Upload
qut
View
1
Download
0
Embed Size (px)
Citation preview
http://asj.sagepub.com
Acta Sociologica
DOI: 10.1177/000169939103400403 1991; 34; 279 Acta Sociologica
Francis G. Castles and Michael Flood the Legal Dissolution of Marriage
Divorce, the Law and Social Context: Families of Nations and
http://asj.sagepub.com/cgi/content/abstract/34/4/279 The online version of this article can be found at:
Published by:
http://www.sagepublications.com
On behalf of: Nordic Sociological Association
can be found at:Acta Sociologica Additional services and information for
http://asj.sagepub.com/cgi/alerts Email Alerts:
http://asj.sagepub.com/subscriptions Subscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.co.uk/journalsPermissions.navPermissions:
http://asj.sagepub.com/cgi/content/refs/34/4/279 Citations
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
279
Divorce, the Law and Social Context
Families of Nations and the Legal Dissolution ofMarriage
Francis G. Castles and Michael Flood
Department of Political Science, Australian National University
This paper seeks to establish the extent to which the incidence of divorce and thecharacter of the law regulating the legal dissolution of marriage can be accounted for byhistorical continuities and cultural traditions in distinctive ’families of nations’ Theresearch brings together the diverse traditions of comparative law and comparativesociology, and uses both to come to grips with the questions of why divorce rates varyfrom country to country and why there has been such a massive increase in the divorcerate in the postwar era. A multivariate model of cross-national divorce outcomes suggeststhe strong influence of historical continuities within distinctive ’families of nations’ ondivorce outcomes in the 1960s and a much enhanced influence of social context variableson the character of the law in the following two decades
Francis G. Castles and Michael Flood, Public Policy Program, Australian NationalUniversity, G.P.O. Box 4, Canberra, ACT 2601, Australia.
1. IntroductionThe cross-national research on the deter-minants of divorce which is reported herehas a very particular focus. It is a con-
tribution to a much wider research projecton whether so-called ’families of nations’are a major force shaping patterns of publicpolicy in advanced capitalist societies.’ Thefamily of nations concept is a new - or abetter word might be, rediscovered - con-cept in sociological and public policy analy-sis. It harks back to an academic tradition,
popular in political science, sociology andlaw in the early decades of this century,which insisted that, in order to comprehendthe variety of policy outcomes to be foundin modern states, it was necessary to traceback their roots to commonalities shared
by groups of nations in virtue of a sharedhistory, culture, legal tradition and
language. In this view, it mattered thatnations were English-speaking, Scandi-navian and Germanic and it matteredbecause that attribution told us somethingabout contemporary policy outcomes.
@Scandinavian Sociological Association, I’)’)/
In the postwar era, comparative researchm all the disciplines concerned with publicpolicy has largely abandoned this approachand adopted a paradigm, the origms ofwhich are more exclusively sociological.From this more recent perspective, the
proper task of comparative analysis is toreduce ’proper names to explanatory vari-ables’ (Przeworski 1987:38-39). The propernames m question are those of nations andthe task, as construed in the literature, hasinvolved demonstrating that cross-nationalvanation is substantially attributable to theimpact of structural variables, whether of asocial, economic, demographic or politicalnature. The guiding hypothesis of the fam-ilies of nations project is that, despite its
many valuable contributions to our under-
standing, this ’sociologizing’of comparativepublic policy analysis has possibly led to anunwarranted neglect of the importance ofhistorical continuities and their attachedcultural and legal dimensions. Proceedingfrom identified policy commonalities withingroups of nations with pn/~a facie sharednational attributes, it seeks to establish the
degree to which these policy outcomes must
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
280
be seen as a consequence of historical con-tinuities. As we shall see, the incidenceof divorce and the character of the laws
governing the legal dissolution of marriageare features of advanced capitalist societiesin which such commonalities are extremelyapparent.
2. Families of nations and thelaw of divorce
In order to comprehend the nature of therelationship between families of nations
and divorce rates, it is necessary first to
define the boundaries of groupings ofnations in terms of the historical de-
velopment of the law relating to the dis-solution of marriage. The notion of cat-egorizmg groups of nations in terms offamilial resemblance is a familiar one in
comparative legal theory and a distinctionbetween Anglo-American, Nordic andRomano-Germanic legal systems is com-
monplace (Glendon 1987).Here we offer a somewhat more dif-
ferentiated categorization of legal systemsbased on a reading of the commonalitiesbetween groups of nations as their provisionfor the legal dissolution of marriage hasevolved over the four centuries or so sincethe Reformation break with the canon lawdoctrine of the indissolubility of marriage.Clearly, this categonzation demands his-torical validation for which there is no spacehere.&dquo; All that is possible here is to note
these legal families’ broad defining charac-teristics and their strong linkage with diver-sity in religious faith and historicalwatersheds in the process of secularization,such as the French Revolution. These legalfamilies and the countries included in themare the following: (1) Common Law (Aus-tralia, Canada, New Zealand, England andWales, the United States). Latest to departfrom canon law, due to the influence ofthe Church of England, allowing divorcegenerally only on grounds of adultery. (2)Scandinavian (Denmark. Fmland, Norway,Sweden). Lutheran influence allows law tobecome secularized and leads to separationas the main ground for divorce. (3) Ger-manic (Austria, Germany, Switzerland).Emerging out of a conflict between Prot-estant common law and canon law, with
separation subject to consent provisions.(4) Code Napoleon (Belgium, France, theNetherlands). The secularization of theFrench Revolution leads to mutual consent
provisions conflicting (and sometimes tem-porarily submerged) by canon law doctrinesin these largely Catholic nations. (5) CanonLaw (Ireland, Italy). Marriage remains
indissoluble.An initial test of the family of nations
perspective is to inquire whether these legalfamilies of nations with a discernible his-torical continuity over several centunes many way correspond with legal provision mthe postwar era. In Table 1, we present atabulation of divorce law provisions in 1960and ensuing reform in the period 1960-76.In the table, we code the liberality of thelaw on the following basis:; 3
3 = No-fault grounds permitting uncon-
tested proceedmgs after three years orless with contestation delaying the pro-cess for no more than a further two
years.2 = Mutual consent with no substantial
restrictions or three years’ separationas grounds for uncontested divorce.
1 = Other more restrictive legislation.0 = Most restrictwe/i.e. no national
divorce legislation.The rationale for this coding is that liberal
access requires a relatively short period toestablish incompatibility or marital break-down, that proceedings should not requirethe necessity of demonstrating the maritalfailings of either party and that divorceshould be available irrespective of the con-sent of both parties. Reasonably unre-
stricted mutual consent and separation onthe basis of agreement constitute halfwayhouses, allowmg a guilt-free dissolution ofmarriage for those who could accommodatetheir differences, although no remedy forothers. Other more restrictive arrange-ments and those resting exclusively on faultpermitted divorce under some circum-stances, but invariably at the expenseof great anguish and often at consider-able financial cost.
In 196~, the correspondence between theliberality of the law and its historical evol-ution in distinct families of nations is extra-ordinarily clear. The Scandinavian group of
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
281
nations is wholly consistent in its liberalismand the Germanic group in its halfwaystatus. Only New Zealand, in the English-speaking group, has sufficiently departedfrom fault principles to allow a designationof partial liberalization. Both the Code
Napoleon and canon law families are whollyconsistent in the degree of liberal accesspermitted in their divorce statutes.By 1976, however, there had been a very
substantial degree of divorce law lib-eralization throughout much of the westernworld, and the distinctiveness of these legalfamilies of nations had to some degree beeneroded. The English-speaking nations
experienced the greatest shifting from
essentially fault-based systems to systemsin which the no-fault element was para-mount or the only ground available. Thisprocess is again, at least in some part,attributable to a diffusion in legal practice,the stimulus to which were the recom-
mendations of a Church of England reportunder the title Pwwzg Asunder: A DivorceLaw for Contemporary Society, which,
departing from the long history of theChurch in England, based its views noton how the ’doctrine of Christ should be
interpreted and applied within the ChristianChurch ... but on what the Church oughtto say and do about secular laws of marriageand divorce’ (Mortimer Commission 1966).The report’s main recommendation wasthat ’the doctrine of breakdown of marriageshould be comprehensively substituted forthe doctrine of matrimonial offence as thebasis for all divorce’. This dramatic changein religious doctrine, issued with the impri-matur of the Archbishop of Canterbury,was a spur to and a platform for reformthroughout the English-speaking world
(except Ireland) and beyond. In Englandand Wales, Australia, Canada and NewZealand, and a number of the Americanstates, starting with New York, relativelyliberal grounds for dissolution of marriageby separation were instituted by the late
1960s. In 1970, the world’s first divorce lawbased solely on irreconcilable breakdownof marriage was introduced in Californiaand by the 1980s exclusively no-fault pro-visions had been adopted in Australia,Canada, the Netherlands, New Zealand,
Sweden, and some 40 per cent of the Ameri-can states. 4
The dramatic reform process in this
largely English-speaking group of nationsmeant that the distinctive liberal characterof the Scandinavian family of nations haddisappeared by 1976, with both groupingsnow being essentially similar in basing mari-tal dissolution substantiallv on no-fault
grounds. Outside the English-speakingcountries, the only liberalization of com-parable magnitude occurred in the Nether-lands. This change made it an atypicalmember of the Code Napoleon legal family,since change in Belgium and France wasmore muted, combmng fault provisionswith a more unrestricted critenon ofdivorce by mutual consent. The German-speaking family of nations also ceased tobe characterized by common provisions,smce Germany adopted liberal separationlaws, whilst Austrian and Swiss lawremained essentially unchanged. Finally,change also occurred in one of the two
remaming canon law nations, when Italyadopted somewhat restricted legislationpermitting divorce in 1970 which was sub-sequently reaffirmed by a hotly contestedpopular referendum in 1974. Of the coun-tnes under survey here, only Ireland hadno law of divorce in 1976 (see Shatter 1981).Indeed, the 1937 constitution forbade the
making of laws for the dissolution of mar-nage and a 1986 referendum to reverse that
position was defeatedThis highly summary account cannot be
interpreted in any other way than as con-firming the existence of quite distinct fam-ilies of nations m respect of the historicaland cultural continuity and development ofthe law of divorce m European nations andnations of European settlement at leastuntil the 1960s. But the historically con-ditioned similanty of the law as an outputof government is no guarantee of a com-
parable similarity of outcomes in terms ofaggregate divorce rates and divorce rate
change. A further important test of the
impact of legal families of nations is
whether there is some further correspon-dence between outputs and outcomes.
Table 2 presents data on average divorcerates for 17 nations for the periods 1961-68and 1976-83 and the change in the divorce
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
282 at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
283
eo.J
’š .Ejo ~omE 06...00..2~
’1J-
[S ~z#« -£m-ciEij 0~.t:ztN Ev 3 0 vz wr n&dquo; c: ~ ...c&dquo;&dquo;!:!sw~ 0E w <#5’1J &dquo;&dquo;0&dquo;c:~ 0
&dquo; &dquo;&dquo; C:.- 0 c:&dquo;&dquo;-
U&dquo;&dquo; Es’ 5S E% 1now<K
r
c<u
r-oa~nLwcn’1J-U c:I«. ENc:~ E ’1J’1J .
2<-n
~ ―’’1J >(..c: u - ’1JJ<
o <o&dquo;&dquo;If)Cc 2«z1w
2 g,.@J &dquo;&dquo;n«3 CUJ c9W cU ea00G~
.aa
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
284
::i.2<t::~)’3#5m~-5c
~...........
~0-
18i...........
*0-RJ%t::Q
-c....
~...of2gn
-~-~
t::Q
r’i$B0B
&dquo;:ca
00~0-.....
£x~
&dquo;’<:3t::Q3bIm
2...0a
QN
vn¡....
<J
-5c =
<IJ
ar~V»
<IJ ~
U ’
c:4J
b~55u
fr©#ftOxU~00
w .
$i %0’> :>7J
-~
e :~ C!
02 0©
....
--- ~~ll ..
o~ e0’> ~vi ~C vO ’E(U ’Ja !!<u
-5 u * *
5 M~ (U:; ~v -oCU
V~ O!J XCJ O
> _
.- ~ ... 0c ~ >oCO .
?3 ~C 0
~r-- 0a_ B§-~ Cd~ë o <IJ >- ~©j I_’’5 QJ U_’0 ~ -c
°o a0:> O ~ ...8+ ODO&dquo;tJ o
- &dquo;’..c: <IJ <IJ E0.-v Ca
N c .©, ji~ <... 00 0.~~ .2g5> %z~2 -o,.U.n ~ a~_r ....
S~ c
t: ...... <IJ :;:) ..o :>:> ... 0 - ~ tJ C~ ~ > 0 EQ -6 c
~ ~3c c
E vQF I4 ~., O
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
285
rate occurring between these periods.Ideally, the measure of the divorce rate
would be the rate per 1000 marriages, butthis is unavailable across such a wide sampleof nations, and the crude rate per 1000 ofthe population is used as the only availableproxy. The periods selected are deliberatelychosen with a view to providing a test ofthe impact of the provisions of the law in1960 and 1976 as set out in Table 1.
Rheinstein argues that, in countries thathave proceeded far along the path of econ-omic modernity, but where the contem-
porary intellectual climate is eitherconservative or pluralistic, the divorce lawof the statute books will be strict, but will
simultaneously tend ’to become a deadletter’ (Rheinstein 1972: 128). This helps toexplain, for the 1960s at least, the mostobvious anomaly we encounter in con-
trasting the provisions of divorce laws withdivorce rates in the countries under surveyhere: the fact that the United States, withlegislation based very largely on variousdefinitions of matrimonial offence, had adivorce rate almost twice as high as anyother western nation throughout the periodunder review. The United States, for muchof this century, has constituted the mostdramatic instance of legal interpretationbeing at variance with statute law.-‘ 5
Collusion, the withholding of informationfrom the courts or the presentation of falseinformation by both parties, became a
standard practice, and although in itself abar to divorce, the evidence presented bythe parties was scarcely ever contested orinvestigated by the courts. This amountedto a practice of divorce by mutual consentin many states and was compounded bythe practice of ’migratory divorce’, allowingdivorces conducted under the liberal inter-
pretations of some states and nations to berecognized under most circumstances evenin states where the grounds for divorce weremuch stricter. In effect, then, by 1960, andindeed for much of this century, the practiceof the law, in contradistinction to its letter,was as liberal as that of any other countryin the western world, although the law gen-erally forced those seeking legal dissolutionto dissemble or travel to obtain the relief it
formally prohibited.
The effect of the American discrepancycan be readily ascertained from the cor-relations between divorce law liberalismand divorce rates in 1961-68 which can beseen at the bottom of Table 2. Includingthe United States, the relationship is onlymarginally significant ; excluding that
country, it is extraordinarily strong,accounting for some 70 per cent of divorcerate variation. Excluding the USA, thecoherence of our five families of nations is
very strong mdeed ; the highest and thelowest in the remaining English-speakinggroup differing only by 0.2, in the German-speaking group by 0.3, and in the Code
Napoleon group by 0.2. Only m the Scandi-navian group is there any significant vari-ation, with Norway registering a divorcerate half that of Denmark.Almost exactly the same story can be
told of the period 1976-83, although thediscrepancy between the correlations
including and excluding the USA is less,at least partly because of that country’sintervening process of legal reform. Theabsolute gaps withm the families of nationshave increased somewhat m line with themore than twofold increase in the averagedivorce rate as between 1961-68 and 1976-83, but only two countnes are out of syn-chronization with the others in their group-mg: once again Norway, now joined byGermany, the only member of the German-speaking family to have substantiallychanged its statutes by 1976. Change overtime, again, follows the same pattern. TheUSA is. on this criterion of evaluation,a typical member of the English-speakingfamily, with New Zealand being furthestfrom the group norm, as one might expectof the nation that had effected the least
legal change in the period. The charge of’English-speaking awfulness’ noted in otherresearches emanating from the family ofnations project with regard to low rates ofeconomic growth and low social security(see Castles 1990) can also be levelled here,with five English-speaking nations with acommon law conception of divorce fea-
turing amongst the six nations to experiencethe greatest mcrease m divorce rate in thisperiod. In contrast, change m the Scandi-navian group was much more moderate,although the pattern of change was equally
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
286
coherent. In the other families, only Ger-many within the German-speaking groupand the Netherlands within the Code Napo-leon family diverged substantially from therank-ordering of other members. Bothwere nations in which legal change hadbeen more rapid than in the rest of therelevant grouping.With the exception of the American case,
the evidence presented here supports thenotion that statute law is an importantdeterminant of the frequency of the legaldissolution of marriage, although it wouldbe mistaken to deduce from that fact thatliberal laws destroy marriage, since there isa well-grounded empirical literature show-ing the inverse relation between divorceand judicial and de facto separation (seeRheinstein 1972: 277-316). What can bededuced from the evidence is a strong primafacie case for the influence of long-termcultural factors transmitted through the his-torical continuities of distinct legal familiesof nation in this arena of domestic policy.The case is only prima facie for at least
two major reasons. First, the lesson ofearlier investigations of apparent familyresemblances amongst nations (Castles1990) tells us that similarity is frequentlydissolved when we come to examine the
impact of social and economic structureson policy outputs and outcomes. Such apossibility is highly consonant with the viewof many of those who study comparativelaw, like Rheinstein, who suggest that thelaw is ultimately a reflection of its socialcontext and is either reinterpreted or sweptaway where it remains too long incongruentwith that context. Second, our account sofar has concentrated on legal outputs andhas not considered the social and economicvariables that may be associated with theindividual decision to seek legal remedies.Even the relatively high correlations re-
corded at the bottom of Table 2 leave suf-ficient scope for explanations of divorcerates and especially divorce rate change interms of the impact of structural variableson individual behaviour. Controlling forsuch contextual variables, and, in particu-lar, for cross-national differences in
religious belief, which given our prioranalysis must be seen as a prime candidatefor a cultural variable simultaneously deter-
mining legal outputs and behavioural out-comes, might well seriously modify ourinitial conclusions as to the causal impactof legal families of nations.
3. On the correlates of divorceThere is a substantial empirically basedsociological literature on the contextual fac-tors associated with divorce in particularnations, a literature which, in recent
decades, has gained much of its impetusfrom a growing awareness of the need todistinguish between the macrostructuraland the microsociological dimensions of theprocess of divorce. It is generally acknowl-edged that monocausal explanations ofrecent divorce trends are insufficient andinaccurate, and that we must seek multi-causal explanations, both because the fac-tors influencing marital stability are manyand because macrostructural and micro-
sociological factors simultaneously impingeon individual decisions to seek a legal dis-solution of marriage (see Hart 1976).However, the increasing sophistication
of the sociological literature has not carriedthrough to the generation of a comparablysophisticated empirically based corpus ofcross-national research. The reasons arenot difficult to discern. Differences and
changes in legal systems have been seenas central variables in explaining nationaldifferences in divorce rates and divorce rate
change, but the very fact that differentnations have different laws has equally beenseen as a major barrier to systematic com-parison using the methods of applied socialresearch. Moreover, the predominant focusof most research in the field has been onthe postwar growth of the divorce rate thathas been a phenomenon of virtually allwestern societies, and this has encouragedthose with an interest in international com-
parisons to stress factors common to manynations - for instance, general attitudinaland ideological shifts associated with secu-larization and modernization (see Goode1963: 81; Ambert 1980: 54-57; Price &
McKenry 1988:7) - rather than factors per-tinent to the substantial differencesbetween nations as revealed in Table 2.
Finally, the awareness of the desirabilityof an approach combining macrostructural
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
287
and microsociological factors has in itselfbeen a discouragement to cross-nationalresearch, which necessarily relies veryheavily on routinely collected aggregatedata that by their nature obscure the subtleinteraction between these classes of caus-ation whilst being inherently biased towardmacrostructural explanation.We acknowledge the very real limitations
imposed by this latter difficulty, but never-theless seek to devise a model for the cross-national analysis of divorce rates anddivorce rate change by combining key socialcontext variables with our earlier cat-
egorization of the legel impediments to
divorce in different nations as presented inTable 1. It is our view that this categoriz-ation, by providing a quantitative mdex ofthe liberality of national divorce laws andtheir liberalization over time, permits us toovercome the problem of the barrier to
systematic cross-national research consti-tuted by the existence of diverse legalhurdles to the dissolution of marriage indifferent nations. The resulting modelshould make it possible to establish, at leastwithin the broad-gauge terms dictated bythe available data, whether a legally definedfamilies of nations approach retains its heu-ristic value when the law is contextualized
by its social setting. The first stage in theprocess of model-building is to test againstcross-national data some of the hypotheseswhich feature most conspicuously innational studies. These hypotheses may bebroadly grouped under the headings of
modernization, secularization, demo-
graphic factors and policy constraints.In accounting for divorce trends in the
last two centuries, long-term economic ormaterialist factors have been widely ident-ified as fundamental in creating the con-ditions for an increasing tendency to
divorce. The three most crucial trends arethe shift in the economic base of house-holds, a growth in married women’s formaland practical economic independence, anda growth in women’s employment oppor-tunities and labour-force participation(Halem 1980; Phillips 1988; Price &
McKenry 1988). This last factor has beenparticularly influential in some interpret-ations of postwar divorce rate change andis an instance where the same variable has
been used to generate both macrostructuraland microsociological interpretations.Thus, quite apart from labour-force par-ticipation’s effect in facilitating femaleeconomic independence, North Americanstudies have established that a husband’s
sporadic employment and low wages, rela-tive to his wife’s employment and wages,are key determinants of marital mstability(Cherlin 1979; Ambert 1980).
’
The three major economic trends ident-ified in the literature are all part of anoverall process of modernization, whichGoode ( 1963), focusing on the joint impactof the processes of industrialization andurbanization, has shown to be highly influ-ential as a force transforming the structureand stability of the family. Socio-economicmodernization variables are the standardfare of sociological and comparative publicpolicy analysis and in what follows weexamine the degree of association betweendivorce rates and GDP per capita (an indi-cator of the shift in the economic base ofhouseholds towards greater affluence), thesize of the service sector (an indicator ofthe expansion of an economic sector par-ticularly associated with female employ-ment), the size of the non-agncuttura)lahour-force ~a broad mdicator of the mod-ernization of the social structure and dis-
ruption of traditional patterns) andurbanization (indicating the shift away fromtraditional ways of living). It should benoted that the model we develop here,which rests on the analysis of successivenational cross-sections and the change tak-ing place between them, does not allow aninvestigation of the impact of short-termeconomic fluctuations on the propensity todivorce (see Cohen 196t~; Rowe & Krishnan
1980). Because of the hypothesized specialimportance of female opportunities for
independence, we utilize female labour-force participation (the most direct indi-cator of women’s potential to maintain theireconomic independence) as a vanable
potentially captunng a separate dimensionof the modernization process.
Religious belief or rather its decline
through the process of secularization is, ofcourse, another factor strongly associatedwith the process of modernity and may beseen as the attitudinal component of that
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
288
process. The most common shift discussedis the erosion of religious sanctions
upholding marriage or negatively sanc-
tioning divorce. We have already noted theformative influence of religious belief in
shaping the law of divorce, and it is no
surprise to find that the major variable
singled out as expressmg this trend is therelative strength of different Christian de-nominations, with the strongest emphasison the basic divide between Protestantismand Catholicism (Chester 1977; Halem
1980). Unfortunately, we do not dispose ofa religiosity variable capable of measuringthe degree of secularization on a cross-
national basis. As a substitute, we use Cath-olic adherence. hypothesized to be a nega-tive predictor of divorce rates and divorcerate change, as the key test variable forthe influence of religious belief on divorcerates. Whilst this choice may be criticizedon the ground that it measures only theimpact of a particular brand of the Chnstianfaith, we plead in some extenuation thatthe long post-Reformation shift away fromthe tenets of the Catholic faith representsa progressive weakening of ecclesiasticalpower m human affairs, i.e. part of thebroad shift towards secularization. Unfor-
tunately, speculation as to the positiveimpact on divorce rates of further nor-
mative shifts of a seculanzing kmd, includ-ing individualism, liberalism and hedonism(Ambert 1980, Price & McKenry 1988),cannot be tested here for lack of suitablecross-national data.
Demographic factors may be related todivorce either quasi-automatically as fac-tors influencing the proportion of the popu-lation eligible to divorce or as factors witha more substantive bearing on marital insta-bility. In the first category, we include thecrude marriage rate (i.e. per 1000 of thepopulation) as a means of controlling forthe fact that the crude divorce rate may wellbe strongly influenced by the proportionof the population that is married. More
substantively, fertility has been hypothe-sized to be linked to divorce in virtue of a
propensity for those with few or no childrento find it easier to escape from the bondsof matrimony. It has also been argued (Hart1976: 77; Norton & Glick 1979; Fergusssonet al. 1984: 539, 542) that this propensity is
likely to be increased if the age at marriageis relatively young. Below we examine thedegrees of association between both fer-
tility rates and early marriage ( percentageof bndes below the age of 20) and divorcerates.
A final factor much discussed has beenthe impact of policy constraints, m par-ticular the availability of welfare paymentsto single mothers and the welfare state pay-ments available to mothers and childrenmore generally. Hart (1976: 71) hypo-thesizes ’that access to this meagre incomeis an important element m the increasingavailability of divorce, particularly for
couples at low income levels’. However,Moles, in a later assessment, reviews bothcensus data studies and longitudinal analy-ses of this possible welfare-dissolution link,and finds either inconsistent or inadequateevidence (Moles 1979:172-78; Albrecht etal 1983:54-55). We would like to test thishypothesis by cross-national comparisonbecause wide differences m national wel-fare systems suggest that any effects are
likely to be more pronounced in such a
context. There are, however, insuperabledifficulties m obtaimng data on the gen-erosaty of expenditure to single parents fora sample of nations anything like as exten-sive as the 17 under examination here.~Thebest proxy we can use is family transfers asa percentage of GDP (available for all ournations except Belgium), but we note that,since some part of such transfer expenditureis, m varymg degrees, intended to preservethe integnty of complete families, any inter-pretation of the resulting correlates wouldhave to be speculative m the extreme.Table 3 presents the bivariate cor-
relations between these 10 contextual vari-ables and divorce rates and between thesevariables as measured in 1960 and sub-
sequent divorce rate change. With the
exception of Catholic religious afhliation,where data is only available for 1970 and1980. the independent variables are lagged,m that 1960 and 1976 data are correlatedwith average divorce rates for the periods1961-68 and 1976-83 respectively. This is
necessary, since some of these variables atleast - most conspicuously, female labour-force participation and family transfers -might, in part, be inferred to be as much
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
289
aOt:~’5
E~l~zo&dquo;
b
xa
rQ&dquo;50i
a<:u
2...O’50.;s2
~Q
aVC~~4z,
i2.gi~L8~
.~zaQ
Q5m.2
sco
v vE ... a.~ .c ~ :a~ E S ~ ~~Ee=ëC’:I2¿j~’ð
0’&dquo;;; 0 E’f:ü2~¡f~’&dquo; (1) - C’:I -a 3g~ ~ ―) -<-
’~
-- < o 11~ ~ C’:I.8&dquo; *
-3~~~:’:.~ <- .° M ’!= ’- >. u >J‘O CU &dquo;3 U ~J~ ~~~V&dquo;jCc:-~...--oU cC(1) ..c: ’&dquo; 0 C’:I_ N- <i _~°<< ~ <2 °- ~ < §£j 32£ e12Z ~~U , _:’J c -p Z E II IIJ2-~J.&dquo;C. y v :: c w ,^ *j’li§cx E o N;2:Goo~-&dquo;,...(1)-J ! ~ ~ E_:J _~
V ~n C rJ ~ N 3 _GO vE-g ~2~~1J : J ‘ .G U E~£’ fl 2=o=’~2lO~&dquo; KëL.ù;-=~
ËO ~~{5:>,E-o-22E~~EO- - >..’7. otE ~~ c~©#i#)< E j 1 ©c0 6’ ~ ~ ~_ :r <
’Ec-~~>Ü E -= ~>L’ . ’L. E ~ ~ 2 oo o g(X;<1)-::-=20~ ~ ~ ? C-E~E5‘ :aZ~~~8@ j- °&r %g* <J: 5’~ -g ðgz- M,~~~3§Ë2L~~~3 ~ 2 C S’’ &dquo;O >&dquo;IE 17L :J C-Ê~2:: ~, #~O #22#8 >
0~~~~~
mth»I
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
290
caused by as causes of divorce. Given thatthe USA’s divorce rate is of a distinctlyhigher order of magnitude than that of theother nations included in the comparison,and that its status as a statistical ’outlier’
might be expected to bias some of the
relationships to a marked degree, cor-
relations are reported both including andexcluding that country. This expectation isstrongly confirmed in respect of the size ofthe service sector and fertility in the earlierperiod and the marriage rate in the later
period.The data in Table 3 strongly affirm a link
between a range of contextual factors andboth the level of and the change in thedivorce rate. With the single exception offamily transfers, every variable is in some
respect statistically significantly related tofeatures of the postwar divorce phenom-enon in this range of advanced capitalistsocieties. Moreover, the direction of the
reported associations with the average levelof divorce in both 1961-68 and 1976-83 is
generally as might be expected from thehypotheses derived from the literature. Themodernization vanables are, with only onevery minor exception (service sector size m1960 excluding the USA), positively associ-ated with the divorce rate. Catholic affili-ation is uniformly a significant negativepredictor of divorce rates and femalelabour-force participation is only slightlyless uniformly a positive predictor. Withthe exception of the sample including theUSA for the earlier period, fertility is
always negatively associated with the levelof divorce. Both marnage rate and earlymarriage correlate positively with divorcerates, although the former relationship is
more consistently statistically significant.Although not quite significant, we note thatthe association for family transfers in 1960is negative, possibly to be counted as evi-dence against the welfare availabilityhypothesis or, equally probable, a reflec-tion of the problematical way in which thathypothesis is operationalized here. Finally,it is extremelv noticeable that the overallimpact of social context variables m far
greater m the later than the former period.For the 1960 variables, only one relation-ship is significant at the 0.01 level m the
sample including the USA and none
achieves that level in the sample excludingthat country. For the 1977 variables, how-ever, four achieve that level in both samplesand two more are significant at the 0.05level. This increasing influence of socialcontext variables is a point to which weshall return later in our analysis.We examine the relationship between
our contextual variables ca. 1960 and
change in the divorce rate in order to estab-lish in what sorts of countnes the divorcerate grew most rapidly. The results are
wholly compatible with the hypothesesoffered above. Most conspicuously, divorcetended to increase in countnes which were
economically modernized (by whichevercriteria), where Catholics constituted a
smaller proportion of the population andwhere there was a high degree of earlymarnage.
4. On paths to divorceGiven the relatively small number of caseson which our comparative analysis is based, .it is not possible to include all 10 contextualvariables in the final multivanate elab-orations of our model, which follow theadvice of Kitson & Raschke (1981:30) thatsuch multivanate designs are the best wayforward for understandmg ’the simulta-neous and relative impact of a number ofvariables’. Our criteria for inclusion are
based on theory and the character of thefamily of nations concept we are seeking toexplore. We wish to include at least oneeconomic modernization vanable, smce
hypotheses linking modernization to chang-ing social behavior are at the very core ofthe sociological enterprise. We choose thesize of the non-agncultural labour-force asour key vanable m this respect becausean examination of correlation matrices forboth penods shows that this vanable aloneis consistently strongly associated ( m excessof 0.70) with all the other economic mod-ernization variables m both periods. In thatsense, it seems appropnate to regard thesize of the non-agricultural labour-force asthe pivotal vanable best expressing the mul-tifaceted aspects of the modernization pro-cess. It is also important to include femalelabour-force participation as an mdicator ofwhat is also, almost certainly, a separate
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
291
dimension of the modernization process. In
support of the view that female labour-force participation is unlike other com-
ponents of modernization, we note a nega-tive relationship (-0.40) between thatvariable and the size of the non-agriculturallabour-force in the earlier period super-seded by a small positive one (0.27) in thelater period. In fact, the relationshipbetween economic modernization andfemale independence is almost certainly nota linear one. Phillips, convincingly, suggestsa three-stage process, whereby the tra-
ditional family economy, in which women’swork on the land created a complementar-ity of tasks and mutual dependence, wasreplaced in the early modernization phaseby women’s dependence based on the per-formance of home tasks, with greaterfemale independence only resulting fromthe later shift of married women into manu-
facturing and service employment (Phillips1988:590-92). Hence, m order to capturethe impact of the major trends presumedby the literature to have shaped the tra-jectory of the divorce rate, it is necessaryfor our model to focus on the special factorsinfluencing women’s economic indepen-dence as well as on broader correlates ofeconomic modernization.At least two additional variables have to
be included in our model m order to makeit possible to confront a legal family ofnations explanation with one based on
social context. On the one hand, we haveto include our operational ization of theextent of the liberality of the law in orderto assess the separate impact of legal pro-visions and, on the other, we need to
include a test for the impact of religiousbelief, the variable that we might mostreadily assume could explain the characterof the law without some reference to a
continuity of legal procedures inheritedfrom the past. As pomted out previously,if our negative proxy indicator of secu-lanzation, the strength of Catholic affili-ation, were capable of accounting for thevast proportion of the cross-national vari-ance in divorce laws, it would at best arguefor a weak variant of the family of nationsconcept, insofar as it would demonstratethat legal policy outputs were a clear reflec-tion of cultural factors. However, it would
simultaneously argue against the strongervariant of the families of nations concept:that these outputs, and the divorce out-comes to which they contnbute, can onlybe understood in terms of the historical
continuity and distinctiveness of legal formsin different groupings of nations.These four factors - economic mod-
ernization, female independence, religiousbelief and the liberality of the law - exhaustthe number of variables it is possible toinclude in a reasonably coherent stat-
istically based model of divorce rates anddivorce rate change. Fortunately, they aresimultaneously the vanables which our cal-culations in Tables 2 and 3 show to be most
strongly and consistently associated withdivorce outcomes. Welfare availability,whether because of the way we havemeasured it or because of the weakness ofthe basic hypothesis, did not manifest anydegree of significant association with thedivorce phenomenon. The demographicvariables are, at best, inconsistent in theirrelationships with the level of and changein divorce rates, being only rarely at all
strongly associated with both level and
change and frequently manifesting quitedivergent associations depending on theinclusion or exclusion of the USA. A fewof the associations between demographicvariables and change are, however, quitestrong. In particular, high levels of earlymarriage in 1960 are highly associated withan increase in the divorce rate and it is
possible that, in consequence, there will besome misspecification of our model insofaras it pertains to change.However, this is rather less probable in
respect of divorce levels, with early mar-nage and the marnage rate only being sig-nificantly positive predictors for the
samples including the USA. We surmisethat the USA’s exceptionally high marriageand early marriage rates throughout theperiod may well have contnbuted to thatcountry’s exceptional divorce rates to somedegree, but note that these factors con-tnbute little to understanding variance inthe remainder of the sample. Fertility inthe later period in the sample excludingthe USA is significantly associated with thedivorce rate, but over time this variable hasbecome markedly more strongly associated
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
292
Figure I Paths to diporce, 1961-68.
with our measure of economic modern-ization, the size of the non-agriculturallabour-force (in 1960 only 0.05, by 19760.66). In other words, fertility only becomesa predictor of the divorce rate once it alsobecomes a part of the syndrome of socio-economic modernization. In this necess-
arily broad-gauge model of the impact ofsocial context, it may be that any inde-
pendent effects of fertility are to some
extent picked up by other modernizationvariables included in the model.
Figures 1-3 present versions of our pre-ferred model of divorce outcomes for the
periods 1961-68 and 1976-83 and for
change over the period as a whole. Thefigures consist of path diagrams showing,by means of standardized regression coef-ficients, the strength of the associationsbetween variables in a theoretically derivedordering of probable causal influences.Given the USA’s clearly established statusas an outlier in both 1960 and 1983, themodels for levels of divorce are specifiedexcluding that country. Since it is not an
outlier in respect of change, all 17 nationsare included in the specification for changein divorce rates.The theoretical argument implicit in
these path diagrams is that economic mod-ernization and religious belief are priorinfluences impacting on the size of thefemale labour-force and that all three
impact on the law of divorce which, in turn,regulates possibility of the legal dissolutionof marriage. Such a causal ordering hasvery important implications for the familiesof nation concept. To the extent that weare able to account for variance in thedivorce rate by the direct influence of thethree social context variables alone, it is
possible to discount explanations resting onthe impact of distinctive legal families ofnations. A similar conclusion would also be
justified if we could offer a full account forthe distinctiveness of variance in the law,even if the law were shown to be closelyassociated with variance in divorce rates.In the first instance, we could argue thatthe law was irrelevant and, in the second,
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
293 at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
294
that it was merely a reflection of socialforces.
In order to establish a prima facie casefor the importance of families of nations asa determinant of policy outcomes, we
would need to demonstrate first that laws,the variance of which we have alreadyshown to be closely associated with a long-term historical development in distinct fam-ilies of nations, are at once a crucial factorin accounting for observed variation indivorce rates and, at the same time, onlyin part themselves accounted for by socialcontext. In the 1961-68 path diagram pre-sented in Figure 1, this does appear to bethe situation we confront. There are three
significant predictors of divorce rates in thegenerally successful model for the smallersample, which accounts for some 84 percent of the variance in divorce rates. Twoare contextual factors, the size of the non-agricultural labour-force and of femalelabour-force participation, with the latterrather the more important. However, evenwhen we control for both these contextualvariables, much the strongest predictor isthe degree of liberality of the law, the lawitself being less than adequately accountedfor by the model variables, with onlyaround 60 per cent of the variance
explained. The only statistically significantpredictor of liberality is the negative impactof Catholic affiliation. One simple indicatorof how important the influence of the lawwas at this date is the decline in explainedvariation of the model from 84 to 57 percent when its impact is removed. So for1961-68 the story is that the law is the singlemost important influence on variance in
divorce, that the law itself only to a veryminor degree reflects the extent of econ-omic modernization, and that only chang-ing religious belief, a vital culturaldimension of modernization, produces a
strong and statistically significant path todivorce via variance in divorce law.
If it is possible to make separate casesfor both a cultural and historical family ofnations approach in the 1960s, the lattercase is much diminished for the later period.In the period 1976-83, as shown m Figure2, it is the law alone which is a significantpredictor of divorce rates and variance inthe law of divorce is now itself adequately
auoounted for by the effects of economicmodernization and religious belief. The bigchanges in the structure of the relationshipsbetween the earlier and the later period arethe replacement of the direct associationbetween economic modernization anddivorce rates by an indirect relationshipmediated by the nature of legal provisionand the disappearance of the somewhatmore tenuous links between female labour-force participation and both divorce ratesand legal provision. The latter set of
changes is interesting in casting some
doubts on the hypothesis of a link betweenwomen’s independence, as conferred bylabour-market position, and the propensityto seek dissolution of marnage. It is poss-ible that the tendency might be more pro-nounced if the model were not lagged,which could suggest that divorce itself con-tributes to enhanced female labour-force
participation, but, as it stands, it would
appear that much of the quite strong bi-variate association between the two vari-ables is accounted for by the increasinglystrong negative relationship between Cath-olicism and women’s labour-force par-ticipation. Although, within the structureof the model, the law is now the only directinfluence on outcomes, it is an influence
mediating these other effects, and removingthe law from the model now only reducesthe degree of explained variance from 81to 71 per cent.
Figure 3 makes it possible for us to assessthe antecedents of liberalization of the lawand subsequent divorce law change. Hereit is shown that liberalization occurred pre-cisely in those nations which were the mosteconomically modernized in 1960 and inwhich the mfluence of Catholicism wasleast, leading to a dramatic shift away fromprior legal forms, so that the liberality ofthe law in 1960 was a strong negative pre-dictor of the change that took place there-after. In other words, with the exception ofthe cultural continuity by which Cath-olicism negatively conditioned legal bar-riers to dissolution of marriage, the
trajectory of legal transformation was
shaped by a reaction to the contradictionof outdated laws persisting under con-
ditions of economic modernity. That con-tradiction, assessed in terms of the
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
295
discrepancy of the size of the non-agri-cultural labour-force in 1960 and the lib-
erality of the law at the same date, was atits greatest in the United Kingdom, theUnited States, Belgium, the Netherlands,Australia and Canada, in descending orderof economic modernization, as measured
by the size of the non-agricultural labour-force. Of these six countries only Belgiumfailed to make the transition from fault-based divorce laws to ones based largely onseparation or irretrievable breakdown ofmarriage in the period 1960-76 and
Belgium is the only one of these countriesin which Catholicism is the predominantChristian denomination. Moreover, giventhat the liberalization of the law is the onlysignificant predictor of change in divorcerates over the period, this concrete identi-fication of the countries in which the con-tradiction was most apparent explains whyit was that the English-speaking nations ledthe way in the growing divorce rate statisticsin the postwar era, for, more thanelsewhere, it was in these countries thata contradiction between divorce law and
modernity existed.The major deficiency of the account
offered here is, of course, its failure to
account for divorce rates in the UnitedStates, the country in which divorce was byfar the most prevalent in both periods. Onthe other hand, the same model which failsto explain the level of the American divorcerate is wholly adequate as an account ofchanges over the period in the sample as awhole. That suggests that the model maynot itself be inadequate, but rather that theexceptionalism of the United States mayrelate either to additional factors specific tothe American experience or some mis-
specification of the character of that experi-ence in terms of the variables constitutingthe model. In respect of the former, demo-
graphic factors, shown to be of importancein the bivariate analysis, and a peculiarlyindividualistic ethos (Weiss 1975), unfor-tunately unmeasurable in a comparativeperspective, may well hold the key. In
respect of the latter, we may hark backto the exceptional dispanty between thepractice and the letter of the Amencan lawof divorce, and note that if the USA wasclassified in terms of the liberality of its
practice ca. 1960 much of its exceptionalismwould disappear.
5. ConclusionThe story that most appropriately seems tofollow from the overall analysis of divorcerates and divorce rate change is one of theexplanatory significance of legal families ofnations in the earlier period and the declineof that concept’s explanatory utility in thesubsequent period. In 1960, the law and itspractice in all the nations other than theUSA reflected both the degree of secu-larization of contemporary populations, asnegatively indicated by the extent of
allegiance to the Catholic faith, and thediversity of barriers to the legal dissolutionof marnage that had developed in a numberot distinct groupings of nations over cen-turies of historical development. Only to aquite marginal degree did it reflect the
impact of more material manifestations ofmodernization, and then it was women’s
capacity to exert their independencethrough employment rather than the gen-eral modernization of the socio-economicstructure which counted. What appears tohave occurred in the 1960s and 1970s was adramatic breakthrough of the impact ofeconomic modernization on the law,although one still somewhat constrained inthe countries where Catholicism was
strongest.The mechanisms of that breakthrough
are not stipulated in the model presentedhere and could only be located with anyprecision by a comparative study of the
ways in which various facets of economicmodernization shaped a normative re-eval-uation that encouraged diverse groupswithin the non-Catholic churches and the
population at large to press for legal reform.It was these reforms which finally under-mined the continuity and persistence of thelegal families of nations that had hithertoexercised so potent an influence on publicpolicy m the domestic arena of marnage.A full account of the sources of national
diversity m divorce outcomes pnor to thosereforms and of the starting-point for the
trajectory of the reform movement itself isnot possible unless we start from an under-standing of the divorce phenomenon which
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
296
allows of the notion of the historical con-
tinuity of distinct families of nations.
Received August 1991Accepted October 1991
Notes1 The senior contributors to this research
endeavour are Francis G. Castles (Public PolicyProgram, Australian National University), Man-fred Schmidt (Political Science, Heidelberg) andGoran Therborn (Sociology, Goteborg).
2 It is to be found in Castles & Flood (1993),’Why Divorce Rates Differ’ in F. G. Castles(ed ), Families of Nations Patterns of PublicPolicy in Western Democracies (forthcoming).The account therein is based on the work ofKitchin (1912), Vernier (1932), Rheinstein
(1968), Rheinstein (1972), Chester (1977),Weitzman (1985), Glendon (1987), Phillips(1988), and Stone (1990).
3 For an alternative categorization of post-1960 divorce laws only, see Glendon 1987.68.
4 If one was to offer a more fine-grainedtypology of the liberality of contemporarydivorce laws, these might feature as a separatecategory. Glendon suggests that by the mid-1980s, the United States, taken as a whole, wassecond only to Sweden, where most divorcesare granted on application, in respect of makingmarriage freely terminable (Glendon 1987:64).
5 Others were Sweden and Denmark, beforethe liberalization of family law in the first decadesof the century, and France, where the law of1884 was progressively interpreted in such a wayas to permit de facto mutual consent. In Italy,the recognition in 1902 of divorces made abroadoffered a channel for legal dissolution,
amounting by the 1920s to 10 per cent of theannual number of judicial separations (seeSgritta & Tufari 1977:258).
6 We do have data for 10 countries on the
percentage of lone families defined as poor post-transfers (see Mitchell 1991) and this hardly sup-ports the welfare availability hypothesis. Quiteoutstandingly, the three of the four most divorce-prone countries of Table 2 - the USA. Canadaand Australia - qualify as those providing theworst deal for lone parents, with poverty ratesranging from 45.7 to 38.7 per cent.
ReferencesAlbrecht, S L., Bahr, H. M & Goodman, K
L. 1983. Divorce and Remarriage Problems,Adaptations and Adjustments Westport,Conn.: Greenwood Press.
Ambert. A.-M. 1980. Divorce in Canada. Onta-rio : Academic Press Canada.
Barrett, D. B. (ed.) 1982. World Christian
Encyclopedia: A Comparative Study ofChurches and Religions in the Modern WorldAD 1900-2000. Nairobi Oxford UniversityPress.
Cohen, A 1968. A Statistical Analysis of Ameri-can Divorce. New York: AMS Press.
Castles, F. G. 1990. The Dynamics of PolicyChange: What Happened to the English-speaking Nations in the 1980s European Jour-nal of Political Research 18, 491-513.
Cherlin, A 1979 Work Life and Marital Dis-solution In G Levinger & O. C Moles (eds.),Divorce and Separation Context, Causes, andConsequences. New York: Basic Books
Chester, R (ed.) 1977 Divorce in EuropeBelgium: Netherlands Interuniversity Demo-graphic Institute.
Fergusson, D. M., Horwood, L J. & Shannon,F. T. 1984. A Proportional Hazards Model ofFamily Breakdown. Journal of Marriage andthe Family 46: 3 (August).
Glendon, M. A 1987. Abortion and Divorcein Western Law Cambridge, Mass.. HarvardUniversity Press
Goode, W J 1963 World Revolution and Fam-ily Patterns New York The Free Press
Halem. L C 1980 Divorce Reform ChangingLegal and Social Perspectives. New York TheFree Press.
Hart, N. 1976 When Marriage Ends A Study inStatus Passage. Great Britain: Tavistock
Kitchin, S. B 1912. A History of DivorceLondon: Chapman and Hall
Kitson, G C. & Raschke, H J. 1981. DivorceResearch What We Know, What We Need toKnow Journal of Divorce 4(3), 1-37.
Law Reform Commission of Canada 1975Studies on Divorce Ottawa Information Can-ada.
Mitchell, D. 1991. Comparing Income TransferSystems: Is Australia the Poor Relation’? In F.G. Castles (ed.), Australia Compared. Sydney:Allen and Unwin.
Moles, O. C. 1979. Public Weltare Paymentsand Marital Dissolution: A Review of RecentStudies In G. Levinger & O. C. Moles (eds.),Divorce and Separation: Context, Causes, andConsequences New York: Basic Books.
Mortimer Commission 1966. Putting Asunder: ADivorce Law for Contemporary, SocietyLondon. SPCK.
OECD 1988 OECD Historical Statistics, 1960-86 Paris
Norton, A. J. & Glick P. C. 1979 Mantal Insta-
bility in America: Past. Present and Future. InG. Levinger and O C Moles (eds.), Divorce
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from
297
and Separation. Context, Causes, and Con-
sequences. New York: Basic Books
Phillips, R. 1988. Putting Asunder: A History ofDivorce in Western Society. New York: Cam-
bridge University Press.Price, S. J. & McKenry, P. C. 1988. Divorce.
Berkeley: Sage Publications.Przeworski, A. 1987 Methods of Cross-National
Research, 1970-83: An Overview. In M Dier-kes, H N. Weiler & A. B. Antal (eds ), Com-parative Policy Research. Aldershot: Gower.
Rheinstein, M. 1968 Divorce Law in Sweden. InP Bohannen (ed. ), Divorce and After NewYork: Doubleday and Company.
Rheinstein M. 1972. Marriage Stability, Divorceand the Law. Chicago: Chicago UniversityPress.
Rowe, G. & Krishnan, P 1980 A ComparativeAnalysis of Divorce Rates in Canada and theUnited States, 1921-1967 Journal of Divorce4(1), 61-72.
Sgritta, G. B. & Tufari, P 1977. Italy In RChester (ed. ), Divorce in Europe Belgium.
Netherlands Interuniversity DemographicInstitute
Shatter, A. J 1981. Family Law in the Republicof Ireland. Dublin. Wolfhound Press, 2nd.edn
Stone, L. 1990. Road to Divorce. England 1530-1987. London: Oxford University Press
Summers, R & Heston, A. 1988. A New Set ofInternational Comparisons of Real Productand Price Levels: Estimates for 130 Countries,1950-85 Review of Income and Wealth, 34(1),1-25
United Nations (various dates). UN Demo-graphic Yearbook. New York.
United Nations 1989. World Population Pros-pects. New York.
Varley, R. 1986. The Government Household
Transfer Base 1960-84. Paris. OECD.Vernier C. G. 1982. American Family Laws.
Stanford: Stanford University Press.Weiss, R S. 1975. Marital Separation New
York: Basic Books.Weitzman. L J 1985 The Divorce Revolution.New York: The Free Press.
at University of Warwick on May 8, 2009 http://asj.sagepub.comDownloaded from