Keywords

1 Introduction

Parenthood is often seen as being a core element of a “normal” adult life (Dykstra and Hagestad 2007). This notion has coloured both scientific and societal views on people who will never make the transition to parenthood. Childless individuals, especially childless women, are depicted as “others”, and even as deviants (Letherby 2002). They are also perceived as being disadvantaged and as having weaker support networks. It is often assumed that childless adults are more likely than parents to suffer from isolation, loneliness, and physical and mental ill health (see for review Dykstra and Hagestad 2007).

Since being a parent is considered to be more central to the life of a woman than to the life of a man (Veevers 1980; Hird and Abshoff 2000; Letherby 2002; Bulcroft and Teachman 2003), the ramifications of not having taken on a parental role are generally assumed to be more disadvantageous for a childless woman than for a childless man. Scholars have often asserted that among men, circumstances and behaviours in the domain of paid employment have a much stronger influence on their identity and wellbeing than those in the domain of family life (e.g., Gilford 1986; Thomson and Walker 1989). Most studies on the impact of childlessness have therefore examined the effects on women only, and have overlooked or simply neglected men (for a review, see Greene and Biddlecom 2000). Recent studies which have investigated the extent to which men’s lives are affected by remaining childless have concluded that the implications of childlessness are no less significant for men than for women, but that the effects may be different (e.g., Eggebeen and Knoester 2001; Keizer et al. 2010).

These studies have revealed that the impact of childlessness among men is conditioned to a much larger extent by partner status than it is among women (Dykstra and Wagner 2007; Kendig et al. 2007; Wenger et al. 2007; Umberson et al. 2010). For example, Kendig et al. (2007) showed that never-married and formerly married childless men were more likely than married childless men to report being in poor physical health, whereas among women there were no significant differences in self-reported health among childless women based on partner status. Other studies have shown that the life outcomes of never-married childless women are much more favourable than those of their married counterparts (Koropeckyj-Cox and Call 2007). Taken together, these findings suggest that the presence of a partner is more important to the wellbeing of childless men than of childless women. If the presence of a partner indeed plays a bigger role in the life outcomes of childless men than of childless women, then are childless men also more affected by couple dynamics than childless women? Moreover, does relationship satisfaction have a greater impact on the overall wellbeing of childless men than on that of childless women?

Understanding the importance of couple dynamics for relationship satisfaction and, subsequently, overall wellbeing is important, especially for middle-aged and elderly couples. With increasing age, the social network of an individual becomes smaller and the relative importance of the partner increases (Carstensen 1992). This may be particularly true for childless couples, whose social networks are already more limited because of the absence of children and grandchildren. By studying the potential gender differences in the effects of couple dynamics and relationship satisfaction, our work addresses the pertinent issue of whether there are particular individuals within the childless population who are “at risk” of maladjustment.

Using a couple perspective, we investigate in the current study the differential importance of couple dynamics for relationship satisfaction among childless couples. Subsequently, we investigate whether relationship satisfaction has different effects on the well-being of the male partner and of the female partner in a given couple. As studies on the impact of parenthood on the wellbeing of adults have shown that the consequences of having or of not having children are not necessarily uniform across life outcomes (Dykstra and Wagner 2007; Kendig et al. 2007; Wenger et al. 2007), we focus on both physical and mental wellbeing. In our analysis, we use of multi-actor data from the Netherlands Kinship Panel Study (NKPS), a nationally representative survey conducted in 2002–2004.

2 Theoretical Background

2.1 Gendered Benefits of Marriage?

Being in a relationship is thought to be beneficial for individuals because a partnership represents an important source of both social support and financial stability, factors which are linked to higher levels of physical and mental health (e.g., Stimpson and Peek 2005). Bernard (1972) was one of the first scholars to explore the idea that relationships are more beneficial for men than for women. Based on her belief that marriage oppresses women because the wife is subordinate to the husband, she concluded that women tend to be less satisfied with marriage than men. Bernard (1972) also argued that men derive greater health benefits from marriage than women, and that marriage is harder on women than on men because women shoulder the majority of the household and childcare tasks.

In Bernard’s work, which was published in the 1970s, gender balance and an equal division of tasks were the key factors used to explain the differences between men’s and women’s levels of satisfaction with marriage. Today, however, there is a much greater degree of gender equality in relationships than was the case in the period in which Bernard wrote her seminal studies (Sullivan 2006). The overwhelming majority of contemporary mothers are no longer confined by the role of housewife, but are actively involved in the labour market. In addition, while women still shoulder the majority of childcare duties, men have taken on a greater share of household and childcare tasks (e.g., Hook 2006). These trends suggest that couples today are much more gender-equal than they were four decades ago. Thus, based on the argumentation of Bernard, we would expect to find that gender differences in marital satisfaction are weaker today than they were in previous generations. Indeed, recent meta-analyses (Jackson et al. 2014) have reported that within non-clinical samples, no gender differences in marital satisfaction could be found. In line with these findings, and given that childless couples are viewed as being more gender-egalitarian than couples with children (e.g., Grunow et al. 2012; Schober 2012), we should not observe any gender differences in relationship satisfaction levels among contemporary childless couples. However, the question of whether experiences within the partnership might affect the two partners differently remains.

2.2 Gender Differences in the Importance of Relationship Characteristics

Following up on the arguments of Bernard, numerous studies have investigated the implications of partner status for wellbeing (e.g., Coombs 1991; Kiecolt-Glaser and Newton 2001). Recently, however, scholars have shifted their attention to investigating the impact on wellbeing of within-relationship dynamics. The majority of these studies have shown that positive marital relations (characterised by support and closeness) are protective for relationship satisfaction and wellbeing, whereas negative marital relations (characterised by disagreement and distress) are associated with poor outcomes for one or both members of the couple (e.g., Acitelli and Antonucci 1994; Ducharme 1997; Miller et al. 2004; Henry et al. 2005; Whisman et al. 2006).

Scholars have argued that what is going on in the relationship tends to have a greater impact on the female partner’s than on the male partner’s satisfaction with the relationship (e.g., McRae and Brody 1989). They often explain this difference by claiming that women tend to do more of the emotional work in the relationship than men (Thomson and Walker 1989), i.e., that women are generally more aware than men of the emotional climate of the relationship, and are more likely to monitor the relationship’s emotional quality. While it has been shown that the perception of problems is associated with lower levels of relationship satisfaction and higher levels of stress among both women and men, women have been found to be more likely than men to perceive problems in the relationship (McRae and Brody 1989). In line with these findings, other studies have revealed that women initiate relationship therapy and file for divorce more frequently than men (e.g., Rokach et al. 2004). Scholars have further observed that a woman tends to be at a double disadvantage in a relationship relative to her male partner, not only because she is more likely to perceive problems in their relationship, but because these problems are more likely to have a detrimental impact on her wellbeing (e.g., Gove and Hughes 1979; McRae and Brody 1989). In other words, as Rae and Brody (1989) have put it: “Women’s marriages are more negative than men’s marriages and the negatives translate into more distress for women than men” (ibid.: 246).

Most studies which have examined the extent to which characteristics of the relationship affect relationship satisfaction have focused on negative marital relations, such as relationship problems and conflicts (for a critique, see Cramer 2004). Although the ways in which partners experience conflict and handle relationship problems are strong predictors of satisfaction with the relationship, recent studies have found that the ways in which partners provide emotional support to one another may be equally or even more important determinants of satisfaction (e.g., Cramer 2004; Hilbert et al. 2013). Findings in this area complement the findings on the impact of conflict and problems in the relationship: i.e., a woman’s level of relationship satisfaction is not only more likely than a man’s to be negatively affected by relationship conflict and problems; it is also more likely to be positively affected by partner support (e.g., Julien and Markman 1991).

Furthermore, scholars have argued that the physical and mental wellbeing of a woman is more strongly affected than that of a man by relationship quality because marriage is considered a more central component of a woman’s than a man’s life (Gilford 1986). Women are socialised to derive their wellbeing from close inter-personal relationships, whereas men are encouraged to derive their sense of self through more autonomous pathways, such as paid labour (Quirouette and Gold 1992). It has therefore been posited that while men tend to benefit from marriage regardless of the quality of the relationship, women may derive mental and physical health benefits from marriage only if the relationship is satisfying (Hess and Soldo 1985). Some scholars have even found that positive relationship characteristics such as high levels of closeness, while beneficial for a woman, are actually detrimental to the wellbeing of a man, as intense intimacy with a partner may interfere with the man’s ability or desire to maintain his autonomy (Quirouette and Gold 1992).

2.3 Is the Picture Different for Childless Couples?

Although the literature has shown that childless couples exhibit higher levels of relationship satisfaction than parents (see Wagner et al. 2015 for a recent exception), it is not yet clear whether the previously described gender differences in the effects of relationship characteristics on relationship satisfaction – and, subsequently, on wellbeing – also apply to childless couples. Although childless couples (perhaps in part because of the greater degree of gender equality in their relationship) may be expected to report having fewer relationship conflicts and problems, there is no basis for assuming that the existence of relationship problems or conflicts would have different effects on the relationship satisfaction levels of childless couples than on those of parents. Therefore, we hypothesise that the link between relationship satisfaction and both positive and negative relationship dynamics will be stronger for childless women than for childless men (H1).

In terms of the effects of relationship satisfaction on physical and mental wellbeing, the literature suggests that previous findings for couples with children should not be extrapolated to childless couples. Compared to childless men, childless women are often better off economically and have substantially larger networks (e.g., Dykstra and Hagestad 2007). This might indicate that the overall wellbeing of childless women is less dependent than that of childless men on what is going on in their romantic relationship. We therefore hypothesise that the link between relationship satisfaction and both physical and mental wellbeing will be stronger for childless men than for childless women (H2).

In the current study we address two main questions: (1) do relationship dynamics have different effects on the relationship satisfaction levels of childless women than of childless men; and, (2) does the link between relationship satisfaction and mental and physical wellbeing differ between childless men and childless women? These questions were investigated by estimating couple-level random effects models using data on 163 Dutch childless couples from the first wave of the nationally representative Netherlands Kinship Panel Study (NKPS). Our work therefore helps to answer the question of whether previous findings on the importance of relationship dynamics and relationship satisfaction for couples with children also apply to the rather distinct population of childless partnerships.

3 Method & Method

3.1 Data

The data used in this chapter come from the first wave of the Netherlands Kinship Panel Study (NKPS; Dykstra et al. 2005). The NKPS is a longitudinal, nationally representative study. In wave one of the NKPS, 8161 individuals aged 18–79 participated. The respondents (also referred to as “anchors”) were selected from a random sample of private addresses in the Netherlands. The first wave was conducted in 2002–2004 and had a response rate of 45% (Dykstra et al. 2005), which is not atypical for the Netherlands. Dutch response rates tend to be lower than elsewhere and have been declining over time, likely because the Dutch are particularly sensitive about privacy issues (De Leeuw and De Heer 2001 (fehlt); Stoop 2005 (fehlt)). The anchor data were collected via computer-assisted face-to-face interviews, as well as through separately completed questionnaires. Data were also collected from a number of significant others (also referred to as “alters”), including the anchors’ current partner.

For our analyses, we focused on anchors who had partners at the time of the first wave of data collection, and whose partners were also participating in the NKPS (51.4% of the wave one sample, n = 4194). We further restricted our sample to couples in which neither partner was a parent (i.e., neither had children, including with an ex-partner) and the female partner was age 40 or older at the time of the interview. This restriction was made because we were interested in the couple dynamics of permanently childless individuals. Earlier research has shown that the proportion of couples who make the transition to parenthood after the age of 40 is small (Landry and Forrest 1995; Garssen et al. 2001). These selections resulted in a final sample of 163 childless couples. In our work, we used the data provided by both partners; thus, our sample consisted of 326 individuals nested in 163 couples.

3.2 Measures

Relationship Satisfaction

Both partners provided answers to the following four items: “We have a good relationship”, “The relationship with my partner makes me happy”, “Our relationship is strong”, and “The relationship with my partner is very stable”. The responses were coded from 1 = strongly agree to 5 = strongly disagree. Developed specifically for the NKPS, the reliability and validity of this scale were tested during pilot studies (Verweij 2002), and it has been used successfully in other studies (Komter et al. 2012). The scale was created based on the mean of the items (α = .95 for anchors and α = .92 for alters). The items were recoded so that a higher value represented higher relationship satisfaction. The correlation between the partners’ answers was r = .55, p < .05.

Self-Reported Health

The partners’ health was assessed based on the following question: “How is your health in general?” The respondents could choose from 1 = excellent to 5 = very poor. Self-assessed health has been shown to be a strong indicator of general health (Ferraro and Farmer 1999; McHorney 2000), and this NKPS item in particular has been validated in previous research on the link between family of origin and health (e.g., Monden 2010). The question was recoded so that a higher value corresponded to better health. The correlation between the partners’ responses was r = .05, p > .05.

Mental Well-Being

The partners’ mood in the past 4 weeks was assessed using the following five questions: “How often have you felt particularly tense in the past 4 weeks?”, “How often have you felt so down in the dumps in the past 4 weeks that nothing could cheer you up?”, “How often have you felt calm and peaceful in the past 4 weeks?”, “How often have you felt downhearted and miserable in the past 4 weeks?”, and “How often have you felt happy in the past 4 weeks?”. The answer categories ranged from 1 = all the time to 6 = never. Two of the items were recoded so that a higher value on this scale indicated a better mental wellbeing. The scale was created based on the mean of the items (α = .82 for anchors and α = .85 for alters). The correlation between the partners’ answers was r = .24, p < .05.

Support from the Partner

Both the anchor and the alter provided information about the level of support they received from their partner by answering the following five questions: “To what extent does your partner support you: (a) in decisions about your work or education; (b) when you have worries or health problems; (c) in your leisure time activities and social contacts; (d) with all kinds of practical things you need to do; and (e) in personal matters that are on your mind?” (1 = no support to 4 = a lot of support). The scale was created based on the mean of the items. The reliability of the measure was high both for the anchors and alters (α = .84 for both). The correlation between the partners’ responses was r = .30, p < .05.

Relationship Conflict

The level of conflict in the relationship was assessed using the following three items: “Please indicate whether the following situations have occurred between you and your partner in the past 12 months: (1) heated discussions between you and your partner; (2) one of you putting down and blaming the other; and (3) you didn’t want to talk to each other for a while”. Both partners responded to these questions on a scale from 0 = not at all to 2 = several times. The scale was created based on the mean of the items. The reliability of the scale in our analytical sample was slightly under the conventionally established .70 threshold, but was still acceptable (α = .63 for anchors and α = .66 for alters). The correlation between the partners’ answers was r = .48, p < .05.

Control Variables

In all of our analyses, we controlled for the age of the reporting partner (in years), the highest attained level of education of the reporting partner (coded as 1 = (incomplete) elementary only/lower vocational/lower general secondary, 2 = intermediate general secondary/upper general secondary/intermediate vocational, and 3 = higher vocational/university/post-graduate), and for the duration of the current relationship in years (from the start of the partnership to the date of the interview).

In the analyses focusing on self-reported health, we also controlled for the reporting spouse’s informal social capital and level of agreement with child-endorsing norms. The informal social capital of each partner was measured based on four questions which referred to the extent to which the anchor/alter was able to rely on his or her friends (“When I am troubled, I can always discuss my worries with my friends”, “I place confidence in my friends”, “Should I need help, I can always turn to my friends”, and “I can always count on my friends”; rated from 1 = strongly agree to 5 = strongly disagree). The items were recoded so that a higher value corresponded to a large amount of informal social capital. A similar scale based on the NKPS data has been successfully used in earlier works on the impact of social contexts on romantic relationships (e.g., Hogerbrugge et al. 2012). The reliability of the scale was high both for the anchor and alter (α = .92 for anchors and α = .93 for alters). Finally, our measure of child-endorsing norms was constructed based on the partners’ responses to the following four statements: “A person’s life is not complete if s/he has not had children”, “People have a duty to society to have children”, “I believe that in this world a person can feel totally at ease only in his or her own family with children”, and “If a person never has children, s/he can never be really happy” (rated on a scale from 1 = strongly agree to 5 = strongly disagree). The items were recoded so that a higher value corresponded to a higher level of agreement with the child-endorsing norm. The reliability of the scale was high for both partners (α = .77 for anchors and α = .87 for alters). Table 15.1 displays descriptive information about all of the variables used in the analyses.

Table 15.1 Descriptive statistics for variables used in the analyses
Full size table

3.3 Analytical Approach

We carried out our analyses in two steps. In the first step, we examined whether gender differences could be observed in the association between aspects of the relationship (i.e., conflict and support) and relationship satisfaction (H1). In the second step, we examined whether the link between relationship satisfaction and self-rated physical and mental health differed between the male and the female partners (H2). The research questions were addressed using linear regression models with couple-level random effects. We did not estimate fixed-effects models because there was little variation in the covariates of interest within couples (e.g., level of conflict, support from the partner), which could have resulted in standard errors which were too large (Allison 2009). The models were fitted using xtreg in STATA, Version 13.1. In all of the models we included control variables for the age of the partner whose relationship satisfaction/wellbeing was being examined, as well as his/her educational level and the duration of the relationship. Additionally, in the models which addressed the link between relationship satisfaction and wellbeing we controlled for the individual’s level of agreement with child-endorsing norms (as a proxy for whether the respondent’s childless status was (in)voluntary) and social capital.

To test the hypotheses, we included interaction terms in our models (i.e., between the gender of the partner and the covariate of interest, such as conflict, support, or relationship satisfaction). To facilitate the interpretation of the significant interaction terms, we used the margins command in STATA to estimate and plot the marginal effects at representative values for the female and the male partners. All of the marginal effects were estimated for the reference categories of the categorical control variables, and the continuous variables were kept at the sample mean.

4 Results

Detailed descriptive information about the measures used in this study is displayed in Table 15.1. The table clearly shows that, by and large, the childless NKPS participants reported rather high levels of relationship satisfaction (M for female partners = 4.57 (SD = .53) and M for male partners = 4.64 (SD = .56)). Furthermore, there were no gender differences in the mean levels of any of the central covariates of interest (i.e., relationship satisfaction, self-reported health, mental wellbeing, support from the partner, and relationship conflict).

In the first step of our analyses we focused on the question of whether there were gender differences in the association between relationship dynamics and relationship satisfaction. The results addressing this question are displayed in Table 15.2. The first two models in the table display the main effects of gender (Model 1) and of support from the partner and relationship conflict (Model 2). As was mentioned earlier, the gender of the partner was not associated with the self-reported level of relationship satisfaction, whereas the self-reported level of support received from the partner and the level of relationship conflict were associated with relationship satisfaction in the manner predicted (i.e., a one point increase in partner support was linked to a .45 point increase in the dependent variable, and a one point increase in relationship conflict was linked to a .19 decrease in relationship satisfaction). Our first research question is, however, addressed in the subsequent models, which included an interaction between the gender of the partner and the indicators of relationship dynamics (Model 3 for support and Model 4 for conflict). All of the results discussed below were found after accounting for the individual and relationship level control variables.

Table 15.2 Estimates from relationship-level Random-Effects Regression Models with relationship satisfaction as dependent variable
Full size table

As can be seen in Model 3, the association between the level of support received from the partner and the level of self-reported relationship satisfaction did not differ between the male and the female childless partners. In other words, we did not find evidence that childless women were more strongly affected than childless men by the positive aspects of their relationship. While we did find evidence of gender differences in the association between conflict and relationship satisfaction, these differences were not in the expected direction (Model 4 of Table 15.2). For ease of interpretation, the estimated marginal effects at representative values are plotted in Fig. 15.1. A post-estimation examination of the slopes for the two groups showed that only the slope for the male partners (b = −.31, SE = .08, 95 % CI [−.47 – −.14]) was significant (slope for female partners: b = −.09, SE = .08, 95 % CI [−.25 – .06]). An additional check demonstrated that the difference between the partners was significant (p < .05) only at particularly high levels of conflict (i.e., two), and that the magnitude of the difference was not large (i.e., a difference of .32 points at frequency of conflict = 2).

Fig. 15.1
figure 1

Plot of estimated values for relationship satisfaction, based on estimates from Model 3, Table 15.2 (control variables at representative value)

Full size image

The subsequent models addressed the second research question: namely, whether there were gender differences in the link between relationship satisfaction and the partners’ wellbeing. Our findings are displayed in Table 15.3, Model 2 and Fig. 15.2 for physical health; and in Table 15.3, Model 4 and Fig. 15.3 for mental health. Model 2 of Table 15.3 and Fig. 15.2 show that relationship satisfaction was found to be positively associated with self-rated health, but only among the childless men. A post-estimation examination of the slopes indicated that whereas the slope for the male partners was significant (b = .25, SE = .11, 95 % CI [.03 -- .47]), the slope for the female partners was not (b = −.08, SE = .12, 95 % CI [−.30 – .15]). Our findings in Model 4 of Table 15.3 also demonstrated that there was a (borderline) significant gender interaction for the link between relationship satisfaction and self-rated mental wellbeing. Once again, whereas the slope for the male partners was significant (b = .33, SE = .10, 95 % CI [.14 – .52]), the slope for the female partners was not (b = .09, SE = .10, 95 % CI [−.11 – .29]; also see Fig. 15.3). In other words, our results indicated that the link between (physical) wellbeing and relationship satisfaction was stronger for the male than for the female childless partners.

Table 15.3 Estimates from relationship-level Random-Effects Regression Models with self-reported health and mental wellbeing as the dependent variables
Full size table
Fig. 15.2
figure 2

Plot of estimated values for self-rated health, based on estimates from Model 2, Table 15.3 (control variables at representative values)

Full size image
Fig. 15.3
figure 3

Plot of estimated values for mental well-being, based on estimates from Model 4, Table 15.3 (control variables at representative values)

Full size image

5 Discussion

In this chapter, our goal was to test to what extent gender differences in couple dynamics exist within childless couples. In our analyses we found that male and female childless partners reported similar levels of relationship satisfaction, a result which is in line with the findings of recent studies on gender differences in marital satisfaction of partners with children (e.g., Jackson et al. 2014). The key contribution of our work however, is our finding that there were gender differences in the link between relationship dynamics and relationship satisfaction and in the link between relationship satisfaction and wellbeing.

First, in contrast to our expectations, we found that the link between relationship conflicts and relationship satisfaction was stronger for the childless men than for the childless women. Interestingly, no gender differences were found in the link between partner support and relationship satisfaction. In other words, the positive aspects of the partnership were equally important for both the male and the female childless partners studied. These findings were surprising, as most of the previous literature has stressed that female partners are more strongly affected by both the positive and the negative aspects of their romantic relationships (e.g., McRae and Brody 1989). Our finding concerning relationship conflict could be interpreted in two ways. As we noted above, compared to childless men, childless women have been reported to be economically better off and, even more importantly, to have larger networks (Dykstra and Hagestad 2007). Therefore, it is possible that childless men depend heavily on their intimate partnerships, and are thus, more sensitive to the internal dynamics of these relationships than childless women. Another possible interpretation of this finding is that it is not the case that childless men are more strongly affected by relationship conflicts, but rather that childless women are less strongly affected by conflicts. In other words, we suggest that previous evidence that mothers are more sensitive than fathers to relationship conflict might be attributable to a heightened preoccupation among mothers with the potential impact of those conflicts on the wellbeing of their children. As childless women do not face this concern, they might be less sensitive to conflicts. These interpretations are, however, highly speculative. Future research may want to examine whether the gender difference we found here is robust, and to investigate to what extent it is driven by a heightened sensitivity to relationship dynamics among childless men or by a lower sensitivity to relationship dynamics among childless women.

The second main finding of our work concerned the link between relationship satisfaction and self-reported mental and physical wellbeing. In line with our hypothesis, we found that the association between relationship satisfaction and health was stronger for the childless men than for the childless women, and that this difference was particularly evident when the levels of relationship satisfaction were low. These results indicate that when they are in unsatisfying romantic relationships, childless men are at greater risk than childless women of physical and mental ill health. Again, future research should investigate in detail the mechanisms underlying this pattern. Are childless men indeed more affected by being in an unsatisfying relationship because they rely on their wife as their chief source of social support (Pugliesi and Shook 1998), and are these men therefore especially vulnerable when that support weakens or dissipates? Or is it the case that compared to fathers and childless women, childless men place a higher value on their romantic relationship than on other domains of life, and are therefore be more affected by what is going on in their relationship? Yet regardless of the underlying mechanisms, our study reveals that when childless men are dissatisfied with their romantic relationship, they are at risk of physical and mental maladjustment.

Some limitations of our study should be mentioned here. First, it is important to note that we did not strictly compare each childless man to his own female partner, who was also childless. As was previously stated, the optimal way to test for possible gender differences in the link between relationship dynamics and relationship satisfaction, and between relationship satisfaction and wellbeing, is to utilize couple-level fixed effects. However, given the very limited variability in the constructs of interest which we observed within our units of analysis (i.e., the partnerships) and due to concerns about the possibility of inflated standard errors, we opted to run random effect models (Allison 2009).

Another methodological concern which might be raised about our work is the fact that we did not address the question of possible reverse causality. In other words, we cannot exclude the possibility that what we are seeing is, for example, a gender difference in the impact of mental and physical wellbeing on relationship satisfaction. We chose to use data from the Dutch NKPS survey because they provide high-quality, dyadic information on the concepts of interest. However, as the data were cross-sectional in nature, we have taken great care throughout our work to avoid implying that we have found evidence of any causal links.

Despite these limitations, our results suggest that childless men may be more affected than childless women by negative couple dynamics. Using a rich couple-level data set, we showed that the link between relationship conflict and relationship satisfaction was stronger among the childless men than among the childless women. In addition, we found that the childless men who reported experiencing low levels of relationship satisfaction were also in worse physical and mental health than the childless women. Currently, approximately one in five men will remain childless (Keizer 2010), and divorce rates remain high. Thus, it appears that entry into fatherhood could become even more selective in the future (e.g., Rønsen and Skrede 2006). Future studies should therefore investigate in greater detail how childless men of middle and older ages function in romantic relationships.