Gender, Turning Points, and Boomerangs: Returning Home in Young Adulthood in Great Britain
- First Online:
- Cite this article as:
- Stone, J., Berrington, A. & Falkingham, J. Demography (2014) 51: 257. doi:10.1007/s13524-013-0247-8
- 2.7k Downloads
The idea of a generation of young adults “boomeranging” back to the parental home has gained widespread currency in the British popular press. However, there is little empirical research identifying either increasing rates of returning home or the factors associated with this trend. This article addresses this gap in the literature using data from a long-running household panel survey to examine the occurrence and determinants of returning to the parental home. We take advantage of the longitudinal design of the British Household Panel Survey (1991–2008) and situate returning home in the context of other life-course transitions. We demonstrate how turning points in an individual’s life course—such as leaving full-time education, unemployment, or partnership dissolution—are key determinants of returning home. An increasingly unpredictable labor market means that employment cannot be taken for granted following university graduation, and returning home upon completion of higher education is becoming normative. We also find that gender moderates the relationship among partnership dissolution, parenthood, and returning to the parental home, reflecting the differential welfare support in Great Britain for single parents compared with nonresident fathers and childless young adults.
KeywordsTransition to adulthoodTurning pointsYoung adultsLife courseReturning home
Recent evidence from cross-sectional data for Great Britain (Stone et al. 2011) suggests a marked increase in coresidence of parents and their young adult children, and the British media have promoted the idea of a generation of young adults “boomeranging” back to the parental home (Bingham 2009; Waite 2008). Returning to the parental home can have significant implications for overall parent/child relations, the risk of conflict between parents and children, opportunities for shared leisure time (Aquilino and Supple 1991), the negotiation of adult roles and identities, and the extent to which the young adult is able to interact with their parent(s) from a position of equality rather than dependence (Sassler et al. 2008).
However, there is virtually no recent empirical work for Great Britain identifying whether the rise in coresidence reflects increasing rates of returning home (as opposed to young adults delaying leaving home) and, if so, what the causes of such a trend might be. The relatively small academic literature on returning home tends to relate either to the North American experience or to the last decades of the previous century (Davanzo and Goldscheider 1990; Ermisch 1999; Gee et al. 1995; Goldscheider and Goldscheider 1998; Sassler et al. 2008; Wang and Morin 2009). Given recent changes in the socioeconomic and institutional contexts within which young adults are making their transitions to adulthood—such as increased higher education enrollment, declining housing affordability, and the continued deterioration and unpredictability of the youth labor market—a more up-to-date assessment is required (Aassve et al. 2002; Coles et al. 1999; Jones 1995; Sassler et al. 2008). Such changes intensify the risk of disruption in the normative life course trajectories of young adults, increasing the frequency with which they experience turning points in these trajectories (Elder 1978). In turn, these key life events—including uncertain destinations on leaving full-time education, becoming unemployed, and partnership dissolution—often increase the need for intergenerational support, prompting a potential return to the safety net of the parental home (Gee et al. 1995; Sassler et al. 2008; Swartz et al. 2011; Wang and Morin 2009).
These processes are likely to differ for men and women. While gender differences in coresidence with parents have been researched extensively across Europe (Chiuri and Del Boca 2010), less is known about gender differences in the pathways and turning points that lead young adults to boomerang back to the parental home. This is particularly relevant in Great Britain, given the relatively high rates of single motherhood and nonresident fatherhood as well as the distinctive situation of those coresident with dependent children, who receive preferential treatment in relation to housing benefit and social housing.1 This article addresses the gaps in the literature, using data from a long-running panel survey to examine how experiencing a turning point in the young adult life course affects the risk of returning home, as well as how this differs by gender.
The article makes both a novel theoretical and empirical contribution to the study of returning home. Although there is much theoretical discussion and empirical evidence regarding leaving the parental home in Great Britain and internationally, little is known about the factors that predict returning. We develop a new conceptualization as to how different turning points in the life course, such as losing a job or experiencing the break-up of a partnership, can result in returning home, highlighting how gender can moderate the effect of these turning points. Empirically, our approach is novel in that the prospective longitudinal design of the British Household Panel Study (BHPS), which follows up on households even when they split to form new ones, allows us to examine the factors predicting returning home from the perspective of multiple actors—the individual and his or her parent(s)—thus providing new insights into this process.
The study of young adulthood in Western countries often reflects discussion of individualization processes (Beck 1992; Giddens 1991), with the transition to adulthood becoming longer, more heterogeneous, and less defined by age norms (Furlong and Cartmel 2007; Furstenberg 2010; Gauthier 2007; Settersten 1998). Transitions from education to work, and then into partnership, parenthood, and residential independence are “late, protracted and complex” (Billari and Liefbroer 2010:60), although there are important differences between and within countries according to gender and social class (Aassve et al. 2002; Furlong and Cartmel 2007; Iacovou 2002). Attainment of adult roles is only part of the complex psychosocial experience of the transition to adulthood (Arnett 2004; Sassler et al. 2008). Nevertheless, residential independence remains an important milestone for young adults, and it is important to increase understanding of how and why this transition is reversed for those who return to the parental home.
The past few decades have seen growing interest in how increased economic uncertainty has affected transitions to adulthood (Furlong and Cartmel 2007; Mills et al. 2005). This has become even more salient in the context of the global recession of the late 2000s (Wang and Morin 2009). In Great Britain, unemployment rates rose disproportionately among young people (Office for National Statistics 2011a), and those employed are often reliant on work that is part-time or otherwise insecure (Furlong and Cartmel 2007; MacDonald 2009; Mills et al. 2005). In addition, partly in response to a heightened need for qualifications to compete in the job market, enrollment in higher education has increased (Chevalier and Lindley 2009). Understandably, residing at an educational institute necessitates a move into residential independence, but it also delays family formation and entry into the labor market (Kneale and Joshi 2008; Liefbroer and Corijn 1999). Hence, across socioeconomic groups, young adults’ dependency, or semi-dependency, on their parent(s) has been extended (Furlong and Cartmel 2007). One consequence of this is continued coresidence with parents (Goldscheider and Goldscheider 1999; Smits et al. 2010), with cycles of leaving and returning to the parental home becoming part of a nonlinear life course that is a key feature of individualization (Beck 1992).
In this article, we see returning home as determined by the characteristics of both young adults and their parents. The young adult’s individual circumstances include family status and material resources. Parental circumstances reflect the opportunities for intergenerational support that could help the young adult maintain residential independence or the conditions to which the young adult might return. These characteristics, which are discussed in more detail in the next section, can have direct effects on the risk of returning but can also work via turning points and may be moderated by gender.
Turning Points and Returning Home
The presence of “turning points” is a key concept within life course theory, referring to an event, an experience, or a change in circumstances that significantly alters the individual’s subsequent life course trajectory (Elder 1978). Turning points can result in life course trajectories becoming derailed, disrupting or limiting future opportunities (Rumbaut 2005). However, turning points can also be positive; for example, entering legitimate employment can be a turning point in the criminal life course, steering former offenders away from recidivism (Uggen 2000). Turning points are often distinguished from normative transitions that routinely occur, but this is not necessarily the case: a planned or anticipated transition can be experienced as a turning point if it involves a change in “significant life roles” (Wethington et al. 1997:217), particularly if the former and new roles contrast strongly in terms of the “re-arrangement of values, time, priorities and responsibilities” (Wheaton and Gotlib 1997:5). Qualitative research suggests that when questioned about their previous experience of turning points, individuals will often cite apparently normative transitions such as marriage, becoming a parent, or entering an occupation as a life event that had the most significant impact on their subsequent life course; thus, “every major role transition can quite reasonably be considered as potentially constituting a turning point” (Clausen 1998:203). Also important is the timing of transitions. In the social context of a “normative timetable” for life events, transitions that occur “off-time” can become turning points if they have consequences for subsequent opportunities or adoption of particular life roles (Elder 1998). This is, of course, highly sensitive to sociohistorical context, and as such, “the interaction of turning points with the varying structural locations and macro-historical contexts in which individuals make the transition to young adulthood” has been highlighted as an important area for research (Sampson and Laub 1996:365).
- Hypothesis 1 (H1):
Leaving full-time education and/or becoming unemployed will be associated with returning to the parental home.
- Hypothesis 2 (H2):
In comparison with beginning or remaining in a stable partnership, partnership dissolution will increase the propensity to return to the parental home among both men and women.
- Hypothesis 3 (H3):
Among women who are consistently unpartnered or who have experienced partnership dissolution, mothers will be less likely to return to the parental home than childless women.
- Hypothesis 4 (H4):
For men who experience a union dissolution, fatherhood will increase their propensity to return to the parental home.
Additional Determinants of Returning Home
A number of other individual and parental characteristics have been postulated in the literature to be associated with returning home. These may have a direct effect on the risk of returning or have an indirect effect via the turning points. We briefly describe these factors and, when possible, include them as controls in our analyses. Individual-level attributes likely to affect the likelihood of returning home include educational experience (Ford et al. 2002; Jones 1995; Stone et al. 2011) and individual income (Ermisch 1999, 2003), which can contribute to determining whether a young adult has the social and financial resources to live independently. Young adults born overseas, especially recent migrants to Great Britain, will be less likely to live with their parents (Stone et al. 2011) because often their parents will not be available in Great Britain for coresidence.
Material circumstances in the parental home can also affect the propensity for coresidence, but via complex and competing mechanisms (Aassve et al. 2002), depending on propensity for the parents to be altruistic, the tastes of the individual and his or her parents for independence and privacy, and the parents’ capacity to make intergenerational transfers (Becker et al. 2010; Ermisch 2003; Goldscheider and Goldscheider 1999; Mulder and Clark 2002). Because we do not have information about parental income, we use parental occupational class as an alternative indicator of parental socioeconomic position.
A number of other factors have been recognized as possible determinants of coresidence with parents, but we cannot address them here. These include parental family structure (Gee et al. 1995; Goldscheider and Goldscheider 1998; Smits et al. 2010), parental income (Aassve et al. 2002; Mulder and Clark 2002), and geographical distance between the young adults’ and parents’ home (Leopold et al. 2012). Furthermore, previous research has found an effect of contextual factors, including local housing and labor markets (Ermisch 1999). As discussed in the Results section, an alternative approach—using the same data source but, in this case, following cohorts of 16-year-olds—allows us to include these additional variables. The findings of these supplementary analyses are reported elsewhere (Berrington et al. 2013).
The British Household Panel Study (BHPS) is an ongoing, nationally representative panel study of individuals from 5,500 households first interviewed in 1991. The annual survey follows individuals from original households even when they divide to form new households. Children from original households are added to the interview sample each year when they reach age 16. We use data from 1991 through 2008. Household grid information is used to identify movement out of and back into the parental home from one year to the next. These annual transitions exclude short-term, temporary changes in living arrangements that occur between panel waves.
We pool the waves of data for any individual who provides valid data from at least two consecutive waves, who is living away from their parents in the first of these waves (t – 1) (i.e., is at risk of returning to the parental home), and who is in our target age range of 20–34 years at t – 1. We then track individuals through the subsequent consecutive waves until they return to the parental home, they are lost to follow-up, or the survey reaches its final wave. This subsample includes 2,273 men with 10,235 person-years of data and 2,900 women with 14,615 person-years. Calculating attrition rates for this sample is not entirely straightforward because the follow-up period is not fixed and there are various reasons for censorship. For example, respondents move out of scope when they exceed the upper age limit. To give an indication of the extent of attrition, we calculate the five-year follow-up rates for original sample members who reach the target age of 20 years before 2003 (and therefore can potentially be followed for five years). In this group, 73 % of men and 77 % of women are followed for at least five years. Respondents are less likely to be lost from this sample if they have a Bachelor’s degree, were born in Great Britain, had at least one parent who was employed when the respondent was aged 14 years, are employed, or are a parent. In general, it appears that those in more advantaged circumstances are more likely to be followed than those who are relatively disadvantaged. If we assume that a residential move is associated with an increased risk of attrition, this would mean that among those in disadvantaged groups, we are less likely to observe those who return to the parental home than those who remain living independently. This, in turn, would mean that any observed association between social disadvantage and returning to the parental home would be made more conservative. To more fully assess the impact of attrition on our analyses, we carry out a series of additional analysis using a probit model with sample selection (Van de Ven and Van Praag 1981), the details of which can be found in Online Resource 1. These analyses suggest that attrition has little impact on our findings.
We classify young adults as living in the parental home if they live in the same household with at least one natural or adoptive parent or one stepparent. Those living outside the parental home at one time point and then coresident with their parents one year later are classified as returning to the parental home. In the BHPS, students living in the parental home during vacations are not enumerated at that address but are treated as members of the household in which they reside at their educational institution’s location. Accordingly, these temporary returns are not included in our analyses. Young adults leaving the parental home to go into an institutional setting are included, however, and hence students in residence halls are included in the follow-up (Taylor et al. 2010).
Distribution of significant variables (% of total person-years)
% in Each Category (n = 24,850 person-years)
Quartile 1 (lowest)
Quartile 4 (highest)
Country of Birth
Outside Great Britain
Change in Economic Activity
Student to employed
Student to unemployed or inactive
Unemployed or inactive to employed
Employed to unemployed or inactive
Stable unemployed or inactive
Change in Partnership Status
New or stable partnered
Already a Parent
Parental Occupational Class
Additional Control Variables
Educational experience is time-varying and coded using self-reported school-leaving age and highest educational qualification: (1) Bachelor’s degree or equivalent, (2) any other postcompulsory education,3 (3) no postcompulsory education, and (4) not known. Individual income is based on total reported income in the month prior to interview and is time-varying. Income is coded in age-specific quartiles at each wave, with Quartile 1 representing the lowest individual income. Country of birth is coded as a dichotomous variable, with those born in Great Britain compared with those born outside Great Britain. Parental occupational class is coded using the conventional approach (Goldthorpe 1983) as a fixed covariate, using the three-category version of the Goldthorpe class schema (Goldthorpe et al. 1987).
We retain only those variables that show a significant association with returning home at the 10 % level in at least one model. We carry out separate analyses for men and women, but when a variable’s contribution is significant in one sex only, we retain the variable for both men and women to allow comparison between the sexes using the same model specification. Where a need for formal testing of key gender differences is indicated, additional analyses are carried out on the total sample. Interactions between variables and gender are then evaluated (analyses not shown but available from the authors upon request). We first present coefficients from a model that contains only the respondent’s individual attributes (Model 1) before adding the turning points and interactions required to test our hypotheses (Model 2).
Annual rate of returning to the parental home by age group and sex, 1991–2008 (percentages, with 95 % confidence intervals shown in parentheses)
Parameter estimates from a discrete-time hazards model of returning to the parental home at ages 20–34 years, by sex
Men (n = 10,235 person-years)
Women (n = 14,615 person-years)
Period (ref. = 1991–1996)
Age Group (ref. = 20–24)
Educational Experience (ref. = bachelor’s degree)
No postcompulsory education
Postcompulsory education (no higher education)
Individual Income (ref. = Quartile 1, lowest)
Quartile 4 (highest)
Country of Birth (ref. = Great Britain)
Outside Great Britain
Parental Occupational Class (ref. = service class)
Change in Economic Activity (ref. = stable employed)
Student to employed
Student to unemployed or inactive
Unemployed or inactive to employed
Employed to unemployed or inactive
Stable unemployed or inactive
Change in Partnership Status (ref. = consistently unpartnered)
New or stable partnership
Parent (ref. = nonparent)
Partner × Parent
Stable unpartnered × parent
Dissolution × parent
Age Group × Period
25–29 × 1997–2002
25–29 × 2003–2008
30–34 × 1997–2002
30–34 × 2003–2008
Degrees of Freedom
Consistent with H3, being a parent reduces the likelihood that women who are unpartnered or who experience a union dissolution will return to the parental home. Women who experience a union dissolution are the most likely to return home among both female parents and nonparents, but those who experience a union dissolution and have no coresident children have a predicted probability of returning of .3, compared with just .06 among mothers who experience a union dissolution. In contrast, parenthood has little effect on returning for women in a new or stable partnership, with predicted probabilities close to zero.
We hypothesize that among men who experience a union dissolution, fatherhood will increase their propensity to return to the parental home (H4). The results are consistent with this hypothesis, but the interaction effect does not reach statistical significance (Table 3). Figure 2 shows that among men who have experienced a union dissolution, the predicted probability of returning is .51 for fathers compared with .31 for nonfathers. Like motherhood, fatherhood has no effect on the probability of returning for those in a new or stable partnership, with predicted probabilities close to zero. Reflecting the predicted probabilities in Fig. 2, additional analysis (not shown) indicates that among parents who experienced a union dissolution, there is a highly statistically significant difference between men and women in their probability of returning (p < .001), whereas there is no significant gender difference for nonparents who experienced a dissolution.
As noted earlier, additional analysis based on a smaller, younger subsample of the BHPS allows the inclusion of various additional parental and contextual control variables (Berrington et al. 2013). These supplementary analyses support the findings reported here. In other words, the inclusion of additional controls for parental family structure, parental income, and contextual effects has no effect on the findings in relation to the other variables, including the turning points, focused on in this article.
This article provides new evidence on the dynamics of returning home in young adulthood in Great Britain. Overall, returning home is a relatively rare event, but among subgroups of the population, returning home is the norm if preceded by particular turning points in the life course. Our analyses suggest that “boomeranging” is not unusual for those in their early 20s, particularly for young adults completing higher education. However, the incidence of returning to the parental home drops rapidly with age. Contrary to media speculation, our empirical evidence suggests only a small increase in the rate of returning home in Great Britain, which is largely confined to women in their early 20s.4 Young women are increasingly likely to leave home to attend higher education rather than to form a partnership and hence are more likely to return home in their early to mid-20s. These findings highlight the importance of gender: the relationship between higher education and period changes in returning to the parental home is clearly linked to the feminization of higher education in Great Britain, with females now outnumbering male undergraduates (Office for National Statistics 2009).
Completing higher education is one of the strongest determinants of returning to the parental home—more so than any other change in economic activity. Consistent with earlier work in North America (Davanzo and Goldscheider 1990; Gee et al. 1995), results from the cohort follow-up show that young adults who leave home for reasons relating to education are the group most likely to subsequently return (Berrington et al. 2013). This relates to the idea of attending college/university as a transitional role with a finite duration, with parents willing to subsidize residential independence only for the duration of this activity (Davanzo and Goldscheider 1990). Although leaving the parental home is cited as a key part of the transition to adulthood, leaving home to attend a higher education institution is different from leaving home, for example, to form a new family, precisely because it is known to be a transitional or temporary state from the outset. This is not to say that other transitions to adulthood, such as marriage and entry into the labor market, are irreversible. New families can break down, and jobs can be lost, potentially resulting in a “refilled nest” (Mitchell 2006). However, such reversals will occur in only a proportion of the relevant group, whereas student status is necessarily transient. Furthermore, it has been argued that young adults who leave home to study are less independent than those who leave for other reasons. They often receive financial support from their parents and may be living in institutionally managed accommodations, which can be regarded as a semiautonomous state (Goldscheider and DaVanzo 1986).
In past decades, the expectation was that upon completion of higher education, young adults would move into employment. However, the increasingly volatile youth labor market in Great Britain represents a barrier to young adults making the normative transition from education to the labor market. Those who do make this transition may often be overeducated for the type of job they secure, while still potentially carrying a burden of student debt (Chevalier and Lindley 2009; Office for National Statistics 2012). Our analysis of turning points clearly shows that this, in turn, has an effect in terms of residential independence. Those who move from being a student to being unemployed are the group most likely to return to the parental home, particularly for men in their late teens and early 20s. Despite this, we do not see any pronounced increase over time in the proportions of young men returning to the parental home. However, this is likely because our data, which cover the period 1991–2008, largely predate the recession in Britain during the late 2000s. At the same time, moving from student status to employment also shows a strong association with returning, and it appears that it is the end of student status rather than subsequent economic activity status that is the most important predictor of returning.
Moving from employment to unemployment or inactivity is also positively associated with returning to the parental home, but less so than ending education. Again, it appears to be the change in economic activity status that is associated with returning rather than the qualitative nature of this change, with moving from unemployment or inactivity into employment also showing a positive association with returning. Moving into low-paid employment may not necessarily help young adults to gain residential independence because by having even a low wage, they may reduce their entitlement to social assistance, such as housing benefit (Lewis 1997; Smith 2005), reflecting the disincentives to work inherent in some welfare policies in Great Britain.
Although previous research has highlighted gender differences in patterns of leaving and returning to the parental home (Gee et al. 2003; Iannelli and Smyth 2008; Widmer and Ritschard 2009), less attention has been paid to the ways in which gender can moderate the effects of other determinants of coresidence with parents. In this article, we address this by analyzing men and women separately and by examining gender differences in the effect of parenthood and partnership transitions. Our results show that union dissolution is a key determinant of returning home. For childless men and women, the effect is similar. However, for mothers, union dissolution has little effect on the propensity to return to the parental home. Conversely, nonresident fathers are even more likely than nonfathers to return to the parental home following a union dissolution. This supports previous research indicating that although there is little gender difference in who remains in the joint home following dissolution among nonparents (Hayes and Al-Hamed 1999), among parents, the female partner will more commonly remain in the joint home (Gram-Hanssen and Bech-Danielsen 2008; Mulder and Wagner 2010; Ongaro et al. 2009).
Our findings reflect the importance of welfare regimes in differentially affecting pathways to residential independence for men and women. In Great Britain, access to means-tested social assistance and social housing determine that young, single mothers are able to maintain an independent household (Lewis 1997). Since the 1977 Housing Act, local authorities have had a duty to house homeless families with dependent children (Smith 2005). As such, social housing is an important safety net for single parents, reducing their likelihood of relying on their parents for accommodation following a union dissolution. Single people without children face more difficulty in accessing such housing and will therefore be more likely to need help from family members—such as their parents—following a union dissolution. Because children are more likely to remain with their mother after union dissolution (Office for National Statistics 2011b), this feature of the British welfare system may partly explain the observed gender difference in returning home following partnership dissolution.
A strength of our analysis lies in the use of prospective data extending over two decades. Using these longitudinal data, we are able to include explanatory variables that require data from multiple time points, such as the turning points that form a key part of our conceptual framework. However, sample sizes tend to be smaller in longitudinal than in cross-sectional surveys, and the sample size of the BHPS is insufficient to examine ethnic differences in returning home, which is an important determinant of leaving (Zorlu and Mulder 2011). The BHPS also provides little insight into the effect of parents’ downsizing of their home following the departure of children, competing demands of siblings on parental resources, the tastes and attitudes of respondents and their families, or social expectations of the wider peer group. We are unable to definitively assume the causal direction of the reported associations, despite the fact that the majority of the explanatory variables were measured prior to the outcome; we could not discount the possibility that these variables may have changed in anticipation of returning home. Our interpretations may, therefore, be oversimplified in some cases. Nevertheless, these limitations are largely offset by the benefits of the panel in providing access to information about the same individuals over an extended period and about the entire household with whom they are coresident at any one time.
Our findings highlight the need to consider the interconnections between transitions and turning points in different domains across life course. In particular, we have shown that union dissolution—a key turning point in the family domain—is an important predictor of returning home in young adulthood. However, this association is highly dependent on both gender and parenthood. After union dissolution, mothers and fathers may find support from different sources, with young single mothers more reliant on the welfare state and single nonresident fathers requiring greater support from their parents. It is also important to consider pathways out of the parental home as a predictor of returning.
Over the past two decades, the postponement of partnership formation and increasing uptake of higher education have meant that women have become more similar to men in their destinations on leaving the parental home. Completion of education continues to be an important catalyst for returning to the parental home, to the extent that it might be perceived as a normative transition for men and women in their early 20s (Billari and Liefbroer 2007; Liefbroer and Billari 2010; Settersten 1998). This is particularly salient in the British context of recession in the late 2000s, increased university tuition fees, and rising student debt.
In Great Britain, social housing refers to lower-rent, affordable accommodation that is provided by local councils and not-for-profit organizations, such as housing associations, and is let to people, often on low incomes, according to their housing needs.
Because of small sample sizes, we do not distinguish between marriage and cohabitation. We recognize that life events following the dissolution of these different types of union can be different, but to some extent, this is accounted for by the variable indicating the presence of children.
During the period under investigation, the minimum school-leaving age in Great Britain was 16 years.
This study has been carried out at the ESRC Centre for Population Change. The Centre for Population Change (CPC) is a joint initiative between the Universities of Southampton, St. Andrews, Dundee, Edinburgh, Stirling, and Strathclyde, in partnership with the Office for National Statistics (ONS) and the National Records of Scotland (NRS). The Centre is funded by the UK Economic and Social Research Council (ESRC), grant number RES-625-28-0001. The British Household Panel Survey (BHPS) was conducted by the Institute for Social and Economic Research (ISER) at the University of Essex and funded by ESRC. The BHPS data were accessed via the UK Data Archive. Neither the original data creators, depositors, or funders bear responsibility for the further analysis or interpretation of the data presented in this study. The authors thank Professor Peter W. Smith for his statistical advice. This article also benefited from constructive comments and advice from three anonymous reviewers and the Editors, which significantly improved the focus of the article and strengthened the arguments therein. Their input is gratefully acknowledged.
Open Access This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.