UNCORRECTEDPROOF
1 Stability and Change: The Structuration of
2 Partnership Histories in Canada, the
3 Netherlands, and the Russian Federation
4 MELINDA MILLS
5 Department of Social-Cultural Sciences, Faculty of Social Sciences, Vrije Universiteit (Free
6 University) Amsterdam, De Boelelaan 1081, 1081 HV Amsterdam, The Netherlands (E-mail:
7 mc.mills@fsw.vu.nl)
8 Received 1 September 2003; accepted in final form 18 February 2004
9 Mills. M. 2004. Stability and change, The structuration of partnership histories in Canada, the
10 Netherlands and the Russian Federation. European Journal of Population.
11 Abstract. This paper explores stability and change in women's partnership histories since the
12 late 1940s in Canada, the Netherlands, and the Russian Federation. Giddens' (1984) theory of
13 structuration is used to understand how the social structure enables or constrains behaviour.
14 Entire partnership histories are examined by applying a Markov and semi-Markov multistate
15 approach to investigate the type, timing, duration, and complexity of partnerships. Results
16 show earlier union formation for younger cohorts in the Russia Federation compared to
17 postponement trends in the other countries. Cohabitation appears to increasingly serve as an
18 alternative to marriage, particularly in Canada. When facilitated by the social structure, di-
19 vorce levels are high (Russian Federation, Canada). Widowhood in the Russia Federation
20 persists even among younger women. Re-partnering is the highest in the Russian Federation,
21 with post-marital cohabitation gaining ground in Canada. Partnership histories are increas-
22 ingly complex in the Netherlands and particularly Canada but remain stable in the Russian
23 Federation.
24 Key words: Canada, cohabitation, divorce, marriage, multistate life tables, remarriage, Rus-
25 sia, structuration, The Netherlands
26 Mills M. 2004. Stabilité et changement. La struturation des histoires conjugales au Canada,
27 aux Pay-Bas et en Russie, Revue Europé enne de Dé mographie.
28 Ré sumé . Cet article s'inté resse aux changements intervenus dans les histoires conjugales des
29 femmes depuis la fin des anné es 1940 au Canada, aux Pays-Bas et en Russie. Il s'appuie sur la
30 thé orie de la structuration de Giddens (1984) pour comprendre comment les structures sociales
31 permettent ou contraignent les comportements. Une approche multi-é tats de type Markov et
32 semi-Markov est appliqué e a` des histoires conjugales comple` tes pour analyser le type, le
33 calendrier, la duré e et la complexité des relations entre partenaires. On observe une formation
34 des unions plus pré coce pour les gé né rations les plus ré centes en Russie alors que dans les
35 autres pays la tendance a` retarder l'entré e en union se poursuit. La cohabitation apparai^t de
36 plus en plus comme une alternative au mariage, notamment au Canada. Quand les structures
37 sociales le facilitent, les niveaux de divorce sont é levé s (Russie, Canada). Le veuvage est encore
European Journal of Population 00: 1­35, 2004.
Ó 2004 Kluwer Academic Publishers. Printed in the Netherlands.
1
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF!
PDF-OUTPUT
UNCORRECTEDPROOF
38 fré quent chez les jeunes femmes en Russie. La remise en couple est particulie` rement fré quente
39 en Russie tandis que la cohabitation apre` s un premier mariage gagne du terrain au Canada.
40 Les histoires de couples deviennent de plus en plus complexes aux Pays-Bas et surtout au
41 Canada a` l'inverse de la Russie ou` elles n'ont pas changé .
42 Mots clé s: Canada, cohabitation, divorce, mariage, tables multi-é tats, remariage, Russie,
43 structuration, Pays-Bas
44
45 1. Introduction
46 Demographers have argued that there is an increasing ``pluralisation'' of the
47 life course such as the increased ``differentiation'' of partnership histories
48 (Lesthaeghe, 1995). But is this true for all societies? How does the social
49 structural context impact the ``pluralisation'' of demographic behaviour? The
50 goal of this paper is to explore stability and change in women's partnership
51 histories since the late 1940s in Canada, the Netherlands, and the Russian
52 Federation. Previous demographic theories are embedded in Giddens' (1984)
53 theory of structuration, which is used as a heuristic to understand how the
54 social structure enables or constrains partnership behaviour across time and
55 in diverse contexts. Entire partnership histories are examined as opposed to
56 one fragmented transition by applying a Markov and semi-Markov multi-
57 state approach to investigate patterns in the type, timing, duration, and
58 complexity of partnerships.
59 To determine the extent of resilience or revolution among partnerships,
60 four phases of the partnership biography are considered: pre-partnership,
61 first union formation, dissolution of first partnership, and re-partnering. Key
62 research questions include: Has there been a postponement of first unions for
63 younger cohorts? Do more women never have a partnership? How does the
64 nature of marriage and cohabitation differ between countries and across
65 time? What is the pattern of new types of partnerships and how are they
66 legitimated? Are there new stages in partnership biographies? How does the
67 social structure enable or constrain certain behaviours such as divorce? Are
68 relationship histories more turbulent among younger women? Who re-part-
69 ners first and if so, what type of relationship do they choose? Has the overall
70 complexity of relationships increased to the extent that it can be labelled as
71 ``pluralised''?
72 Due to the complexity of the multistate models across three countries,
73 only two cohorts of women born in the late 1940s and early 1960s, or roughly
74 a ``mother'' and ``daughter'' cohort, were selected for the analysis. The
75 ``mother'' cohort, born between 1946 and 1950 (1950 and 1954 in the
76 Netherlands due to data limitations), entered the partnership market in the
77 mid- to late-1960s and early 1970s. The ``daughter'' cohort, born between
78 1961 and 1965, entered the partnership market in the late 1970s and early
MELINDA MILLS2
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
79 1980s. The assumption is that birth cohorts share similar conditions of the
80 social structure (e.g., economic, cultural, legal, labour, marriage, and housing
81 markets).
82 This study builds on, yet contributes to previous research in the field of
83 partnership studies. As Lesthaeghe (1998) recently argued, demographic
84 studies of family formation have persistently been examined with the aid of
85 three basic frameworks: the theory of increased female economic autonomy
86 (Becker, 1981), relative economic deprivation (Easterlin, 1976), and idea-
87 tional shift (Lesthaeghe and Meekers, 1986; Preston, 1986). This study de-
88 parts from the sole use of demographic theories to embrace an alternative
89 theoretical framework from the discipline of sociology (Giddens, 1984). In
90 addition to similarities with demographic theories, structuration theory adds
91 fresh insights beyond the habitual economic-based assumptions, particularly
92 in its attention to power, norms and sanctions, cultural frameworks, and
93 bridging the interplay between macro-level institutional context and micro-
94 level individual action.
95 Second, most multistate applications to partnerships use vital statistics or
96 census data and estimate Markov models (Willekens, 1987). Since only
97 officially registered events are available in the aforementioned data, ``mari-
98 tal'' life tables can only be estimated. The use of individual-level survey data
99 allows the true richness of partnership biographies to surface by capturing
100 ``unofficial'' events such as cohabitation formation and dissolution and more
101 intricate stages in marital dissolution (i.e., separation). A decisive factor that
102 impacts the transition from one partnership state to another is the duration
103 in the state of origin. For example, divorce is closely related to the duration
104 spent in a marriage. Going beyond a time-homogeneous (i.e., time-station-
105 ary) Markov process, the introduction of a semi-Markov model allows us to
106 examine not only age, but also duration in a state and thus the prediction of
107 age-duration-specific probabilities (Hoem, 1972; Namboodiri, 1991; Rajul-
108 ton, 1992).
109 Third, previous studies mainly focus on one transition, such as first
110 partnership formation or divorce, thus adopting an outcome, instead of a
111 process-oriented approach. Yet life events are part of an underlying trajec-
112 tory where outcomes are consequences of earlier conditions, events, and
113 experience (Mills, 2000). The examination of entire partnership histories al-
114 lows us to envision transitions in context and understand the entire part-
115 nership process. Finally, in-depth comparisons of entire partnership histories
116 between countries from disparate regions remain rare. Comparisons across
117 diverse contexts draw attention to what is unique and provides a measure of
118 relative importance. It likewise furnishes insight into what is contextually
119 bound or more universal across individuals. Three diverse contexts were
120 chosen due to known variations in partnership behaviour, divergent social
121 structures, and to serve as a general proxy for partnership behaviour in
THE STRUCTURATION OF PARTNERSHIP HISTORIES 3
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
122 different regions of the world, in this case: North America (Canada), Western
123 (the Netherlands), and Eastern Europe (Russian Federation).
124 The ensuing discussion is ordered as follows. Section 2 defines and de-
125 scribes how structuration theory is used to interpret continuity and change in
126 partnership histories. Each country is then placed within this framework in
127 Section 3, followed by a summary of hypotheses in Section 4. Section 5
128 describes the data sources used in the empirical analysis followed by a brief
129 description and specification of the multistate methods in Section 6. Detailed
130 results are presented and interpreted in Section 7, concluding in Section 8
131 with a discussion of the implications of these findings and suggestions for
132 further research.
133 2. The Structuration of Partnerships
134 Structuration theory (Giddens, 1984) provides a novel and adaptable
135 framework to understand how partnership behaviour varies between contexts
136 and has changed or remained stable over time. It provides a bridge between
137 agency and structure, often referred to in demography as the nexus between
138 micro-level individual behaviour and macro-level social institutions. A key
139 concept is the ``duality of structure'', which defines the mechanisms of the
140 social structure as being ``the medium and outcome of the conduct it
141 recursively organises'' (Giddens, 1984, p. 374). In other words, the social
142 relations that determine partnership behaviour are in fact the instruments
143 used by individuals to reproduce prevalent partnership ``institutions'' such as
144 marriage or invent innovative types of partnerships like non-marital
145 cohabitation. These social relations are in turn enabled or constrained by the
146 context of the social structure itself. The process of structuration thus de-
147 scribes how the social structure is either reproduced through the repetition of
148 routine social practices or transformed through the enactment of new
149 behaviour or as a result of unintended consequences. By adopting this per-
150 spective, observed partnership outcomes are viewed as the result of micro-
151 level individual action and interaction that influences and is influenced by
152 meso-level networks (friends, family) and macro-level context (policies, legal
153 regulation).
154 Another hallmark of this theory is the operationalisation of the social
155 structure into the three dimensions of domination, signification, and legiti-
156 mation. When making partnership decisions, individuals draw on the domi-
157 nation structure, which consists of rules and resources that in turn influence
158 the power or capacity to act. Rules may be formal such as the legal restrictions
159 regarding divorce or informal such as religious or cultural norms. In order to
160 act according to the norm or conversely, engage in innovative behaviour,
161 individuals must posses the facility or power for action, which is defined by
MELINDA MILLS4
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
162 their level of ``allocative'' (i.e., material) and ``authoritative'' (i.e., power)
163 resources. Resources may be economic or other forms of social capital, such
164 as educational credentials, which have the potential to augment bargaining
165 power within a relationship. Giddens' attention to ``allocative'' resources is
166 reminiscent of previous demographic theories such as Easterlin (1976), Becker
167 (1981), and Butz and Ward (1979). Since these theories are generally under-
168 pinned by economic-based explanations, they have scant development or
169 make no references to power, culture, norms, or values. The addition of
170 ``authoritative'' power complements previous demographic theories.
171 A second component is the signification structure. This consists of the
172 interpretative schemes or mental frameworks that individuals draw upon,
173 which subsequently regulate everyday activity (Giddens, 1984, p. 31). These
174 are the ``semantic rules'', ``stocks of knowledge'', or ``cultural frames'' that
175 individuals refer to when they are trying to make sense of reality. It may be
176 based on, for instance, religious beliefs or rituals that manifest themselves in
177 customs of behaviour. This echoes the work of Heiner (1983), who argued
178 that cultural traditions, social institutions, or norms serve as rule-mecha-
179 nisms that restrict the flexibility to choose potential courses of actions, or
180 which produce a selective alertness to information. The cumulative history
181 and collective memory of partnership behaviour constitutes the interpretative
182 scheme that in turn equips us with a mode to understand and organise
183 everyday activities. Individuals often act, Giddens (1984) argues, via routine
184 behaviour of the re-enactment of values and norms. However, routinisation
185 is not the only type of action that exists.
186 There is a potential for change when individual action evolves into
187 aggregate collective notions about what types of behaviour are acceptable.
188 Consider, for example, the evolution of cohabitation from a ``deviant''
189 relationship to a viable ``alternative'' to, ``trial'' stage on the path to marriage
190 or a selection process to ``weed out'' weak unions (Oppenheimer, 1988;
191 Rindfuss and Vandenheuvel, 1990; Axinn and Thornton, 1992; Lillard et al.,
192 1995). Lesthaeghe (1995) described the emergence of cohabitation in older
193 cohorts as the desire to behave in a deviant manner to protest against
194 authority, conformity, and conventions. Yet even as early as the 1980s in
195 Sweden, Trost (1980, p. 19) suggested ``far from being deviant, cohabitation
196 has become a social institution''. The meaning ascribed to partnership
197 behaviour differs across time and between countries (Manting, 1996). The
198 ``structuration'' of new partnership types or stages occurs when relatively
199 uncommon behaviour by a comparative few, such as cohabiting couples or
200 widows that remarry, evolves into a new social practice, which in turn dif-
201 fuses and transforms the social structure itself. This occurs when a particular
202 threshold is reached and the evolution of behaviour ultimately overturns
203 existing values and norms.
THE STRUCTURATION OF PARTNERSHIP HISTORIES 5
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
204 The final element of the social structure is legitimation, which is the moral
205 constitution of interaction and action, which materialises in the form of
206 ``informal'' moral or ``formal'' legal regulations (Giddens, 1984, p. 21).
207 Norms constitute rules, which indicate how values surrounding partnerships
208 are realised. Rules refer to rights and obligations. The structure is reproduced
209 when sanctions (via norms) are imposed during interaction. Families or other
210 informal institutional bodies impose norms that are experienced in the form
211 of sanctions, which have formal counterparts in, for instance, religious or
212 legal regulations. Norms often materialise in policy legislation, which may
213 attempt to dissuade innovative or non-traditional behaviour. For example,
214 when new forms of family behaviour emerge, they are often coined with
215 terms that reflect moral sanctions or their representation as the antithesis of
216 traditional behaviour such as ``non-marital'' or ``pre-marital'' unions and
217 ``illegitimate'' children or ``out-of-wedlock'' births. These terms have clear
218 normative connotations reflecting what the new behaviour is not.
219 The legitimation structure works not only to deter, but may also
220 endeavour to stimulate behaviour via tax incentives or legal recognition of
221 certain types of partnerships. Yet policy regulations do not cause shifts in
222 demographic behaviour, rather they constrain or enable it. Individuals
223 making partnership decisions thus draw on these structures and enact the
224 rules often through routine behaviour. If they deviate, they may be con-
225 strained by sanctions such as social stigma or lack of recognition of their
226 situation (e.g., no benefits or rights for a cohabiting partner).
227 The core contribution of structuration theory is that it adds attention to
228 the: (1) interplay between individual behaviour and the social structure; (2)
229 notion of authoritative (and not merely material economic) power; (3) cul-
230 turally based interpretative mental frameworks; and, (4) sanctions moral and
231 legal norms that regulate behaviour. It is useful for this study as it can be
232 applied in a more general sense across the entire partnership biography
233 among different institutional and temporal contexts.
234 3. The Context of Partnerships
235 The theoretical concepts are now embodied by a description of the pertinent
236 aspects of the social structure within each of the study countries since the late
237 1940s that impacts partnership behaviour. Formal legitimation and domi-
238 nation structures are operationalised by social policies and legal regulations.
239 The assumption is that these are a reflection of values or the ``informal''
240 signification structure and the norms and sanctions that enable or constrains
241 action.1
This overview is not intended as an exhaustive exploration of partner
242 or family-related policies and regulations in each country, but rather as a way
243 to operationalise the theoretical framework and as a basis to develop more
244 specific research hypotheses.
MELINDA MILLS6
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
245 3.1. THE RUSSIAN FEDERATION
246 There are several key elements of the post 1950s Russian Federation social
247 structure that differentiate it from Canada and the Netherlands: the
248 stronghold of marriage, more lenient divorce laws, high levels of male
249 mortality, and a turbulent history. Moskoff (1983) argued that the institution
250 of marriage and re-marriage remained strong in the former Soviet Union.
251 Early and universal marriage was an enduring feature for Russians (Vish-
252 nevsky, 1996; Scherbov and Van Vianen, 2001). Avdeev and Monnier (1994)
253 furthermore argue that the acceptance and ability to remarry has increased
254 over time, with around 25% remarrying after divorce in 1993. High remar-
255 riage rates are attributed to a general attachment to marriage, but also to the
256 younger age at which divorcees experience dissolution in comparison to
257 widows.
258 A comparatively lenient history of divorce laws in the former Soviet
259 Union created an atmosphere that was both legal and morally conducive to
260 divorce. This meant greater social acceptability and authoritative power and
261 support for women to leave a union. In 1944, more restrictive measures were
262 introduced to create obstacles to divorce, but these ``cumbersome'' proce-
263 dures were again relaxed in 1965 (Von Frank, 1979; Moskoff, 1983). In fact
264 by 1968, further liberalisation meant that couples with no children could
265 obtain a divorce by sending a postcard to the local registration office and wait
266 60 days (Von Frank, 1979). This was manifested in a high divorce rate, to the
267 extent that in 1993 approximately 50% of marriages ended in a divorce
268 (Avdeev and Monnier, 1994). In fact, the divorce rate in the former USSR
269 was already one of the highest in the world in 1971, with a crude divorce rate
270 of 2.63, which rose by 29% from 1971 to 1990 to reach 3.39 (United Nations,
271 1997).
272 Early and formal support by the state for women's participation in the
273 labour force also increased their allocative resources.2
The early Family Law
274 Code of 1919, likewise declared the ``equality of sexes'', which was again
275 reaffirmed in 1977 to increasingly liberalise divorce and abortion on demand.
276 Imbrogno and Imbrogno (1989, p. 3) argue: ``a Soviet citizen is legally
277 guaranteed autonomy in marriage and family.'' This fostering of higher levels
278 of authoritative power likely also contributed to the ability to leave a part-
279 nership. The combination of lenient divorce regulations and little stigma,
280 coupled with a history of female labour force participation, left couples with
281 relatively fewer constraints to divorce. They could also enter marriage with
282 the advanced knowledge that divorce was relatively easy.
283 A final distinguishing factor in the Russian Federation is high levels of
284 male mortality. From 1990 to 1995, the life expectancy at birth in the Russian
285 Federation was 75 years for women and almost 10 years lower at 66 years
286 for men (United Nations, 1997, pp. 23­27). In Pskov, where the sample for
THE STRUCTURATION OF PARTNERSHIP HISTORIES 7
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
287 this analysis is taken (see Section 5), life expectancy was somewhat lower at
288 71.9 for women and 58.9 for men (Goskomstat of Russia, 1994). This likely
289 related to a sharp increase in male deaths from non-natural causes attributed
290 primarily to alcoholism, which was slightly curbed by Mikhail Gorbachev's
291 anti-alcohol campaign in the mid-1980s, but soon rebounded after the
292 campaign's abandonment (DaVanzo and Adamson, 1997). Russia's homi-
293 cide and suicide rates are the highest in the world (DaVanzo and Adamson,
294 1997, p. 4). A final aspect is the turbulent historical experiences in the Rus-
295 sian Federation, particularly for the older cohort of women (Scherbov and
296 Van Vianen, 2001). This undoubtedly impacts women's everyday lives and
297 their partnership histories.
298 3.2. CANADA
299 Relevant aspects within the Canadian context are a changing signification
300 structure with respect to relationships, comparatively early policies that
301 provided women with more allocative and authoritative resources, and
302 substantial changes in the legitimation structure in the form of divorce laws.
303 As in many other Western countries, attitudes towards non-marital sexual
304 behaviour and non-marital cohabitation became progressively more accept-
305 able, with less emphasis placed on marriage, and more liberal values towards
306 alternate relationship behaviour (Turcotte and Goldscheider, 1998; Wu,
307 2000). There has also been a trend towards cohabitation as a stable union in
308 itself or ``alternative'' to marriage (Le Bourdais and Marcil-Gratton, 1996;
309 Wu, 2000).
310 Canadian women experienced a shift in increased allocative and authori-
311 tative resources, which occurred later and in a different manner than for
312 women in the former Soviet Union. Early Canadian family policy was di-
313 rected towards encouraging women to stay at home by offering incentives
314 such as the family allowance program (Gauthier, 1996). But employment
315 equity and parental leave laws beginning in the 1960s reduced the opportu-
316 nity costs of entering a union and subsequently having children for women
317 (Baker, 1995). In 1990, the percentage of the female population aged 15­64
318 that participated in the labour market was 68.2% in Canada, compared to
319 53.0 in the Netherlands (OECD, 1996 in O'Connor et al., 1999, p. 68). In
320 general, these were also women in full-time positions. Previous Canadian
321 studies (e.g., Belanger and Turcotte, 1999) have found that increased female
322 autonomy has not resulted in an overall decline of first unions, but rather a
323 change in their type and timing, similar to Oppenheimer's expectations
324 (1988).
325 Canadian policy was more conservative than Soviet measures with respect
326 to contraceptive use and divorce, with both being virtually normatively and
MELINDA MILLS8
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
327 legally unattainable before 1969. Divorce was informally and formally
328 sanctioned by virtue of being expensive, legally complex, and socially unac-
329 ceptable. After 1969, women could leave a relationship if there was adultery
330 or physical cruelty, or a legally enforced separation period of three to five
331 years. In 1985, the separation period was shortened to one year and provi-
332 sions relating to custody and support of children were changed (Baker, 1995).
333 After 1985, alimony was based on financial need instead of life-long support
334 that further reduced the impediments to divorce and long periods of financial
335 co-dependence.
336 3.3. THE NETHERLANDS
337 The Dutch social structure consists of an enigmatic blend of ``non-inter-
338 ventionist'' approaches to family related policy (Jonker, 1990), low full-time
339 female labour market participation, a strong welfare regime, coupled with
340 early formal acceptance of relationships beyond legal marriage. Kamerman
341 and Kahn (1978 in Baker, 1995, p. 37) maintain that Dutch family policy has
342 been largely implicit. There is a tendency to emphasise autonomy and per-
343 sonal accountability, thus shifting the burden of family-related decisions and
344 responsibility from the institutional level to the individual or family unit
345 (Baker, 1995). Due to the long-term governance of the Christian Democratic
346 Party, religion played a role in the development of the social structure. This
347 was reflected in the reluctance to provide childcare services directly, which in
348 turn encouraged or even compelled Dutch women to remain at home. This
349 was complemented with an extensive social assistance benefit for mothers
350 with pre-school children and divorcees, which permitted women to stay at
351 home and afforded them basic financial independence (Poortman and Kal-
352 mijn, 2002). In fact, the total Dutch social assistance benefit is more than
353 three times the value of the Canadian one (Gauthier, 1996, p. 166).
354 This is in stark contrast to Soviet laws, which promoted the integration of
355 women and mothers into the labour force as early as the 1940s and Canadian
356 policies that attempted to actively integrate women in the labour force in the
357 1960s. In fact, it was not until the mid-1970s and late 1980s that several laws
358 promoting equality in the workplace were initiated (ARPL, 2000). Together,
359 these factors have attributed to low full-time labour force participation of
360 Dutch women. However, since the early 1990s, there has been new legislation
361 to improve and provide subsidies for childcare. This context translates into
362 comparatively lower levels of allocative and authoritative resources for wo-
363 men, combined with more restrictive norms regarding the acceptability of
364 divorce. In fact, previous research has shown that a low number of Dutch
365 couples actually discuss divorce or consider it as an option (Janssen et al.,
366 1998).
THE STRUCTURATION OF PARTNERSHIP HISTORIES 9
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
367 Conversely, in comparison to the two other study countries, the Nether-
368 lands was the first to adopt formal legal regulations that recognised alternate
369 forms of partnerships. There has been a long tradition of high levels of
370 cohabitation (Manting, 1994). In 1992, a Decree that recognised other forms
371 of relationships besides marriage came into force allowing co-residing part-
372 ners to be treated in the same way as married couples if they submitted a
373 notarised agreement asserting that they were cohabiting (APRL, 2000). The
374 Registered Cohabitees Act of 1997 created legal recognition for the status of
375 cohabitees or partners and the dissolution of registered cohabitation was also
376 formalised as a court decision. The registered partnership enacted in 1998
377 allowed it to become virtually legally equivalent to marriage (Ministry of
378 Justice, 1997).
379 4. Research Hypotheses
380 On the basis of the previous theoretical and contextual discussion, 12 central
381 research hypotheses are formulated according to partnership phase. The first
382 two hypotheses examine the first pre-partnership phase.
383 H1 ­ Postponement hypothesis. Due to transformations in the three areas
384 of the social structure, in comparison with older cohorts, younger cohorts
385 will postpone entry into a first union.
386 H2 ­ Remaining never in a partnership hypothesis. Due to the greater
387 allocative and authoritative resources of Canadian women and the turbulent
388 historical experiences of older Russian cohorts, both groups are expected to
389 have higher levels of remaining never in a partnership.
390 The second stage of partnership histories is the examination of first union
391 formation, which is formulated within three hypotheses:
392 H3 ­ Marriage attachment hypothesis. Both younger and older Russian
393 women will be more attached to the institution of marriage than in the other
394 countries. This will be evident by: (a) little or no cohabitation and (b) overall
395 higher entry into marriages in comparison with Canada and the Netherlands.
396 H4 ­ Type of union hypothesis. The transformation of the signification
397 structure entails that women from younger cohorts will have a higher
398 probability of choosing cohabitation as a first union over marriage in Can-
399 ada, and due to earlier acceptance, even more enhanced in the Netherlands.
400 H5 ­ Nature of cohabitation hypothesis. As a result of conversions in the
401 type of union (H4), cohabitation is expected to increasingly take the form of
402 an alternative (as opposed to trial) marriage in both Canada and the Neth-
403 erlands.
404 The third stage of first union dissolution is divided into four hypotheses.
MELINDA MILLS10
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
405 H6 ­ Structural support of divorce hypothesis. Where the social structure
406 enables divorce, the levels will be higher, which is expected in the Russian
407 Federation, followed by Canada and the Netherlands. This will be empiri-
408 cally observed by: (a) higher levels of divorce, but also (b) a faster rate of
409 divorce and subsequent shorter duration of time in first marriage, and (c) a
410 larger proportion of women's lives spent in the ``divorced'' state.
411 H7 ­ Separation stage hypothesis. Due to the formally enforced separation
412 period of three to five years up to 1985, Canadian women who dissolve
413 marital unions are expected to have a clear ``separation stage'' in their
414 partnership history, which is artificially created by the legitimation structure.
415 H8 ­ Widow hypothesis. As a result of high levels of male mortality, the
416 expectation is that there will be a larger number of widows in the Russian
417 Federation, particularly among the older cohort.
418 H9 ­ Dehabitation hypothesis.3
Due to the selection process and less
419 attachment of cohabitors to sanctions, norms, and legal specifications, co-
420 habitors are expected to have: (a) higher levels of dissolution compared to
421 marital unions; (b) relationships of a shorter duration; and (c) higher levels of
422 dissolution among younger cohorts. Levels of dehabitation are expected to be
423 lower in the Netherlands than in Canada.
424 The final phase of re-partnering consists of two hypotheses.
425 H10 ­ Marital re-partnering hypothesis. In light of higher attachment to
426 marriage, higher divorce and widowhood levels, Russian women will have
427 higher remarriage rates, particularly divorced women.
428 H11 ­ Cohabitation re-partnering hypothesis. In lieu of the prospect that
429 younger cohorts are more likely to cohabit (H4) and that first cohabiting
430 unions are anticipated to be increasingly more fragile due to dehabitation
431 (H9) and divorce (H6), it is interesting to speculate further about the re-
432 partnering experience of younger cohorts. The expectation is that after dis-
433 solution of first partnerships, younger cohorts are: (a) more likely to enter
434 higher order (second) cohabiting relationships, and due to less sanctions and
435 constraints (b) will do so at a faster rate than older cohorts.
436 A final and general hypothesis compares entire partnership histories in
437 general.
438 H12 ­ Complexity hypothesis. Due to less sanctions, more individual re-
439 sources and shifting values, more complex relationship histories will be found
440 among: (a) younger cohorts and (b) Canadian and Dutch women. This is
441 operationalised by: (a) the pluralisation of relationships (represented
442 empirically by more partnership states and stages) and (b) multiple rela-
443 tionships (represented by an increase in the number of partnerships).
THE STRUCTURATION OF PARTNERSHIP HISTORIES 11
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
444 5. Data
445 Three different individual-level data sources were used in the analysis.
446 Readers who require more detailed information can refer to the sources listed
447 below. The 1995 General Social Survey is used for Canada, taken from the
448 Fertility and Family Survey (FFS) recode file, which contains a sample of
449 4166 women between the ages of 15 and 54 years (Statistics Canada, 1997).
450 The female sample of 4516 women from the 1993 Netherlands Family For-
451 mation survey (also from the FFS) includes women aged 18­42 years (Latten
452 and De Graaf, 1997). Finally, a selection of data from the Russian Federa-
453 tion Microcensus of 1994 of the oblast (region) of Pskov was used. The
454 Microcensus is a 5% sample of the entire population (excluding the Chechen
455 Republic). The sample was reduced to 9631 women between the ages of 15
456 and 49 years. The Russian data used in this study are taken from the oblast
457 (region) of Pskov in the Northwest. It was chosen due to its homogeneity of
458 Russian language (96.3%) and ethnicity (95.3%) and in consultation with
459 Russian researchers (Volkov, 1999; Scherbov and Van Vianen, 2001). As
460 Scherbov and Van Vianen (2001) state, this data has somewhat of a selection
461 bias in that it contains only survivors. Although this is true of all of the data
462 sources, the high mortality and catastrophic events that have occurred in
463 recent Russian history raise the pertinence of this issue. As described in
464 Section 1, only two cohorts of women born between 1946 and 1950 (1950 and
465 1954 in the Netherlands due to data limitations) and 1961 and 1965 were
466 selected for the analysis. This represents roughly a ``mother'' cohort who
467 entered the partnership market in the mid to late 1960s and early 1970s and a
468 ``daughter'' cohort who entered in the late 1970s and early 1980s.
469 6. Methods and Models
470 A technique amenable to the examination of partnership histories is the
471 multistate (increment-decrement) life table (Willekens, 1987). Figures 1a­c
472 illustrate the models with the parameters, lij, denoting the rate of transition
473 from state i (e.g., never married) to state j (e.g., first marriage). All transient
474 states are discrete, with the exception of final absorbing states specified within
475 each model (Namboodiri and Suchindran, 1987). Transition rates (until the
476 survey date) are estimated by age or duration [x, x + n) in the Markov and
477 Semi-Markov models, respectively, using LIFEHIST (Rajulton, 1992) and
478 the author's own calculations, thereby examining the temporal axis of both
479 individual and process time.
480 The underlying assumption is that a stochastic process generates the
481 events in the partnership histories of women, which are inferred from random
482 variables (Namboodiri, 1991). In the more commonly applied Markov
483 model, the probability of transition from state i to j is not only dependent on
MELINDA MILLS12
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
484 the origin state i, but also on the age of the individual, denoted by x. It fills
485 the homogeneity assumption by disregarding the pathway in which the
486 previous event was reached. Thus, past history of state occupancy and
487 duration since entry in the origin state are not taken into account. It is thus
488 time-homogeneous (or time-stationary).
489 One way to remove time-homogeneity is to consider the impact of dura-
490 tion on the outcome of events. For example, divorce is closely related to the
2 5
1
st
cohabitation
1
st
dehabitation
1
Neverin
Union
(age 15)
3 4 6
1st
marriage
1st
divorce/widow
2nd
marriage
(b)
13
23
12
5342
46
25
34
56
2 4 7
1
st
cohabitation 1
st
dehabitation 2
nd
cohabitation
1
Never in
union
(age 15) 6
1
st
Divorce/Widow
3 5 8
1st
Marriage 1
st
Separation 2nd
Marriage
56
62
47
43
(61-65
only)
36
(46-50
only)
12
13
23
24
35
68
78
(46-50
only)
28
(46- 50
only)
73
(61- 65
only)
1 3
Never
in union
2 Separated
/Divorced 5
(age 15) Marriage 2
nd
Marriage
4
Widowed
(a)
(c)
2312
24
35
45
1
st
Figure 1. Multistate models of partnership status categories and transitions: (a) Russian
Federation (five-state), (b) The Netherlands (six-state) and (c) Canada (eight-state).
THE STRUCTURATION OF PARTNERSHIP HISTORIES 13
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
491 duration spent in a marriage. Semi-Markov models are therefore introduced
492 to reflect both age and duration in a state via the estimation of age-duration-
493 specific probabilities. For a detailed mathematical description of these life
494 table calculations, readers can refer to standard sources such as Rogers
495 (1975), Namboodiri and Suchindran (1987), or Schoen (1988).
496 The origin state of never being in a union is set at age 15. The occurrence
497 of a partnership event (e.g., entering into first marriage) signals a transition
498 from one discrete state to one or more discrete states within a specified
499 interval. As Figure 1a­c illustrate, first union formation occurs as the tran-
500 sition to first marriage (Russia) or into marriage or cohabitation (Nether-
501 lands and Canada), which may be followed by various other stages of union
502 formation. Union dissolution from marriage occurs in the form of a shift to:
503 separated (in Canada only), divorced (combined with separation in Nether-
504 lands and Russia), or widowhood (Russia only). In common-law unions,
505 dissolution occurs due to the death of a partner or as ``dehabitation'' (break
506 of relationship). Due to the small number of women who experienced the
507 death of a partner within common-law unions in Canada and the Nether-
508 lands, these categories were collapsed into one defining state called ``de-
509 habitation.''
510 Figure 1a shows the five-state model for the Russian Federation, with
511 four transient (one of which is the origin state) and one absorbing state of
512 second marriage and five non-repeatable transitions.4
The model for the
513 Netherlands (Figure 1b) contains six states with five transient and one
514 absorbing state (second marriage) for a total of nine transitions. The most
515 complex model is the eight-state Canadian model, shown in Figure 1c, which
516 has seven transient and one absorbing state (second marriage). Due to
517 variations in union transitions experienced by older and younger cohorts in
518 Canada, slightly different transitions were modelled, for a total of 12 and 11
519 transitions for the older and younger cohorts, respectively. For both Canada
520 and the Netherlands, due to small numbers, the analysis goes up to second-
521 order relationships and does not include detailed reasons for union disso-
522 lution (the majority are divorce and for dehabitation ``break of relation-
523 ship''). First cohabitation in the Canadian model refers to cohabiting unions
524 that were experienced for the first time by the individual. In other words, a
525 woman may directly marry (first union), divorce, and then enter a cohabiting
526 union for the first time ­ but it is actually her second relationship and first
527 cohabiting union. The impact of this categorisation is discussed in more
528 depth shortly.
529 The multistate model offers several analytical advantages. First, it pro-
530 vides a multitude of ways to interpret similarities and differences in the
531 timing, intensity, tempo, complexity, and type of life course histories of
532 individuals. This basic and rich information to describe partnership biogra-
533 phies is often overlooked when advanced regression analyses are used.
MELINDA MILLS14
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
534 Armed with these techniques, we can answer many compelling questions such
535 as: What is the probability that a woman who has never had a relationship at
536 age 25 will remain single at older ages? What percentage of their lives can
537 women expect to spend in different types of partnership states? Is the
538 probability of dissolving a cohabiting union higher than that for a marital
539 union? How long will women remain married or cohabiting before the union
540 ends? What is the probability that they will enter a second cohabiting or
541 marital union?
542 The second key benefit is that it enables the conceptualisation and
543 examination of women's entire union histories up to the point of the survey
544 instead of one fragmented transition. Using the multistate model, events in
545 the union career are dynamically defined as a part of a staging process or
546 sequence of cumulative experiences (Willekens, 1991). It is only through the
547 examination of the entire partnership process that we can determine which
548 transitions are more meaningful to pursue in detail, rather than arbitrarily
549 choosing just one. Finally, it is a general analytical method useful for a cross-
550 national and cross-temporal comparative study.
551 A drawback of the multistate approach is that it neglects the importance
552 of heterogeneity within the population under study. For this reason, the life
553 table is often a starting point of a basic risk model for many analyses. The
554 position taken in this study, which is likely consistent with critics, is that the
555 multistate method is a powerful descriptive and exploratory method that can
556 be used to uncover basic behavioural patterns. It is a tool that returns us to
557 the basic. Or, as Hannan (1984, p. 43) argued, multistate demography has a
558 ``power'' for deriving long-run implications of rates and probabilities of
559 demographic behaviour.
560 7. Results
561 The results are discussed in relation to the 12 major hypotheses. Table 1
562 encompasses entire partnership histories and will therefore be referred to
563 throughout. It shows the expected (or life table) percentage of time that
564 women can expect to spend in various partnership states over their lifetime.
565 These ``life expectancy'' statistics are linked to the time spent in all part-
566 nership states. As an aid for interpretation, consider the example of Canadian
567 women born from 1946 to 1950, who are at the age of 40 (first column).
568 During interpretation it is essential to bear in mind the ``disposable time''
569 lived by individuals within each cohort, which in this case is restricted to
570 information up to age 48. A typical woman in this group would spend an
571 average of 10.24% of her life before entering a first partnership, 57.45 in a
572 first marriage, 3.53 in first cohabitation, 5.69 in dehabitation, and could
573 expect to be separated for 6.94 and divorced for 14.79% of her lifetime (up to
THE STRUCTURATION OF PARTNERSHIP HISTORIES 15
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
Table 1. Life table percentage of lifetime to be spent in various partnership states, women,
Canada, The Netherlands and Pskov, Russian Federation, by selected ages and cohort*
Age Canada The Netherlands Pskov, Russian Fed.
b1946 and
1950
b1961 and
1965
b1950 and
1954
b1960 and
1964
b1946
and
1950
b1961
and
1965
Before entering a union
15 30.17 44.34 29.03 45.92 29.01 42.24
20 19.07 28.58 14.39 27.25 16.46 22.49
25 13.22 17.00 6.55 13.43 9.73 12.09
30 11.43 12.07 4.47 8.75 7.81 8.84
35 10.58 ­ 3.55 ­ 7.12 ­
40 10.24 ­ 3.03 ­ 6.62 ­
45 10.10 ­ ­ ­ 6.42 ­
First marriage
15 52.39 32.96 59.66 35.07 60.44 52.78
20 60.41 43.11 72.01 47.90 70.94 70.58
25 62.91 51.32 78.77 61.09 75.31 78.58
30 61.49 53.32 79.67 68.35 75.12 79.59
35 59.86 ­ 79.72 ­ 74.14 ­
40 57.45 ­ 79.46 ­ 72.86 ­
45 54.55 ­ ­ ­ 70.61 ­
First cohabitation and dehabitation
Canada 1st
cohabitation
The Netherlands
1st cohabitation
Canada 1st
dehabitation
The Netherlands
1st dehabitation
46­50 61­65 50­54 60­64 46­50 61­65 50­54 60­64
15 2.89 9.66 4.19 12.81 2.73 4.73 0.99 5.09
20 3.37 11.23 4.90 16.38 3.23 6.12 0.98 6.91
25 3.57 9.72 4.33 14.45 3.83 7.49 0.90 8.75
30 3.44 7.65 3.77 10.77 4.50 7.85 0.69 9.43
35 3.60 ­ 3.42 ­ 5.03 ­ 0.45 ­
40 3.53 ­ 3.03 ­ 5.69 ­ 0.17 ­
45 3.28 ­ ­ ­ 6.31 ­ ­ ­
MELINDA MILLS16
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF574 age 48). The sum of these transient states is 98.64. From this, we can calculate
575 the expected percentage of time to be spent in the final absorbing state of
576 second marriage (for this model), which is 1.36%.
577 7.1. NEVER IN A UNION
578 The first stage in the partnership biography is the period before a woman
579 enters into a first partnership. The ``postponement hypothesis'' (H1) gains
580 mixed support. Although there are a higher percentage of women in the
581 younger cohort postponing entry into first union in Canada and the Neth-
582 erlands, the difference between younger and older cohorts in the Russian
583 Federation is not as large. Younger women in Canada having the highest
584 percentage (11.6) who have never entered a first partnership compared to
585 9.4% in the Netherlands and 7.7% in Pskov.5
For older cohorts these figures
586 are 6.8% (Canada), 5.3% (Pskov), and somewhat lower at 2.8% in the
587 Netherlands. As Table 1 also illustrates, younger women spent a larger
588 amount of their lives being single before entering a first union, a finding
589 confirmed in other countries (e.g., Toulemon, 1997).
590 Based on the results presented above, the expectation that Canadian
591 women have overall higher levels ``remaining never in a partnership'' (H2) is
592 sustained. In light of these findings, it is interesting to pursue the question of
Table 1. (Continued)
Age Canada The Netherlands Pskov, Russian Fed.
b1946 and
1950
b1961 and
1965
b1950 and
1954
b1960 and
1964
b1946
and
1950
b1961
and
1965
First marital dissolution states
Canada
1st marital
separation
Canada 1st
marital
dissolution
Netherlands
1st marital
dissolution
Pskov
1st marital
separation
Pskov 1st
widowhood
46­50 61­65 46­50 61­65 50­54 60­64 50­54 60­64 46­50 61­65
15 4.41 2.32 6.88 2.72 3.54 1.17 8.34 4.52 2.21 0.41
20 4.89 3.03 8.12 3.63 4.32 1.55 9.99 6.28 2.61 0.57
25 5.64 4.05 9.74 5.16 5.24 2.15 11.79 8.41 3.16 0.79
30 6.27 5.43 11.60 7.04 6.67 2.69 13.26 10.20 3.75 1.36
35 6.45 ­ 13.05 ­ 7.73 ­ 14.24 ­ 4.50 ­
40 6.94 ­ 14.79 ­ 8.75 ­ 14.94 ­ 5.45 ­
45 8.33 ­ 15.91 ­ ­ ­ 15.88 ­ 7.09 ­
*Percentage of lifetime to be spent in each state at exact age x.
THE STRUCTURATION OF PARTNERSHIP HISTORIES 17
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
593 how many women remain without a partner at certain moments in their lives.
594 Table 2 shows the probability that a woman who has never had a partnership
595 at age x will remain without a partner at the later age of x + n.6
This table
596 illustrates that women are increasingly less likely to form first partnerships as
597 they age.
598 The results from Table 2 further confirm that Canadian women show the
599 highest probability of remaining without a partner at any age. The proba-
600 bility that a Canadian woman from the older cohort who has never had a
601 partnership at age 35 will remain without a partner at age 45 is 79%. This is
602 compared to 64% in the Netherlands and a substantially lower figure of 54%
603 in Pskov. The proportion that never enters a union reflects the historical
604 period and marriage market availability (Scherbov and Van Vianen, 2001),
605 but may also be attributed to a changing signification structure that accepts
606 singlehood. As Forsyth and Johnson (1995) contend, the shift from the
607 attitude that those who remain single are deviant or inadequate has
608 increasingly been replaced by an emerging new style of singlehood. Certain
609 women gain identity via singlehood or some may have desired a partner but
610 were unable to find a match. Considering Canadian women's history of
611 relative equality and participation in higher education and the workforce,
612 they may have more allocative and authoritative resources, which affords
613 them with the power to remain single. Although Russian women have also
614 participated in the labour force, their resources have remained comparatively
615 lower, which coupled with housing constraints, and a general norm regarding
616 the importance of marriage has restricted the growth of singles. Another
Table 2. Probability that a woman who has never entered a union at age x will remain never in
a union at age x + n, Canada, The Netherlands, and Pskov, Russian Federation*
Probability of remaining never in a union at age
Country
25 for those never in a
union at age 15
35 for those never in a
union at age 25
45 for those
never in a un-
ion at age 35
b1946
and 1950
b1961
and 1965
b1946
and 1950
b1961
and 1965
b1946
and 1950
Canada 0.2561 0.3129 0.4569 0.3202 0.7923
Netherlands 0.1714 0.2619 0.2343 0.1972 0.6374
Russian
Federation
0.2343 0.1972 0.3239 0.3556 0.5388
Note: *Due to data restrictions, the cohorts for the Netherlands are represented by those born
between 1950 and 1954 and 1960 and 1964. For this reason, the results shown in the third and
fourth column for the Netherlands represents the probability of remaining never in a union at
age 33 (and not 35) and in the last column, those at age 43 (and not 45).
MELINDA MILLS18
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
617 possibility is that these women may have had or continue to have non co-
618 residing or legal relationships (e.g., LAT relationships), which are not reg-
619 istered by the survey data.
620 7.2. FIRST UNION FORMATION
621 Table 3 provides the transition probabilities to first union by type and se-
622 quence by selected single years of age. The estimates are interpreted as fol-
Table 3. Probabilities of transition to first union as marriage and/or cohabitation by various
sequences, women, Canada, The Netherlands and Pskov, Russian Federation, by selected ages
and cohort
Selected
age and cohort
Canada The Netherlands* Pskov,
Russian
Federation
nu1m nu1c 1c1m nu1m nu1c 1c1m nu1m
Numbers and proportions of women ever experiencing the transition
n 46­50 356 57 29 578 245 151 1285
(%) (78.6) (8.1) (50.9) (68.1) (28.9) (61.6) (94.7)
61­65 328 307 125 311 563 321 1310
(45.6) (42.7) (40.7) (32.2) (58.3) (57.0) (91.6)
Conditional probability of experiencing the transition before the next birthday
20 46­50 0.1835 0.0092 0.0000 0.1751 0.0400 0.2680 0.1753
61­65 0.0632 0.0556 0.1304 0.0521 0.1047 0.2177 0.2342
25 46­50 0.0948 0.0345 0.0556 0.0538 0.0847 0.2616 0.2013
61­65 0.0889 0.0622 0.0761 0.0850 0.1222 0.1498 0.1844
30 46­50 0.0290 0.0290 0.0000 0.0206 0.1341 0.1457 0.1088
61­65 0.0521 0.0313 0.0787 0.0390 0.1159 0.2488 0.0892
35 46­50 0.0377 0.0000 0.0000 0.0422 0.1216 0.1342 0.0485
61­65 ­ ­ ­ ­ ­ ­ ­
40 46­50 0.0222 0.0000 0.0000 0.0000 0.0000 0.0000 0.0366
61­65 ­ ­ ­ ­ ­ ­ ­
45 46­50 0.0000 0.0274 0.0000 ­ ­ ­ 0.0317
61­65 ­ ­ ­ ­ ­ ­ ­
Notes: *Cohorts for the Netherlands are b1950­54 and b1960­64. nu1m = never in a union to
first marriage. nu1c = never in a union to first cohabitation. 1c1m = first cohabitation to
first marriage. The categories `nu1c' and `1c1m' are not mutually exclusive, `nu1c' contains
both those who may convert cohabitation to marriage, dissolve cohabitation or are censored
by the interview date (i.e., remain cohabiting).
The sample sizes (N) and number of censored cases that remained never in a union (nu) for
each cohort are as follows: Canada, 46­50, N = 453 (nu = 31); 61­65, N = 719 (nu = 84);
The Netherlands, 50­54, N = 849 (nu = 24); 60­64, N = 965 (nu = 91); Pskov, Russian
Federation: 46­50, N = 1357 (72); 61­65, N = 1430 (110).
THE STRUCTURATION OF PARTNERSHIP HISTORIES 19
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
Figure2.Lifetablepercentageinunionstate(s)atexactagex,women,bycountryandcohort:(a)firstunionstate,oldest
cohort,(b)firstunionstate,youngestcohort,(c)firstmaritaldissolutionstate,oldestcohort,(d)firstmaritaldissolution
state,youngestcohort,(e)secondunionstate,oldestcohort,(f)secondunionstate,youngestcohort.Note:Scaleof
expectedpercentagevariesaccordingtounionstage.
MELINDA MILLS20
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
623 lows. For example, provided that women had not entered a marriage by the
624 age of 30, the probability of entering a first marriage between age 30 and 31
625 for Dutch women born between 1950 and 1954 is 0.0206. We can interpret
626 this statistic in another way: A typical unmarried, 30 year-old Dutch woman
627 in the older cohort had a 2.9% chance of marrying between her 30th and 31st
628 birthday. Figure 2a and b depict how the timing of entry into first marriage
629 has shifted between older (2a) and younger (2b) cohorts in addition to be-
630 tween-country differences.
631 The expectation that the institution of marriage remains strong and stable
632 in the Russian Federation (H3) is supported with one unexpected twist. Levels
633 of cohabitation remain low and Russian women have a higher probability of
634 entering first marriage throughout their lifetime than their Dutch or Canadian
635 counterparts. A striking finding is that the younger cohort of Russian women
636 actually enters marriage at a younger age than the older cohort. This is
637 demonstrated in the last column of Table 3 that shows a higher probability of
638 entry into first marriage by age 20 for the younger cohort. In fact, by calcu-
639 lating the mean timing of first marriage, we find that the younger cohort
640 marries at 23.6 years, compared to the mean age of 26.2 for the older cohort.
641 DaVanzo and Adamson (1997, p. 2) also report that between 1960 and 1995
642 the average age of marriage for women in Russia fell by 4.2 years, from 26.2 to
643 22.0. This suggests that the attachment to marriage became even more en-
644 hanced over time. However, it is important to note that the younger cohort
645 (born 1961 and 1965) entered the partnership market in the late 1970s and
646 early 1980s. More recent data show that there has been a decline in marriage
647 and fertility in younger cohorts (e.g., Roberts et al., 2003).
648 As Table 3 demonstrates, younger Dutch and Canadian cohorts increas-
649 ingly choose cohabitation as a first union over marriage, a trend which started
650 earlier in the Netherlands, providing support for the ``type of union''
651 hypothesis (H4). Figure 2a and b likewise illustrate this shift to cohabitation
652 for younger cohorts. The expected percentage of time spent within first
653 cohabitation is shown in Table 1 under the heading ``First cohabitation and
654 dehabitation''. The younger Canadian, and particularly Dutch cohort, spends
655 a substantially longer period in the first cohabitation phase than the older
656 cohort. We see a remarkable increase in the young Dutch cohort that spent
657 14.5% of their lives (up to age 35) cohabiting compared to their mother's
658 generation that spent only 4.3% (up to a somewhat higher age of 43). This
659 concurs with previous studies such as Manting (1994), who attributed a de-
660 cline in Dutch marriages in the 1970s and early 1980s to the fact that marriage
661 is delayed by cohabitation. Cohabitation has become an integral early stage in
662 many partnership biographies. Further support from Table 3 demonstrates
663 that almost half (42.7%) of the youngest cohort in Canada and even more
664 (58.3%) in the Netherlands had a cohabiting relationship for the first time,
665 compared to only 8.1% and 28.9%, respectively, of their older counterparts.
THE STRUCTURATION OF PARTNERSHIP HISTORIES 21
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
666 However, cohabitation is not merely a phenomenon of the young, but also
667 a product of the historical period, a finding that challenges the type of union
668 hypothesis (H4). Cohabitation appears to be a growing choice for older
669 cohorts, particularly after marital dissolution. Recall that the model for
670 Canadian women allows entry into first cohabitation from a first marital
671 dissolution state, provided there was no previous consensual union (see
672 Figure 1b). Figure 3 shows the percentage of women from the older Cana-
673 dian cohort who entered a first consensual partnership, illustrating that there
674 are clearly two waves and two groups of women. This concurs with Tou-
675 lemon's (1997) recent study of cohabitation in France, which notes: ``the
676 probability of beginning a union outside of marriage increases with age,
677 because of the ever-growing population of cohabitation from year to year.''
678 It likewise corresponds with Lesthaeghe and Moors (2000), who argued that
679 post-marital cohabitation has begun to replace marriage.
680 The ``nature of cohabitation'' hypothesis (H5) gains mixed support. Al-
681 though younger cohorts in both Canada and the Netherlands progressively
682 opt for cohabitation, it appears to take the form of an ``alternative'' to
683 marriage to a stronger degree in Canada, as many Dutch cohabiting unions
684 are eventually transferred into marital ones. However, as discussed shortly
685 (H9), there is also the function of an early ``weeding out'' of bad matches,
686 which is higher in the Netherlands. As the percentages in Table 3 show, the
687 probability of transition from first cohabitation to first marriage is higher in
688 the Netherlands, with 61.6 and 57.0% of older and younger Dutch women
689 turning their cohabiting unions into a marriage compared to 50.9 and 40.7%
15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
expectedpercentage
age
Canada,b46-50, 1Coh
Figure 3. Two waves of entering into a cohabiting union for the first time, women,
Canada, cohort b1946­50.
MELINDA MILLS22
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
690 in Canada. There is thus a 5% point decrease between the two cohorts in the
691 intensity to marry after cohabitation in the Netherlands, which is twice the
692 amount in Canada (10% point decrease). The propensity to enter a marital
693 union is declining in both countries. An essential point, however, is that
694 although there is considerable change in the type and nature of partnership
695 formation, there is still great stability as the majority of individuals still enter
696 partnerships.
697 7.3. FIRST UNION DISSOLUTION
698 Figure 2c and d and Table 4 demonstrate that the ``structural support of
699 divorce'' hypothesis (H6) sustains the empirical test. With the exception of
Table 4. Probabilities of transition to first union dissolution by type, women, Canada, The
Netherlands and Pskov, Russian Federation, by selected ages and cohort
Selected age
and cohort
Canada The
Netherlands*
Pskov, Russian
Federation
1c-
1dehab
1msep 1mdis 1c-
1dehab
1m-
1mdis
1m1div
sep
1m1wid
Numbers and proportions of women ever experiencing the transition
n 46­50 41 151 118 89 128 380 114
(%) (71.9) (42.4) (33.2) (36.3) (22.2) (29.6) (8.9)
61­65 143 119 78 139 59 257 30
(46.6) (36.8) (23.8) (24.7) (19.0) (19.6) (2.3)
Conditional probability of experiencing the transition before the next birthday
20 46­50 0.1429 0.0085 0.5000 0.1271 0.0193 0.0193 0.0055
61­65 0.1087 0.0349 0.6667 0.0912 0.0000 0.0375 0.0000
25 46­50 0.1111 0.0205 0.1538 0.1327 0.0042 0.0161 0.0064
61­65 0.0870 0.0132 0.3684 0.0852 0.0185 0.0257 0.0010
30 46­50 0.1875 0.0293 0.1250 0.0965 0.0200 0.0200 0.0030
61­65 0.1102 0.0394 0.3500 0.0569 0.0110 0.0169 0.0027
35 46­50 0.0000 0.0177 0.2083 0.0914 0.0147 0.0140 0.0075
61­65 ­ ­ ­ ­ ­ ­ ­
40 46­50 0.0000 0.0230 0.3333 0.0000 0.0131 0.0126 0.0034
61­65 ­ ­ ­ ­ ­ ­ ­
45 46­50 0.1111 0.0238 0.1455 ­ ­ 0.0304 0.0076
61­65 ­ ­ ­ ­ ­ ­ ­
Notes: *Cohorts for the Netherlands are b1950­1954 and b1960­1964. 1c-1dehab = first
cohabitation to first dehabitation, 1msep = first marriage to first marital separation, 1msep-
div = first marital separation to first divorce, 1m1mdis = first marriage to first marital
dissolution, 1m1wid = first marriage to first widowhood. For the sample sizes of each cohort
in the analysis, refer to notes in Table 3.
THE STRUCTURATION OF PARTNERSHIP HISTORIES 23
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
700 women over the age of 41 in the older cohort in Canada, Russian women
701 have the highest probability of divorce for both cohorts. This confirms the
702 assumption that a strong legal and social legitimation of divorce persists in
703 Russian society. The last section of Table 1 confirms that women at age 45 in
704 the older Canadian and Russian cohorts can expect to spend almost 16% of
705 their lifetime in the divorced state. This is double the time that a Dutch
706 woman in this category would spend (8.3%), which may be attributed to
707 faster rates of remarriage after marital dissolution (see Uunk, 1999).
708 The sizeable increase in divorce for Canadian women from the older co-
709 hort after the age of 41 would have occurred roughly between 1987 and 1991.
710 This is thus likely attributed to the shift in divorce laws in 1985, higher social
711 acceptability, and increased resources and ability to leave a relationship.
712 Marital dissolution in Canada has also been institutionalised into a two-stage
713 process, which concurs with the ``separation stage'' hypothesis (H7). This is
714 not to suggest that separation prior to divorce does not take place in other
715 countries, rather that it is formally institutionalised in the social structure.7
716 The last column of Table 4 and Figure 2c and d display the extraordinarily
717 high proportion of Russian widows, particularly for the older cohorts, thereby
718 confirming the ``widow'' hypothesis (H8). As anticipated, older cohorts were
719 more likely to become widows, even in the younger ages from 20 to 30 where
720 the two cohorts can be compared directly (see Table 4). Logically, the per-
721 centage of widows increases with age. It remains difficult to relate widowhood
722 directly to the framework of structuration theory. While the theory is capable
723 of embracing changes in individual power/resources, formal and legal rules or
724 legislation, cultural frameworks, and values and norms among other things, it
725 appears to exclude exogenous factors. Increases in widowhood are related to
726 male mortality, which arguably has a connection to structural change, yet it
727 clearly lacks any element of choice on behalf of the widow.
728 The ``dehabitation'' hypothesis (H9) gains mixed support. The probability
729 of dehabitation compared to first marital dissolution is higher for both
730 Canada and the Netherlands. Yet older cohorts actually have a higher
731 probability of dissolving first consensual unions, and not younger cohorts as
732 anticipated. This is likely related to the fact that early cohabitors were
733 ``innovators'' and thus engaged in more ``deviant'' types of behaviour cou-
734 pled with added external social and family pressure on the relationship. The
735 cohabiting unions that women entered in the 1970s were apparently more
736 fragile (Toulemon, 1997). Placing this in the structuration framework, this
737 appears to be evident of a tangible transition period during structural change
738 where innovators were effectively ``punished'' or sanctioned by their inno-
739 vative behaviour. Cohabitation for younger birth cohorts is more resilient,
740 which can be attributed to increased acceptability, enabling factors within the
741 social structure (e.g., formal legitimisation demanded by older cohorts),
742 which in turn places less constraints and stress on the relationship. Con-
MELINDA MILLS24
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
743 sensual unions of the past were thus vastly different than contemporary
744 cohabitation, both in the context of the partnership and in the type of person
745 who engaged in that behaviour (Manting, 1996).
746 Figure 4a and b contrast dehabitation and marital dissolution patterns for
747 the oldest and youngest cohorts in Canada and the Netherlands. Dehabita-
748 tion appears to be on the rise in Canada, but is increasingly evident among
749 the younger Dutch cohort. Again, this disputes another aspect of the de-
750 habitation hypothesis that dehabitation would be lower in the Netherlands.
751 This lends support for the ``weeding out'' function of cohabitation. Couples
752 may use a period of living together to learn what a marriage would be like,
753 with those who turn their unions into a marriage considered as the best
754 matched unions. Others have argued that cohabitors are a select group of
755 individuals that are less committed to marriage and relationships in general.
756 Bennett et al. (1988, p. 128) maintain that the relationships of those who
757 cohabit are ``characterised by a lack of commitment and stability.'' In other
758 words, cohabitors attach much less importance to traditional institutions and
759 are less influenced by the social structure. Cohabitation also represents a
760 more flexible union of personal choice, has less formal constraints, and does
761 not require formal legal approval (Villeneuve-Gokalp, 1990). Higher disso-
762 lution rates in the Netherlands may also be attributed to the earlier move
763 toward cohabitation as a legitimate type of union. Another theory is that
764 considering the relatively high stigma of divorce in this country, individuals
765 may have a stronger inclination to leave a relationship that may not result in
766 a marriage or that may potentially end in a divorce.
767 Another element of the ``dehabitation'' hypothesis was that cohabiting
768 relationships would be of a shorter duration, which demands a Semi-Markov
769 approach. A selection of these results is shown in Table 5 and Figure 5.
770 Figure 5 shows the survival (or duration stay) probabilities for all non-
771 absorbing partnership states and offers a useful visual depiction of the
772 duration that women remain in the different phases of their entire partnership
773 biography. Table 5 shows the cumulative probabilities of first partnership
774 dissolution by the type of first union and duration of the union by selected
775 years from 0 to 20.8
776 Since there is evidence of an early selection or weeding out process for
777 women who cohabit in comparison to marriage, the last aspect of the ``de-
778 habitation'' hypothesis is confirmed. Referring to Table 5, the probability of
779 first dehabitation during the first five years for the younger Dutch cohort was
780 0.2239 in comparison to 0.0830 for first marital dissolution. However, as
781 Figure 5 illustrates, after initial selection, dissolution rates remain relatively
782 constant with many cohabiting unions remaining intact. This supports Brines
783 and Joyner (1999), who recently argued that we should focus on cohesion and
784 what unites cohabiting partners over time rather than persistently linking
785 pre-marital cohabitation to higher marital instability.
THE STRUCTURATION OF PARTNERSHIP HISTORIES 25
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
786 7.4. RE-PARTNERING
787 As expected in the ``marital re-partnering'' hypothesis (H10), Table 6 and
788 Figure 2e and f exhibit that levels of remarriage for Russian women stand
789 apart. Just as divorce appears to carry little stigma and is largely enabled by
790 the social structure, so too is remarriage, which appears to be viable, even in
(a)
(b)
0
2
4
6
8
10
12
14
16
15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49
age
expectedpercentage
Netherlands,b50-54, 1MarDiss Netherlands,b50-54 1CohDiss
Canada,b46-50, 1MarDiss Canada,b46-50, 1CohDiss
First Marital Dissolution, Canada
First Marital
Dissolution,
Netherlands
First Dehabitation,
Canada
First Dehabitation,
Netherlands
0
2
4
6
8
10
12
14
16
15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49
age
expectedpercentage
Netherlands,b60-64, 1MarDiss Netherlands,b60-64, 1CohDiss
Canada,b61-65,1MarDiss Canada, b61-65,1CohDiss
First Marital Dissolution, Canada
First Marital Dissolution, Netherlands
First Dehabitation, Canada
First Dehabitation, Netherlands
Figure 4. Life table percentage in all marital dissolution and dehabitation states at exact
age x, women, Canada and the Netherlands, by cohort: (a) oldest cohort and (b)
youngest cohort.
MELINDA MILLS26
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
Table5.Cumulativeprobabilitiesoffirstuniondissolutionbytypeandduration,women,Canada,TheNetherlandsandPskov,Russian
Federation,bycohort
Durationinyears
Unionstate
Cohortborn1946­1950Cohortborn1961­1965
n05101520n05101520
Canada
1c-1dehab390.01750.25340.30250.37440.37441430.06200.37560.47340.4930­
1m-1sep1510.01020.10460.19200.27880.33021190.01030.16180.27510.4036­
1m1sep-1mdis1180.28860.77490.83340.85190.8519780.32140.78140.84700.8470­
TheNetherlands*
1c-1dehab860.08200.27030.31360.33680.33681380.05270.22390.26970.2697­
1m-1mdis1280.00340.05080.10230.15540.1959590.00000.08300.13740.1846­
Pskov,RussianFederation
1m-1divsep3800.01010.12730.20010.24990.28062570.01070.13230.22250.2985­
1m-1wid1140.00390.01870.03290.05060.0676300.00080.01070.02630.0963­
Notes:*
BirthcohortsshownfortheNetherlandsareb1950­1954andb1960­1964.Forcohortborn1960­1964inNetherlands,Duration`15'is
`13'years.InCanadaandtheNetherlandsalmostallmaritaldissolutionsareduetodivorceandnotwidowhood.1c-1dehab=firstcohabitation
tofirstdehabitation,1m-1sep=firstmarriagetofirstseparation,1m-1mdis=firstmarriagetofirstmarriagedissolution,1m-1divsep=first
marriagetofirstdivorce/separation,1m-1wid=firstmarriagetofirstwidowhood.
THE STRUCTURATION OF PARTNERSHIP HISTORIES 27
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
Figure5.Semi-Markovestimatesofsurvival(duration-stay)probabilitiesfornon-absorbingstatesbybirthcohortand
country,women:(a)b1946­50,Canada,(b)b1961­65,Canada,(c)b1950­54,Netherlands,(d)b1960­64,Netherlands,(e)
b1946­50,Pskov,RussianFederationand(f)b1961­65,Pskov,RussianFederation.
MELINDA MILLS28
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
791 the face of high male mortality. Re-partnering may also reflect a need to
792 consolidate resources and is related to additional factors such as housing.
793 Further confirmation of H10 appears in Table 6, where we see that the
794 probability to remarry is higher for divorcees across all durations. The
795 probability that Russian women enter a second marriage after divorce/sep-
796 aration within the first year (duration 0) is 0.1303 compared to 0.0457 for
797 widowed women in the oldest cohort. Remarriage prospects differ greatly
798 among the divorced and the widowed, likely related to factors such as the age
799 of the respondent and personal ``marriageability'' characteristics that differ
800 for divorced and widowed women. In addition, the longer a woman spends in
801 the ``single'' state after first marriage dissolution, the lower the rate of
802 transition to a second marriage. This confirms that, methodologically
803 speaking, both duration and origin state is relevant. Marriage also remains as
804 a strong institution in the Netherlands, demonstrated by the high percentage
805 of women who remarry in the older Dutch cohort (60.9%).9
806 The category of ``second cohabitation'' is rarely included in previous
807 multistate partnership life tables. As Figure 2f and Table 6 illustrate, second
808 cohabitation is the overwhelming choice of higher-order relationships for the
Table 6. Cumulative probabilities of second marriage and second cohabitation by type and
duration, women, Canada, The Netherlands and Pskov, Russian Federation, by cohort
Duration in years
Union
States
Cohort born 1946­1950 Cohort born 1961­1965
n 0 5 10 n 0 5 10
Canada
1c-2mz 18 0.0349 0.1579 0.1678 ­ ­ ­ ­
1mdis-2m 27 0.0756 0.1433 0.1708 14 0.0706 0.2003 0.2003
1dehab-2c 14 0.0213 0.2918 0.3272 72 0.1453 0.5424 0.6213
The Netherlands*
1dehab-2m 2** 0.0099 0.0202 0.0202 4** 0.0000 0.0357 0.0357
1mdis-2m 78 0.0316 0.3124 0.4023 17 0.0530 0.3639 0.4235
Pskov, Russian Federation
1sepdiv-2m 212 0.1303 0.3951 0.5377 129 0.1844 0.5828 0.7384
1wid-2m 43 0.0457 0.3452 0.4753 15 0.1091 0.7697 0.8849
Notes: *Birth cohorts shown for the Netherlands are b1950­1954 and b1960­1964. **
Small
numbers should be judged with caution. 1c-2m = first cohabitation to second marriage (z -
Canadian cohort b1946­1950 only), 1mdis-1m=first divorce/widowhood to first marriage,
1dehab-2c=first dehabitation to second cohabitation, 1dehab-2m = first cohabitation
dissolution to second marriage, 1mdis-2m = first marriage dissolution to second marriage,
1sepdiv-2m = first separation/divorce to second marriage, 1wid-2m = first widowhood to
second marriage.
THE STRUCTURATION OF PARTNERSHIP HISTORIES 29
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
809 younger Canadian cohort. This confirms the ``cohabitation re-partnering''
810 hypothesis (H11). This was not the case in the Netherlands, where numbers
811 for higher-order cohabitation were too small for inclusion in the analysis.
812 Younger cohorts also entered into second relationships at a much faster pace.
813 Referring to Table 6, we see that the probability of re-partnering in the early
814 phases (0, 5 years) is uniformly higher for the younger cohorts. This is par-
815 ticularly the case in Canada. Once again, this likely reflects a transformation
816 in the social norms surrounding re-partnering and marriage after a divorce,
817 softening of legal restrictions related to dissolution, coupled with an increase
818 in women's power and monetary resources.
819 The final ``complexity'' hypothesis (H12) anticipated that relationships
820 would remain relatively stable in the Russian Federation. Complexity would
821 thus arise among younger cohorts and particularly Dutch and Canadian
822 women. As Figure 1b and c confirm, there has been a pluralisation of rela-
823 tionships, represented by more partnership states and stages such as the
824 complex six-state, but principally the eight-state model for Canada that was
825 necessary to capture the majority of transitions. The analysis also found an
826 increase in multiple relationships among younger cohorts, particularly mul-
827 tiple consensual unions. In fact, 50% of young Canadian women who
828 experienced their first dehabitation go on to a second cohabiting union,
829 compared to 34% of younger Dutch women.
830 8. Conclusion and Consequences
831 This study offers a more complex description of partnership processes in
832 three different countries via the implementation of classic multistate life ta-
833 bles and the application of structuration theory. Giddens' (1984) structur-
834 ation theory worked as an encompassing framework to interpret how
835 partnership behaviour is enabled or constrained across various contexts. It
836 allowed the operationalisation of the social structure into three domains of:
837 domination (economic and power resources), signification (cultural, mental
838 frameworks), and legitimation (informal moral and formal legal regulations,
839 rules, values, and sanctions). It likewise takes us beyond a static theory of
840 stability to embrace the mechanisms of change via the duality of structure,
841 enabling us to recognise how new types of demographic behaviour emerge via
842 individual action and interaction to transform the existing social structure.
843 This study illustrated the spectrum of partnership behaviour across the
844 Russian Federation, Canada, and the Netherlands. Twelve research
845 hypotheses confronted our expectations about the partnership biographies of
846 women. Results confirmed that the younger cohort of Dutch and Canadian
847 women postpone union formation, while Russian counterparts do the
848 opposite. As discussed previously, more recent Russian data suggest that this
MELINDA MILLS30
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
849 drop in the average age of first marriage for younger cohorts reversed into a
850 rise during the 1990s. Canadian women have the highest probability of
851 remaining without a partner at any age, which hints to emerging acceptance
852 of singlehood. Patterns of first union formation in the Russian Federation
853 remain relatively stable, with a steadfast attachment to marriage. Younger
854 Canadian and particularly Dutch cohorts choose cohabitation as a first un-
855 ion, yet the nature of cohabitation appears to differ in each context.
856 Cohabitation appears to increasingly take the function of an alternative to
857 marriage in Canada (see also Wu, 2000). In contrast, in the Netherlands,
858 longer-term cohabiting unions are not as common as in Canada with these
859 partnerships having a higher likelihood of being transformed into a marriage,
860 suggesting that it serves more as a trial marriage function. However, there is
861 also larger proportion of women whose cohabiting unions are ``weeded out''
862 at an early stage in the Netherlands. The examination of first union disso-
863 lution confirmed that when divorce is ``enabled'' by the social structure, levels
864 are higher, particularly in the Russian Federation and after the age of 40 in
865 the older Canadian cohort. In comparison with the other two study coun-
866 tries, the Russian Federation has an extraordinarily high number of widows,
867 particularly in the older cohort. The growing form of union dissolution in
868 Canada and the Netherlands is dehabitation, which was higher and occurred
869 earlier than marital dissolution, particularly among older cohorts and those
870 in the Netherlands.
871 Finally, re-marriage is strikingly higher in the Russian Federation, with
872 cohabitation gaining ground in Canada after a first marital or cohabiting
873 dissolution. We can conclude that marriage and high levels of divorce and
874 widowhood have remained relatively stable in the Russian Federation. Con-
875 versely, partnership histories have become increasingly complex and plura-
876 lised in the Netherlands and to an even greater extent in Canada. Yet in all
877 countries, individuals still virtually universally form partnerships; it is merely
878 the type of union and timing of partnership formation that has altered. This
879 detailed analysis demonstrates that it is essential to not only look for change,
880 but also search for stability in partnership histories in modern societies.
881 The results presented here provide many answers, but also raise questions.
882 Due to the scope of the analysis of entire partnership histories in three
883 countries, it examined only inter-cohort, cross-country, and duration-stay
884 differences. It was impossible to empirically pursue further aspects of het-
885 erogeneity within the confines of one paper. A complement to this study
886 would be an examination of the impact of additional characteristics on
887 partnership patterns (e.g., education, labour force participation). The mul-
888 tistate method would not be effective, as it would likely produce erratic
889 estimates due to disaggregation of data. Rather, regression techniques would
890 be more amenable and efficient. This would empirically answer the more
891 substantive questions regarding why change and stability has occurred.
THE STRUCTURATION OF PARTNERSHIP HISTORIES 31
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
892 The consequence of these findings raises some intriguing issues. The
893 postponement of unions in Canada and the Netherlands and earlier age of
894 union formation in the Russian Federation is certainly related to fertility
895 patterns in these countries. The timing of fertility has clear implications for the
896 total number of children. Whereas the peak age of childbirth in many Western
897 countries is between 25 and 29, it occurred in the early 1920s for the younger
898 Russian cohort observed in this study. Fertility levels of young Russian wo-
899 men between the ages of 15­19 and 25­29 actually exceeded those of the over
900 30 group (DaVanzo and Adamson, 1997). However, as noted previously,
901 Russian union formation and fertility patterns have now dropped to much
902 later and lower levels for younger cohorts in the 1990s and early 2000.
903 Oppenheimer (1988) proposed that cohabiting unions are a main mech-
904 anism in the postponement of marriages. This study shows that not only is
905 this the case, but that the increased complexity of partnerships via multiple
906 relationships and the dissolution of first unions may also serve as an addi-
907 tional factor to postpone entry into marriage or other long-term stable
908 relationships. The sheer amount of union disruptions, such as high levels of
909 divorces in Canada and the Russian Federation and higher rates of dehab-
910 itation in Canada and the Netherlands, raise questions about how these
911 turbulent life changes impact individuals' lives. The striking number of
912 widows, even in the younger cohort of women in the Russian Federation
913 likewise begs the question of how individuals cope with these radical frac-
914 tures in their everyday lives. The turbulent union formation and dissolution
915 patterns observed in this study have far reaching consequences not only for
916 the individuals involved, but also for the children involved and the society as
917 a whole. The consequences of these findings and deeper analysis into varia-
918 tion within these populations would be a future stage of research. This study
919 erects the foundations to isolate which aspects are useful to pursue.
920 Acknowledgements
921 The author wishes to thank Frans Willekens in particular, but also Fernando
922 Rajulton, the anonymous reviewers and the Advisory Group of the FFS
923 programme of comparative research for its permission, granted under iden-
924 tification number 7, to use the FFS data on which this study is partially based.
925 Notes
926 1
An in-depth study of interpretative cultural frames (i.e., signification structure) in each of
927 the countries surpasses the scope of this study and is more appropriate for qualitative eth-
928 nographic research (see for e.g., Hutter, 1994).
929 2
It is, however, difficult to know how these resources are distributed within the household.
930 3
``Dehabitation'' is a concise term for the dissolution of cohabiting unions first used by
931 (Nelissen, 1992 in Prinz, 1995).
MELINDA MILLS32
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
932 4
In addition to small numbers, there was also a lack of information on the timing of
933 cohabiting relationships; they are therefore not included in the model. The seemingly non-
934 discrete states of separation and divorce are collapsed into one category due to small numbers,
935 inconsistent reporting and often simultaneity of timing.
936 5
The `remaining never in a partnership' statistic is calculated by dividing the number of
937 censored cases in the ``never in a union'' (nu) category by the total cohort sample size. Both of
938 these figures are listed in the notes of Table 3. For example, for the youngest cohort in Pskov
939 this figure is calculated by: 100  (110/1430)  7.7%. This statistic represents those never in a
940 union until the survey date, therefore this overall figure still has the potential to decrease over
941 time.
942 6
This is calculated by dividing the life table number of survivors at exact age x + n by the
943 life table number of survivors at exact age x.
944 7
The fact that the timing of separation is not collected in the other countries is a reflection of
945 less formal importance attributed to this stage.
946 8
The high probability of transition from first separation to first marital dissolution for
947 Canadian women in Table 5 is predictable due to the fact that most marital separations end in
948 divorce.
949 9
One aspect that is important to note when examining the re-partnering results for the
950 Netherlands, and particularly Canada, is that the women may enter a second marriage from
951 two different origin states. Thus, to obtain the entire picture of remarriage one should combine
952 these transitions for interpretation.
953 References
954 ARPL, 2000. Annual Review of Population Law Database. Harvard Law School: UN Popu-
955 lation Fund <<http://www.law.harvard.edu/Programs/>>
956 Avdeev, A. and Monnier, A., 1994. `La nuptialité russe, une complexité mé connue', Popula-
957 tion 54(4­5): 635­676.
958 Axinn, W. G. and Thornton, A., 1992. `The Relationship between cohabitation and divorce:
959 Selectivity or causal influence?', Demography 29: 357­374.
960 Baker, M., 1995. Canadian Family Policies: Cross-National Comparisons. Toronto: University
961 of Toronto Press.
962 Becker, G. S., 1981. A Treatise on the Family. Cambridge: Harvard University Press.
963 Bennett, N. G., Blanc, A. K. and Bloom, D. E., 1988. `Commitment and the modern union:
964 Assessing the link between premarital cohabitation and subsequent marital stability',
965 American Sociological Review 53: 127­138.
966 Belanger, A. and Turcotte, P., 1999. `L'influence des caracté ristiques sociodé mographiques sur
967 le dé but de la vie conjugale des Qué bé coises', Cahier Québécois de Démographie 28(1­2):
968 173­197.
969 Brines, J. and Joyner, K., 1999. `The ties that bind: Principles of cohesion in cohabitation and
970 marriage', American Sociological Review 64: 333­355.
971 Butz, W. P. and Ward, M. P., 1979. `The emergence of countercyclical US fertility', American
972 Economic Review 69(3): 318­328.
973 DaVanzo, J. and Adamson, D., 1997. `Russia's Demographic `Crisis' How Real Is It?' RAND
974 Issue Paper. Center for Russian and Eurasian Studies. Labor and Population Program.
975 Easterlin, R., 1976. `The conflict between aspirations and resources', Population and Devel-
976 opment Review 2(3): 417­425.
977 Forsyth, C. J. and Johnson, E. L., 1995. `A sociological view of the never married', Interna-
978 tional Journal of Sociology of the Family 25(2): 91­104.
THE STRUCTURATION OF PARTNERSHIP HISTORIES 33
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
979 Gauthier, A. H., 1996. The State and the Family. Oxford: Clarendon Press.
980 Giddens, A., 1984. The Constitution of Society: Outline of the Theory of Structuration. Cam-
981 bridge: Polity Press.
982 Goskomstat of Russia, 1994. The Demographic Yearbook of the Russian Federation, 1993.
983 Goskomstat of Russia, Moscow.
984 Hannan, M. T., 1984. `Multistate demography and event history analysis', in A. Diekmann
985 and P. Mitter (eds), Stochastic Modelling of Social Processes. London,: Academic Press,
986 Inc. 39­88.
987 Heiner, R. A., 1983. `The origin of predictable behaviour', The American Economic Review 73:
988 560­595.
989 Hoem, J. M. 1972. `Inhomogeneous semi-Markov processes, select actuarial tables, and
990 duration dependence in demography', in T. N. Greville (ed), Longitudinal Analysis of
991 Labour Market Data. Cambridge: Cambridge University Press, 251­296.
992 Hutter, I., 1994. Being Pregnant in Rural South India. Thela Thesis, Amsterdam.
993 Imbrogno, S. and Imbrogno, N. I., 1989. `Soviet women and the autonomous family',
994 International Journal of Sociology of the Family 19: 1­20.
995 Janssen, J. P. G. et al., 1998. `The instability of marriages and cohabiting relationships in the
996 Netherlands' [in Dutch] Mens en Maatschappij 73(1): 4­26.
997 Jonker, J. M. L., 1990. Under the Spell of the Family: Some Considerations over the Meaning of
998 Primary Networks as a Safety Net [in Dutch]. Catholic University of Nijmegen, Nijmegen.
999 Latten, J. and De Graaf, A., 1997. Fertility and Family Surveys in countries of the ECE region.
1000 Standard Country Report. The Netherlands. UN/ECE, Economic Studies No. 10c. United
1001 Nations, Geneva.
1002 Le Bourdais, C. and Marcil-Gratton, N., 1996. `Family transformations across the Canadian/
1003 American border: When the laggard becomes the leader', Journal of Comparative Family
1004 Studies 27(3): 415­436.
1005 Lesthaeghe, R., 1995. `The second demographic transition in Western countries: An inter-
1006 pretation', in K. O. Mason and A. -M. Jensen (eds), Gender and Family Change in
1007 Industrialized Countries. New York: Oxford University Press 17­62.
1008 Lesthaeghe, R., 1998. `On theory development: Applications to the study of family formation',
1009 Population and Development Review 24(1): 1­14.
1010 Lesthaeghe, R. and Meekers, D., 1986. `Value changes and the dimensions of familism in the
1011 European community', European Journal of Population 2: 225­68.
1012 Lesthaeghe, R. and Moors, G., 2000. `Recent Trends in Fertility and Household Formation in
1013 the Industrialized World', IPD-Working Paper, no 2000:2. Belgium: Vrije Universiteit
1014 Brussels.
1015 Lillard, L. A., Brien, M. J. and Waite, L. J., 1995. `Premarital cohabitation and subsequent
1016 marital dissolution: A matter of self-selection?' Demography 32(3): 437­457.
1017 Manting, D., 1994. Dynamics in Marriage and Cohabitation. Thesis Publishers, Amsterdam.
1018 Manting, D., 1996. `The changing meaning of cohabitation and marriage', European Socio-
1019 logical Review 12: 53­65.
1020 Mills, M., 2000. `Providing space for time: The impact of temporality on life course research',
1021 Time & Society 9(1): 91­127
1022 Ministry of Justice, The Netherlands, 1997. Registered Partnerships [in Dutch] Information
1023 Department, Internal and External Communication Section, The Hague.
1024 Moskoff, W., 1983. `Divorce in the USSR', Journal of Marriage and the Family 45(2): 419­425.
1025 Namboodiri, K. and Suchindran, C. M., 1987. Life Table Techniques and Their Applications.
1026 Orlando: Academic Press.
1027 Namboodiri, K. 1991. Demographic Analysis. A Stochastic Approach. San Diego, Academic
1028 Press.
MELINDA MILLS34
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4
UNCORRECTEDPROOF
1029 O'Connor, J., Orloff, A. S. and Shaver, S., 1999. States, Markets, Families. Cambridge:
1030 Cambridge University Press.
1031 Oppenheimer, V. K., 1988. `A theory of marriage timing', American Journal of Sociology
1032 94(3): 563­591.
1033 Poortman, A. -R. and Kalmijn, M., 2002. `Women's labour market position and divorce in the
1034 Netherlands: Evaluating economic interpretations of the work effect', European Journal of
1035 Population 18(2): 175­2002.
1036 Preston, S. H., 1986. `Changing values and falling birth rates', Population and Development
1037 Review 12: 176­195.
1038 Prinz, C., 1995. Cohabiting, Married or Single. Avebury, Aldershot.
1039 Rajulton, F., 1992. Life History Analysis: Guidelines for Using the Program LIFEHIST.
1040 Discussion Paper 92­5, Canada: University of Western Ontario.
1041 Rindfuss, R. R. and Vandenheuvel, A., 1990. `Cohabitation: A precursor to marriage or an
1042 alternative to being single?' Population and Development Review 16(4): 703­726.
1043 Roberts, K. et al. 2003. `Economic conditions and the family and housing transitions of young
1044 adults in Russia and the Ukraine', Journal of Youth Studies 6(1): 71­88.
1045 Rogers, A., 1975. Introduction to Multiregional Mathematical Demography. New York: Wiley.
1046 Scherbov, S. and Van Vianen, H., 2001. `Marriage and fertility of Russian women born
1047 between 1900 and 1960: A cohort analysis', European Journal of Population 17: 281­294.
1048 Schoen, R., 1988. Modelling Multigroup Populations. New York: Plenum Press.
1049 Statistics Canada, 1997. Public Use Microdata File Documentation and User's Guide, 1995
1050 General Social Survey. Ottawa: Statistics Canada.
1051 Toulemon, L., 1997. `Cohabitation is here to stay', Population 9: 11­46.
1052 Trost, J., 1980. `The concept of the one-parent family', Journal of Comparative Family Studies
1053 11(1): 129­138.
1054 Turcotte, P. and Goldscheider, F., 1998. `Evolution of factors influencing first union forma-
1055 tion in Canada', Canadian Studies in Population 25(2): 145­173.
1056 United Nations, 1997. Demographic Yearbook, 1970­1995. United Nations, New York.
1057 Uunk, W., 1999. `Remarriage in the Netherlands. Sociodemographic determinants of married
1058 and unmarried cohabitation after divorce' [in Dutch] Mens en Maatschappij, 74(2): 99­118.
1059 Villeneuve-Gokalp, C., 1990. `Du mariage aux unions sans papiers: Histoire ré cente des
1060 transformation conjugales', Population, 49(2): 265­297.
1061 Vishnevsky, A. G., 1996. `Family, fertility and demographic dynamics in Russia: Analysis and
1062 forecast', in J. DaVanzo (ed), Russia's Demographic ``Crisis''. Rand Conference Pro-
1063 ceedings, Santa Monica, CA, 1­35.
1064 Volkov, E., 1999. Methodology and Organization of the 1994 Microcensus in Russia. PRC
1065 Working Paper 99­5, University of Groningen, The Netherlands.
1066 Von Frank, A. A., 1979. Family Policy in the USSR since 1944. R&E Research Assoc, Palo
1067 Alto.
1068 Willekens, F. J., 1987. `The marital status life table', in J. Bongaarts, T. K. Burch and K. W.
1069 Wachter (eds), Family Demography: Methods and Their Applications. Clarendon Press,
1070 Oxford, 125­149.
1071 Willekens, F. J.,1991. `Lifetableanalysisof stagingprocesses', inH. A.Becker (ed),LifeHistories
1072 and Generations. ISOR, Volume II. University of Utrecht, The Netherlands, 477­518.
1073 Wu, Z., 2000. Cohabitation: An Alternative Form of Family Living. Toronto: Oxford University
1074 Press.
1075
THE STRUCTURATION OF PARTNERSHIP HISTORIES 35
Journal : EUJP SPS Article No. : 619 Dispatch : 15-4-2004 Pages : 35
PIPS No. : 5274081 h LE h TYPESET
MS Code : EUJP 619 h CP h DISK4 4