Living Apart Together and Cohabitation Intentions in Great Britain

Abstract

A growing number of studies examine how, why, and when people form and maintain living apart together (LAT) relationships. Although this literature shows that LAT is a diverse and ambiguous practice, little is known about whether people live apart together in particular ways under distinct constellations of life course circumstances. Moreover, it is unclear how intentions to convert LAT into cohabitation are configured by life trajectories. Drawing on data from an unprecedentedly large survey of people in LAT partnerships, we construct a fourfold typology of individuals in LAT relationships and show that each of the identified profiles is characterized by a distinctive position in the life course and different cohabitation intentions. These results indicate that LAT is a flexible way to practice partnership within the context of life course circumstances.

Keywords

cohabitation intentions intimate relationships life course living apart together partnership

Introduction

Rapid changes in patterns of partnership formation and dissolution, such as the rise of unmarried cohabitation, mean that intimate relationships in Britain have become more diverse in recent decades (Jamieson & Simpson, 2013). This trend has stimulated interest in those partnerships that are not recorded by data on de jure marital status or de facto household structures. Particular attention has been directed toward those unmarried individuals who identify themselves as part of a couple but who live in a different household from their partner (Roseneil, 2006; United Nations Economic Council for Europe [UNECE], 2011). These relationships are often termed “living apart together” (LAT) partnerships (Haskey, 2005).¹ Recent estimates suggest that 25% of unmarried Britons aged 16 to 59 years, or roughly 10% of all adults, are LAT (Duncan & Phillips, 2011; Haskey, 2005). Similar figures have been reported for other Western countries (Castro-Martín, Domínguez-Folgueras, & Martín-García, 2008; Régnier-Loilier, Beaujouan, & Villeneuve-Gokalp, 2009; Reimondos, Evans, & Gray, 2011; Strohm, Seltzer, Cochran, & Mays, 2009; Turcotte, 2013).

The “discovery” of LAT has sparked debate about its sociological significance. Several authors contend that LAT is an innovative type of relationship characterized by the voluntarism, flexibility, and gender egalitarianism associated with “individualized” social relations or a Second Demographic Transition (Levin, 2004; Upton-Davis, 2012, 2015). In this view, LAT is part of the broader shift from conventionally structured linear life paths to more diverse “choice biographies” that has been made possible by rising affluence and the detraditionalization of norms, attitudes, roles, and lifestyles (De Jong Gierveld & Merz, 2013; Jamieson & Simpson, 2013). In contrast, other scholars caution that LAT does not seem to be becoming more prevalent and that partners often live apart when they are unready or unable to cohabit (Ermisch & Siedler, 2009; Haskey & Lewis, 2006; Régnier-Loilier et al., 2009; Turcotte, 2013). Work by Duncan (2015) has nuanced this debate by showing that LAT is often an uncertain and ambiguous practice that can be as much a pragmatic and taken-for-granted response to life course constraints as a reflexively innovative act (Duncan, Carter, Phillips, Roseneil, & Stoilova, 2013; Duncan, Phillips, Carter, Roseneil, & Stoilova, 2014; Lyssens-Danneboom & Mortelmans, 2015). For example, while most women value the autonomy of LAT and report that it enhances the quality of their relationships (Funk & Kobayashi, 2016), many also perform gendered tasks for their partner (such as cooking or cleaning) and desire to ultimately cohabit (Duncan, 2015; Stoilova, Roseneil, Crowhurst, Hellesund, & Santos, 2014).

In view of the diverse ways LAT is practiced and experienced, many studies have examined how, why, and when people live apart together in Western societies. This work highlights the value of conceptualizing LAT as a “flexible practice” rooted in three intersecting dimensions of the life course (De Jong Gierveld & Merz, 2013; Duncan et al., 2013). In terms of individual biographies research shows that LAT is particularly common when people feel unready to cohabit (Duncan et al., 2013; Duncan & Phillips, 2011). This partly explains why LAT is common among young adults (Ermisch & Siedler, 2009). LAT also enables people to protect their independence, autonomy, own space, and assets when (re)partnering later in life (Funk & Kobayashi, 2016; Karlsson & Borell, 2002; Upton-Davis, 2015). LAT is therefore a strategy for people to maintain intimate relationships when they feel that cohabitation could place them or their dependents in a vulnerable position (Duncan, 2015; Roseneil, 2006). However, the ambiguity and lack of legal recognition of LAT can also create new vulnerabilities, for example, regarding care support in the event of ill health or financial arrangements following separation or bereavement (Duncan, Carter, Phillips, Roseneil, & Stoilova, 2012). Flexibility and informality can thus be a double-edged sword.

The above issues demonstrate how LAT is embedded within the demands, exchanges, and obligations generated by relationally “linked lives” (Elder, Johnson, & Crosnoe, 2003). Although popular perceptions often focus on highly skilled couples LAT to study or work in different places (Neustatter, 2013), most people LAT live near to their partner and few cite employment as a reason for living apart (Duncan et al., 2013; Duncan et al., 2014). Much more important is the way in which LAT allows people to maintain intimate relationships when their agency is bounded or restricted by their ties to others. For example, caring responsibilities and concern for the welfare of children from previous relationships are common explanations for LAT rather than cohabiting (De Jong Gierveld & Merz, 2013; Levin, 2004). LAT is thus configured by the needs of others and the power dynamics of interpersonal ties (Roseneil, 2006).

The importance of resources and power relations also emerges in the ways contextual conditions shape LAT. Although normative shifts mean that LAT is socially acceptable in many Western countries, cohabitation and in particular marriage remain many people’s “gold-standard” ideal and long-term aspiration (Duncan et al., 2014; Levin, 2004). In many cases, LAT is thus a compromise strategy when economic constraints inhibit cohabitation (Duncan et al., 2013). This is especially pertinent for young adults as the transition out of the parental home has become increasingly precarious, protracted and reversible due to greater participation in higher education, job insecurity, youth unemployment, and high housing costs (Castro-Martín et al., 2008; Stone, Berrington, & Falkingham, 2014). Difficulties in leaving the parental home may affect the (re)partnership practices of parents as well as young adults, as the presence of “boomerang children” or a “cluttered nest” makes it tricky for both generations to form new coresidential unions (De Jong Gierveld & Merz, 2013). Similarly, economic inequalities generated by the welfare system can also configure LAT. For example, some couples live apart in order to ensure that their incomes remain below benefit eligibility thresholds, while others are unable to live together due to social housing regulations (Neustatter, 2013). This means that housing market trends and British welfare reforms like the “bedroom tax” or benefit cap could strongly affect how people form and practice LAT relationships.

Although existing literature amply documents the heterogeneity of LAT, life course theories challenge scholars to also describe and explain patterns and structural regularities in partnership experiences (Elder et al., 2003). However, instead of examining the whole gamut of LAT practices, many studies focus on particular types of LAT relationship. This often involves discarding young adults and people who recently began LAT on the grounds that these might not be “proper LAT” (Lyssens-Danneboom & Mortelmans, 2015; Strohm et al., 2009), or else sampling individuals of particular ages who are thought to practice LAT in distinct ways for specific sets of reasons (De Jong Gierveld & Merz, 2013; Funk & Kobayashi, 2016; Upton-Davis, 2015). These approaches not only conflate “age with stage” and fit poorly with how people perceive LAT to be flexible and ambiguous (Duncan, 2015); they also assume that short relationships and those formed in young adulthood are casual, transitory, and thus of little significance for behaviors like fertility or residential mobility (Stoilova et al., 2014).

Recent analyses of Australian (Reimondos et al., 2011) and French (Régnier-Loilier et al., 2009) surveys show that age is only one of several life course characteristics that distinguish different types of LAT relationship. Instead, these studies identified four broader demographic profiles of individuals LAT. These were (a) childless young adults with little history of coresidential partnership, (b) individuals who have left the parental home but who have rarely been married and are often childless, (c) single parents in midlife, and (d) older individuals with a history of coresidential partnership. Each of these profiles was found to practice LAT in distinct ways (Régnier-Loilier et al., 2009; Reimondos et al., 2011).

In order to better understand whether different factors shape LAT in different countries, this study’s first objective is to identify whether there are distinct life course profiles of individuals LAT in Great Britain. This is important for two reasons. First, debates about the legal status of British LAT partnerships require detailed knowledge of who lives apart from their partner and how LAT is practiced across the life course (Duncan et al., 2012). Second, although population-level description and analysis cannot tell us about micro-level family processes, it can provide representative evidence about how inequalities and vulnerabilities are linked to LAT (Irwin, 2009). This is important for designing policies to support individuals for whom LAT is a long-term choice as well as those who would prefer to live together.

As LAT relationships have implications for housing demand and the provision of services, formulating household projections also requires knowledge about who intends to convert LAT into cohabitation. Although some studies have described cohabitation intentions, showing that these are not simply a function of age or stated reasons for LAT (Duncan et al., 2014; Ermisch & Siedler, 2009; Turcotte, 2013), far less is known about how different factors may influence who intends to move in with their partner at different points in the life course. Both Régnier-Loilier et al. (2009) and Reimondos et al. (2011) only briefly examined the cohabitation intentions of their four groups of individuals in LAT relationships.

As previous work shows that how people perceive the future of LAT partnerships is ambiguous and context-dependent (Roseneil, 2006; Stoilova et al., 2014), this study’s second objective is to examine how the factors associated with intending to convert LAT into cohabitation vary with life course position. Fulfilling this objective will yield insights about who perceives LAT as more of a durable family form than a transitional step toward coresidence (Levin, 2004). Examining cohabitation intentions using representative data will also shed light onto how partnership biographies, resources, relationally “linked lives,” and contextual factors come together to configure anticipated pathways out of LAT.

In light of the above, the next section of the article describes the quantitative data and methods used to fulfil the research objectives. We then present and discuss the results. Finally, the article concludes by reflecting on the broader implications of the findings.

Data and Method

Data and Definitions

This article uses data from the United Kingdom Household Longitudinal Study (UKHLS; Buck & McFall, 2012). In the first wave, a nationally representative sample of over 50,000 individuals aged 16 years and older from 30,000 households completed face-to-face interviews throughout 2009 and 2010 (McFall, 2013). Although UKHLS gathers annual data on marital status and cohabitation, information on whether people have a partner residing in another household was first gathered in the third sweep (conducted in 2011-2012). We therefore draw on data from this wave and use cross-sectional response weights to adjust for unequal selection probabilities, nonresponse, and sampling error (McFall, 2013). We discarded respondents from Northern Ireland where the sampling strategy differed from that used in Great Britain (McFall, 2013).

We began by extracting data from the 37,900 UKHLS respondents providing valid partnership data. These respondents’ partnership status was then coded into five groups using information about their de facto marital status (including cohabitation) and their response to a question about noncoresidential relationships. Using these two pieces of information, we classified respondents into (a) married; (b) cohabiting; (c) LAT; (d) single and never married (no current partner and no previous marriage); and (e) single but previously married (no current partner and either separated, divorced, or widowed). Disaggregating the nonpartnered according to their marital history takes into account that singletons are not a homogenous group. Same-sex couples were retained and civil partners were treated as married.² A very small number of married individuals LAT were discarded as it is difficult to ascertain their circumstances. Duncan et al. (2012) suggest that many may have partners who are in prison or resident in other institutions; but it is also possible that respondents have concurrent partners, are estranged from their legal spouse and LAT with a new partner, are temporarily living apart, or are restricted from living together by immigration/visa regulations.

Specifically, we identified LAT relationships using the question: Do you have a steady relationship with someone you are not living with here, whom you think of as your “partner”? This question was asked to all adults without a coresident partner using a confidential self-completion extension to the main questionnaire. The question asks people to only report committed intimate relationships, while avoiding the potentially unfamiliar term “living apart together” (Haskey, 2005). However, one drawback with the phrasing may be that some people do not use the term “partner” to describe their relationship, perhaps preferring boy/girlfriend or even husband/wife (Duncan et al., 2014). Even if this is the case, we anticipate that the first portion of the question guides respondents to interpret the word “partner” in a general sense, regardless of the appellation they use to describe their own circumstances (Ermisch & Siedler, 2009; Haskey, 2005, for prior examples).³

In contrast to some studies, we do not exclude young adults or people who have only been in a steady relationship for a short period of time from our definition of LAT (Lyssens-Danneboom & Mortelmans, 2015). We do this for two reasons. First, prior research suggests that people view LAT as a flexible practice rather than a specific type of living arrangement (Duncan et al., 2013). This means that rather than arbitrarily defining “proper LAT” we need to examine how life course circumstances correlate with LAT in different ways. This is best achieved by letting people decide whether or not they have a partner before examining variation in their relationship characteristics (UNECE, 2011). Second, qualitative studies show that the duration of LAT is often a poor proxy for perceived commitment to the union (Stoilova et al., 2014). Taken together, these points suggest that the commitment of couples who live apart should be regarded as an empirical question rather than a definitional issue.

Analytic Approach

To provide contextual information about LAT in Great Britain, we began by analyzing how a range of life course attributes are associated with partnership status using a multinomial logistic regression model. This required accepting the Independence of Irrelevant Alternatives assumption. This states that adding or deleting outcome categories does not affect the ratio of the probabilities among the remaining alternatives (Long & Freese, 2006). This is most likely to hold when the outcomes are conceptually distinct (Long & Freese, 2006), making Independence of Irrelevant Alternatives a plausible assumption when modelling partnership status (cf. Turcotte, 2013).

The multinomial model includes a range of covariates. To capture variation in partnership practices by life course stage we include a cohort (categorical age) variable. As a proxy for cultural attitudes, religious beliefs, and the like, we include an ethnicity dummy distinguishing respondents who self-identify as White from those with other ethnic identities.⁴ We also include a dummy variable indicating whether one was born in the United Kingdom as immigration and acculturation experiences may matter for relationships beliefs, orientations, and practices.

To take into account that living arrangements and caring responsibilities may affect people’s freedom and inclination to form coresidential unions, we include dummy variables to identify respondents living with their parents and those with dependent children. As LAT may be more common among the highly educated, we distinguish individuals with no qualifications or only qualifications below school level from those with high school level qualifications (such as General Certificate of Secondary Education or A-Levels) and those with higher qualifications (university degrees or equivalent). Self-reported labor force status is included to capture the impact that job location and economic resources may have on partnership decisions. Finally, we include variables to capture the effects of housing tenure and the spatial context, distinguishing urban settlements with a population of over 10,000 from rural areas. Before running the models only 623 cases (1.6%) were discarded due to missing data. We term the remaining 37,277 cases to be the full sample.

As our first objective was to assess whether we can discern distinct life course profiles of individuals LAT, after running the multinomial model we retained only those 3,358 persons in LAT relationships. Of this sample, 246 cases with missing values on key covariates were discarded to give our LAT sample of 3,112 individuals.⁵ Using this sample, we examined whether there are distinct and relatively homogenous subgroups of individuals in LAT relationships using multiple correspondence analysis (MCA; Régnier-Loilier et al., 2009; Reimondos et al., 2011). MCA is an effective method for detecting and representing the underlying structure of data using multiple categorical variables (Greenacre & Blasius, 2006). It works by transforming contingency tables into low-dimensional spaces from which the proximities between cases can be identified and interpreted. Specifically, the coordinates generated from MCA can be fed into cluster analysis to identify mutually exclusive groups of respondents with similar attributes. In this article, we used the joint correspondence analysis (JCA) method of MCA (Greenacre & Blasius, 2006).

Two sets of variables were entered into the JCA. The set of active variables used to calculate the modalities and coordinates in each dimension were gender, cohort,⁶ the dummy for living with parents, the dummy for living with dependent children, and de facto marital status. These were selected to capture the demographic characteristics of respondents which the multinomial model and previous studies suggested were particularly influential predictors of partnership status or types of LAT relationship (Régnier-Loilier et al., 2009; Reimondos et al., 2011). We experimented with various other combinations of variables (e.g., urban residence and ethnicity), but as these made little contribution to the dimensions in the JCA, we excluded them from our analysis. In addition to the active variables, we included a series of illustrative variables which are not used in the calculations of modalities or dimensional coordinates. These “round out” the observations by fitting into the divisions defined by the active variables.

The JCA results and Ward’s method of cluster analysis were then used to group respondents. We identified four exhaustive groups that stood out from the first two dimensions of the JCA, of which the first dimension explained 74.9% and the second 19.8% of the inertia. To address the first objective, we descriptively compared these four groups in order to characterize different forms of LAT and how they relate to life course circumstances.

Our second objective was to analyze how the predictors of cohabitation intentions vary by life course position. To do this we fitted four logistic regression models predicting cohabitation intentions: one for each of the LAT profiles identified by the JCA and cluster analysis. The dependent variable in these models was constructed using a question asking: Do you intend to start living with your current partner during the next 3 years? Although there will inevitably be some uncertainty and ambiguity in people’s responses, this question allows us to broadly distinguish those who do and do not plan to live with their partner in the near future.

We included all illustrative variables as predictors in the logistic regression models. These capture characteristics of the life course and the relationship that might be associated with cohabitation intentions. These attributes include the respondent’s caring duties, their partner’s labor force status, the travel time between partners’ residences, the frequency of meeting each other, the duration of the relationship, and whether the respondent reported having deliberately “decided” to live apart. As the active variables were used to construct the LAT typologies, they are excluded from the logistic regression models (with the exception of age and gender).

Results

Objective 1: Categorizing LAT Over the Life Course

In line with previous estimates (Duncan et al., 2013), around 9 % of the adults in the full sample report LAT. Men are slightly less likely to report LAT (8.7 %) than women (9.2 %). The appendix shows LAT is concentrated early in the life course after which it becomes a minority practice as most people enter cohabitation and (then) marriage. Individuals LAT are disproportionately likely to be living with their parents, to be in work or full-time education, and to be living in rented accommodation (see the appendix).

Table 1 disentangles the life course factors associated with LAT using a multinomial model of partnership status where the reference category is LAT. The pseudo-r² and reduction in log likelihood indicate that the model fits well. As anticipated, cohort is a strong predictor of partnership status. Older cohorts are more likely than those aged 16 to 24 years to be married or previously married relative to LAT. The odds of cohabiting rather than LAT are higher in the 25 to 39 years cohorts than for those aged 16 to 24 years. Interestingly, the over-70s are less likely than the youngest cohort to be cohabiting relative to LAT. This fits with evidence that people tend to be reluctant to enter coresidential unions later in life (De Jong Gierveld, 2004). Studying later life trends in (re)partnership by analyzing de jure marital status or de facto household structures may therefore underestimate the proportion of older people in intimate relationships.

Table 1.

Multinomial Logistic Regression Model of Partnership Status (ref = LAT).

	Married		Cohabiting		Previously married		Never married
	β	SE	β	SE	β	SE	β	SE
Cohort (years; ref = 16-24)
25-29	1.80	0.15***	0.62	0.12***	3.15	0.75***	0.22	0.10*
30-39	2.57	0.15***	0.56	0.11***	4.70	0.72***	0.47	0.10***
40-49	2.77	0.15***	0.06	0.11	5.85	0.72***	0.15	0.10
50-59	3.48	0.15***	−0.04	0.12	6.73	0.72***	−0.10	0.11
60-69	3.98	0.16***	−0.13	0.15	7.46	0.72***	−0.02	0.14
70+	3.66	0.18***	−1.14	0.20***	8.21	0.72***	−0.16	0.17
Sex (ref = male)	0.00	0.07	0.00	0.08	1.13	0.08***	−0.16	0.08*
Lives with dependent child(ren) (ref = no)	3.89	0.24***	3.08	0.24***	2.18	0.30***	−0.96	0.39*
Female × lives with dependent child(ren)	−2.95	0.25***	−2.94	0.26***	−1.81	0.31***	0.51	0.40
Lives with parent (ref = no)	−3.11	0.17***	−3.60	0.22***	−0.34	0.23	0.34	0.10***
Female × lives with parents	0.47	0.22*	0.32	0.28	−0.74	0.30*	−0.23	0.11*
Ethnic minority (ref = no)	0.31	0.13*	−0.46	0.16**	0.38	0.17*	0.59	0.09***
Foreign-born (ref = U.K.-born)	0.09	0.11	−0.12	0.13	0.15	0.14	−0.12	0.13
Ethnic minority × foreign-born	0.48	0.21*	−0.63	0.27*	0.04	0.25	−0.12	0.19
Education (ref = none)
GCSE/A-level (high school)	−0.22	0.08**	−0.23	0.09*	−0.29	0.09***	−0.21	0.09*
Higher degree (college and above)	−0.24	0.08**	−0.38	0.10***	−0.39	0.09***	−0.21	0.09*
Self-reported economic activity (ref = FT employed)
PT employed	0.30	0.07***	0.11	0.09	0.08	0.10	0.38	0.08***
Unemployed	−0.59	0.12***	0.01	0.12	0.21	0.14	0.69	0.10***
FT student	−0.71	0.20***	−0.66	0.17***	−0.42	0.41	0.66	0.09***
Inactive	0.44	0.08***	0.20	0.10*	0.52	0.10***	0.73	0.09***
Housing tenure (ref = homeowner)
Social renter	−1.53	0.07***	−0.74	0.08***	−0.07	0.08	0.05	0.07
Private renter	−1.12	0.08***	−0.35	0.08***	0.11	0.10	0.25	0.08**
Urban residence (ref = rural)	−0.42	0.06***	−0.22	0.07**	−0.10	0.07	0.11	0.07
Constant	−0.36	0.17*	1.11	0.16***	−6.47	0.73***	0.10	0.15
McFadden’s pseudo-r²		.291
Log likelihood (LL null model)		−35131.3 (−49512.4)
N		37,277

Note. FT = full-time; PT = part-time; SE = standard error; GCSE = General Certificate of Secondary Education.

p < .05. **p < .01. ***p < .001.

As women are more likely than men to be LAT with dependent children (see the appendix) and as the timing and meaning of leaving home may differ by gender, we included interaction terms between gender and the dummies for living with dependent children and living with (a) parent(s). These parameters show that men with children are more likely than those without children to be married, cohabiting, or previously married relative to LAT. The significant interaction between gender and children shows that this effect is attenuated for women. Living with parents is associated with lower odds of marriage and cohabitation as compared with LAT, in the case of marriage somewhat more strongly among men than women. However, in general, coresidential partnership is closely tied to independent living.

The ethnicity and place of birth parameters suggest that cultural and contextual factors intersect to condition LAT. British-born ethnic minorities are somewhat more likely to be currently, never, or previously married and less likely to cohabit relative to LAT than their White peers. As these effects are conditional on age, they imply that cultural factors may lead ethnic minorities to eschew cohabitation in favor of LAT, while also encouraging marriage and staying single. Importantly, the significant interaction of foreign birth with ethnicity indicates that the effects of ethnicity on marriage and cohabitation are stronger for minorities born outside the United Kingdom. This could be due to cultural orientations or because immigration policy privileges the “family reunification” of married couples.

Although the descriptive statistics show a weak association between educational qualifications and LAT (see the appendix), Table 1 shows that people with higher degrees, and to a lesser extent school-level qualifications, have significantly lower odds of being in all partnership statuses relative to LAT than those with no qualifications. The high propensity of well-qualified persons to be living apart from a partner may be a way to cope with the spatial dispersion of highly skilled work, or it may be a consequence of the acquisition and performance of particular lifestyle and partnership preferences. Educational participation is associated with lower odds of cohabitation and marriage than LAT when compared with people in full-time employment. Similar effects are visible for unemployment (marriage only), while full-time employees are less likely to be never married relative to LAT than those in other labor force states. Inactive individuals also have a low propensity to be LAT.

The housing tenure parameters indicate that the odds of being married or cohabiting relative to LAT are lower for renters than homeowners. This could be a selection effect caused by better-off individuals selecting into coresidential unions and then homeownership. Interestingly, the odds of being married or cohabiting relative to LAT are lower in urban than rural areas. This could be because LAT is more difficult in rural areas (Jamieson & Simpson, 2013), although it is also possible that individuals looking to “settle down” with a partner have previously chosen to leave urban areas for a more rural environment.

The large number of LAT relationships observed by UKHLS offers a unique opportunity to investigate whether we can discern distinct profiles of individuals who are LAT in different ways under particular life course circumstances. To do this, Table 2 presents the results of the MCA conducted using the LAT sample of 3,112 individuals. The MCA classified individuals in LAT relationships into four sizeable and well-defined groups that closely resemble the clusters derived from French and Australian data by Régnier-Loilier et al. (2009) and Reimondos et al. (2011). This indicates that LAT is practiced at similar junctures of the life course across Western countries with different welfare regimes, labor, and housing markets.

Table 2.

A Fourfold Typology of LAT.

	Nested young adults (NYAs)	Independent adults (INAs)	Single parents (SPTs)	Seniors (SENs)	All
Variable
N (unweighted)	1,158	1,146	384	424	3,112
Percentage	44.1	32.1	11.1	12.7	100.0
Active variables (column %)
Sex
Male	50.4	48.5	2.0	70.6	47.0
Female	49.6	51.5	98.0	29.5	53.0
Cohort (years)
16-19	43.0	3.0	1.2	0.0	20.1
20-24	42.2	12.8	14.6	0.0	24.4
25-29	14.8	14.1	16.0	0.0	12.8
30-39	0.0	24.7	34.0	0.2	11.7
40-49	0.0	29.4	33.6	1.6	13.4
50-59	0.0	16.0	0.7	34.7	9.6
60+	0.0	0.0	0.0	63.4	8.1
Lives with parents
No	0.0	86.4	96.7	98.0	50.9
Yes	100.0	13.6	3.3	2.0	49.1
Lives with dependent children
No	100.0	94.7	0.0	99.8	87.2
Yes	0.0	5.3	100.0	0.2	12.8
De facto marital status
Never married	100.0	70.4	62.3	21.3	76.3
Separated	0.0	6.5	11.5	3.7	3.8
Divorced	0.0	22.5	26.2	43.5	15.6
Widowed	0.0	0.6	0.0	31.5	4.2
Illustrative variables (column %)
1. Individual characteristics
Education
No qualification	5.0	11.8	14.4	37.3	12.3
GCSE/A-level (high school)	73.3	45.8	58.0	33.3	57.7
Higher degree (college and above)	21.7	42.5	27.6	29.4	30.0
Self-reported economic activity
FT employed	43.5	66.7	27.4	28.4	47.3
PT employed	21.7	12.1	30.9	10.0	18.1
Unemployed	8.2	9.0	12.3	3.6	8.3
Inactive	26.7	12.2	29.5	58.0	26.3
Housing tenure
Homeowner	69.9	49.2	29.2	70.1	58.8
Social renter	22.5	21.3	44.7	20.3	24.3
Private renter	7.7	29.5	26.1	9.6	17.0
Care provision within household
No	95.4	95.5	92.0	98.1	95.4
Yes	4.6	4.5	8.0	1.9	4.6
Care provision outside household
No	94.6	86.6	88.2	82.1	89.7
Yes	5.4	13.5	11.8	17.9	10.3
Intend to stay in neighborhood
No	65.9	52.9	45.8	27.4	54.6
Yes	34.1	47.1	54.2	72.6	45.4
2. Relationship characteristics (column %)
Partner’s economic activity
Employed	55.4	76.9	81.5	44.1	63.8
Unemployed	6.9	5.5	12.7	3.6	6.7
FT education	32.6	8.1	0.2	0.7	17.1
Inactive	5.0	9.5	5.6	51.6	12.4
Travel time to partner
<30 Minutes	68.6	54.3	65.7	63.6	63.1
30 Minutes to 1 hour	14.5	21.6	19.0	14.5	17.3
1 Hour to 2 hours	7.6	10.6	9.8	9.3	9.0
>2 Hours	9.4	13.5	5.5	12.6	10.7
Frequency of seeing partner
Daily	36.0	25.2	32.3	21.4	30.3
Several times a week	40.1	42.6	43.7	40.6	41.4
At least once a week	12.3	16.0	15.6	21.1	14.9
Several times a month or less	11.6	16.2	8.4	16.9	13.4
Duration of LAT relationship
<1 Year	23.6	15.7	23.7	8.5	19.1
1-2 Years	32.4	27.4	26.8	11.0	27.4
2-3 Year	28.8	23.6	20.8	16.0	24.6
>3 Years	15.3	33.3	28.7	64.6	28.8
Definite choice not to live together
No	68.0	61.4	47.0	39.7	60.0
Yes	32.0	38.6	53.0	60.3	40.0
3. Intention to cohabit in 3 years
No	30.0	34.7	36.3	75.5	38.0
Yes	70.0	65.3	63.7	24.5	62.0

Note. LAT = living apart together; FT = full-time; PT = part-time; GCSE = General Certificate of Secondary Education. Column percentages may not sum to 100 due to rounding.

The first profile (44% of individuals) epitomizes early life “dating” or “going steady” relationships. We call them “nested young adults” (NYAs) as all are younger than 30 years and living in the parental home. The second profile (32 %), we label “independent adults” (INAs) as they have an older age profile and rarely live with a parent. Unlike NYAs, in a minority of cases INAs live with children and nearly 30% have previously married. Although other researchers have discerned a gender-balanced (Reimondos et al., 2011) or female-dominated (Régnier-Loilier et al., 2009) cluster of single parents, our profile of “single parents” (SPTs, 11% of individuals) is fewer than 2% male. This is not surprising as 90% of lone parent households in England and Wales are headed by single mothers (Office for National Statistics, 2013). SPTs are clustered in midlife, rarely live with parents, and more have never been married than have experienced widowhood or marital dissolution. Our final group of “seniors” (SENs, 13% of individuals) are so named as almost all are older than 50 years and most have been previously married. Few seniors are living with parents or dependent children and there is a preponderance of men in this group. Taken together, these results show how LAT is practiced in a wide range of life course circumstances.

The lower panels of Table 2 compare the attributes of individuals in the four LAT profiles, subdividing these into individual and relationship characteristics. The first rows show that INAs are the most highly qualified, while seniors tend to have fewer educational qualifications. The majority of seniors are inactive (retired), while inactivity rates are also high among SPTs and NYAs (many of the latter are in education). Interestingly, more than 50% of NYAs are employed, implying that LAT is a way for young people to maintain intimate relationships during the increasingly difficult and turbulent transition to residential independence (Castro-Martín et al., 2008). The economic context and social distribution of resources may thus be as much of an influence on LAT as more frequently discussed changes in attitudes, roles, values, and demography (Upton-Davis, 2012). By contrast, most INAs are in full-time work, while SPTs have the greatest propensity to be in part-time work or unemployment. The socioeconomic disadvantage of SPTs is further reflected in their reliance on privately rented and especially subsidized socially rented housing. Due to their younger profile, INAs are more likely to be private tenants and less likely to be homeowners than SENs, while NYAs have high homeownership rates as they are living with their parents.

The four groups also have distinctive behaviors and intentions. Although Levin (2004) argues that care provision is a powerful motive for LAT, the extent of intrahousehold and interhousehold care exchanges varies across the profiles. SPTs are the most likely to provide care within the household, while SENs are the most likely to care for others outside the household. As anticipated from the age profiles, NYAs are the most likely to intend to move home in the future, while SENs typically intend to stay. Moving intentions are also quite common among INAs.

To examine whether people practice LAT in distinct ways under different life course circumstances, the lower panel of Table 2 compares a range of partnership attributes and practices across the four groups.⁷ Although the labor force status of NYAs, INAs, and SENs resembles those of their partners, SPTs (only 58.3% of whom are employed) tend to have employed partners. This suggests that LAT relationships may be a valuable but frequently “hidden” source of social support for economically vulnerable single mothers.

Long-distance relationships comprise a small minority of LAT partnerships as most partners live less than 30 minutes apart (Duncan et al., 2014). Furthermore, the frequency of meeting seems fairly consistent across the life course, with SENs slightly more likely than the other groups to see their partner infrequently. SENs are also the most likely to be in long-term relationships or LAT out of choice, while SPTs and especially NYAs have shorter relationship durations. By contrast, NYAs and INAs are the least likely to report having chosen LAT.

Taken together, these results suggest two preliminary conclusions. First, the balance of motives for LAT seems to broadly shift from “constraints” toward “choice” over the life course. Second, there are few grounds to dismiss or discard young adult relationships (over 75 % of which have lasted over a year) as not “proper LAT.” Instead, it is theoretically and empirically plausible to think of LAT as a practice used in different ways across the life course in response to intersecting biographical, “linked life,” and contextual factors.

Objective 2: Cohabitation Intentions

Whether people intend to convert LAT into a coresidential union is not only important for predicting behavior but also provides insights into how LAT can have heterogeneous meanings and roles across the life course. In our sample 62% of LAT individuals intend to cohabit with their current partner in the next 3 years. This closely resembles estimates derived using Australian (64%) and French (70%) data (Régnier-Loilier et al., 2009; Reimondos et al., 2011). In line with these studies, cohabitation intentions are most common among NYAs (70%) followed by INAs (65%), SPTs (64%), and finally SENs (25%).These findings reinforce the preliminary conclusion that LAT is a heterogeneous practice which typically shifts over the life course from being perceived as transitional toward being seen as a durable arrangement. The results also provide further evidence that LAT is practiced in a similar way across a range of Western contexts.

To assess whether the factors affecting cohabitation intentions vary over the life course, Table 3 presents the results of four logistic regression models predicting the cohabitation intentions of each of the four LAT groups using the illustrative variables shown in Table 2. Each of the models fits well although possible differences in unobserved heterogeneity preclude direct comparison of effect size across models (Mood, 2010). We therefore focus on interpreting the general direction and statistical significance of effects.

Table 3.

Logistic Regression Models of Intending to Convert LAT Into Cohabitation Within 3 Years (0 = No, 1 = Yes).

	Nested young adults (NYAs)		Independent adults (INAs)		Single parents (SPTs)		Seniors (SENs)
Variable	β	SE	β	SE	β	SE	β	SE
Age	0.11	0.04**	−0.07	0.01***	−0.07	0.02***	−0.01	0.02
Sex (ref = male)	−0.00	0.18	0.12	0.18	0.39	1.12	−0.20	0.35
Education (ref = no qualification)
GCSE/A-level (high school)	0.34	0.40	0.13	0.28	−0.16	0.39	0.67	0.40
Higher degree (college and above)	0.64	0.44	0.14	0.29	−0.05	0.42	0.30	0.40
Self-reported economic activity (ref = FT employed)
PT employed	−0.57	0.23*	−0.27	0.30	0.09	0.33	0.88	0.55
Unemployed	−0.54	0.34	−0.38	0.37	0.07	0.51	0.31	1.18
Inactive	−0.54	0.24*	−0.57	0.28*	0.40	0.43	0.13	0.40
Housing tenure (ref = homeowner)
Social renter	0.50	0.21*	−0.07	0.23	−0.35	0.38	−0.59	0.46
Private renter	0.09	0.32	−0.22	0.22	0.46	0.39	−0.77	0.61
Care provision within household (ref = no)	0.18	0.38	−0.44	0.43	−0.44	0.48	−3.86	1.52*
Care provision outside household (ref = no)	−0.13	0.34	0.54	0.24*	−0.67	0.44	−0.11	0.39
Intends to stay in neighborhood (ref = no)	−0.80	0.17***	−0.70	0.18***	−0.73	0.26**	−1.45	0.34***
Partner’s economic activity (ref = employed)
Unemployed	0.14	0.33	−0.33	0.37	0.58	0.46	−0.57	0.76
Inactive	−0.49	0.19**	−1.02	0.26***	0.53	0.55	−0.50	0.35
Travel time to partner (ref < 30 minutes)
30 Minutes to 1 hour	−0.24	0.24	0.24	0.22	0.43	0.38	1.16	0.41**
1 Hour to 2 hours	0.88	0.34**	0.99	0.30***	0.65	0.57	0.87	0.50
>2 Hours	1.11	0.40**	2.08	0.39***	0.99	0.64	2.37	0.61***
Frequency of seeing partner (ref = daily)
Several times a week	−0.45	0.20*	−0.80	0.23***	−0.45	0.33	−0.84	0.39*
At least once a week	−1.15	0.29***	−1.39	0.31***	−1.35	0.46**	−2.50	0.57***
Several times a month or less	−1.45	0.34***	−1.69	0.37***	−1.25	0.64*	−2.24	0.62***
Relationship duration (ref = less than 1 year)
1-2 Years	0.63	0.21**	−0.08	0.26	0.40	0.40	−0.93	0.55
2-3 Years	0.76	0.22***	0.08	0.26	0.49	0.41	−1.31	0.53*
> 3 Years	1.50	0.34***	−0.00	0.25	0.14	0.39	−1.78	0.44***
Decided to LAT (ref = no)	−0.43	0.17*	−0.97	0.17***	−0.99	0.28***	−1.28	0.29***
Constant	−1.07	0.98	4.56	0.61***	3.45	2.48	3.18	1.41*
McFadden’s pseudo-r²	.163		.206		.172		.273
Log likelihood (LL null)	−717.5 (−857.5)		−522.6 (−658.4)		−190.8 (−230.4)		−163.6 (−225.1)
N (unweighted)	1,158		1,146		384		424

Note. LAT = living apart together; FT = full-time; PT = part-time; GCSE = General Certificate of Secondary Education.

p < .05. **p < .01. ***p < .001.

A range of biographical attributes affect the cohabitation intentions of individuals LAT. Age increases the odds of intending to cohabit among NYAs and reduces it for INAs and SPTs. Gender has no significant links to cohabitation intentions and further tests rejected the inclusion of any gendered interactions (results not shown). Although educational qualifications are associated with LAT, they have no significant links with cohabitation intentions. Part-time employment and inactivity are associated with a lower propensity to intend to cohabit among NYAs, while inactivity is negative and significant for INAs. This may be because LAT allows people to combine partnership with the spatial mobility common during tertiary education. The only significant tenure parameter indicates that NYAs living in socially rented dwellings are slightly more likely to intend to cohabit than those living in owner-occupied housing. This may indicate that children from less affluent families anticipate leaving home or becoming economically independent earlier than those from more privileged backgrounds.

In support of Levin (2004) and Duncan’s (2015) notion of “obligated preferences,” “linked life” factors in the form of caring for household members reduce the odds of SENs intending to cohabit. Caring duties can, however, have the opposite effect elsewhere in the life course as INAs have higher odds of intending to cohabit with their partner if they are caring for someone who does not live in their household. Intending to stay in the current neighborhood is universally associated with lower odds of intending to convert LAT into a coresidential arrangement. This indicates that a variety of possible factors, such as one’s attachment to neighbors and the local area, can compete with one’s willingness to cohabit and one’s commitment to intimate relationships. For some individuals, LAT could be a compromise between these competing priorities.

The lower half of Table 3 shows that relationship attributes are, in general, strongly associated with cohabitation intentions. Having an inactive rather than employed partner is associated with lower odds of intending to cohabit among NYAs and INAs, perhaps because labor force participation is important for obtaining the economic resources necessary to move in together. Except for SPTs, the odds of intending to cohabit rise with greater distances between partners. This suggests that couples are willing to maintain long-distance LAT relationships only when they perceive them to be temporary arrangements. By contrast, individuals are less likely to intend to cohabit with their partner if they meet them infrequently. In part, the frequency of meeting seems to act as a proxy for commitment to the relationship. Although it is unsurprising that people are less likely to intend to cohabit if they perceive having made a deliberate choice to live apart, the association between union duration and cohabitation intentions varies strongly by life course position. Longer NYA relationships are associated with a greater intention to cohabit, while the opposite is true for SENs. This indicates that LAT is a flexible practice which tends to only be seen as a durable arrangement later in life.

Discussion and Conclusions

Scholars are becoming increasingly interested in how and why people live apart together and what this tells us about intimate relationships in the 21st century. Existing research has richly documented the diversity, ambiguity, and complexity of LAT across a range of Western countries (Duncan et al., 2014; Funk & Kobayashi, 2016; Levin, 2004; Lyssens-Danneboom & Mortelmans, 2015; Stoilova et al., 2014; Upton-Davis, 2015). However, although most studies agree that a little under 10% of adults are LAT (Duncan et al., 2013), little is known about whether people typically live apart together in similar ways under particular constellations of life course circumstances (Régnier-Loilier et al., 2009; Reimondos et al., 2011). In addition, few studies have systematically examined how intentions to convert LAT into cohabitation are configured by the biographical, “linked life,” and contextual dimensions of life course trajectories. These lacunae leave policy makers poorly informed about how LAT might affect household projections and have legal and administrative implications for family law and family policy (Duncan et al., 2012). This article therefore sought to reexamine LAT using a life course framework and rich data gathered from an unprecedentedly large sample of Britons.

The results suggest several conclusions. On a basic level, our findings support the notion that LAT is both heterogeneous and relatively ambiguous (Duncan et al., 2014; Duncan & Phillips, 2011; Jamieson & Simpson, 2013; Roseneil, 2006). Young adults, individuals living with their parents, the highly educated and those in full-time education are disproportionately likely to be LAT rather than living with a partner. While few men are LAT when living with dependent children, roughly 11% of individuals in LAT partnerships are single mothers. LAT is also common later in life, often following a previous partnership. Importantly, cohabitation intentions and relationship attributes such as whether LAT is perceived as a “choice” and the length of the partnership vary considerably with life course position.

Taken together, these findings provide support for conceptualizing LAT less as a family form (Levin, 2004) and more as a practice used flexibly to combine intimacy with other demands of life within the context of life course constraints (De Jong Gierveld & Merz, 2013; Duncan et al., 2013; Stoilova et al., 2014). This implies that it may be futile to try to define a priori what constitutes a “true” LAT union and how this differs from a dating, casual, or fleeting relationship. Given that conventional ideas of commitment and fidelity are highly valued by the vast majority of people reporting themselves to be LAT (Duncan et al., 2014), a more fruitful approach may be to let people subjectively define whether they are LAT and then examine the full spectrum of their relationship practices (UNECE, 2011). Conceptualizing LAT as a practice thus turns an intractable definitional issue into a set of empirical questions about how and why people form and maintain relationships across households in different ways.

Embedding LAT within a life course framework and analyzing cohabitation intentions provides several insights about LAT in Great Britain. First, we find that Britons LAT have four well-defined life course profiles distinguished by age, gender, living arrangements, and partnership history. This indicates that we need to take care not to conflate “age with stage” and that researchers should examine LAT over the entire life span as well as honing in on specific life phases (Karlsson & Borell, 2002; Upton-Davis, 2015). This is particularly crucial because the four profiles reveal that expectations of LAT vary with life course position. While young adults living with their parents tend to intend to convert LAT into cohabitation, seniors typically live apart for longer periods with no intention to cohabit. Furthermore, the balance of reasons for LAT tends to shift from “constraint” early in life toward LAT being a more deliberate “choice” as the life course progresses, although constraints and choices are often ambiguously defined and interwoven at all life stages. As the size and composition of the four profiles closely resembles those described for other Western countries (Régnier-Loilier et al., 2009; Reimondos et al., 2011), it appears that LAT may be used in a similar fashion under broadly similar life course circumstances regardless of national specificities.

While it is tempting to interpret cross-national similarity as evidence that societal affluence and the “individualization” of values are reconfiguring intimate relationships, our findings demonstrate that LAT is profoundly shaped by the relationality of “linked lives.” While single parents LAT appear disadvantaged in terms of employment and housing, a surprisingly large proportion have employed partners who could be providing hidden forms of practical, financial, or emotional support (Duncan et al., 2012). Consistent with the view that caring and the perceived needs of dependents can bound individual agency (De Jong Gierveld & Merz, 2013), caring for people within and outside the household has different effects on the cohabitation intentions of individuals in LAT relationships at different life stages. Caring for someone outside the household is linked to a higher likelihood of intending to cohabit among independent adults. In contrast, seniors are less likely to intend to move in with a partner when caring for someone with whom they live. When taken with evidence that relationship duration has the opposite effect on the cohabitation intentions of nested young adults and seniors, these results highlight how the determinants of anticipated pathways out of LAT vary with life course position.

A final contribution of the life course approach lies in highlighting how biographical and contextual factors intersect to condition LAT. The interactions between ethnicity and country of birth suggest that the relative odds of LAT may be conditioned by the differing cultural orientations and values of native Whites, ethnic minority immigrants, and native-born minorities. Given high levels of immigration and the growth of ethnic minority populations, this finding should be developed through qualitative analysis of the processes linking ethnicity and migration to partnership practices. Furthermore, the results show that LAT is especially common in young adulthood but a clear majority of nested young adults also intend to cohabit. This indicates that the increasingly lengthy and reversible transition to adulthood may be as much an influence on the prevalence of LAT as changing social values (Castro-Martín et al., 2008). While perceived vulnerability can lead individuals to opt for LAT rather than cohabitation (Roseneil, 2006), economic precarity can also constrain partnership options and thus generate intergenerational and intragenerational inequality in LAT. Future research could profitably consider how the restructuring of labor/housing markets and the welfare state are (re)configuring LAT.

Our study poses several methodological challenges for future research. Conceptualizing LAT as a flexible practice indicates that surveys need to gather richer data about noncoresidential relationships by asking a wide range of questions about how people conduct and experience relationships with nonresident individuals (UNECE, 2011). Analysts can then, if desired, use these answers to subdivide intimate relationships without being bound by prescriptive definitions of LAT. Fortunately, a growing number of surveys, such as the Gender and Generations Programme, are adopting this approach (UNECE, 2011).

A second challenge is to develop the life course approach by collecting and analyzing longitudinal data capable of linking intentions to subsequent experiences by capturing the timing, ordering, and duration of relationships. At present, this is difficult as many longitudinal surveys only infrequently ask respondents about LAT. As LAT relationships are often quite short (Ermisch & Siedler, 2009), future longitudinal research may require innovative primary data collection strategies where both partners are tracked and contacted more frequently than once per year. Such data could provide information about how partnership biographies are dynamic and shaped by prospective intentions as well as retrospective experiences.

Finally, collecting data from both partners is essential for understanding more about the relational dimension of LAT. Dyadic information would allow us to analyze how relative resources and (dis)agreements in norms, values, preferences, intentions, and attitudes interact to configure subsequent cohabitation.⁸ This could yield rich insights about gendered relations within LAT partnerships, as well as how power and resource imbalances configure current and subsequent relationship practices. While we are developing a good understanding of why people live apart together, the challenge is to develop a more dynamic and relational perspective capable of assessing how people practice LAT in different ways as their life courses unfold.

Footnotes

Appendix

Sample Characteristics.

Variable (column proportions)	Male						Female
Variable (column proportions)	All	Married	Cohabiting	Previously married	Never married	LAT	All	Married	Cohabiting	Previously married	Never married	LAT
Cohort (years)
16-24	.13	.00	.08	.00	.48	.42	.12	.01	.12	.00	.43	.43
25-29	.07	.02	.16	.00	.13	.14	.07	.04	.19	.01	.12	.11
30-39	.15	.15	.30	.04	.12	.11	.16	.18	.27	.03	.16	.13
40-49	.18	.21	.22	.11	.10	.11	.19	.23	.21	.12	.13	.16
50-59	.17	.21	.14	.18	.08	.11	.16	.21	.12	.17	.07	.10
60-69	.15	.21	.07	.23	.06	.06	.15	.20	.06	.21	.04	.04
70+	.15	.19	.03	.44	.04	.06	.15	.13	.02	.46	.05	.03
Lives with dependent child(ren)	.23	.32	.38	.04	.01	.02	.28	.34	.42	.10	.20	.23
Lives with parent(s)	.16	.01	.02	.05	.59	.51	.12	.01	.03	.02	.46	.43
Ethnic minority	.08	.08	.03	.04	.14	.09	.08	.08	.03	.05	.15	.08
Foreign-born	.11	.12	.08	.08	.11	.12	.11	.12	.10	.09	.12	.09
Level of education
No/other qualification	.22	.24	.17	.43	.17	.13	.26	.25	.15	.50	.15	.12
GCSE/A-level	.44	.38	.50	.34	.58	.56	.40	.36	.44	.27	.54	.56
Higher degree	.34	.39	.33	.23	.25	.30	.34	.39	.40	.23	.31	.31
Self-reported labor force status
Full-time employed	.52	.56	.71	.23	.38	.53	.31	.31	.45	.17	.31	.41
Part-time employed	.08	.07	.06	.06	.13	.11	.22	.26	.23	.11	.20	.24
Unemployed	.06	.03	.10	.05	.14	.09	.04	.02	.06	.03	.11	.08
Full-time education	.05	.00	.01	.00	.19	.14	.04	.00	.02	.00	.17	.11
Other/inactive	.28	.34	.11	.65	.15	.13	.39	.41	.24	.69	.21	.16
Housing tenure
Homeowner	.71	.81	.57	.61	.58	.62	.69	.82	.55	.61	.51	.56
Social renter	.16	.10	.20	.28	.24	.21	.19	.10	.21	.30	.30	.27
Private renter	.13	.08	.23	.11	.18	.17	.12	.08	.24	.09	.19	.17
Urban area	.77	.74	.80	.77	.83	.79	.77	.73	.79	.78	.84	.82
Unweighted N	16,436	9,129	2,079	1,121	2,673	1,434	20,841	10,351	2,435	3,224	2,907	1,924

Note. GCSE = General Certificate of Secondary Education. Proportions may not sum to 1 due to rounding.

Acknowledgements

The authors would like to thank Maria Iacovou, Nitzan Peri-Rotem, and two anonymous reviewers for their insightful comments on earlier versions of this article.

Authors’ Note

The United Kingdom Household Longitudinal Study (UKHLS) is conducted by the Institute for Social and Economic Research at the University of Essex. UKHLS data were accessed via the U.K. Data Service. Neither the original collectors of the data nor the U.K. Data Service bear any responsibility for the analyses or interpretations presented in this study.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Rory Coulter’s contribution was partially supported by an Economic and Social Research Council Future Research Leaders award [ES/L009498/1]. Additional support was provided by the Isaac Newton Trust. Yang Hu gratefully acknowledges the Gates Cambridge Trust for funding his research.

Notes

References

Buck

McFall

(2012). Understanding society: Design overview. Longitudinal and Life Course Studies, 3, 5-17.

Castro-Martín

Domínguez-Folgueras

Martín-García

(2008). Not truly partnerless: Non-residential partnerships and retreat from marriage in Spain. Demographic Research, 18, 443-468.

De Jong Gierveld

. (2004). Remarriage, unmarried cohabitation, living apart together: Partner relationships following bereavement or divorce. Journal of Marriage and Family, 66, 236-243.

De Jong Gierveld

Merz

. (2013). Parents’ partnership decision making after divorce or widowhood: The role of (step)children. Journal of Marriage and Family, 75, 1098-1113.

Duncan

(2015). Women’s agency in living apart together: Constraint, strategy and vulnerability. Sociological Review, 63, 589-607. doi:10.1111/1467-954X.12184

Duncan

Carter

Phillips

Roseneil

Stoilova

(2012). Legal rights for people who “Live Apart Together”? Journal of Social Welfare and Family Law, 34, 443-458.

Duncan

Carter

Phillips

Roseneil

Stoilova

(2013). Why do people live apart together? Families, Relationships and Societies, 2, 323-338.

Duncan

Phillips

(2011). People who live apart together (LATs): New family form or just a stage? International Review of Sociology, 21, 513-532.

Duncan

Phillips

Carter

Roseneil

Stoilova

(2014). Practices and perceptions of living apart together. Family Science, 5, 1-10.

10.

Elder

G. H.

Johnson

M. K.

Crosnoe

(2003). The emergence and development of life course theory. In Mortimer

J. T.

Shanahan

M. J.

(Eds.), Handbook of the life course (pp. 3-19). New York, NY: Kluwer Academic/Plenum.

11.

Ermisch

Siedler

(2009). Living apart together. In Brynin

Ermisch

(Eds.), Changing relationships (pp. 29-43). Abingdon, England: Routledge.

12.

Funk

L. M.

Kobayashi

K. M.

(2016). From motivations to accounts: An interpretive analysis of “Living Apart Together” relationships in mid- to later-life couples. Journal of Family Issues, 37(8): 1101–1122

13.

Greenacre

Blasius

(2006). Multiple correspondence analysis and related methods. London, England: CRC Press.

14.

Haskey

(2005). Living arrangements in contemporary Britain : Having a partner who usually lives elsewhere and Living Apart Together (LAT). Population Trends, 122, 35-45.

15.

Haskey

Lewis

(2006). Living-apart-together in Britain: Context and meaning. International Journal of Law in Context, 2, 37-48.

16.

Irwin

(2009). Locating where the action is: Quantitative and qualitative lenses on families, schooling and structures of social inequality. Sociology, 43, 1123-1140.

17.

Jamieson

Simpson

(2013). Living alone: Globalization, identity and belonging. Basingstoke, England: Palgrave Macmillan.

18.

Karlsson

S. G.

Borell

(2002). Intimacy and autonomy, gender and ageing: Living apart together. Ageing International, 27, 11-26.

19.

Levin

(2004). Living apart together: A new family form. Current Sociology, 52, 223-240.

20.

Long

J. S.

Freese

(2006). Regression models for categorical dependent variables using Stata (2nd ed.). College Station, TX: Stata Press.

21.

Lyssens-Danneboom

Mortelmans

(2015). Living apart together: Longing for the couple, enjoying being single. Family Science, 6, 11-22.

22.

McFall

S. L.

(Ed.). (2013). Understanding Society—UK Household Longitudinal Study: Wave 1-3, 2009-12, User Manual. Colchester, England: University of Essex.

23.

Mood

(2010). Logistic regression: Why we cannot do what we think we can do, and what we can do about it. European Sociological Review, 26, 67-82.

24.

Neustatter

(2013, April 22). Divided we stand: Committed couples who live apart. The Daily Telegraph. Retrieved from http://www.telegraph.co.uk/women/sex/relationship-advice-and-romance/10003381/Divided-we-stand-committed-couples-who-live-apart.html>

25.

Office for National Statistics. (2013). Families and households in England and Wales 2011. Newport, Wales.

26.

Régnier-Loilier

Beaujouan

É.

Villeneuve-Gokalp

(2009). Neither single, nor in a couple: A study of living apart together in France. Demographic Research, 21, 75-108.

27.

Reimondos

Evans

Gray

(2011). Living-apart-together (LAT) relationships in Australia. Family Matters, 87, 43-55.

28.

Roseneil

(2006). On not living with a partner: Unpicking coupledom and cohabitation. Sociological Research Online, 11, 1-14.

29.

Stoilova

Roseneil

Crowhurst

Hellesund

Santos

A. C.

(2014). Living apart relationships in contemporary Europe: Accounts of togetherness and apartness. Sociology, 48, 1075-1091.

30.

Stone

Berrington

Falkingham

(2014). Gender, turning points, and boomerangs: Returning home in young adulthood in Great Britain. Demography, 51, 257-276.

31.

Strohm

Seltzer

Cochran

Mays

(2009). “Living Apart Together” relationships in the United States. Demographic Research, 21, 177-214.

32.

Turcotte

(2013). Living apart together. Insights on Canadian Society. Retrieved from http://www.statcan.gc.ca/pub/75-006-x/75-006-x2013001-eng.htm

33.

Upton-Davis

(2012). Living apart together relationships (LAT): Severing intimacy from obligation. Gender Issues, 29, 25-38.

34.

Upton-Davis

(2015). Subverting gendered norms of cohabitation: Living Apart Together for women over 45. Journal of Gender Studies, 24, 104-116.

35.

United Nations Economic Council for Europe. (2011). Measurement of different emerging forms of households and families. Retrieved from http://www.unece.org/fileadmin/DAM/stats/publications/Families_and_Households_FINAL.pdf