Marriage

Abstract

Along with most Western nations, Australia has witnessed a decline in marriage, an increase in divorce, and a growth in the number of couples choosing to cohabit prior to, or instead of, marriage. In the mid-1970s, just 16 % of couples lived together before marrying. Now, more than three-quarters of couples do so. Same-sex marriage is commonly debated, and most states in Australia recognize same-sex relationships as a legitimate form of intimate partnering. Marriage is no longer considered an essential foundation for raising children with growing numbers of people raising children outside marriage, either as single parents or in cohabiting relationships. Despite these substantial social shifts however, marriage remains an aspiration for many young Australians and most marry at some point in their lives. This chapter commences by describing trends in the marriage rate in Australia over several decades using data from the Australian Bureau of Statistics, providing international comparisons where appropriate. We then review theoretical perspectives concerned with explaining change in marriage patterns and trends that provide insights into the reasons why most still aspire to marriage. Data from the Household, Income and Labour Dynamics in Australia (HILDA) survey are used to investigate who gets married in Australia, while in-depth interviews with men and women from a smaller study investigate the meaning of marriage in people’s lives. We conclude the chapter with a discussion of possible future developments and implications.

Appendices

Appendix 3.1: HILDA Data and Measures

3.1.1 The Data

HILDA provides excellent data for our purposes (see Technical Appendix). In the current study we focus on all participants who were never married, or were currently cohabiting but never married at Wave 1 (2,098 men and 1,881 women) and follow them through to wave 6, collected in 2006, to examine the characteristics of those who transitioned into their first marriage. We further restrict our sample to those aged between 18 and 40, excluding 343 men and 258 women, and those who had responded to the attitudinal measures, excluding a further 476 men and 377 women. The final analytic samples comprise 1,279 men with an average of 4.5 wave observations and 1,246 women with an average of 4.6 wave observations.

3.1.2 Measures and Covariates

The dependent variable indicates whether or not a respondent married after wave 1 and before wave 6. This is scored 1 if the respondent married between waves 2 and 6, and scored 0 if they did not. In our final analytic sample we observe a total of 444 transitions into marriage by 205 men and 239 women. It should be noted that while cohabitation is not the primary subject of interest, it remains the main pathway into marriage and people who are cohabiting have a much greater likelihood of getting married than those living alone (de Vaus 2004). To help account for this we include a dummy variable for whether or not the respondent was cohabiting (1 = yes), with a referent of not cohabiting (single). This measure is time varying and if the respondent transitions from single into cohabitation they are given a score of 1. It is also lagged by 1 year and therefore indicates the probability of getting married given that a respondent was cohabiting in the previous year.

Age of respondent is categorized into groups: 18–25 years (reference); 25–30 years, 31–35 years, and 36–40 years. We also include a measure indicating whether or not the respondent has a child under the age of 5 in the household. Our final control is for ethnic background and is coded 1 = Australia-born, 2 = Migrant – English speaking country and 3 = Migrant – non-English speaking country.

We include highest level of education, which is scored: 1 = Year 12 or less (reference); 2 = TAFE/Certificate; 3 = Diploma; and 4 = Bachelor degree or higher. We also include a measure for employment status, indicating 1 = employed full time (reference), 2 = employed part time and 3 = not in the labour force. Due to the age range of the sample many people who were employed part time, or not in the labour force, were studying. We therefore include a dummy control for full time study (1 = yes). We also include a continuous measure of respondents’ annual individual income from wages and salary; for inclusion in the models this is scaled to $10,000 (i.e. income/10,000). Finally, we have a measure of home ownership which indicates 1 = buying or owns home (reference), 2 = renting, 3 = other (i.e. living rent free/life tenure).

We include five measures capturing different normative attitudes towards work and family that might influence the decision to marry. The first measure indicates the importance of family (1 = very important). The original variable was measured on a scale from 1 (not important at all) to 10 (the most important thing), but in preliminary analysis the average score for men was 9.13 and for women 9.5, so we dichotomised this measure to indicate those who rated the importance of family between 8 and 10 (1 = yes), relative to the rest. This measure was only asked in Wave 1 and these responses were used for all 6 waves. The next two measures capture attitudes towards the gendered division of paid and unpaid household labour in households. Respondents were asked to indicate their level of agreement with the statement “It is much better for everyone involved if the man earns the money and the woman takes care of the home and children” and “If both partners in a couple work, they should share equally in the housework and care of children”. Responses to both questions ranged from 1 (strongly disagree) to 7 (strongly agree). The questions were asked in Waves 1 and 5. Responses from Wave 1 were used for Waves 1–4 and responses from Wave 5 were used for Waves 5 and 6. We also include a measure for the importance of a respondent’s employment and work situation on a scale of 1 (not important at all) to 10 (the most important thing). This question was only asked in Wave 1 and responses are used for Wave 1–6. The final attitudinal measure indicates the importance of religion to the respondent on a scale of 1 (not at all important) to 10 (the most important thing). This was collected in Wave 1 and Wave 4. Responses at Wave 1 were used for Waves 1–3, and responses at Wave 4 were used for Waves 4–6.

For our analysis we use a mixed effect (multilevel) model with a random intercept that takes into account the multiple observations for each respondent using xtlogit in STATA (StataCorp 2008). The results are based on a model that includes all measures and interactions between age and the socioeconomic measures. The models are run separately for men and women, however all models are re-estimated on a pooled sample of men and women with gender interactions to test for significant gender differences.

Appendix 3.2: Descriptive Statistics of Model Variables, Pooled Sample, Waves 1–6 HILDA (Hewitt and Baxter 2011)

	Men		Women
	Mean %c	SD	Mean %c	SD
Transition to Married (1 = yes)	4		5
Economic measures
Education
Yr 12 or Less	52		49
TAFE/Cert	22		16
Diploma	7		8
Bachelor degree+	19		26
Employment Status
Full time	67		44
Part time	18		32
Not in labour force	15		24
Full time study (1 = yes)	14		17
Earnings	26,868.16	17,706	22,302.27	15,098
Home Ownership
Own/Buy	51		47
Rent	46		50
Other (i.e. rent free)	3		3
Attitudinal Measures
Importance of family (1 = very important)	90		94
Attitudes to male breadwinnerb	3.16	1.7	2.53	1.7
Attitudes to sharing housework/childcareb	5.64	1.3	6.28	1.11
Importance of employment and worka	7.66	2.0	7.31	2.4
Importance of religiona	2.45	3.1	3.66	3.4
Controls
Child under 5 (1 = yes)	6		14
Cohabiting (lagged 1 = yes)	24		31
Age cohort of respondent
<25	51		54
26–30	22		21
31–35 (ref)	15		15
36–40	12		10
N	1,279		1,746
Person-years	5,756		5,732

1. SD Standard Deviations
2. ^a0 = not at all important to 10 = the most important thing
3. ^b1 = strongly disagree to 7 = strongly agree
4. ^cMean scores for continuous measures and percentages (%) for categorical variables. SD only reported for continuous measures

Appendix 3.3: Mixed Effect Logit Model of the Odds of Marriage for Men (Hewitt and Baxter 2011)

	Model 1	Model 2	Model 3	Model 4
	Odds ratio	Odds ratio	Odds ratio	Odds ratio
Economic measures
Education
Yr 12 or Less (reference)	1.00	1.00		1.00
TAFE/Cert	1.04	0.92		0.92
Diploma	2.38†c	2.04		1.93
Bachelor degree or higher	3.04**	3.07**		2.81
Employment Status
Full time (ref)	1.00	1.00		1.00
Part time	0.27**	0.16*		0.16**
Not in labour force	0.40*	0.56		0.52
Full time study (1 = yes)	1.29c	1.46		1.40
Earnings (scaled $10,000)	1.16†	1.48*		1.49*
Home Ownership
Own/Buy (reference)		1.00		1.00
Renting		0.47**		0.49**
Other (i.e. rent free)		0.29		0.37
Attitudinal Measures
Importance of family (1 = very important)			1.50	1.30
Attitudes to male breadwinnerb			1.00	1.05
Attitudes to sharing housework/childcareb			1.08	1.10
Importance of employment and worka			0.95c	0.92
Importance of religiona			1.10**	1.12**
Significant Interactions (Models 2 & 4)
26–30*Earnings		0.70†		0.69*
36–40*Earnings		0.63†		0.66†
Controls
Child under 5 (1 = yes)	2.42*c	2.24*c	1.74c	2.00†
Cohabiting (lagged 1 = yes)	6.60***	6.91***	8.07***	6.75***
Age of respondent
<25 (reference)	1.00	1.00	1.00	1.00
26–30	3.32***	8.73**	5.72***c	8.56*
31–35	2.46*	3.81	4.22***	3.70†
36–40	1.37c	4.29	2.39*c	3.90
Ethnic Background
Australia-born (reference)	1.00	1.00	1.00	1.00
Migrant – English speaking	2.18	2.12	2.49*	2.14†
Migrant – non-English speaking	2.00†	2.30*	1.55	1.58

1. SD Standard Deviations
2. ^† p < .10; ^* p < .05; ^** p < .01; ^*** p < .001
3. ^a0 = not at all important to 10 = the most important thing
4. ^b1 = strongly disagree to 7 = strongly agree
5. ^cThis association for men is significantly different from that for women (Appendix 3.4) at p < 0.05

Appendix 3.4: Mixed Effect Logit Model of the Odds of Marriage for Women (Hewitt and Baxter 2011)

	Model 1	Model 2	Model 3	Model 4
	Odds ratio	Odds ratio	Odds ratio	Odds ratio
Economic measures
Education
Yr 12 or less (reference)	1.00	1.00		1.00
TAFE/Cert	1.02	0.99c		0.95
Diploma	0.88c	0.80c		0.83
Bachelor degree or higher	1.40	1.36		1.46
Employment status
Full time (reference)	1.00	1.00		1.00
Part time	0.63*	0.60		0.63
Not in labour force	0.59†	1.30		1.28
Full time study (1 = yes)	0.37**c	0.35**c		0.35**c
Earnings (scaled $10,000)	1.13	1.36*		1.38*
Home ownership
Own/buy (reference)		1.00		1.00
Rent		0.41***		0.43***
Other (i.e. rent free)		0.17*		0.17*
Attitudinal measures
Importance of family (1 = very important)			3.16*	2.83*
Attitudes to male breadwinnerb			1.03	1.08
Attitudes to sharing housework/childcareb			0.91	0.94
Importance of employment and worka			1.08†c	1.00
Importance of religiona			1.07**	1.08**
Significant interactions (models 2 & 4)
26–30*Not in the labour force		0.21**c		0.21**c
31–35*Not in the labour force		0.28†		0.24†
26–30*Earnings		0.57**c		0.56**c
Controls
Child under 5 (1 = yes)	0.90c	0.89c	0.71c	0.88c
Cohabiting (lagged 1 = yes)	7.02***	6.78***	8.21***	7.33***
Age of respondent
<25 (reference)	1.00	1.00	1.00	1.00
26–30	1.72*	9.43***c	2.47***c	10.17***c
31–35	1.22	1.41	1.68†	1.52
36–40	0.39*c	0.14†c	0.55c	0.13†c
Ethnic background
Australia-born (reference)	1.00	1.00	1.00	1.00
Migrant – English speaking	1.29	1.19	1.43	1.23
Migrant – non-English speaking	1.68	1.66	1.19	1.43

1. SD Standard Deviations
2. ^† p < .10; ^* p < .05; ^** p < .01; ^*** p < .001
3. ^a0 = not at all important to 10 = the most important thing
4. ^b1 = strongly disagree to 7 = strongly agree
5. ^cThis association for women is significantly different from that for men (Appendix 3.3) at p < 0.05

Appendix 3.5: Qualitative Thematic Analysis

Thematic codesa	Sourcesb	Referencesc
Commitment (main theme)	20	49
Security/stability (sub- theme)	17	21
Love	9	11
	Total	81
Ritual	17	29
Public ritual – religious	9	16
Public ritual – non-religious	8	10
	Total	26
Capstone	12	14
Natural progression	6	8
	Total	22

1. ^aNVivo v.9 used to code data
2. ^bSources are the number of interviews in which the topic/theme was raised
3. ^cReferences are the number of times the topic/theme was raised across all interviews