Attitudes on marriage and new relationships: Cross-national evidence on the deinstitutionalization of marriage.

BACKGROUND: Consistent with the deinstitutionalization-of-marriage thesis, studies report a decline in support for marital conventions and increased approval of other relationship types. Generalizations are limited by the lack of cross-national research for a broad domain of attitudes on marriage and alternative arrangements, and by the lack of consensus on what counts as evidence.
OBJECTIVE: Acknowledging the conceptual distinction between expectations for behavior inside and outside marriage, we address the deinstitutionalization debate by testing whether support for marital conventions has declined for a range of attitudes across countries.
METHODS: Based on eleven International Social Survey Program items replicated between the late 1980s and the 2000s, OLS regressions evaluate attitude changes in up to 21 countries.
RESULTS: Consistent with the deinstitutionalization argument, disapproval declined for marital alternatives (cohabitation, unmarried parents, premarital and same-sex sex). For attitudes on the behavior of married people and the nature of marriage the results are mixed: despite a shift away from gender specialization, disapproval of extramarital sex increased over time. On most items, most countries changed as predicted by the deinstitutionalization thesis.
CONCLUSIONS: Attitude changes on 'new relationships' and marital alternatives are compatible with the deinstitutionalization of marriage. Beliefs arguably more central to the marital institution do not conform as neatly to this thesis. Because results are sensitive to the indicators used, the deinstitutionalization of marriage argument merits greater empirical and conceptual attention.

1. Introduction

While they are not without precedent, a host of ‘new’ relationships offer couples options for achieving sexual and emotional intimacy outside heterosexual marriage. From same-sex couples to unmarried cohabiters to committed partners who live apart, these forms of dyadic relationships have invited inevitable comparisons to heterosexual marriages. Researchers studying cohabitation, for example, have asked whether the arrangement is a genuine, long-term substitute for marriage or simply a new courtship stage on the path to the altar (Heuveline and Timberlake 2004; Raymo, Iwasawa, and Bumpass 2009; Smock 2000). Unmarried cohabiters have often been compared to their married counterparts. Studies report that cohabiters are characterized by lower relationship quality (Skinner et al. 2002), equivalent domestic outsourcing expenditures (Treas and de Ruijter 2008), similar economic benefits (Light 2004), comparable health and well-being (Soons and Kalmijn 2009; Wu et al. 2003), higher sexual frequency (Forste and Tanfer 1996; Yabiku and Gager 2009), and higher risk of relationship dissolution (Hohmann-Marriott 2006; Poortman and Lyngstad 2007). Same-sex couples have been measured against heterosexual marriages on children’s academic success (Rosenfeld 2010), the division of housework (Solomon, Rothblum, and Balsam 2005), and relationship outcomes (Kurdek 1998).

If ‘new relationships’ are benchmarked against marriage, marriage has also been judged in the context of other relationship options. According to Cherlin (2004:888), the growing acceptance of relationship alternatives is one indicator that marriage is undergoing a “deinstitutionalization”, described as “a weakening of the social norms that define partners’ behavior.” This theme is also seen in the Second Demographic Transition’s emphasis on individualism (Lesthaeghe and Surkyn 1988), the post-material valuation of self-actualization over conformity (Inglehart 1997), the second modernity’s “institutionalization of individualization” and “normalization of diversity” (Beck and Beck-Gernsheim 2004), and the insistence in “pure relationships” on satisfying emotional needs (Giddens 1992). Whether contemporary marriage is viewed as a casualty of social changes or a resilient institution evolving with the times, the study of relationships cannot ignore questions about the ways in which marriage itself is changing.

To understand marital change we evaluate shifts over time in attitudes toward marriage and its alternatives in 21 countries. We situate this analysis in the theoretical framework of the deinstitutionalization marriage (Cherlin 2004), distinguishing empirically between two conceptions of deinstitutionalization. Framing marriage as a hegemonic ideal, the first conception assumes that deviations from the conventional model of marriage, such as acceptance of non-marital cohabitation, are evidence of the deinstitutionalization of marriage. The second definition describes the marriage institution as a set of rules governing the behavior of spouses and rejects the notion that it makes sense to evaluate marriage in terms of other relationship options (Lauer and Yodanis 2010). This perspective views deinstitutionalization more narrowly in terms of changes in the core normative expectations held for married people.

This paper expands on previous research to make two important contributions. First, building on these conceptual distinctions in the definition of the marital institution, this analysis is the first to marshal a wide battery of attitude items to gauge whether the deinstitutionalization-of-marriage thesis is supported by changes in attitudes toward both marital and non-marital behavior. Second, moving beyond single-country studies, our cross-national analysis considers whether the changes in attitudes about marriage and relationship alternatives indicate that the deinstitutionalization of marriage is broadly characteristic of advanced industrial societies.

2. Background

The growing tolerance of new relationships is often described as a commentary on the institution of marriage. Some see new relationships as a rebuke to marriage — evidence that the marital institution is no longer fulfilling societal requirements nor meeting individual needs. Pointing to such behavioral indicators as the increase in non-marital fertility, the decline in household gender specialization, and the rise in cohabitation, Cherlin (2004) concludes that marriage is on the road to deinstitutionalization. Having resonance as a cultural ideal and even a status symbol, marriage may still be desired by most people, but the increasing tolerance of non-marital arrangements implies a weakening of the normative expectations for couples.

New relationships testify that marriage is not the only agreed-upon standard for organizing partners’ behavior. With declining expectations that sex, co-residence, and childbearing be linked only to marriage, the institution has lost its hegemonic position as a principle for bringing order to adult lives. Indeed, the increasing similarities between marriage and its alternatives — what has been termed the “blurred boundaries” between cohabitation and marriage in function, social acceptance, and legal treatment (Musick and Bumpass 2011) — can be read as offering additional evidence for the deinstitutionalization of marriage.

Others question whether new relationships really matter for marriage. Rather than signaling a decline of marriage, both new relationship options and new expectations for marriage can be said to demonstrate the resilience of the marital institution (Amato 2004). According to Lauer and Yodanis (2010), developments in non-marital arrangements are irrelevant to marriage as an institution. Conceptually, they distinguish between the weakening of the rules for behavior within marriage and the increase in the alternatives outside of marriage. What is supposed to go on in marriage and what marriage means constitute evidence for deinstitutionalization, they argue, but what happens outside marriage does not. Drawing from institutional economics and theories of new institutionalism, they contend that the marital institution provides the normative framework and cognitive scaffolding that governs the behavior of married partners. Even if there is greater social acceptance of non-marital arrangements, these changes do not speak to changes in the popular understanding of marriage or expectations for behavior within marriage itself, and it is these core norms for married partners that matter for the deinstitutionalization argument (Lauer and Yodanis 2010).

This distinction between the more expansive reach of the marital institution discussed by Cherlin and the narrower case argued by Lauer and Yodanis is useful for evaluating the deinstitutionalization argument and assessing the relationship between marriage and its alternatives. Cherlin’s broader account, which considers behavior inside and outside marriage, is a representation of marriage as a hegemonic institution with broad—if declining—influence. Lauer and Yodanis’ definition of marriage as a governance structure for married people presumes that deinstitutionalization relates only to the core of behavior and attitudes most central to marriage as it is understood and practiced.

2.1 Evidentiary challenges

Historically, the norms and ideals surrounding marriage have been reconstituted from time to time (Bernard 1972). Changes in what is expected of married persons are central to this redefinition. Early in the 20th Century patriarchal authority receded in importance, to be replaced by the modern ideal of the companionate marriage — a gendered partnership founded on respect, affection, and camaraderie (Burgess and Locke 1945). As Amato (2004) notes, companionate marriage was a departure from earlier ideals of marriage, because it rested on personal ties as opposed to social obligations. Today, the late modernity viewpoint holds marriage to be an unapologetic tool of personal fulfillment (Cancian 1987; Giddens 1992). Demanding intensive communication and emotional intimacy, marriage must meet expressive needs, not just functional requirements to maintain the family. Indeed, recent cohorts of divorced Dutch women are more likely than earlier ones to point to relationship difficulties as the cause of the break-up, even as they are less likely to cite the husband’s poor conduct (e.g., violence) (de Graaf and Kalmijn 2006). Given changing expectations, Gilding (2010) faults “a new orthodoxy in the sociology of personal life, formerly the sociology of the family” for assuming that institutions impose only constraints instead of fostering individual choice and personal development.

The distinction between broad hegemonic and narrower institutional definitions of marriage raise questions of what should count as proof of deinstitutionalization. How much change adds up to an erosion of the normative framework for marriage? Cherlin (2004) points to more egalitarian gender roles as illustrating deinstitutionalization. Although consistent with their what-married-partners-ought-to-do criterion, Lauer and Yodanis (2010) point to the remaining gender inequality in marriage to question whether gender specialization has gone far enough to count as significant change in marital expectations.

Some Second Demographic Transition trends, such as later marriage and smaller family size, are not incompatible with what most of us think marriage is. Other trends challenge the hegemony of marriage as the only acceptable arrangement for partnering and parenting, but they do not call on husbands and wives to change how they go about ‘doing’ marriage. Three decades ago Kingsley Davis (1983) described the rise in non-marital births, life-long singlehood, one-parent households, and cohabiting unions. None of these conflicted with taken-for-granted institutional understandings of the meaning of marriage or the behavior of married people. Nor were these demographic developments necessarily unprecedented, often being a revival of historical patterns that had fallen out of favor.

As Coontz (1992) observes, what we think of as ‘traditional’ American family life (universal marriage, full-time homemakers) had its heyday only among the white middle-class in the middle of the 20th Century. While cohabitation and non-marital fertility are today taken as proof of deinstitutionalization, we do not talk of marriage as having become more thoroughly institutionalized in the 1950s. The rise and fall of this mid-century American marriage model show the danger of accepting an historical anomaly as proof of a sweeping deinstitutionalization. Declining marriage rates may signal lasting change, but recent declines may prove only a short-run response to a global financial crisis. Furthermore, contradictory evidence is often ignored in discussions of deinstitutionalization. Divorce rates have been declining for decades in the U.S. (Kreider and Ellis 2011). To our knowledge, no one today cites this ‘marriage-friendly’ trend to challenge deinstitutionalization – much less to argue that marriage is being reinstitutionalized. Of course, Talcott Parsons (Parsons and Bales 1953) described divorce not as a threat to marriage, but rather as a benign “adjustive process” that occurred as the family focused its functions on socializing children and stabilizing adult personalities (Gilding 2010).

In the past the norms and ideals for marriage have undergone changes without an unraveling of the marital institution. Whether recent developments add up to a worrisome decline or deinstitutionalization of marriage is unclear, in part because of uncertainties that cloud discussion and research. These include 1) conceptual differences on the scope of the marriage institution, 2) questions about how much change constitutes deinstitutionalization, 3) the possibility of confusing a short-run anomaly with a long-run trend, 4) the potential neglect of contradictory findings, and 5) the failure to consider the normative strengthening or institutionalization of new or revived practices. This lack of clarity on what counts as evidence shows the need for a systematic approach to the question of deinstitutionalization.

2.2 The role of attitudes

Attitudes (i.e., positive or negative opinions about marriage and its alternatives) are important for two reasons. First, they predict behavior, albeit imperfectly. The association between attitudes and behavior is central to a long line of theorizing which stretches from symbolic interactionism (Blumer 1969) through the theory of planned behavior (Ajzen 1991) to efforts to relate specific perceptions to actual marital decisions (Carroll et al. 2007; Clarkberg, Stolzenberg, and Waite 1995). Second, as the internalization of cultural understandings, attitudes speak to culturally based theories that motivate contemporary discussions of family change. To explain a new regime of family behavior, the theory of the Second Demographic Transition points to the diffusion of a cultural ethos of individualism traced back to the Enlightenment (Lesthaeghe and Surkyn 1988). Reducing the need for supportive social institutions such as marriage (Esping-Andersen 1999), greater material security is said to foster post-material values that valorize personal fulfillment over conformity to societal expectations (Inglehart 1977). Criticizing marriage as a site of patriarchal oppression, feminist theory demands that relationships be organized around a new moral foundation of gender equality (Budig 2004). Consistent with cultural theories, marriage is no longer viewed as just a social obligation but rather as a tool of personal realization (Beck and Beck-Gernsheim 2004; Cancian 1987; Giddens 1992). Although persuasive analyses point to material, not cultural, roots for family change (Perelli-Harris and Gerber 2011), cultural arguments give weight to attitudes and values.

Culture invites cross-national comparisons. Despite single-country studies of changing attitudes toward marriage and other relationships (Kraaykamp 2002; Mynarska and Bernardi 2007; Thornton and Young-DeMarco 2001), a systematic, cross-national approach is lacking. Comparative studies have typically focused on one dimension of marriage-related attitudes, such as beliefs about the value of marriage (Gubernskaya 2010), acceptance of divorce (Rijken and Liefbroer 2012), or approval of non-marital sex (Widmer, Treas, and Newcomb 1998). Where there are multiple items they have often been combined into a single measure (Gubernskaya 2010; Liefbroer and Fokkema 2008), a strategy that improves on measurement reliability and facilitates explication but may obscure differences in change across attitudes. No empirical investigation has asked whether changes in the core beliefs about behavior inside marriage have changed as much as attitudes about relationships outside marriage.

Drawing on 1988–2008 International Social Survey Program (ISSP) data for 21 countries, this paper compares changes over time in 11 variables. The items are limited to those available in the ISSP for multiple time points. One subset of attitudes addresses marital alternatives. These beliefs on behavior permissible outside the marriage institution include unmarried cohabitation, non-marital childbearing, single parenthood, and premarital and same-sex sexual relations. Another subset of attitudes speaks directly to the definition of marriage, as well as core beliefs on how partners should behave inside marriage. These attitudes cover beliefs about the superiority of marriage and expectations of permanence, gender specialization, and sexual fidelity.

2.3 Previous research

Prior studies have investigated public opinion on marriage and other arrangements. Consistent with the deinstitutionalization thesis, these studies generally report a liberalization of opinion on alternatives to marriage, as well as on some expectations for behavior in marriage itself, such as views of divorce and gender specialization. Cross-national evidence is limited, but country-to-country differences in attitudes are evident, and some results remain inconclusive.

Non-marital Cohabitation

Between and often within countries, there is no firm consensus on what cohabitation means. European experience points to the evolution from a rebellion against convention to a testing ground for marriage to a substitute for marriage altogether (Manting 1996). In some countries (e.g., Sweden, Denmark), cohabitation is the norm, widely accepted and largely indistinguishable from married unions. In the U.S. cohabitation does not have the same cachet or durability as marriage (Cherlin 2009), but 61% of American women aged 25–44 report having cohabited at some point in their lives (Goodwin, Mosher, and Chandra 2010).

Public acceptance of cohabitation is higher in, say, Norway, where the arrangement is well established (Syltevik 2010), than in the Czech Republic where it is still rare (Mynarska and Bernardi 2007). Even in Southern and Eastern Europe younger adults hold positive views of non-marital cohabitation (Liefbroer and Fokkema 2008; Poortman and Liefbroer 2010). American acceptance of cohabitation rose fairly consistently from the 1970s through the 1990s (Thornton and Young-DeMarco 2001). British cohorts are more likely than earlier ones to endorse the practice (Haskey 2001). Considering young adults in 1994 and 2002, 12 out of 35 countries saw a significant growth in acceptance of alternative living arrangements; only two registered a significant decline (Liefbroer and Fokkama 2008).

Non-marital Childbearing

Moral and practical arguments frame heterosexual marriage as the optimal arrangement for having and rearing children. These arguments hinge on beliefs about the wrongness of non-marital sex, the stigma of illegitimacy, and the disadvantages for children of not being raised by two parents. Disapproval of non-marital childbearing is highest in Southern and Eastern Europe and lowest in Scandinavia (Aassve, Siron, and Bassi 2011). Disapproval has waned over time. Respondents in five European countries and the U.S. became less likely to insist that getting married should be a prerequisite for having children (Gubernskaya 2010). From the 1970s into the 1990s Americans expressed increasing tolerance toward those who have a child outside of marriage (Pagnini and Rindfuss 1993; Thornton and Young-DeMarco 2001).

Sexual Relations

Heterosexual marriage has lost its monopoly as the only socially approved context for sexual relations. Into the 1990s the declining disapproval of premarital sex had not extended to same-sex relations. Thereafter North American disapproval of homosexuality fell off sharply (Andersen and Fetner 2008), led by liberalizing attitudes among the more secular and better educated (Ohlander, Batalova, and Treas 2005; Treas 2002). Growing acceptance of same-sex relations does not imply a rejection of marriage, as shown by the legalization of same-sex marriage. In the EU, agreement that homosexual marriage should be legal ranges from 11% in Romania to 82% in the Netherlands (Gerhards 2010). Same-sex marriages are only recognized in some U.S. states, but disapproval declined from 71% in 1988 to 52% in 2006 (Baunach 2011). Between 1973 and 1998, however, Americans became less tolerant of marital infidelity (Treas 2002), even as legal changes around the globe decriminalized adultery (Frank, Camp, and Boucher 2010).

Gender Specialization

Holding women responsible for homemaking and men for breadwinning, the gendered division of labor was once the functional basis for marriage. The benefits of gender specialization were undermined by fertility declines, female employment, and the pragmatic necessity of husbands and wives substituting or augmenting one another’s labor (Treas 2008). The trend has been away from support for strict gender specialization (Cherlin and Walters 1981; Lee, Tufis, and Alwin 2010; Mason and Lu 1988; Pampel 2011; Scott, Alwin, and Braun 1996), but some countries are more conservative than others (Baxter and Kane 1995; Treas and Widmer 2000). In Western nations gender egalitarianism diffused from more to less educated groups, especially for women (Pampel 2011). No such pattern was seen for Eastern Europe, perhaps because socialism imposed near-universal female labor force participation on societies with traditional gender attitudes.

Divorce

Married partners are expected to stay married. Permanency is a defining characteristic of the institution, as evidenced by the formal steps required to end a marriage (Lauer and Yodanis 2010). Divorce rates, however, are only loosely associated with marriage rates (Kalmijn 2007). At least among young adults, Southern Europe, where divorce rates are low, shows the highest approval of divorce, and the English-speaking liberal countries show the lowest (Liefbroer and Fokkema 2008). Furthermore, approval of divorce depends on whether the marriage falls short of the ideal. Divorce is approved when a marriage is childless, the relationship poor, and both partners unfaithful (Diefenbach and Opp 2007). Attitudes vary cross-nationally. As for trends, U.S. tolerance for ending a bad marriage leveled off, if at high levels (Thornton and Young-DeMarco 2001). From 1994 to 2002 young Europeans became more accepting of divorce as a solution for troubled marriages (Liefbroer and Fokkema 2008).

The Superiority of Marriage

Perhaps the best evidence for the hegemony of marriage is the extent to which it is regarded as the best option. In the U.S. most question whether anyone, especially men, can have a satisfying life without being married, but the young are less likely to share this view (Kaufman and Goldscheider 2007). Between the 1960s and the end of the 20th century U.S. trends were small and inconsistent for the happiness, importance, and superiority of marriage (Thornton and Young-DeMarco 2001). In four European countries and the U.S. (but not Germany) the belief that married people were generally happier declined from 1988 to 1994, but this trend later flat-lined or rebounded in Great Britain, Ireland, and the Netherlands (Gubernskaya 2010).

Although much attitude change is consistent with the deinstitutionalization thesis, some trends run in the opposite direction and some attitudes seem largely unchanged. Furthermore, previous research does not establish whether public opinion in all countries, or just some, has moved toward deinstitutionalization of marriage. Undertaking a cross-national analysis we test the deinstitutionalization hypothesis by comparing change in 1) beliefs about appropriate behavior for unmarried people and 2) core beliefs on what marriage is and how married people should behave.

3. Data and methods

This paper analyzes 11 attitude items on marriage and relationships replicated in the International Social Survey Program (ISSP). Fielded and collected by independent survey organizations in participating countries, the ISSP provides cross-nationally comparable data on representative national samples. From 1988 to 2008 intermittent modules asked about views on marriage, sexual relationships, and related topics. Repeated cross-section data describe changes over time in attitudes across 21 industrialized countries: Australia, Austria, Bulgaria, Czech Republic, (East) Germany, (West) Germany, Great Britain, Hungary, Ireland, Israel, Italy, Japan, Netherlands, New Zealand, Northern Ireland, Norway, Poland, Russia, Slovenia, Sweden, United States. No item is available in every year or for every country, but the ISSP provides arguably the richest cross-national evidence on the direction of attitude change.

Table 1 describes the 11 dependent variables, as well as the years and countries for which they appear. Five items address the narrower core definition of marriage. Reflecting beliefs about what marriage is, two statements endorse the superiority of marriage (“It is better to have a bad marriage than no marriage at all” and “Married people are generally happier than unmarried people”). Other core items gauge the strength of norms for what married people should do, that is, stay married, specialize by gender, and be sexually faithful: 1) “Divorce is the best solution when a couple can’t seem to work out their marriage problems.” 2) “All in all, family life suffers when the woman has a full-time job.” 3) “What about a married person having sexual relations with someone other than his or her husband or wife?” Except for sexual relations, the response categories were “strongly agree” (1), “agree” (2), “neither agree nor disagree” (3), “disagree” (4), and “strongly disagree” (5). Where necessary, responses are reverse recoded, so higher values signify stronger support for ‘traditional’ marriage. For the sexual relations questions, responses are reverse coded to range from “not at all wrong” (1), “wrong only sometimes” (2), “almost always wrong” (3), to “always wrong” (4).

Table 1

Dependent variables

	Item	Survey Years	Countries	N
Outside Marriage
Cohabitation
Cohabit plan to wed	It is a good idea for a couple who intend to get married to live together first.	1994, 2002	Australia, Austria, Bulgaria, Czech Republic, East Germany, Great Britain, Hungary, Ireland, Israel, Japan, Netherlands, New Zealand, Northern Ireland, Norway, Poland, Russia, Slovenia, Sweden, United States, and West Germany	47368
Cohabit no plan to wed	It is all right for a couple to live together without intending to get married.			47819
Parenting
Marriage if child	People who want children ought to get married.R	1988, 1994, 2002	Austria, Great Britain, Hungary, Ireland, Netherlands, United States, and West Germany	29667
Single parent	One parent can bring up a child as well as two parents.	1994, 2002	Australia, Austria, Bulgaria, Czech Republic, East Germany, Great Britain, Hungary, Ireland, Israel, Japan, Netherlands, New Zealand, Northern Ireland, Norway, Poland, Russia, Slovenia, Sweden, United States, and West Germany	47548
Sex
Premarital sex	Do you think it is wrong or not wrong if a man and a woman have sexual relations before marriage?R	1991, 1994, 1998, 2008	Australia, Austria, East Germany, Great Britain, Hungary, Ireland, Italy, Netherlands, New Zealand, Norway, Poland, Slovenia, United States, and West Germany	69820
Same sex sex	And what about sexual relations between two adults of the same sex, is it…R			63801
Inside Marriage
Superiority of Marriage
Bad marriage better	It is better to have a bad marriage than no marriage at all.R	1988, 1994, 2002	Austria, Great Britain, Hungary, Ireland, Netherlands, United States, and West Germany	29450
Married happier	Married people are generally happier than unmarried people.R			27988
Permanence
Divorce	Divorce is usually the best solution when a couple can’t seem to work out their marriage problems.	1994, 2002	Australia, Austria, Bulgaria, Czech Republic, East Germany, Great Britain, Hungary, Ireland, Israel, Japan, Netherlands, New Zealand, Northern Ireland, Norway, Poland, Russia, Slovenia, Sweden, United States, and West Germany	46598
Gender Specialization
Wife’s job bad	All in all, family life suffers when the woman has a full-time job.R	1988, 1994, 2002	Austria, Great Britain, Hungary, Ireland, Netherlands, United States, and West Germany	29577
Sex
Extramarital	What about a married person having sexual relations with someone other than his or husband or wife, is it…R	1991, 1994, 1998, 2008	Australia, Austria, East Germany, Great Britain, Hungary, Ireland, Italy, Netherlands, New Zealand, Norway, Poland, Slovenia, United States, and West Germany	70533

Notes:

Reverse coded to strongly disagree (1)…strongly agree (5) or to not at all wrong (1)…always wrong (4).

Six items offer direction to the unmarried and thus speak to support for marital alternatives. Two Likert items gauge tolerance of non-marital cohabitation: “It’s a good idea for a couple who intend to get married to live together first” and “It is alright for a couple to live together without intending to get married”. Two items consider rearing children outside of marriage: “One parent can bring up a child as well as two parents together” and “People who want children ought to get married”. Two questions ask about tolerance for unmarried people having sex: “Do you think it is wrong or not wrong if a man and a woman have sexual relations before marriage? And what about sexual relations between two adults of the same sex?”

Respondent characteristics are control variables suggested by prior research. We control for respondent’s sex (female=1, male=0). Except for premarital and extramarital sex, women are less supportive than men of ‘traditional’ marriage (Gubernskaya 2010; Treas 2002). We control for age, because younger adults hold less conventional attitudes toward sex and marriage (Aassve, Siron, and Bassi 2011; Gubernskaya 2010; Haskey 2001; Treas 2002). Given more conservative attitudes for the married (Gubernskaya 2010), we include dummy variables for four marital statuses (never married; divorced or separated; widowed; and married, the omitted reference category). Educational attainment, harmonized cross-nationally into a dummy variable (post-secondary=1, else=0), is linked to less conventional views (Ohlander, Batalova, and Treas 2005). Dummy variables for respondent’s and spouse’s work status (employed=1, else=0) are included, given the less traditional views of the employed (Aassve, Siron, and Bassi 2011). Descriptive statistics appear in Appendix A.

Appendix A

Descriptive statistics for pooled sample

	Mean	SD	Minimum – Maximum
Bad marriage better	1.62	.83	1 – 5
Married happier	3.08	1.17	1 – 5
Divorce solution	2.47	1.14	1 – 5
Marriage if child	3.47	1.29	1 – 5
Single parent okay	3.04	1.22	1 – 5
Wife’s job bad	2.96	1.39	1 – 5
Cohabit plan to wed	2.45	1.17	1 – 5
Cohabit no plan to wed	2.44	1.21	1 – 5
Premarital sex	1.65	1.03	1 – 4
Same sex sex	2.76	1.34	1 – 4
Extramarital sex	3.39	.86	1 – 4
Female	.55	.50	0 – 1
Age	45.89	17.02	15 – 98
Married	.61	.49	0 – 1
Widowed	.09	.28	0 – 1
Divorced separated	.08	.27	0 – 1
Never married	.22	.42	0 – 1
College education	.26	.44	0 – 1
Employed	.53	.50	0 – 1
Partner Employed	.39	.48	0 – 1

Source: 1988, 1991, 1994, 1998, 2002, and 2008 International Social Survey Program. Sample size differs by dependent variable and years.

The Philippines was dropped from analysis altogether because it did not collect data on partner’s employment. Because Israel, Northern Ireland, and Russia did not collect partner information in survey years having sexual attitude items they are excluded from analyses on these three items. For the Netherlands we use 1988 baseline data to impute spouse’s employment for subsequent waves based on regressions on married respondents’ sex, age, education, employment, work hours, and religious affiliation (Protestant, Catholic, other, none). Cases missing data on dependent variables were dropped from analyses. This sample reduction ranged from 2.6% for the item on the family effects of the wife’s job to 14.7% for same-sex attitudes. Information was more complete for control variables, which were collected as part of a core questionnaire, than for dependent variables, which were often obtained from a supplement such as the mail-back questionnaire left with U.S. respondents after their interview. Because missing data on individual-level control variables were generally under 1%, listwise deletion was used. Pooled over countries and years, the samples analyzed range from 27,988 for the married-are-happier item to 70,533 for the extramarital sex attitude.

The omitted country reference category in analyses is the U.S. Having data on each of the items, it provides a point of comparison for country differences across all of the attitudes. The deinstitutionalization argument was initially formulated for the U.S., its trends are well known, and the country remains a distinctive outlier on most items (as we demonstrate). The reference year is the baseline survey year when the item was first asked. Supporting the deinstitutionalization thesis, a negative coefficient for survey year indicates a significant decline in support for traditional marriage over the baseline year. Adding the individual-level control variables, a second model ascertains whether the change in attitudes holds after adjusting for differences in demographic composition. While highlighting country differences in attitudes, this model also demonstrates how respondent characteristics are associated with attitudes toward marriage. Lastly, based on a model adding year-by-country interactions, we calculate the country-specific change for each item, adjusted for the control variables, to demonstrate which countries and which attitudes showed changes consistent with the deinstitutionalization thesis.

In analyses we pool data across countries and survey years. For each dependent variable we estimate an initial OLS model, including only the dummies for countries and survey years. Much like fixed effects hierarchical linear models, OLS models with dummy variables for countries adjust the standard errors for clustering of observations within the countries. For our problem OLS models are preferable, because we have a manageable number of country dummies, want to estimate country differences directly, and do not theorize particular country-level variables as level-2 predictors of attitudes. Random effects models require stricter assumptions that are not often met, but as a sensitivity test we also obtained estimates of the year coefficients based on random effects models. The results were virtually identical.

4. Results

Across attitudes, is there evidence of deinstitutionalization? Pooling data and controlling for country, we estimate OLS models for each of 11 attitudes. The survey year coefficients capture the change over time since the omitted baseline year. For the 11 attitude items individually, Table 2 summarizes the change for each survey year. Consistent with the deinstitutionalization of marriage thesis, a negative sign indicates a decline over the baseline survey in beliefs privileging traditional marriage.

Table 2

Pooled OLS coefficients for survey year change from baseline for attitudes by type (standard errors in parentheses)

	1988	1991	1994	1998	2002	2008	Countries
Adjusted for Country
Outside Marriage
Cohabit plan to wed			B		−.14*** (.01)		20
Cohabit no plan to wed			B		−.12*** (.01)		20
Marriage if child	B		−.29*** (.02)		−.55*** (.02)		7
Single parent			B		−.12*** (.01)		20
Premarital sex		B	−.09*** (.01)	.03* (.01)		−.15*** (.01)	14
Same sex sex		B	−.07*** (.01)	−.22*** (.01)		−.55*** (.01)	14
Inside Marriage
Bad marriage better	B		−.04*** (.01)		−.01 (.01)		7
Married happier	B		−.08*** (.02)		−.11*** (.02)		7
Divorce			B		−.08*** (.01)		20
Wife’s job bad	B		−.02 (.02)		−1.10*** (.02)		7
Extramarital		B	.15*** (.01)	.06*** (.01)		.05*** (.01)	14
Adjusted for Country and Respondent Characteristics
Outside Marriage
Cohabit plan to wed			B		−.16*** (.01)		20
Cohabit no plan to wed			B		−.14*** (.01)		20
Marriage if child	B		−.31*** (.02)		−.56*** (.02)		7
Single parent			B		−.12*** (.01)		20
Premarital sex		B	−.09*** (.01)	−.04*** (.01)		−.16*** (.01)	14
Same sex sex		B	−.05*** (.01)	−.26*** (.01)		−.43*** (.01)	14
Inside Marriage
Bad marriage better	B		−.05*** (.01)		.01 (.01)		7
Married happier	B		−.09*** (.02)		−.09*** (.02)		7
Divorce			B		−.06*** (.01)		20
Wife’s job bad	B		−.03 (.02)		−1.06*** (.02)		7
Extramarital		B	.14*** (.00)	.06*** (.00)		.12*** (.01)	14

Notes: Respondent characteristics=sex, age, marital status, education, employment, and partner’s employment. B=baseline year;

p<.05,

p<.01,

p<.001; (two-tailed).

The top panel of Table 2 shows regression coefficients for year, controlling only for the countries. Of the 21 comparisons with a baseline year, 15 show a decline at the .001 significance level in public support for traditional attitudes supporting marriage. Compared to 1991, disapproval of premarital sex was only marginally lower in 1998 (p<.05), but the 1994 and 2008 surveys did show decreases at the .001 level. Leaving aside items on non-marital behavior, only half of the ten comparisons for beliefs about marriage and married people show a significant decline in support for marital conventions. Unlike 1994, the belief that a bad marriage is better than no marriage proves to be no less popular in 2002 than when it was first asked in 1988. Although belief in gender specialization (the family suffers when the woman works full-time) was no weaker in 1994 than in 1988, a significant decline (p<.001) was evident by 2002. Compared to the absolute changes for most items, which were quite small, the declining approval of gender specialization (about one point on the five point scale) was impressive. Contrary to the deinstitutionalization thesis, disapproval of married people having extramarital sex was significantly greater, not smaller, in 1994, 1998, and 2008 than in 1991. Thus, the variables that prescribe behavior outside marriage for single people offer support for the deinstitutionalization thesis, but changes in core beliefs focused on what marriage is and how people should behave inside marriage were inconclusive.

Next, we estimate a full model which includes survey year, country, and the individual-level control variables. By including the controls we ascertain whether the changes in attitudes hold after we take account of changes in the demographic composition of the countries. The results, shown in the lower panel of Table 2, do not suggest that attitude shifts were driven by shifts in population characteristics. The results for one item on behavior outside marriage (premarital sex in 1998) increased in significance to the .001 level when respondent’s age, sex, education, marital status, employment status, and spouse’s employment status were added to the model. The significance, direction, and general magnitude of other time trends remain unchanged. Controls for respondent characteristics change neither the overall picture of increasing support for marital alternatives, nor the fact that evidence is mixed for the core indicators of the strength of the marital institution. In particular, the controls do not explain the contradictory finding of growing disapproval of extramarital sex.

4.1 Country differences and individual characteristics

Table 3 demonstrates the results for all the independent variables in the full models discussed above. We discuss the results for the individual-level controls (their association with the dependent variables net of survey year and country) – demonstrating that the results are consistent with prior research. In the interest of brevity we report the full models for only five attitudes, although the results are similar for others. The items address not only approval of marital alternatives (i.e., cohabitation and childbearing outside marriage), but also beliefs about marriage itself (its happiness) and appropriate behavior inside marriage (gender specialization, sexual fidelity).

Table 3

Pooled OLS results for selected attitudes (standard errors in parentheses)

	Married Happier	Wife’s Job Bad	Extramarital	Cohabitation No Wed	Marriage if Child
	B	B	B	B	B
Constant	3.32*** (.04)	2.79*** (.04)	3.47*** (.02)	2.46*** (.03)	3.42*** (.04)
Survey Year (ref=baseline year)
1994	−.09*** (.02)	−.03 (.02)	.14*** (.01)		−.31*** (.02)
1998			.06*** (.01)
2002	−.09*** (.02)	−1.06*** (.02)		−.14*** (.01)	−.56*** (.02)
2008			.12*** (.01)
Country (ref=US)
Austria	−.21*** (.03)	.79*** (.03)	−.30*** (.02)	−.92*** (.03)	−.47*** (.03)
Great Britain	−.48*** (.02)	.08** (.03)	−.06*** (.02)	−.69*** (.03)	−.40*** (.03)
Hungary	.09*** (.02)	.71*** (.03)	−.31*** (.02)	−1.05*** (.03)	−.50*** (.03)
Ireland	−.43*** (.03)	.18*** (.03)	.01 (.02)	−.28*** (.03)	−.26*** (.03)
Netherlands	−.89*** (.02)	.19*** (.03)	.17*** (.01)	−1.10*** (.03)	−1.04*** (.03)
West Germany	−.35*** (.02)	.53*** (.03)	−.35*** (.01)	−.82*** (.03)	−.28*** (.02)
Australia			−.07*** (.01)	−.60*** (.03)
East Germany			−.34*** (.02)	−.85*** (.04)
Italy			−.25*** (.02)
New Zealand			.06*** (.02)	−.48*** (.03)
Norway			.05*** (.01)	−.76*** (.03)
Poland			−.03* (.02)	−.42*** (.03)
Slovenia			−.38*** (.02)	−.57*** (.03)
Northern Ireland				−.30*** (.04)
Sweden				−1.19*** (.03)
Czech Republic				−.49*** (.03)
Bulgaria				−.38*** (.03)
Russia				−.55*** (.03)
Israel				−.27*** (.03)
Japan				−.05 (.03)
Female	−.26*** (.01)	−.09*** (.02)	.17*** (.01)	.03** (.01)	−.10*** (.01)
Age	.01*** (.00)	.01*** (.00)	.00*** (.00)	.02*** (.00)	.02*** (.00)
Marital Status (ref=Married)
Widowed	−.09*** (.03)	−.10*** (.03)	−.01 (.01)	−.12*** (.02)	−.07* (.03)
Divorce/Separated	−.54*** (.03)	−.23*** (.03)	−.23*** (.01)	−.47*** (.02)	−.38*** (.03)
Never Married	−.55*** (.02)	−.23*** (.02)	−.18*** (.01)	−.25*** (.02)	−.25*** (.02)
College education	.00 (.02)	−.24*** (.02)	−.14*** (.01)	−.11** (.01)	−.10*** (.02)
Employed	−.09*** (.01)	−.15*** (.02)	−.10*** (.01)	−.12*** (.01)	−.12*** (.02)
Partner Employed	−.09*** (.02)	−.13*** (.02)	−.08*** (.01)	−.13*** (.01)	−.09*** (.02)
R2	.20	.23	.07	.19	.23
N	27988	29577	70533	47819	29667

Note:

p<.05,

p<.01,

p<.001; (two-tailed).

Controlling for survey year and respondent characteristics, the country dummy captures its underlying support for marriage compared to the U.S., an often-studied outlier that provides a convenient reference. Consider the core variables on marriage itself. Consistent with the American idealization of marriage, all countries except Hungary are less likely (p<.001) than the U.S. to agree that married people are happier than single ones. The Dutch are the most skeptical. By contrast, other countries are more approving of gender specialization, believing that women’s full-time employment is harmful to the family. Great Britain comes closest to the U.S. position on the drawbacks of a full-time job. As for extramarital sex, Ireland is not significantly different from the U.S., and Poland is different only at the .05 level. Norway, New Zealand, and the Netherlands are more disapproving (p<.001) of extramarital sex, but most countries show greater tolerance. Regarding unmarried people, all countries except Japan voice significantly less disapproval of cohabitation (absent marriage plans) than does the U.S. Similarly, all countries, especially the Netherlands, are less insistent that people should marry if they want children. Thus, the U.S. — which inspired the deinstitutionalization thesis — is a conservative outlier in attitudes; it does not favor all conventional views of marriage, however, as indicated by its comparatively low endorsement of gender specialization. All things considered, countries differ in their attitudes but they are not necessarily consistent across attitude domains in their support for the beliefs taken as evidence of deinstitutionalization.

Table 3 also shows how individual characteristics are associated with conventional views of marriage. Results are largely consistent with prior research. Note the item on whether the married are happier. Controlling for survey year and country, respondents who are older and married are significantly (p<.001) more likely to view married people as happier than singles. Women and those who are employed or have an employed spouse are less likely to see marriage as foretelling happiness. Education is not significantly related to believing the married are happier, although better-educated people are less traditional on the other attitudes. Other attitudes parallel the results for marital happiness, except that women are more disapproving of extramarital sex and cohabitation. On extramarital sex, the widowed do not differ significantly from the married.

4.2 Country-specific change

Lastly, to the full model we add the interaction between the country and the latest survey year in which the item was asked. For each item we are interested in determining how much a country’s attitude changed between the baseline survey and the most recent survey year. Where the country-by-year interaction is significant at the .05 level we present the total country-specific change (adjusted for individual-level controls), calculated by summing the year coefficient and the country-year interaction term. Shown in Table 4 for each country, the changes ascertain how widespread the deinstitutionalization was across countries.

Table 4

Change in attitudes supporting marriage: 21 countries, baseline to last survey year

	Outside Marriage
	Cohabitation plans to wed	Cohabitation no plans wed	Marriage if child	Single Parent	Premarital Sex	Same Sex Sex
	1994–2002	1994–2002	1988–2002	1994–2002	1991–2008	1991–2008
U.S.	−.33***	−.12**	ns	−.13**	−.14***	−.34***
Australia	−.08*	−.22***		−.23***	−.17***	−.44***
West Germany	−.16***	−.23***	−.69***	−.12**	ns	−.14**
East Germany	−.26***	ns		ns	ns	−.35***
Great Britain	−.15***	−.18***	−.49***	−.14**	−.19***	−.83***
Northern Ireland	−.46***	−.48***		−.25***
Austria	ns	−.24***	−.72***	ns	ns	−.54***
Hungary	−.34***	ns	−.51***	ns	−.29***	−.45***
Ireland	−.34***	−.28***	−.70***	ns	−.59***	−.70***
Netherlands	−.26***	−.18***	−.67***	ns	ns	−.16***
Norway	−.13***	−.23***		−.11**	−.17***	−.63***
Sweden	ns	ns		−.16**
Czech Republic	−.09*	ns		−.47***
Slovenia	−.26***	−.33***		ns	ns	−.28***
Poland	−.24***	−.27***		−.20***	−.18***	ns
Bulgaria	ns	−.16**		ns
Russia	.09*	.38***		−.27***
New Zealand	−.18***	−.20***		ns	−.19***	−.60***
Israel	.19***	.24***		−.14**
Japan	−.37***	−.23***		ns
Italy					−.32***	−.67***

Note: Change [Last Survey Year + (Country X Year)] adjusted for sex, age, marital status, education, employment, and partner’s employment, as based on OLS model for each attitude. ns=change not significantly different from zero;

p<.05,

p<.01,

p<.001; (two-tailed).

The overall pattern of attitude change indicates the deinstitutionalization of marriage occurred across most countries on most items. With the exception of only three attitudes, a majority of surveyed countries showed a decline in conventional views. Most of the exceptions to changes consistent with deinstitutionalization spoke to core issues of the qualities of marriage and how people should behave within marriage—not to the items about acceptable behavior outside marriage. 1) Only half of 20 countries surveyed became less disapproving of divorce; 3 registered small increases in disapproval while 7 witnessed no significant change. 2) The 7 countries polled on whether a bad marriage is better than no marriage were about evenly split; agreement with the statement declined for 1 country, increased for 3, and showed no significant change for 2. 3) Of the 14 countries reporting on extramarital sex, 8 adopted a more conservative position in direct contradiction to the deinstitutionalization thesis.

The overall picture is one of attitude shifts toward deinstitutionalization, but the countries are hardly uniform in their changes. Only on gender specialization did all the surveyed countries show a trend consistent with a deinstitutionalization of marriage; for each of the 7 countries, there was a large and highly significant (p<.001) decline in the belief that a family suffered from a woman’s full-time employment. In general, the change toward deinstitutionalization was more widespread for attitudes about behavior outside than inside the marital institution.

Some individual countries, of course, deviated from deinstitutionalization. Even on non-marital alternatives Russia and Israel became less accepting of living together outside marriage, as evidenced by the fact that they broke ranks with other countries on both of the two measures of cohabitation. Across a number of items Ireland was among the countries showing the greatest movement away from conservative views of both marriage and non-marital arrangements. These changes reflected other developments in Irish society, including the 1995 divorce referendum and the convergence of women’s labor force participation toward EU rates (Russell et al. 2009). While reporting on fewer items, Northern Ireland also stood out in its declining disapproval of cohabitation and divorce. Except for growing acceptance of cohabitation with plans to marry, the U.S. was not at the forefront of attitude change toward deinstitutionalization of marriage. Showing few statistically significant changes across items, East Germany stands out, but this may be due to its relatively small sample — about 500 or only half that of other countries.

5. Conclusions

In the decade since the idea of the deinstitutionalization of marriage was introduced by Cherlin (2004), the concept has gained great currency as an intuitively satisfying description of a host of observed changes in family life and couple relationships. This generalization about the weakening of the normative expectations for couples suggests a powerful explanation consistent with cultural approaches ranging from theories of the Second Demographic Transition (Lesthaeghe and van der Kirk 1986) to late modernity (Giddens 1992; Beck and Beck-Gernsheim 2004). A spirited analysis of the concept of deinstitutionalization by Lauer and Yodanis (2010), however, challenges the adequacy of much of the evidence that has been cited to demonstrate that marriage is being deinstitutionalized — evidence such as the growing acceptance of cohabitation and other ‘new relationships’. Other hazards stymie the inference that deinstitutionalization has occurred. These difficulties include the lack of consensus on how much change is necessary to establish deinstitutionalization, the risk of mistaking a short-run anomaly for a long-run trend, the occasional blind-spot regarding contradictory evidence, and the inattention to the possible institutionalization of new or revived practices. Furthermore, the piecemeal nature of the empirical research, illustrated by the reliance on single-country studies of a few attitudes, has made it hard to say how widespread any deinstitutionalization might be.

Taking a systematic approach to the question of the deinstitutionalization of marriage and using the richest data available, this paper makes two contributions. First, utilizing repeated cross-sectional surveys that permit an assessment of recent changes in attitudes over time, we evaluate the deinstitutionalization thesis with comparable data for 21 countries. This allows us to determine whether deinstitutionalization is a broad-based phenomenon or is limited to only a few countries. Second, rather than considering a handful of attitudes or some summary indicator, we analyze 11 replicated items that tap various dimensions of beliefs about marriage and its alternatives. The results of this analysis offer new insights on the association of marriage and new relationships.

The breadth of the indicators allows us not only to determine whether all attitudes move in the same direction, but also to ascertain whether the conceptual distinctions posed by Lauer and Yodanis (2010) find empirical support. They argue against taking changes in approval of marital alternatives (e.g., cohabitation) as evidence and insist that only core changes in the governance structure for marriage itself can demonstrate a deinstitutionalization of marriage. Thus, we investigate whether beliefs about marriage and marital behavior have changed or whether change is limited to attitudes toward behavior outside marriage.

Analyzing the 11 attitude items for the pooled ISSP samples, we find a change in public opinion toward greater acceptance of all sorts of non-marital arrangements. Over time respondents became more approving of unmarried cohabitation, single parenting, and sex between single people. The deinstitutionalization thesis stands on shakier ground for indicators of attitudes about the nature of marriage and appropriate behavior for married people. On the one hand, declining support for gender specialization was far and away the biggest change away from marital conventions. On the other hand, the public became even more disapproving over time of married people having sex with someone besides the marriage partner. Furthermore, comparing the baseline year with the most recent survey, there was no change in the view that a bad marriage is better than no marriage at all.

Considering country-specific changes, most countries show a pattern of deinstitutionalization for most attitudes, but country-to-country differences are apparent. As with the pooled sample, the evidence is stronger for attitudes about behavior outside marriage than inside marriage. Three items did not see a majority of countries shifting toward views consistent with deinstitutionalization. All three items expressed attitudes toward marriage and married persons’ behavior. In short, across countries the findings for the core aspects of the marriage institution were, at best, mixed.

If marriage is viewed as a hegemonic institution defining appropriate behavior for both married and unmarried people, the growing acceptance of cohabitation and other marital alternatives offers solid support for the deinstitutionalization of marriage. If, following Lauer and Yodanis (2010), the marriage institution is a governance structure applying only to married people, a more skeptical or at least nuanced judgment on the deinstitutionalization thesis is called for. Even our expanded list of indicators may be too few for sweeping generalizations, but the analysis identifies evidence that contradicts the deinstitutionalization argument. The tentative conclusion is that there is more normative change regarding alternatives to marriage than regarding marriage itself. Some core values on marriage have resisted change. The erosion in social expectations for gender specialization points to two-earner couples as the centerpiece of a changing institution. Coupled with the reinforcement of traditional norms of sexual fidelity, however, this development seems more in line with an adaptive institution than a declining one. Rather than weakening, norms for marriage may simply be consolidating around a new normal.

This paper provides the most ambitious and systematic analysis to date of the attitudinal evidence on the deinstitutionalization of marriage. Unsurprisingly, given the large sample sizes, our results are highly significant, but the extent of attitude change is usually quite modest — again raising questions about how much change counts as deinstitutionalization. Admittedly, we are constrained by the available data. Different countries, different attitude items, or different years might well have led to different conclusions. The results of our analyses and their limitations suggest practices to guide future research. Although the deinstitutionalization of marriage is a powerful idea, it merits further empirical testing, refinement, and specification. Where there are sufficient countries to justify hierarchical linear modeling it would be useful to investigate just what characteristics of countries are associated with deinstitutionalization. To further content validity, the collection and analysis of survey data would benefit from greater conceptual clarity and more thoughtful consideration of attitude domain coverage. According to our own limited analysis the distinction between behavioral norms inside and outside marriage is a useful one, but the heterogeneous changes inside marriage itself call for serious attention to what constitutes the key dimensions defining contemporary marriage.