological affinity, not
a definitional cntail111ent. It is true so1nctin1es, but not all the time.