ArticlePDF Available

Trust in Intimate Relationships The Increased Importance of Embeddedness for Marriage in the United States


Abstract and Figures

Due to the rise of a market for casual relationships, investing in a serious relationship now requires more trust than it did four decades ago. We develop a theory of trust and embeddedness in intimate relationships. One implication of the theory is that given the increased importance of trust, the effects of embeddedness on marriage chances should have increased. Analyses were performed on data from the Chicago Health and Social Life Survey. Various measures of embeddedness were employed. The hypothesis on the increasing importance of embeddedness for marriage found empirical support in the effects of each of these measures.
Content may be subject to copyright.
Rationality and Society
DOI: 10.1177/1043463106063319
2006; 18; 123 Rationality and Society
Arnout van de Rijt and Vincent Buskens Embeddedness for Marriage in the United States
Trust in Intimate Relationships: The Increased Importance of
The online version of this article can be found at:
Published by:
can be found at:Rationality and Society Additional services and information for Email Alerts: Subscriptions:
SAGE Journals Online and HighWire Press platforms):
(this article cites 57 articles hosted on the Citations
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Arnout van de Rijt and Vincent Buskens
Due to the rise of a market for casual relationships, investing in a
serious relationship now requires more trust than it did four decades
ago. We develop a theory of trust and embeddedness in intimate
relationships. One implication of the theory is that given the
increased importance of trust, the effects of embeddedness on
marriage chances should have increased. Analyses were performed
on data from the Chicago Health and Social Life Survey. Various
measures of embeddedness were employed. The hypothesis on the
increasing importance of embeddedness for marriage found empiri-
cal support in the effects of each of these measures.
KEY WORDS .embeddedness .marriage .social networks .trust
During the last decades of the 20th century, a market for short-term
intimate relationships emerged. This emergence originated in the
sexual revolution, during which technological advancements and
changes in legal and normative systems came to jointly facilitate a
new standard of sexuality (Allyn 2001). The advent of birth-control
technology made it easier for couples to have sex without commit-
ting to a durable relationship, illustrated by the more common
reports of premarital sex (Forste and Tanfer 1996; Treas and
Giesen 2000). Individualization introduced values of independence
and self-exploration that are less compatible with commitment to
a durable relationship (Lesthaeghe and Surkyn 1988; Giddens
1991; Beck and Beck-Gernsheim 1994). And through secularization,
Rationality and Society Copyright &2006 Sage Publications. Vol. 18(2): 123–156. DOI: 10.1177/1043463106063319
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
normative systems that used to enforce premarital abstinence eroded
(Espenshade 1985; Waite et al. 1986; Clarkberg et al. 1995).
The sexual liberalization of the late 20th century has had an
adverse effect. The rise of the market for short-term relationships
has jeopardized the market for long-term relationships. While find-
ing a date has never been easier, finding a future spouse has become
much more challenging. Those who desire more than casual inti-
macy can no longer be certain that the other has similar desires.
Committing to a relationship is now risky. It may thus happen
that a couple, because of a lack of trust, unnecessarily postpones
family formation or even gives up on it altogether.
Following Granovetter (1985), various researchers have empha-
sized the importance of embeddedness – the temporal, network and
institutional environment of a relationship – in overcoming trust
problems in durable relations (Raub and Weesie 1990, 2000; Baum
and Oliver 1992; Portes and Sensenbrenner 1993; Romo and
Schwartz 1995; Uzzi 1996, 1997; Montgomery 1998). At the
dyadic level, relationship investments serve as indications of serious
intentions and make it more costly to leave the relationship. In this
way, trust can increase within a relationship over time. At the net-
work level, shared friends and acquaintances function as sources
of information on the trustworthiness of the partner and as sanc-
tioning potential if trust is violated.
In this article, we develop a theory of embeddedness in intimate
relationships. From this theory, we deduce the hypothesis that the
increased importance of trust in intimate relationships has made
embeddedness a more crucial factor, increasingly distinguishing
those who marry from those who do not or do not yet marry.
Data from the Chicago Health and Social Life Survey (CHSLS)
allow us to test this hypothesis for the period 1950–1997.
Theory and Hypotheses
We present a theory on the importance of embeddedness for explain-
ing marital decision-making. We argue that marriage constitutes a
twofold trust problem in which embeddedness facilitates trust. The
theory is of a level of generality that allows for the accommodation
of many of the explanatory mechanisms proposed earlier in the
marriage literature. Much of the language we will be using stems
from the sociological rational choice tradition, which means that
the reader can expect to find terms such as ‘utility’ and ‘love’ in the
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
same sentence. The choice for such a theoretical framework is not
motivated by a desire to reduce the social processes operating in
the most intimate of human relations to egoistic calculation – to
be sure, sociological rational choice theory does not preclude altru-
ism in any way (Coleman 1992; Heckathorn 2001) – but rather to
underscore our belief that decisions with such potential emotional
repercussions as placing trust in a loved one and committing oneself
to the happiness of a single other cannot reasonably be argued to be
made in pure naivete
Marriage as a Twofold Trust Problem in Intimate Relationships
Before we can develop a theory on the role of trust and embedded-
ness in marriage decisions, we have to define what we mean by trust.
We define a trust situation as a choice situation of two partners that
takes on the form of a trust game (Dasgupta 1988). Figure 1 is a
game-theoretic representation of a trust game (Kreps 1990). The
following analysis of the game shows that the placing of trust by
the trustor is not necessarily rational. In a single encounter, a
rational trustee will abuse trust if it is placed, because the utility
from abusing trust is higher than the utility from honoring trust
(T2>R2). Therefore, a rational trustor will not place trust because
the utility from no trust is higher than the utility from abused trust
(P1>S1). The two actors will as a consequence arrive at the single
equilibrium outcome where they obtain P1and P2. Since placing and
honoring trust would have yielded a higher utility for both actors
(R1>P1and R2>P2), the equilibrium outcome is Pareto-ineffi-
cient. The actors face a social dilemma because individually rational
behavior leads to a collectively sub-optimal result. This formaliza-
tion of a trust problem allows us to define trust as the extent to
which a trustor dares to place trust in the trustee. More trust is
needed if the trustor loses more if trust is abused (i.e., S1is smaller)
and if the trustee has a larger temptation to abuse trust (T2is larger).
In accordance with earlier research, we argue that if two partners
have initiated a traditional long-term relationship, specialization in
household labor by the woman constitutes a trust problem.
If the
woman chooses not to specialize in household labor, both partners
keep their jobs and share household activities. If she decides to
specialize in household labor, she quits her job and makes the
man fully responsible for the provision of income. The man then
has more time to work, being freed from household labor. If the
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
man finds another woman, he can choose to quit the relationship or
continue the relationship with his current partner.
Figure 1 fits this decision-making situation. Many factors make
honored intimate trust preferable to no trust at all, R1>P1and
R2>P2: love, the possibility of having one’s children grow up in
a two-parent family, expectations from parents and family to have
a serious relationship and regular intimacy. Economic factors are
also important. The lower earning capacity of a woman in a tradi-
tional relationship makes trading the income from his hours of
market work for hours of household labor mutually beneficial,
since it results in a higher household income combined with the
same amount of performed household activities. Other economic
factors are the generally better payment of full-time jobs as com-
pared to part-time jobs and needing only one car. As the man further
specializes in earning money, the woman’s value on the labor market
drops. Not participating in the labor market, she becomes less and
less attractive to potential future employers. If the woman were to
separate from her husband after having specialized in household
labor, she would find herself with devalued human capital and pos-
sibly children to take care of as well, being less able to make a living.
She would be worse off than if she had not placed trust (P1>S1).
Moreover, it would be disadvantageous for a woman to divorce
after specializing in household work (Smock et al. (1999) estimate
the negative effects of divorce on women’s economic well-being
and find these to be substantial yet somewhat smaller than generally
assumed; see Rusbult and Martz (1995) for an analysis of the likeli-
hood of married women remaining in abusive relationships). That is
why no action on behalf of the woman is considered after trust has
been placed.
Unlike the woman, the man has specialized in earning
Figure 1. Extensive Form of a Trust Game. R1>P1>S1and T2>R2>P2. Double
Lines Indicate Equilibrium Path of Play
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
money. Thus, if he were confronted with a new love, it could be
rational for him to quit (T2>R2). Anticipating this, a rational
woman would withhold trust and (continue to) participate on the
labor market during the course of the relationship.
The marriage contract solves to some extent the trust problem of
household specialization. The contract specifies the legally enforce-
able payment of alimony/spousal support and child support.
payment decreases the temptation (T2) to abuse trust for the man
as a husband. Also, the abuse of trust is now less costly (S1is
larger) for the woman because she is legally guaranteed financial
support. See England and Farkas (1986), Frank (1988) and Raub
and Weesie (1993) on the uses and limitations of the marriage con-
tract in the context of trust problems, and Feld (2003) for an analysis
of the covenant marriage, which makes divorce very difficult and is
available to couples in some southern states of the United States.
Although a marriage contract can alleviate the trust problem that
concerns the possibility that the financially independent partner
leaves, there is a second trust problem that is not covered by the
marriage contract. If two partners start dating, they are not sure if
their relationship will develop into a serious one. And, even if one
partner is sure about his or her own feelings and would be willing
to commit to a long-term relationship, it remains questionable
whether or not the other is willing to commit as well. One of the part-
ners might unfaithfully be having a relationship with a third person
without the other partner being aware of this. In contrast to the
trust problem of household specialization, the partners face this
second trust problem from the moment of romantic involvement.
If both partners do not intend the relationship to develop into a
serious one, they take any opportunity of infidelity and the relation-
ship will not last long. However, if at least one of them thinks he
or she would like to further invest in the current relationship,
there is a problem. It is hard to tell whether the other wants the
same, but even if so, the other faces the same uncertainty. One
may show good intentions to get the other to share his or her bed,
but cheat when an alternative partner turns up. Before the marriage
decision even comes into question, both partners have to be con-
vinced that they are dealing with someone who intends to form a
long-term relationship.5
We will now investigate what the roles of trust and embeddedness
are in the decision to start building a durable relation with a partner.
Women and men in the market for relationships pursue certain goals
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
with a future partner. Laumann et al. (2004) distinguish between
transactional goals (such as sexual pleasure) that can be pursued in
a short-term relationship and relational goals (such as love and rais-
ing children) that can also be pursued in long-term relationships.
Correspondingly, we distinguish between relational-goal oriented
and transactional-goal oriented partners. If two partners pursue only
transactional goals, neither will invest in a long-term relation. They
are satisfied with the short-term relation. However, if someone is
interested in a long-term relationship, he or she has to choose
whether or not to invest in it. As in the trust game, the problem is
that the worst outcome for a relational goal-oriented partner is to
invest in a long-term relationship with a partner who does not.
Returns on the investment would never come. The best outcome is
reached if a relational goal-oriented person meets another relational
goal-oriented person and they both invest in the long-term relation-
ship. Therefore, the decision to invest depends on three main
1. One’s estimate of the probability that one’s partner pursues
relational goals.
2. One’s own interest in a marriage and potential gains of such
3. Losses as a consequence of investments in a long-term relation-
ship, if the partner eventually turns out not to be interested in
the relationship.
We thus hypothesize that the probability that a person invests in a
long-term relationship increases with both her estimate of the likeli-
hood that her partner also invests and her gains from a marriage and
decreases with the losses she would incur from being deceived.
Role of Embeddedness
We have argued that the marriage contract solves to a considerable
extent the trust problem of household specialization but cannot
solve the preceding trust problem of durable partner selection. How-
ever, there are other possibilities for solving trust problems. Raub
and Weesie (1993, 2000) distinguish three general solutions to
trust problems: dyadic embeddedness,network embeddedness and
institutional embeddedness. Embeddedness refers to the altering of
the structure of the decision-making situation by its environment.
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
The institution of marriage is an example of institutional embedded-
ness. It operates through a reduction in the utility of trust abuse by
attaching a legally enforceable sanction to this action. Apart from
marriage (and the cohabitation contract, common in Western
Europe) no institutional arrangement exists that solves trust prob-
lems in intimate relationships. In the remainder of this section, we
focus on how dyadic and network embeddedness affect behavior
in the particular cases of intimate partner selection and long-term
commitment decisions. Recall that we have defined trust as the
extent to which a trustor dares to place trust in a trustee. By reducing
the utility of trust abuse, embeddedness makes placing trust for
trustors less risky. Thus, there is more trust if there is more embed-
dedness. A trustee can therefore be considered trustworthy even if he
would abuse trust without a doubt in the absence of embeddedness.
Dyadic embeddedness refers to the past and potential future inter-
actions between the same actors. The effect of dyadic embeddedness
operates through two mechanisms: learning and control (for a more
elaborate discussion see Buskens 2002; Buskens and Raub 2002).
Learning refers to the experiences partners have had in the past. If
one has invested in the relationship in the past, one is more likely
to be interested in a long-term relationship and can thus be expected
to do so as well in the future. Similarly, if one’s partner has invested
more in the relationship, one will become more convinced that he
or she is interested in a long-term relationship. Control is related
to the sanctioning ability in the case of opportunism. The more
two partners have invested in their relationship, the more they can
lose if opportunistic behavior leads to the other breaking up. In
this way, sanctioning potential and thus the probability that they
will choose each other as long-term partners increases as relation-
ship investments are made, possibly leading to a marriage.
Network embeddedness refers to third parties that are connected
to two partners in a durable relationship. These third parties affect
decision-making within the relationship. As with dyadic embed-
dedness, a learning mechanism and a control mechanism can be
identified. Third parties who have positive or negative experiences
with a person might inform potential future partners of this person
about these experiences. Being embedded in a network, a person
learns about the trustworthiness of a potential partner from third
parties. Partners control each other’s behavior, by being able to
complain to other potential partners or friends of the partner about
his or her behavior. A person may control a partner not only through
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
avoice network, as just described, but also through an exit network;
i.e. if a person has many alternatives on the marriage market, this
person might quite easily leave the present partner and find a new one.
There is an ambivalent relationship between marriage and
embeddedness. Marriage implies the solution of the trust problem
of durable partner selection and is less likely to occur if there are
no solutions at hand to this trust problem. Marriage, at the same
time, solves to some extent the trust problem of household special-
ization and is therefore more likely to occur if this trust problem
cannot be solved in an alternative way. Dyadic embeddedness and
network embeddedness, however, are potential solutions to both
problems. Therefore, they substitute for marriage if they solve the
trust problem of household specialization, but complement it where
the trust problem of durable partner selection is concerned. Two
partners will be less inclined to marry if their non-contractual
embeddedness is strong enough to solve the trust problem of house-
hold specialization and more inclined to marry if it is strong enough
to solve the trust problem of durable partner selection.
Therefore, we expect dyadic and network embeddedness to have a
positive effect on the probability that partners decide to build a long-
term relationship, but a negative effect on whether or not they marry
subsequently. This implies that the net effect of embeddedness on
marriage chances is unclear. However, as we stated in the intro-
duction of this article, the partner selection problem has increased
over time. Therefore, the relative importance of embeddedness to
solve this problem has increased as well. The advent of birth control
technology and the wider acceptance that people have sexual
relations before they marry have led to higher uncertainty about
the intentions of a partner one starts a relationship with. This un-
certainty implies that learning about a partner becomes more and
more important for making a partner-selection decision. In addition,
having other control options that decrease the possibility that a
partner will leave becomes more important as the general societal
sanctions against changing partners decrease. As a consequence,
dyadic embeddedness (e.g. spending free time together or cohabit-
ing) and network embeddedness (e.g. sharing friends and acquain-
tances) are more likely to help partners solve the trust problem of
durable partner selection in the present than in the past.
There are additional reasons why one can expect an increase in the
size of the embeddedness effect over time. Traditional relationships
have become less common and women’s economic opportunities
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
have improved. In many current durable intimate relationships,
both partners work. Some household chores are facilitated by new
technology (Nickols and Fox 1983; Weinberg and Winer 1983; Kim
1989), others are outsourced (De Ruijter et al. 2003; De Ruijter
2005), and the remainder is usually largely taken care of by the
female partner (Coverman 1985; Schor 1992). Hence, the trust
problem of household specialization has become smaller. The nega-
tive part of the embeddedness effect on marriage that operates
through the trust problem of household specialization has therefore
Hypothesis 1. The effect of dyadic embeddedness on marriage
chances has increased during the last four decades, i.e. it has
become less negative or more positive.
Hypothesis 2. The effect of network embeddedness on marriage
chances has increased during the last four decades, i.e. it has
become less negative or more positive.
The theory we have just presented is not meant as a competing alter-
native to established theories of marital decision-making. On the
contrary, our intention is to combine a theory of embeddedness
developed elsewhere with the existing literature, thereby adding a
new dimension to it. Before we turn to the empirical analyses, we
show that our theory can easily accommodate factors that are
known to affect marriage. The effect of each factor can be argued
to operate through a change in one or more of the three before-
mentioned variables: probability of dealing with a short-term
oriented partner, potential gains from marriage and losses from
failed investment.
Accommodation of Established Theoretical Mechanisms
Religious persons are more likely to marry (Thomas and Cornwall
1990; Thornton et al. 1992; Goldscheider and Goldscheider 1993;
Thornton and Camburn 1993; Stolzenberg et al. 1995; Call and
Heaton 1997; Bumpass 2002). This fact in itself makes being reli-
gious a strong signal of truly pursuing relational goals. A religious
person has an easier time convincing the partner of his or her true
intentions. Religiosity, however, does more than simply facilitating
the identification of partners with serious intentions. The gains
from a successful marriage are higher to a religious person than
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
to a non-religious person. In all major religions, marriage has an
intrinsic value and the religious environment shows approval to
married couples and disapproval to non-married couples. Because
relationship failure will also be more costly to religious persons
due to disapproval from the religious community, we expect a reli-
gious person to be more careful in starting romantic involvement,
but as soon as this person is involved he or she is more willing to
invest in a long-term relationship, in particular, if the partner is
religious as well.
One’s belief that sex should come after marriage is also expected
to increase one’s odds to marry (Laumann et al. 1994; Gottman
1998; Anderson 1999; Furstenberg 2001). First of all, having this
belief increases the gains from marriage. Moreover, it serves as a
signal that one is less interested in short-term transactional goals
and, therefore, that one has the intention to develop a long-term
Both the traditional specialization hypothesis (Becker 1973, 1981)
and the career-entry hypothesis (Oppenheimer 1988) about the effect
of education on marriage chances can be accommodated. Whether
it is the gains from specialization that make marriage attractive or
the joint earning capacity of the spouses, either concerns relational
Persons who cheat on their partners before marriage are clearly
less willing to invest in a long-term relationship with the focal
partner, since they invest in alternative relations at the same time
(Laumann et al. 1994; Treas and Giesen 2000). Even if we neglect
the possibility that the focal partner finds out about the deceit,
this will decrease marriage chances.
Unmarried couples with children as well as couples in which the
woman is pregnant are more likely to marry. Pregnancy and
having kids are relationship investments in themselves and are
indications that partners are willing to commit to a long-term
Thus, effects of factors that are known to influence marriage
decisions are consistent with our theory. In our empirical analyses,
we control for these factors, making sure time-dependent embedded-
ness effects cannot be a byproduct of parallel changes in these
factors. We have argued that since the 1960s it has become
much more commonplace that partners pursue transactional goals.
Therefore, the theory is also compatible with the observation that
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
partner selection decisions take more time and marriage is increas-
ingly being postponed (Marini 1978; Rodgers and Thornton 1985;
Norton and Moorman 1987; Schoen 1987; Oppenheimer 1988;
Kalmijn 1991; Qian and Preston 1993; Oppenheimer et al. 1997;
Goldstein and Kenney 2001). This extra time is used for convincing
one’s partner that one pursues relational goals and for obtaining a
better estimate of the probability that one’s partner does too. The
partners may nevertheless fail to convince one another. Whether
or not they succeed, we argue, crucially depends on the embedded-
ness in the relationship.
The embeddedness indicators we will be using in the empirical
analyses – the amount of shared free time, the extent to which
social networks overlap, and a timely measure of cohabitation –
are not novel either. Just like the indicators of marriage gains,
costs, and belief in the partner’s serious intentions, which we dis-
cussed above, the effects we have argued these indicators to have
are not at odds with those that have previously been attributed to
them. Shared free time and family-oriented social networks have
been said and found to facilitate the solution of relationship prob-
lems and reinforce family-related norms (Stolzenberg et al. 1995;
Furstenberg 2001; Wilcox 2002). And cohabitation has been argued
to help in screening unsuitable mates (Becker 1973, 1981; Clarkberg
et al. 1995; Lillard et al. 1995).
Lastly, the relationship between cohabitation and marriage has
been thoroughly studied elsewhere. The facilitation of trust through
a relationship investment by both sides is just one of various
mechanisms linking the two. For a minority of American couples,
cohabitation is an independent family arrangement (Bumpass et al.
1991; Smock 2000). Furthermore, several studies have shown that
couples that cohabited prior to marriage have higher odds of divorce
(Fergusson et al. 1984; Burch and Madan 1986; Balakrishnan et al.
1987; Bennett et al. 1988; Schoen 1992), which would be a puzzling
finding if the only thing cohabitation did were to facilitate trust.
Also, Brines and Joyner’s (1999) finding that cohabiting couples
typically do not specialize suggests that the extent to which
cohabitation-based embeddedness facilitates the solution of the
trust problem of household specialization is limited. We should
once more underscore that we are not primarily interested here in
the net effect of cohabitation. Hypothesis 1 concerns the change in
this effect. If the net effect used to be negative, it should have
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
become less so; if it were positive in the past, it should be even more
positive in the present.
Data and Methods
For the test of our hypotheses we draw on data from the CHSLS,
a retrospective survey containing questions about past and current
relationships (National Opinion Research Center 1997). Respon-
dents are from different birth cohorts and their relationships start
and end in different years. Moreover, data is included on time
changes in the embeddedness in the relationships between respon-
dents and their partners. This allows for the estimation of a time-
dependent effect of embeddedness on couples’ marriage chances.
The CHSLS was performed between 1995 and 1997 on a stratified
random sample of the non-institutionalized Chicago population
(i.e. excluding people living in group quarters, such as barracks,
college dormitories and prisons). This sample consists of five sub-
samples of two geographical levels. The first is the city level, and
consists of respondents from Cook County as a whole including
the inner suburban ring. The other four are targeted neighborhood
areas within the city of Chicago. Each neighborhood area is charac-
terized by the dominant presence of one race. A total of 2114 respon-
dents, 890 from the city level sub-sample and 1224 from the four
neighborhood sub-samples, answered detailed questions on their
past sexual experiences. Response rates were 71 percent for the city-
level sub-sample and ranged from 60 to 78 percent for the neighbor-
hood sub-samples.
The survey was executed using Computer Assisted Personal Inter-
view (CAPI) technology, meaning that the answers of respondents
were directly entered into lap-top computers. Respondents were
surveyed in person by experienced interviewers from the National
Opinion Research Center (NORC), who matched respondents on
various master statuses such as race and ethnicity, for an interview
averaging 90 minutes. Spanish-speaking interviewers were employed
so that recent migrants lacking facility in English could be included.
Because the survey asked very intimate questions, special care was
taken of response validity. For example, a part of the questionnaire
was self-administered (see Laumann et al. 2004: ch. 2, for details on
these validity issues).
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
The multi-level structure of the dataset allows for inter-
neighborhood comparison. Since differences between Chicago neigh-
borhoods are not relevant here, all analyses are performed on the
pooled sample. Because the master status variables with distribu-
tions that notably differed from those for the overall population
(mainly gender and race) are included as independent variables in
our models, issues of biased effect estimation are partially addressed.
We further investigated the extent to which the problem of poten-
tially biased estimation due to heterogeneity between neighborhoods
could play a role. For all models shown in the results section, we ran
a parallel model including neighborhood dummy variables. In addi-
tion, we ran the models for each neighborhood as well as the repre-
sentative city sample separately. Results were very similar and all of
our substantive findings were reproduced.
The data include information on many features of the sexual
relationships of the respondents; demographic characteristics and
sexual histories of both partners, information on the way they
became acquainted, the physical and social venues of meeting
places, the social networks surrounding relationships, in-bed activ-
ities and the subjective quality of relationships. A substantial part
of the interview concerned questions about the respondent’s two
most recent sexual relationships – if these existed. This gives us
particularly detailed information on the timing of events during
the relationships of the respondent with her/his two most recent
sexual partners. The presence of second most recent relationships
in the data reduces the usual bias due to selection on successful rela-
tionships. We can therefore expect more variance between relation-
ships in characteristics that distinguish successful from unsuccessful
Operationalization of the Variables
The questionnaire asked respondents whether or not they had ever
been married to any of their two most recent sexual partners and,
if so, at what time that marriage took place. This is our dependent
The following independent variables were employed.
The rela-
tionships together spanned the period between 1950 and 1997. We
created a calendar time variable with years as measurement unit
in order to control for base line changes in marriage propensities
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
over time. Respondents reported their race and that of their partner
by picking one of the following alternatives: White/Caucasian,
Black/African American, Alaskan native/native American/American
Indian, Asian/Pacific Islander/Oriental, indigenous Latin American,
Hispanic, other (specified) and Mestizo. To control for racial
marriage trends, we assigned couples to either of three categories:
White–White, African American–African American, and other.
The ‘other’ category thus includes all couples in which at most one
partner is White and at most one is African-American (less than
10% of CHSLS relationships are racially mixed). The numbers of
years of education completed by the respondent and by the partner
were used as indicators for long-term earning power. We created a
three-category relative education variable by subtracting the number
of years of education for the woman from those for the man to
incorporate the effect of earning power differences.
Three measures of embeddedness are employed in the analyses.
Respondents were asked to indicate the amount of free time shared
with their partner on a six-point scale ranging from ‘none’ to ‘all’.
This variable straightforwardly measures the extent to which part-
ners collect information about each other and are willing to invest
in the relationship.
Cohabitation is an additional measure for the time that partners
spend together. Moreover, sharing a household is an explicit
relationship investment. In our analysis, cohabitation is a time-
dependent variable. It is coded 1 for each premarital month during
which the partners were living together and 0 for each premarital
month during which they were not.
Respondents were further asked to list up to three adults whom
they spent much free time with during the 12 months preceding
the interview and three adults whom they discussed important
matters with. The respondent was asked to provide some additional
information about these people, including their relation to the
partner of the respondent at the time of the romantic involvement.
We computed the proportion of these social relations that the
partner shared as friends or acquaintances and that had known the
partner before the respondent. In this way the causality problem
of marriage and shared friends is accounted for: only shared friends
that were shared at the time of romantic involvement are included.
Note also that as a consequence children are automatically excluded.
Of course, we measure only a part of the network overlap between
the respondent and his or her partner, but given that the partner
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
does not complete a questionnaire him or herself and questions have
to be answered retrospectively, this is close to what is practically
possible. Because this variable was highly skewed we transformed
it by taking the logarithm of the original scores. This transformation
had no effect on the results.
The following control variables were employed. Respondent’s age
at the start of the relationship was used in the analyses in both linear
and quadratic form. The quadratic variable was constructed as the
square of the original variable centered around the mean (25.5)
such that the main effect is interpretable as the effect at the mean
age. The gender of the respondent was also controlled for. The
corresponding variable has value 1 if the respondent was male.We
created a couple education variable by averaging the lengths of
education of the man and the woman. The religiosity of respondents
was measured by their frequency of church attendance, a seven-
point scale variable ranging from no attendance (0) to daily atten-
dance (6). A four-category variable measuring respondents’ value
towards pre-marital sex was included as well, ranging from ‘not
wrong at all’ (1) to ‘always wrong’ (4). Respondents were asked
whether they and their partner had children or whether they had
ever tried to become pregnant. Further, information was available
on whether the two most recent sexual relationships overlapped,
in our analysis included as concurrent relationships variable.
Invalid missing data points for independent variables – respon-
dent refused or did not remember – were imputed by best-subset
regression (118 cases). Excluding these cases with imputed values
does not change the results. Cases with valid missing data – the
CAPI program skipped the question due to an answer to a previous
question that made the current question irrelevant or unnecessary –
were removed (22 cases). Table 1 presents the mean, standard devia-
tion, minimum and maximum value, and the number of cases prior
to imputation for all variables used in the analysis of the CHSLS
Statistical Model
We employ the Cox proportional hazards model (Yamaguchi 1991:
ch. 2; Blossfeld and Rohwer 1995: ch. 9, including applications on
marriage and divorce).
It estimates marriage chances per non-
married relationship time unit without making any assumption on
the general change of marriage chances over the course of the
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
relationship, and it allows for time-varying variables. The dependent
variable, a couple’s marriage hazard, is analyzed over the course of
each relationship.
Each time unit that does not end in marriage is
assigned a value 0, and is assigned 1 otherwise. This means that for
couples that have not married (yet) all time units take on value 0,
and that for couples that have married only the last time unit
takes on value 1. The Cox model compares the values of the first
time units of all cases and the second time units of all cases, and
so on, and addresses the differences in occurrence to the independent
variables included in the model. Coefficients reported on in the
results section are natural logarithms of hazard ratios. These para-
meter estimates should be interpreted as follows. Logged effects
are additive, so that a value 0 for the logarithm of the estimate b
of the effect of covariate x
indicates no effect. Positive values repre-
sent variables that make a couple’s monthly marriage chance
increase and negative values represent variables that make a couple’s
monthly marriage chances decrease.
The Cox model has several important advantages over alternative
models for current purposes (see again, e.g. Blossfeld and Rohwer
1995). The simplest way to estimate marriage chances is by per-
forming a probit or logistic regression model on the chance that a
relationship eventually develops into marriage without including
relationship time in any way in the model. One problem with such
an analytical approach is that relationships that started rather
shortly before the moment of interview have not had a fair chance
to develop into marriage. This problem is fairly pronounced here
because the comparison of young with old relationships plays a
central role in the test of the hypotheses. A second problem is that
the marriage chances of persons who would have had higher
marriage chances than others if only they could initiate the same
number of relationships as those others are underestimated. Includ-
ing relationship time as an independent variable in the model as a
solution creates new problems. The model then assumes a specific
relationship between the general marriage chances in early stages
and those in later stages. This could result in the biased estimation
of variables that are correlated with relationship time or variables
that are mediated by relationship time in their effects on marriage
chances. A more severe limitation of such an adjusted logistic regres-
sion model is that it does not allow for time-varying independent
variables. It would, for example, not be possible to distinguish
between couples that cohabited during almost their entire non-
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
married relationship and those that started to cohabit just shortly
before marriage. If these kinds of couples structurally differ in
their marriage chances, estimates of the effect of prior cohabitation
on a couple’s chance to marry could be wrong.
CHSLS respondents reported on their two most recent relation-
ships if they had at least two relationships. We accounted for this
type of respondent-level interdependence of observations by using
a clustered estimation of standard errors (Huber 1967).
The risk period ends in one of three possible ways: Through
marriage, through relationship dissolution, or because the moment
of interview was reached. We treated both the second and the
third ending as censors. The moment of relationship dissolution
was measured by the moment of last sex. The start of the relation-
ship was measured by the moment respondents indicated they got
first romantically involved with their new partner. Events that
took place in the same month (‘ties’) were handled using the ‘Efron’
method (Efron 1977).
Relationship months do not fall within the scope of our theory
and hence are not included in the empirical analysis in the following
cases: (1) if the moment of romantic involvement was reported to
fall less than three months before marriage; (2) if the respondent
or partner was married with someone else during that month or
had been married before;
(3) if they had had sex only once (in
this case the CAPI computer program skipped questions that are
crucial for our analyses); or (4) if the relationship months took
place before age 18 of either the woman or the man.
If the starting
or ending year of the risk period was missing, the case was dropped.
This case selection procedure yielded 2496 CHSLS relationships.
As was mentioned before, missing values on independent measures
were imputed.
The Cox model allows for time-varying variables. For some of
the independent variables described before, information on timing
was available for the construction of a time-varying version of
the variable. The following time-varying variables were employed:
(1) four variables indicating whether the relationship month con-
sidered belonged to the 1950s and 1960s,
1970s, 1980s or 1990s;
(2) a variable indicating whether cohabitation took place before
this month; and (3) a variable indicating whether one of the partners
had another, concurrent intimate relationship. All are coded 1 for
relationship months during the period of the relationship to which
they apply, and 0 otherwise. One thing that should be kept in
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Table 1. Descriptive Statistics for All Variables Used in the Analysis
Variable Mean St. dev. Minimum Maximum N
Ever married to partner 0.298 0.458 0 1 2496
Relationship start 1986 9.530 1950 1997 2496
In 1950s or 1960s 0.085 0.279 0 1
In 1970s 0.144 0.351 0 1
In 1980s 0.304 0.460 0 1
In 1990s 0.467 0.499 0 1
Racial composition couple 2479
both White 0.376 0.485 0 1
both African American 0.294 0.456 0 1
other 0.330 0.470 0 1
Religiosity respondent 3.882 2.067 1 8 2496
Age respondent at relationship start 25.083 7.924 13 57 2496
Education female partner (in years) 12.762 3.751 6 21 2468
Education male partner (in years) 12.772 3.935 6 21 2479
Pre-marital value respondent 1.879 1.169 1 4 2489
Ever pregnant 0.399 0.490 0 1 2495
Relationships respondent are concurrent 0.309 0.462 0 1 2496
Shared free time 3.930 1.385 1 6 2495
Ever cohabited before marriage 0.295 0.456 0 1 2496
Shared friends 0.060 0.171 0 1 2441
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
mind is that effects for later years tend to yield higher levels of
statistical significance – but not larger sizes – because there are
more cases for these years in the dataset, as Table 1 shows.
In Table 2, the results of six Cox models are shown. Models 1
through 5a are nested and the fit of each is a significant improvement
of that of the previous model. In these models, calendar time is
measured in years and its effect is estimated linearly. Model 5b
reveals details on trend effects and has calendar time measured
with decade dummies.
Model 1 includes only an estimate for the annual effect of calen-
dar time. The effect is negative and highly significant. A clear decline
in marriage is visible. More precisely, a couple’s chance of marrying
in a given year is about 96 percent (e0:044 ¼0:96) of their marriage
chance in the previous year. This indicates that the overall US
marriage decline after WWII can at least partially be attributed to
a decline in the marriage chances of Americans during months in
which they are already romantically involved.
The marriage decline
could then not solely be due to a more picky search for romantic or
sexual partners (Oppenheimer 1988).
In model 2, the race variables and their interactions with calendar
time are included. The main effect of calendar time remains un-
altered, meaning that White–White couples, the baseline category
for race, have baseline marriage chances. African-American couples
have lower overall marriage chances, 42 percent of those of White–
White couples, and this difference increases over time (roughly
e0:019 ¼2% each month). Thus, the diverging racial marriage
trend discussed in recent marriage literature is visible in our data
(Guttentag and Secord 1983; Espenshade 1985; South 1986, 1993;
South and Messner 1988; Bennett et al. 1989; Ellwood and Crane
1990; Lichter et al. 1991, 1992; Raley 1995; Wood 1996; Brien
1997). The marriage chances of couples in the ‘other’ category
differ from those of White couples as well, but to a lesser degree
and this difference does not increase with calendar time.
The effects of the control variables are added in model 3.
All remain constant throughout models 4–7 and mostly confirm
well-known correlates with marriage chances. Couples in which
the respondent attends church daily are per month 1.8 times
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Table 2. Results of Cox Proportional Hazard Models for Effects on Couples’ Monthly Marriage Chance
Independent variables Dependent variable: couples’ monthly marriage chance
Model 1 Model 2 Model 3 Model 4 Model 5a Model 5b
Calendar time (years) 0.044*** 0.043*** 0.013** 0.018*** 0.027***
1970s (vs. 1950s & 1960s) 0.402*
1980s (vs. 1950s & 1960s) 0.754***
1990s (vs. 1950s & 1960s) 0.887***
African American 0.871*** 1.099*** 1.040*** 1.019***
calendar time 0.019* 0.026** 0.024* 0.022*
In 1950s & 1960s 0.602*
In 1970s 1.072***
In 1980s 0.865***
In 1990s 1.004***
Other than non-Hispanic White 0.115 0.386*** 0.430*** 0.433***
calendar time 0.005 0.009 0.012 0.012
In 1950s & 1960s 0.429*
In 1970s 0.008
In 1980s 0.443**
In 1990s 0.512**
Age at relationship start 0.012* 0.007 0.008 0.007
Age at relationship start squared 0.002*** 0.002** 0.002** 0.002**
Respondent is male 0.059 0.067 0.073 0.082
Sum couple education 0.015* 0.019** 0.020** 0.020**
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Difference couple education 0.007 0.000 0.001 0.000
Religiosity respondent 0.094*** 0.099*** 0.099*** 0.100***
Pre-marital sex value 0.105** 0.107** 0.105** 0.108**
Ever pregnant while unmarried 1.375*** 1.215*** 1.196*** 1.214***
Relationship concurrency 1.115*** 1.028*** 1.034*** 1.040***
Shared free time 0.129*** 0.155***
calendar time 0.006*
In 1950s & 1960s 0.026
In 1970s 0.108
In 1980s 0.133**
In 1990s 0.192***
Cohabiting 0.487*** 0.549***
calendar time 0.022*
In 1950s & 1960s 0.593
In 1970s 0.186
In 1980s 0.414**
In 1990s 0.699***
Shared friends 0.941*** 1.112***
calendar time 0.079**
In 1950s & 1960s 0.741
In 1970s 0.694
In 1980s 0.743
In 1990s 1.559***
Log-likelihood 4967.845 4919.550 4738.396 4704.707 4695.588 4689.916
Degrees of freedom 1 5 14 17 20 32
Number of cases 2496 2496 2496 2496 2496 2496
Notes: Coefficients are logarithms of hazard ratios; * ¼p<0:05; ** ¼p<0:01; *** ¼p<0:001 (one-tailed tests).
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
(e0:094 6¼1:8) more likely to marry than couples in which the
respondent does not attend church at all. Couples in which the
respondent has conservative values concerning sex and marriage
marry at a rate that is 37 percent higher than the rate for non-
conservative couples. Partners with more education are also more
likely to marry. Monthly odds are 1.5 percent higher per year of
schooling for any of the partners. The difference in education
between the partners, however, does not seem to matter. The effect
is virtually 0 in all models.
These findings match those of Mare
(1991) and Sweeney (2002), who show that marriage chances increase
with both female and male earning capacities. Respondents with
concurrent relationships have 33 percent lower chances of marrying
than those without concurrent relations. Couples in which the
female partner has been pregnant during the relationship have
four times the marriage chances of other couples. Whether the
respondent is male or female does not affect the marriage chances
of couples in any of the models. The age of respondents at the
start of the relationship affects couple’s marriage chances in a curvi-
linear way. Marriage chances initially rise but go down after age 28.
It should be emphasized that the effect of calendar time is controlled
for in the model and that relationship time is the time dimension in
the event history model. Therefore, the effects of age do not partially
represent effects of the cohort of the respective partners or of the
stage of the relationship that they are at.
In model 4, the effects for the three embeddedness indicators are
additionally estimated. All three embeddedness indicators affect
marriage chances strongly and significantly. Couples that share
more free time over the course of their relationship have significantly
higher chances of marrying. Those who share all their free time are
per month 1.9 times more likely to marry than those who share prac-
tically no free time. Furthermore, months during which partners are
cohabiting are more likely to include a wedding than months during
which partners are dating. This factor is 1.63. A net positive effect of
cohabitation has been found before and also makes sense in the light
of Brines and Joyner’s (1999) findings that suggest that the extent to
which cohabitation-based embeddedness alleviates the trust prob-
lem of specialization is limited. And lastly, sharing all of one’s
most important friends at the start of one’s relationship versus
sharing no friends increases marriage chances by a factor of 1.92.
While the theory could not determine the net direction of the
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
effects of embeddedness, the empirical analyses clearly show that it
is positive. What these results imply for our theory is that overall the
complementary effect of embeddedness through the durable partner
selection mechanism outweighs the substituting effect of embedded-
ness through the household specialization mechanism. One could
argue that spending more time together, starting to live together,
or getting to know one’s partner’s friends and acquaintances are
not only factors that increase trust but might also be general indica-
tors for relationship quality. Therefore, we will not make a strong
claim about the main effects of embeddedness. The main confirma-
tion of our theory should come from the tests of interaction effects
with calendar time.
In model 5a, interaction effects of the three embeddedness indica-
tors with calendar time are added. All interaction effects between
embeddedness and calendar time are in the hypothesized direction
and significant. The effect of free time increases each calendar year
with 0.6 percent. This makes the free time effect in 1997 33 percent
higher than in 1950. This interaction effect is significant at the 5 per-
cent level. The effect of cohabitation increases each year with 2.2 per-
cent and this interaction effect is significant at the 5 percent level as
well, resulting in a 181 percent increase over the 47-year calendar
time span of our data. These results provide strong support for
hypothesis 1. The effect of sharing friends increases each calendar
year with 8.2 percent implying that the effect in 1997 is no less
than 41 times as large as the effect in 1950. This interaction effect
is significant at the 1 percent level. Hence, hypothesis 2 also finds
support. We want to emphasize that the support of the interaction
effects with calendar time for the embeddedness indicators for our
theory is less ambiguous than the support provided by the main
effects in model 4. There is no alternative explanation for why our
embeddedness indicators are more important for marriage in later
years than in earlier years (we come back to this point in the next
section). For the case of cohabitation, one would rather expect the
opposite. While cohabitation used to be tightly connected to
marriage until the 1970s, it has become more of a distinct way of
life in the 1980s and thereafter. Model 5b makes these counteracting
effects of cohabitation somewhat visible. The effect first declines
between the 1960s and 1970s, but reverses in trend afterwards. The
effects of the other two embeddedness indicators, shared free time
and shared friends, however, increase monotonically over time.
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
In sum, the results tell us that, in agreement with our theory,
the effects of embeddedness on couples’ marriage chances have
increased over the last four decades. More specifically, spending
free time together, knowing one another’s best and closest friends,
and living together have become more crucial factors in the determi-
nation of whether two intimate partners marry or not.
Some Remarks on the Validity of the Results
We have to realize that we work with retrospective data that might
be biased toward the present. One implication is that the reported
time partners spent together may partly be due to the fact that
they spend more time together now, possibly as a consequence of
the marriage. However, this alternative explanation does not apply
to the other two embeddedness variables. The other two embedded-
ness variables were constructed such that they specifically measure
premarital embeddedness. The shared friends variable includes
only friends who met the respondent before the respondent met
the partner. Hence, the network triangle was quite likely already
in place before the moment of romantic involvement and is not an
indicator for anything that was going on between the partners
between the start of romantic involvement and marriage. The co-
habitation variable is a time-dependent variable, so it also only
measures premarital embeddedness.
A second worry concerns the downward bias that exists in reports
on common friendships before romantic involvement. If this bias did
not depend on relationship cohort, there would hardly be a problem.
However, there is reason to believe that the bias is larger for older
relationships than for recent relationships. For example, respondent
memory regarding which friends were shared at the start of the
relationship may vary with relationship length. And the proportion
of friends that are currently shared and were shared in the past
naturally declines as these friends move to another city or die. We
therefore find it important to stress that such cohort-dependent
bias only strengthens our results. Suppose our hypotheses were
false and the effect of network embeddedness were time-constant
(or even decreasing with time). Then, this bias would lead to a nega-
tive interaction effect with time: In the past, the same difference in
marriage chance would be attributed to a smaller difference in net-
work embeddedness. In spite of, not due to, the downward bias we
do find an increasing effect.
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
It is possible that our embeddedness variables are indicators of
something else in addition to premarital embeddedness. Another
latent variable they may represent is relationship satisfaction. One
could argue that the higher the satisfaction with the relationship,
the higher marriage chances are and the higher levels of shared
free time, shared friends and the more likely one is to cohabit.
Such an argument would be compatible with our theory. One could
then propose that relationship quality may have become more
important a predictor of marriage chances, since relationship com-
mitment is nowadays more emotion-based while it used to be based
on economic necessity. These arguments together would form an
alternative explanation of the increased effects of the embeddedness
indicators. We tested this alternative explanation. We constructed
a relationship satisfaction variable that averaged answers to the
questions ‘How emotionally satisfying do/did you find your
relationship?’ and ‘How physically satisfying do/did you find your
relationship?’, interacted this variable with calendar time, and
added it to the fifth model. The results supported the idea that rela-
tionship satisfaction has become a better predictor of marriage
chances. The interaction effects of embeddedness and time, however,
did not disappear. On the contrary, they hardly changed at all.
One could also question whether or not younger cohorts finding
embeddedness more important than older cohorts causes the inter-
actions with calendar time. We ran models without the other two
measures of time as well as models with embeddedness interacted
with the two other measures. This did not lead to nontrivial estima-
tion differences, indicating little interference. As mentioned before,
the data allow us to properly separate out these three time effects.
The event history model disconnects ‘calendar time’ from ‘age at
the moment of interview’, and when relationships start is by
nature largely independent from the other two measures.
In principle, our analysis should be based on all relationships
respondents have had. Therefore, one could wonder whether the
bias created by having only the two latest relationships has caused
the effects we find. This seems implausible. In most similar data
sets, researchers only have information about the current relation-
ship. We ran an analysis to see what we would find if we had only
information on the current relationship. The effects weakened
substantially. Therefore, we expect that having even more infor-
mation about unsuccessful relationships would only strengthen the
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Conclusion and Discussion
There exists an ambiguous relation between marriage and trust.
Marriage forms a confirmation of mutual trustfulness. At the
same time, the marriage contract contains indications for a lack of
trust. Among other things, it specifies the legally enforceable pay-
ment of alimony and child support in the case of a divorce. This
ambiguity is resolved as soon as one identifies the different types
of trust involved here. The marriage contract covers the risk that a
preferable alternative partner shows up long after relationship
investments have been made. It does not cover the risk that one’s
partner is not willing to make considerable investments in a relation-
ship. Since such opportunists do not marry, marriage forms an indi-
cation that the trust problem of durable partner selection has been
These two trust problems are explicitly distinguished in the theory
we have presented. The theory constitutes an application of a more
general theory of trust and embeddedness in durable relations,
which has been successfully applied in the analysis of coalition
government stability as well as relations between and within firms
(Weesie and Raub 2000; Buskens et al. 2003). It allows for the incor-
poration of core elements of established theories of marriage and
known effects on marriage chances. Moreover, it adds the dimension
of embeddedness to the marriage literature. Since embeddedness
helps to solve trust problems, it substitutes for marriage as far as
the problem of household specialization is concerned, but com-
plements marriage in the problem of durable partner selection.
During the last four decades, decreased economic dependence of
women, the advent of birth control technology and secularization
have reduced the trust problem of household specialization, but
augmented the trust problem of durable partner selection. The net
effect of embeddedness has as a consequence become more positive
over time.
Data from the Chicago Health and Social Life Survey, containing
information on the timing of relationship events, allowed us to
model the behavior in 2496 relationships dynamically. Time series
analyses on these data proved the correctness of the above-
mentioned prediction. Indeed, the effect of embeddedness at the
dyadic as well as the network level turned out to be more positive
for more recent relationships.
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Our theory is consistent with the idea that the presumed decreas-
ing social embeddedness of American society (Putnam 2000) is an
explanation for decreasing marriage trends. However, our data do
not indicate decreased embeddedness of couples in intimate relation-
ships. The embeddedness variables could therefore not explain
away marriage trends. Alternatively, the increased trust problem
of partner selection might be part of the explanation of downward
marriage trends. Due to the fact that we did not have a direct
measurement of the size of trust problems we could not test this
Although the theory can be applied to the explanation of marriage
trends, marriage is just one of many realizations of trust in intimate
relationships. Its explanatory scope extends to variation in the
prevalence of other events in intimate relationships that involve
trust. It may help explain differences and trends in premarital task
division, in money management such as shared bank accounts, or
even in fertility and divorce rates (see Oppenheimer (1988) for
evidence that after marriage network embeddedness discourages
divorce and Esser (2003) that this effect has increased since the
1960s). All of the above examples concern actions that involve the
trust of one partner in some future actions of the other and hence
opportunities for embeddedness to differentiate between couples
who successfully pursue relational goals and those who fail to do so.
We wish to thank Damon Centola, Marin Clarkberg, Edward O.
Laumann, Ineke Maas, Werner Raub, Chris Snijders, Jeroen Weesie,
and anonymous reviewers for comments and helpful suggestions.
Buskens’ contribution is part of the project ‘Third-Party Effects in
Cooperation Problems’ of the Royal Dutch Academy of Arts and
Sciences (KNAW).
1. The notion of decreased trust in intimate relationships is in line with the hypo-
thesis of a decline in general trust in American society, central in the work of a
number of scholars (see, e.g. Fukuyama 1995; Putnam 1995; Seligman 2000).
2. Recently, the trust game is used as a description for trust situations in many con-
texts. See, e.g. the contributions of Gibbons (2001), Messick and Kramer (2001),
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
and Miller (2001) in Cook (2001). In addition, many others have argued that
marriage decisions in combination with household specialization have an incen-
tive structure that is similar to the one proposed here, see e.g. Ben-Porath (1980),
Becker (1973, 1981), England and Farkas (1986), Frank (1988), Huber and Spitze
(1988), England and Kilbourne (1990), Raub and Weesie (1993), Dufwenberg
(1996), Sweeney (2002), Youm and Laumann (2003) and Youm (2004).
3. One could object that men also face a risk. Just as specialization in household
work is costly in terms of foregone human capital investment, specialization in
paid work is costly in terms of foregone investment in household work efficiency.
Nevertheless, without the intention to trivialize household work, we think we can
safely assume that a lack of experience in household work after a potential future
break up is a negligible factor in a man’s decision-making on whether to special-
ize or not.
4. What exactly the contract covers and what not varies from time to time and place
to place.
5. The different structure of the second trust game does not change the formal
analysis (see Van de Rijt 2002).
6. Three theoretical accounts for this finding predominate in family studies: (1)
People who cohabit tend to be divorce-prone in the first place (Bennett et al.
1988; Booth and Johnson 1988); (2) cohabitation produces relationships,
values and attitudes that increase susceptibility to divorce (Booth and Johnson
1988; Thomson and Colella 1991; Axinn and Thornton 1992); and (3) relation-
ship dissatisfaction increases after the moment two persons start living together.
Higher divorce rates for those who cohabited before marriage would then be a
statistical artifact (Weston et al. 2003).
7. All non-dummy variables are centered around their means.
8. Educational level is generally regarded as a better indication of earning capacity
than short-term income, because it is not subject to temporal fluctuation. More-
over, relative income was a problematic variable for the CHSLS since it was
differently measured for relationships that were ever cohabiting and for those
that were not. Nevertheless, we ran models that included relative income vari-
ables. The effects were weaker and had the same direction.
9. The Cox proportional hazards model is part of a class of regression models,
referred to as event history analysis (EHA) models. We ran other EHA
models, including ones that assume a monthly baseline hazard of marriage.
Results turned out to be highly robust across these models.
10. A hazard is conceptually close to a ‘probability’ and is the ratio of the uncon-
ditional probability function to the survival function at a certain time t. In the
remainder of this article we will nevertheless refer to hazards with the more intui-
tive terms ‘probabilities’ and ‘chances’, ignoring the slight conceptual difference.
11. We checked whether other methods to handle ties changed our results and they
did not.
12. We realize that, as a reviewer pointed out, union history might be an important
predictor for trustworthiness in subsequent relationships and we cannot test this
effect limiting ourselves to first marriage. Nevertheless, we limited the scope of
our analysis to first marriage, because we have data only on the two last sexual
partners of a respondent. This does not provide us with enough information to
operationalize union history in a way that is precise enough for this purpose.
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
13. Relationship time before age 18 is ignored because marriage cannot occur. The
consequence is that the relationship year directly after the second partner
turned 18 is treated as the first year of the relationship. We ran models not
shown in the results section including a dummy variable coded 1 for such cases
and it did not affect the results.
14. Even though we feel these steps were necessary for a valid analysis of first
marriage chances, we did run parallel models that included cases we did not
want to generalize our theory to. The embeddedness time interaction effects
were even stronger and more significant in these analyses.
15. We decided to combine these two decades, because the theory predicts that
changes in the effect occur only after the 1960s. Also, there are rather few
cases for the 1950s and 1960s and separating them out confirmed that these
periods were similar.
16. One should not be too conclusive about the sizes of these effects because they
could in part be a result of case selection on relationship success. Relationships
that started longer ago are likely to have been ‘better’ relations because the
couples stayed together until the present day.
17. We do not claim this to be a strong rejection of Becker’s specialization hypothesis
(Becker 1973, 1981). The models do not take into account whether one or both
partners are currently in school, because no adequate information on this issue
was available. Since the minimum age for inclusion in our study is 18, we can
be sure some are. This omission may disturb the education effects in certain
ways. If young students delay family formation, both the sum effect and the
differential effect of education may be biased downward.
18. In model 5b, with calendar time increasing levels of statistical significance of the
embeddedness effects are in part the result of there being more recent than old
relationships in the sample, and in part the result of the rarity of premarital co-
habitation in the 1950s and 1960s. But again, our hypothesis concerns the
increase in effect size, not the increase in significance level.
Allyn, D. 2001. Make Love, not War: The Sexual Revolution: An Unfettered History.
New York, NY: Routledge.
Anderson, E. 1999. Code of the Street. New York: Norton.
Axinn, W.G. and A. Thornton. 1992. ‘The Relationship between Cohabitation and
Divorce: Selectivity or Causal Influence?’ Demography 29: 357–74.
Balakrishnan, T.R., K.V. Rao, E. Lapierre-Adamcyk and K.J. Krotki. 1987.
‘A Hazard Model Analysis of the Covariates of Marriage Dissolution in Canada.’
Demography 24: 395–406.
Baum, J.A.C. and C. Oliver. 1992. ‘Institutional Embeddedness and the Dynamics of
Organizational Populations.’ American Sociological Review 57: 540–59.
Beck, U. and E. Beck-Gernsheim. 1994. Riskante Freiheiten. Individualisierung in
modernen Gesellschaften. Frankfurt/Main, Germany: Suhrkamp.
Becker, G.S. 1973. ‘A Theory of Marriage: Part I.’ Journal of Political Economy 81:
Becker, G.S. 1981. A Treatise on the Family. Cambridge, MA: Harvard University
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Ben-Porath, Y. 1980. ‘The F-Connection: Families, Friends, and Firms and the
Organization of Exchange.’ Population and Development Review 6: 1–30.
Bennett, N.G., A.K. Blanc and D.E. Bloom. 1988. ‘Commitment and the Modern
Union: Assessing the Link between Premarital Cohabitation and Subsequent
Marital Stability.’ American Sociological Review 53: 127–38.
Bennett, N.G., D.E. Bloom and P.H. Craig. 1989. ‘The Divergence of Black and
White Marriage Patterns.’ American Journal of Sociology 95: 692–722.
Blossfeld, H. and G. Rohwer. 1995. Techniques of Event History Modeling: New
Approaches to Causal Analysis. Mahwah, NJ: Lawrence Erlbaum.
Booth, A. and D.R. Johnson. 1988. ‘Premarital Cohabitation and Marital Success.’
Journal of Family Issues 9: 255–72.
Brien, M.J. 1997. ‘Racial Differences in Marriage and the Role of Marriage Markets.’
Journal of Human Resources 32: 741–78.
Brines, J. and K. Joyner. 1999. ‘The Ties that Bind: Principles of Cohesion in
Cohabitation and Marriage.’ American Sociological Review 64: 333–55.
Bumpass, L.L. 2002. ‘Family-Related Attitudes, Couple Relationships, and Union
Stability.’ In Meaning and Choice – Value Orientations and Life Cycle Decisions,
ed. R. Lesthaeghe, pp. 161–84. The Hague: Netherlands: Netherlands Interdisci-
plinary Demographic Institute.
Bumpass, L.L., J.A. Sweet and A. Cherlin. 1991. ‘The Role of Cohabitation in Declin-
ing Rates of Marriage.’ Journal of Marriage and the Family 53: 913–27.
Burch, T.K. and A.K. Madan. 1989. ‘National Estimates of Cohabitation.’
Demography 26: 615–25.
Buskens, V. 2002. Social Networks and Trust. Boston, MA: Kluwer.
Buskens, V., J. Weesie, W. Raub and C. Snijders. 2003. The Governance of Relations in
Markets and Organizations. Research in the Sociology of Organizations, vol. 20.
Oxford: JAI/Elsevier.
Buskens, V. and W. Raub. 2002. ‘Embedded Trust: Control and Learning.’ In Group
Cohesion, Trust, and Solidarity. Advances in Group Processes, vol. 19, eds
E.J. Lawler and S.R. Thye, pp. 167–202. Oxford: Elsevier Science.
Call, V.R.A. and T.B. Heaton. 1997. ‘Religious Influence on Marital Stability.’
Journal for the Scientific Study of Religion 36: 382–92.
Clarkberg, M., R. Stolzenberg and L. Waite. 1995. ‘Attitudes, Values, and the
Entrance into Cohabitational Unions.’ Social Forces 74: 609–34.
Coleman, J.S. 1992. ‘Rational Choice Theory.’ In Encyclopedia of Sociology, vol. 4,
eds E. F. Borgatta and M.L. Borgatta, pp. 1619–24. New York: MacMillan.
Cook, K.S. 2001. Trust in Society. New York: Russell Sage Foundation.
Coverman, S.W. 1985. ‘Explaining Husbands’ Participation in Domestic Labor.’
Sociological Quarterly 26: 81–97.
Dasgupta, P. 1988. ‘Trust As a Commodity.’ In Trust: Making and Breaking Coopera-
tive Relations, ed. D. Gambetta, pp. 49–72. Oxford: Blackwell.
De Ruijter, E. 2005. ‘Household Outsourcing.’ PhD thesis. Utrecht University.
De Ruijter, E., T. van der Lippe and W. Raub. 2003. ‘Trust Problems in Household
Outsourcing.’ Rationality and Society 15: 473–507.
Dufwenberg, M. 1996. Time Consistent Matrimony with Endogenous Trust. Uppsala:
University of Uppsala.
Efron, B. 1977. ‘The Efficiency of Cox’ Likelihood Function for Censored Data.’
Journal of the American Statistical Association 72: 557–65.
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Ellwood, D.T. and J. Crane. 1990. ‘Family Change among Black Americans: What
Do We Know?’ Journal of Economic Perspectives 4: 65–84.
England, P. and G. Farkas. 1986. Households, Employment, and Gender: A Social,
Economic, and Demographic View. New York: Aldine.
England, P. and B.S. Kilbourne. 1990. ‘Markets, Marriages, and Other Mates:
The Problem of Power.’ In Beyond the Marketplace, eds R. Friedland and
A.F. Robertson, pp. 163–88. New York: Aldine.
Espenshade, T. 1985. ‘Marriage Trends in America: Estimates, Implications, and
Underlying Causes.’ Population and Development Review 11: 193–245.
Esser, H. 2003. ‘Soziale Einbettung und eheliche (In-)Stabilita
¨t.’ In Blickrichtung
Familie. Viefalt eines Forschungsgegenstandes. Festschrift fu
¨r Rosemarie Nave-
Herz anla
¨sslich Ihrer Emeritierung, eds M. Feldhaus, N. Logemann and M. Schlegel,
pp. 115–39. Wu
¨rzburg: Ergon.
Feld, S.L. 2000. ‘Covenant Marriage: A New Alternative for Traditional Marriage.’
Journal of Comparative Family Studies 31: 385–94.
Fergusson, D.M., L.J. Horwood and F.T. Shannon. 1984. ‘A Proportional Hazards
Model of Family Breakdown.’ Journal of Marriage and the Family 46: 539–49.
Forste, R. and K. Tanfer. 1996. ‘Sexual Exclusivity among Dating, Cohabiting, and
Married Women.’ Journal of Marriage and the Family 58: 33–47.
Frank, R.H. 1988. Passions within Reason. New York: Norton.
Fukuyama, F. 1995. Trust. The Social Virtues and the Creation of Prosperity. New
York: Free Press.
Furstenberg, F. 2001. ‘The Fading Dream: Prospects for Marriage in the Inner City.’
In Problem of the Century, eds E. Andersson and D. Massey, pp. 224–46. New
York: Russell Sage Foundation.
Gibbons, R. 2001. ‘Trust in Social Structures: Hobbes and Coase Meet Repeated
Games.’ In Trust in Society, ed. K. S. Cook, pp. 332–53. New York: Russell Sage
Giddens, A. 1991. Modernity and Self-Identity.Self and Society in Late Modern Age.
Cambridge: Polity Press.
Goldscheider, F. and C. Goldscheider. 1993. ‘Whose Nest? A Two-Generational View
of Leaving Home during the 1980s.’ Journal of Marriage and the Family 55: 851–62.
Goldstein, J.R. and C.T. Kenney. 2001. ‘Marriage Delayed or Marriage Forgone?
New Cohort Forecasts of First Marriage for US Women.’ American Sociological
Review 66: 506–19.
Gottman, J.M. 1998. ‘Toward a Process Model of Men in Marriages and Families.’
In Men in Families, eds A. Booth and A. C. Crouter, pp. 167–92. Mahwah, NJ:
Lawrence Erlbaum.
Granovetter, M.S. 1985. ‘Economic Action and Social Structure: The Problem of
Embeddedness.’ American Journal of Sociology 91: 481–510.
Guttentag, M. and P.F. Secord. 1983. Too Many Women? The Sex Ratio Question.
Beverly Hills, CA: SAGE.
Heckathorn, D.D. 2001. ‘Sociological Rational Choice.’ In Handbook of Social
Theory, eds G. Ritzer and B. Smart, pp. 273–84. London: SAGE.
Huber, J. and G. Spitze. 1988. ‘Trends in Family Sociology.’ In Handbook of Sociol-
ogy, ed. N. Smelser, pp. 425–48. Newbury Park, CA: SAGE.
Huber, P.J. 1967. ‘The Behavior of Maximum Likelihood Estimates under Non-
Standard Conditions.’ Proceedings of the Fifth Berkeley Symposium on Mathe-
matical Statistics and Probability 1: 221–33.
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Kalmijn, M. 1991. ‘Shifting Boundaries: Trends in Religious and Educational
Homogamy.’ American Sociological Review 56: 786–800.
Kim, C. 1989. ‘Working Wives’ Time-Saving Tendencies: Durable Ownership,
Convenience Food Consumption, and Meal Purchases.’ Journal of Economic
Psychology 10: 391–409.
Kreps, D.M. 1990. ‘Corporate Culture and Economic Theory.’ In Perspectives on
Positive Political Economy, eds J.E. Alt and K.A. Shepsle, pp. 90–143. Cambridge,
MA: Harvard University Press.
Laumann, E.O., J.H. Gagnon, R.T. Michael and S. Michaels. 1994. The Social Orga-
nization of Sexuality in the United States. Chicago, IL: University of Chicago Press.
Laumann, E.O., M. Van Haitsma, Y. Youm, A. Paik and S. Ellingson. 2004. The
Sexual Organization of the City. Chicago, IL: University of Chicago Press.
Lesthaeghe, R. and J. Surkyn. 1988. ‘Cultural Dynamics and Economic Theories of
Fertility Change.’ Population and Development Review 14: 1–45.
Lichter, D.T., F.B. LeClere and D.K. McLaughlin. 1991. ‘Local Marriage Markets
and the Marital Behavior of Black and White Women.’ American Journal of
Sociology 96: 843–67.
Lichter, D.T., D.K. McLaughlin, G. Kephart and D.J. Landry. 1992. ‘Race and the
Retreat from Marriage: A Shortage of Marriageable Men?’ American Sociological
Review 57: 781–99.
Lillard, L.A., M.J. Brien and L.J. Waite. 1995. ‘Pre-Marital Cohabitation and Subse-
quent Marital Dissolution: Is It Self-Selection?’ Demography 32: 437–58.
Mare, R.D. 1991. ‘Five Decades of Educational Assortative Mating.’ American Socio-
logical Review 56: 15–32.
Marini, M.M. 1978. ‘The Transition to Adulthood: Sex Differences in Educational
Attainment and Age at Marriage.’ American Sociological Review 43: 483–507.
Messick, D.M. and R.M. Kramer. 2001. ‘Trust as a Form of Shallow Morality.’ In
Trust in Society, ed. K.S. Cook, pp. 89–117. New York: Russell Sage Foundation.
Miller, G. 2001. ‘Why is Trust Necessary in Organizations? The Moral Hazard of
Profit Maximization.’ In Trust in Society, ed. K. S. Cook, pp. 307–31. New
York: Russell Sage Foundation.
Montgomery, J.D. 1998. ‘Toward a Role-Theoretic Conception of Embeddedness.’
American Journal of Sociology 104: 92–125.
National Opinion Research Center. 1997. The Chicago Health and Social Life Survey
[MRDF]. Chicago, IL: National Opinion Research Center.
Nickols, S.Y. and K.D. Fox. 1983. ‘Buying Time and Saving Time: Strategies for
Managing Household Production.’ Journal of Consumer Research 10: 197–208.
Norton, A.J. and J.E. Moorman. 1987. ‘Current Trends in Marriage and Divorce
among American Women.’ Journal of Marriage and the Family 49: 3–14.
Oppenheimer, V.K. 1988. ‘A Theory of Marriage Timing.’ American Journal of
Sociology 94: 563–91.
Oppenheimer, V.K., M. Kalmijn and N. Lim. 1997. ‘Men’s Career Development and
Marriage Timing during a Period of Rising Inequality.’ Demography 34: 311–30.
Portes, A. and J. Sensenbrenner. 1993. ‘Embeddedness and Immigration: Notes on
the Social Determinants of Economic Action.’ American Journal of Sociology 98:
Putnam, R.D. 2000. Bowling Alone: The Collapse and Revival of American Community.
New York: Simon & Schuster.
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Qian, Z. and S. Preston. 1993. ‘Changes in American Marriage. 1972 to 1987: Avail-
ability and Forces of Attraction by Age and Education.’ American Journal of
Sociology 58: 482–95.
Raley, K.R. 1996. ‘A Shortage of Marriageable Men? A Note on the Role of
Cohabitation in Black-White Differences in Marriage Rates.’ American Socio-
logical Review 61: 973–83.
Raub, W. and J. Weesie. 1990. ‘Reputation and Efficiency in Social Interactions: An
Example of Network Effects.’ American Journal of Sociology 96: 626–54.
Raub, W. and J. Weesie. 1993. The Management of Matches. Utrecht: ISCORE.
Raub, W. and J. Weesie. 2000. ‘The Management of Matches: A Research Program
on Solidarity in Durable Social Relations.’ The Netherlands’ Journal of Social
Sciences 36: 71–88.
Rodgers, W.L. and A. Thornton. 1985. ‘Changing Patterns of First Marriage in the
United States.’ Demography 22: 256–79.
Romo, F.P. and M. Schwartz. 1995. ‘The Structural Embeddedness of Business
Decisions: The Migration of Manufacturing Plants in New York State, 1960–
1995.’ American Sociological Review 60: 874–907.
Rusbult, C.E. and J.M. Martz. 1995. ‘Remaining in an Abusive Relationship: An
Investment Model Analysis of Nonvoluntary Dependence.’ Personality and
Social Psychological Bulletin 21: 558–71.
Schoen, R. 1987. ‘The Continuing Retreat from Marriage: Figures from the 1983 U.S.
Marital Status Life Tables.’ Social Science Research 71: 108–9.
Schoen, R. 1992. ‘First Unions and the Stability of Marriage.’ Journal of Marriage and
the Family 54: 281–84.
Schor, J. B. 1992. The Overworked American. The Unexpected Decline of Leisure. New
York: Basic Books.
Seligman, A. 2000. The Problem of Trust. Princeton, NJ: Princeton University Press.
Smock, P.J. 2000. ‘Cohabitation in the United States: An Appraisal of Research
Themes, Findings, and Implications.’ Annual Review of Sociology 26: 1–20.
Smock, P.J., W.D. Manning and S. Gupta. 1999. ‘The Effect of Marriage and Divorce
on Women’s Economic Well-being.’ American Sociological Review 64: 794–812.
South, S.J. 1986. ‘Sex Ratios, Economic Power, and Women’s Roles: A Theoretical
Extension and Empirical Test.’ Journal of Marriage and the Family 50: 19–31.
South, S.J. 1993. ‘Racial and Ethnic Differences in the Desire to Marry.’ Journal of
Marriage and the Family 55: 357–70.
South, S.J. and S.F. Messner. 1988. ‘The Sex Ratio and Women’s Involvement in
Crime: A Cross-national Analysis.’ Sociological Quarterly 28: 171–88.
Stolzenberg, R.M., M. Blair-Loy and L.J. Waite. 1995. ‘Religious Participation over
the Early Life Course: Age and Family Life Cycle Effects on Church Membership.’
American Sociological Review 60: 84–103.
Sweeney, M.M. 2002. ‘Two Decades of Family Change: The Shifting Economic Foun-
dations of Marriage.’ American Sociological Review 67: 132–47.
Thomas, D.L. and M. Cornwall. 1990. ‘Religion and Family in the 1980s: Discovery
and Development.’ Journal of Marriage and the Family 52: 983–92.
Thomson, E. and U. Colella. 1991. ‘Cohabitation and Marital Stability: Quality or
Commitment?’ NSFH Working Paper 23, Center for Demography and Ecology,
University of Wisconsin.
Thornton, A., W.G. Axinn and D.H. Hill. 1992. ‘Reciprocal Effects of Religiosity,
Cohabitation, and Marriage.’ American Journal of Sociology 98: 628–51.
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
Thornton, A. and D. Camburn. 1987. ‘The Influence of the Family on Premarital
Sexual Attitudes and Behavior.’ Demography 24: 323–40.
Treas, J. and D. Giesen. 2000. ‘Sexual Infidelity among Married and Cohabiting
Americans.’ Journal of Marriage and the Family 62: 48–60.
Uzzi, B. 1996. ‘The Sources and Consequences of Embeddedness for the Economic
Performance of Organizations: The Network Effect.’ American Sociological
Review 61: 674–98.
Uzzi, B. 1997. ‘Social Structure and Competition in Interfirm Networks: The Paradox
of Embeddedness.’ Administrative Science Quarterly 42: 35–67.
Van de Rijt, Arnout. 2002. ‘Why Marry if You Trust Your Partner?’ Utrecht:
Waite, L.J., F.K. Goldscheider and C. Witsberger. 1986. ‘Nonfamily Living and the
Erosion of Traditional Family Orientations among Young Adults.’ American
Sociological Review 51: 541–54.
Weesie, J. and W. Raub 2000. The Management of Durable Relations. Amsterdam:
Thela Thesis.
Weinberg, C.B. and R.S. Winer. 1983. ‘Working Wives and Major Family Expendi-
tures: Replication and Extension.’ Journal of Consumer Research 10: 259–63.
Weston, R., L. Qu and D. de Vaus. 2003. ‘Premarital Cohabitation and Marital
Stability.’ Paper presented at the HILDA Conference, University of Melbourne,
Wilcox, W.B. 2002. ‘Religion, Convention, and Paternal Involvement.’ Journal of
Marriage and the Family 64: 780–92.
Wood, R.G. 1995. ‘Marriage Rates and Marriageable Men: A Test of the Wilson
Hypothesis.’ Journal of Human Resources 30: 163–93.
Yamaguchi, K. 1991. Event History Analysis. Newbury Park, CA: SAGE.
Youm, Y. 2004. ‘The Sex Market and Its Implications for Family Formation.’ In The
Sexual Organization of the City, eds E. O. Laumann, M. Van Haitsma, Y. Youm,
A. Paik and S. Ellingson, pp. 65–194. Chicago, IL: University of Chicago Press.
Youm, Y. and E.O. Laumann 2003. ‘The Effect of Structural Embeddedness on the
Division of Household Labor: A Game-theoretic Model Using a Network
Approach.’ Rationality and Society 15: 243–69.
ARNOUT VAN DE RIJT is a PhD candidate in the Department
of Sociology at Cornell University. His areas of interest include
economic sociology, immigration, game theory and social networks.
In his dissertation he studies the dynamics of immigrant networks.
ADDRESS: Department of Sociology, Cornell University, 358 Uris
Hall, Ithaca, NY 14853, USA [email:]
VINCENT BUSKENS is a Research Fellow of the Royal
Netherlands Academy of Arts and Sciences at the Department of
Sociology/ICS, Utrecht University. His research interests focus on
social networks, cooperative relations and trust. He combines
(game-)theoretic approaches and social network analysis with
empirical tests using surveys as well as experiments.
ADDRESS: Heidelberglaan 2, room 1418B, 3584 CS Utrecht, The
Netherlands []
© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at University of Groningen on February 29, 2008 http://rss.sagepub.comDownloaded from
... The capability of insecurity to ruin romantic relationships is predicated upon the negative externalities that it births, like paranoia, cynicism, and aggressive disposition aimed at producing defenses in the face of confrontations (Miga et al. 2010). It threatens trust and since trust is a fundamental ingredient in human interactions (Roy et al. 2016) and an important factor for a healthy and stable romantic relationship (Rijt and Buskens 2006), this could pose a threat to the relationship. Furthermore, Campbell and Stanton (2019) describe the importance of trust in a relationship by alluding to the role it plays in reinforcing perceptions of the partner's dependability and beliefs regarding the future of the relationship. ...
Full-text available
This research investigated the extent to which low self-esteem, relationship dissatisfaction, and relationship insecurity exacerbate intention to break up in romantic relationships. Results from the study indicated that low self-esteem contributed to individuals having negative thoughts, emotions, and evaluation of their romantic relationships. Quite importantly, findings in the study showed that insecurity and relationship dissatisfaction partially mediate the relationship between low self-esteem and intention to break up in romantic relationships. Findings also indicated that due to low self-esteem, individuals may begin to doubt the level of trust, love and care accorded them by their romantic partners.
... Thompson and Snyder (1986), as well as Benson, Arditti, de Atiles and Smith (1992) applied attribution theory and showed strong evidence of the importance of attributional processes in determining spousal 5 interactions and relationship satisfaction. The issue of trust in adult intimate relationships is also addressed mainly by social theory (Van de Rijt & Buskens, 2000). Social theorists who continue to perceive traditional family values as crucial to social and economic stability interpret this cultural shift in terms of loss, decline and gradual degeneration. ...
Close relationships are essential to health and well-being. There is clear evidence that problems in intimate relationships contribute to both health and psychological problems (Cassidy, 2001). In order to understand and treat adult intimate relationships effectively, there needs to be a theory that describes and explain adult intimate relationships. A review of the literature shows that the field has lacked a framework or a broad theory. A theory that proposes to be a framework for the understanding of adult intimate relationship, is John Bowlby's attachment theory. A review of the literature on adult intimate relationships, shows that attachment theory has become prominent in the field. Research suggests that attachment theory offers clinicians a way to grasp and help clients shape their attachment bond, transforming their marriages and their families (Johnson, 2003). It is therefore researchers' view that attachment theory offers the clinician a guide to understand adult intimate relationships and to treat and transform them. The objective of this review is to explore the applicability of attachment theory as a theory of adult intimate relationships. This review focuses on: A brief review of theories used in the area of adult intimate relationships; the central concepts of attachment bond and attachment styles and how these are carried through to adulthood via internal working models; the application of the concepts of attachment theory to the understanding and treatment of adult intimate relationships; the evaluation of EFT as a treatment model; and finally, the evaluation of attachment theory as a model of adult intimate relationships. Although there are valid concerns about the extensions and applications of the theory, it is concluded that the use attachment theory contributes to clinicians' understanding of adult intimate relationships. It was also suggested that extensive research on the use of this theory in other countries and specifically South Africa is critically needed to make well-grounded conclusions about the usefulness of this theory. Thesis (MA (Psychology))--University of Stellenbosch, 2007.
Full-text available
We present a meta-analytic review of the literature on sex differences in the trust game (174 effect sizes) and the related gift-exchange game (35 effect sizes). Based on parental investment theory and social role theory we expected men to be more trusting and women to be more trustworthy. Indeed, men were more trusting in the trust game (g = 0.22), yet we found no significant sex difference in trust in the gift-exchange game (g = 0.15). Regarding trustworthiness, we found no significant sex difference in the trust game (g = −0.04), and we found men, not women, to be more trustworthy in the gift-exchange game (g = 0.33). These results suggest that men send more money than women do when their money is going to be multiplied, thereby creating an efficiency gain. This so-called “male multiplier effect” may be explained by a stronger psychological tendency in men to acquire resources.
Full-text available
The abstract for this document is available on CSA Illumina.To view the Abstract, click the Abstract button above the document title.
Full-text available
The neural basis of decision-making has been an elusive concept largely due to the many subprocesses associated with it. Risky decision making involves weighing good and bad outcomes against their probabilities in order to determine the relative values of candidate actions. Although human decision-making sometimes conforms to rational models of how this weighing is achieved, irrational patterns of risky choice, including shifts between risk-averse and risk-seeking choices involving equivalent-value gambles, are frequently observed in adolescents. Recent efforts involving neuroimaging, neuropsychological studies, and animal work indicate that cortical and subcortical structures play a central role in several of the subprocesses involved in decision-making and risk-taking. The identified mechanism is now known to include a multi-component valuation stage, implemented in ventromedial prefrontal cortex and associated parts of striatum, and a choice stage, implemented in lateral prefrontal and parietal areas. The frontal lobes are involved in tasks ranging from making binary choices to making multi-attribute decisions that require explicit deliberation and integration of diverse sources of information. The lateral and medial orbitofrontal, and ventromedial prefrontal areas are most relevant to deciding based on reward values and contribute to affective information regarding decision attributes and options. Lateral prefrontal cortex is critical in making decisions that call for the consideration of multiple sources of information, and may recruit separable areas when making well defined versus poorly defined decisions. The anterior and ventral cingulate cortex appear especially relevant in sorting among conflicting options, as well as signalling outcome-relevant information. Goal-directed decisions have their basis in a common value signal encoded in ventromedial prefrontal cortex, and exercising self-control involves the modulation of this value signal by dorsolateral prefrontal cortex. Emotion and feelings appear to influence the decision-making process. Decision-making often occurs in the face of uncertainty and it has been claimed that such decisions can be aided by emotions. Serotonin activity plays a significant role in nonnormative risky decision-making under conditions of uncertainty. There are neurobiological differences in structural and functional neuroanatomy in adolescents with substance abuse disorders and healthy controls. Risk-taking behaviors in adolescents have been associated with a high level of morbidity and mortality and might be explained by relative immaturity of brain structures. This is especially true for the prefrontal cortex that is linked with higher-order cognitive functions and emotional regulation. The anterior cingulate cortex and orbital prefrontal cortex along with medial prefrontal cortex have been involved in cognitive processes of conflict monitoring and response-inhibition that are pivotal for decision-making in adults. Increased vulnerability to risk-taking during adolescence may be a reflection of relatively higher tendency to seek rewards and still maturing capacities for self-control. Thus, neuromaturational differences in prefrontal cortex during adolescence might contribute to frequent underestimation of risks and increased risk-taking behaviors. In this review, we will focus on the neuroanatomical aspects of decision-making and risk-taking behaviours, two cognitive patterns which are intimately linked.
Using Simmel’s notion of sociation, the way in which rural elders in England and Wales relate to, or connect with, each other and others within their community, can be seen to be conflictual as well as consensual. As a vehicle for exploiting this relationship, social capital also can be antithetic as well as convergent and an important element of social capital – trust – can be freely given but also, at the extreme, instrumental and enforced. Measuring trust as a descriptive variable amongst rural elders suggests that personal trust is very high but trust in organisations (system trust), less so. All types of trust tend to be higher amongst the ‘better off’ measured by a number of socio economic and locational variables. Exploring trust as a process suggests that losses of trust and not trusting do damage social capital and connectivity, but rural elders will try and overcome trust loss in a variety of ways. Building personal trust through bridging capital is important here, where a ‘leap of faith’ in trusting and reciprocity are significant. System trust is built around helpfulness, honesty, reasonableness and civility. Instrumental and system trust start with scepticism but can be built through personal experiences and the ‘word of mouth’ of those already trusted, where there is strong bonding capital. ‘Localness’ is seen as an important aspect of building instrumental trust and here familiarity is more important than reassurance or ‘warrants of trust’. Successive government polices promoting trust as a means of building both social capital and connectivity are seen to have limitations as the factors that influence convergent trust largely fall outside of the scope of policy manipulation.
Couples seeking counseling often present with concerns regarding intimacy. Emotional risk-taking is an integral component of creating intimacy, yet little is known about the subjective, gendered experiences of emotional risk-taking within marital relationships. The purpose of this phenomenological study is to understand subjective emotional risk-taking for couples within marital relationships. This phenomenological study provides a glimpse into the experiences of male and female participants in regard to emotional intimacy. Through qualitative analysis, several common themes emerged with varying twists in regard to how these themes are experienced differently by the two genders. Implications for marital therapy and research are discussed.
Examines educational attainment as a dimension of assortative mating. Barriers to marriage between persons with unequal amounts of formal schooling increased between the 1930s and the present. These increases may be the result of trends in average educational attainment, age at leaving school, and age at marriage. The degree to which schools affect the selection of marriage partners is dictated by the degree to which leaving school and marriage occur closely together and by the educational attainments of marriage partners. Variation in the average age at leaving school and marriage and in educational attainment induce variation in educational assortative mating. Trends in age at marriage affect both the structure of marriage and inequality within and between generations.
Cohabitation has risen dramatically in the United States in a very short time. So, too, has the amount of sociological research devoted to the topic. In the span of a bit more than a decade, family sociologists and demographers have produced a large and rich body of research, ranging from documentation of cohabitation to assessment of its various consequences and implications. I first review basic descriptive findings about cohabitation as well as common explanations for its striking increase over recent decades. I next identify the central questions motivating most of the extant research and provide an assessment of past research as a whole. Finally, I speculate about themes that will be central to future research on cohabitation and consider the implications of cohabitation for gender equality in the United States and social science research on families.