Fathers’ and Mothers’ Expectations about Child Rearing after Divorce: Does Anticipating Difficulty Reduce the Chance of Divorce? *
Anne-Rigt Poortman, Vrije Universiteit Amsterdam, Amsterdam Judith A. Seltzer, University of California, Los Angeles
* Direct correspondence to Anne-Rigt Poortman, Faculty of Social Sciences (FSWSCW),Vrije Universiteit Amsterdam, De Boelelaan 1081, 1081 HV Amsterdam, The Netherlands, Email: [email protected]
, Phone: +31 20 4446720
Acknowledgements This research was supported by travel grants from the Netherlands Organization for Scientific Research (NWO, no. SIR 12-3720, R 50-436 & R 50-457 ). The authors wish to thank Rob Mare and Larry Bumpass for their helpful comments.
Abstract A common, but rarely tested, assumption is that children deter divorce because mothers anticipate great economic difficulties and fathers expect parenting difficulties after divorce. This study examines the role of mothers’ and fathers’ expected economic and parental costs for their chances to divorce. Using couple data from the National Survey of Families and Households, our results show first that mothers expect their standard of living to be worse than fathers, whereas fathers expect more difficulties in parenting after divorce. However, the majority of both mothers and fathers expect high economic as well parenting costs. Comparing the reports of the mother and the father of one family, the results show that a large minority of families consists of parents who disagree about the expected costs, which indicates that there may be a conflict of interests when deciding to divorce. The multivariate analyses show that only parents’ expected parental costs deter divorce, not their economic concerns. In addition, the mother’s parental concerns are found to be stronger deterrents of divorce than the father’s in case of conflicting interests. Finally, the relation between parents’ expected costs and divorce is moderated by whether the couple is happily married or not; expected parental costs are stronger deterrents of divorce for unhappy couples. Overall, our study shows that parents’ expected costs deter divorce, but that their role depends upon which and whose costs are considered as well as upon the stage of the divorce process.
Introduction A divorce is usually associated with costs and these costs are presumed to be higher in case of greater relational investments. Having children typically implies large investments in the relationship and it has therefore been argued that children decrease the risk of divorce, because of the associated higher costs (Becker, Landes & Michael, 1977; Waite & Lillard, 1991; Kalmijn, 1999). Many studies have indeed shown that couples with children, particularly young ones, are less likely to split up than childless couples (e.g., Heaton, 1990; Remez, 1990, Waite & Lillard, 1991). To account for this association between children and divorce, economic and parenting costs are most commonly argued to deter divorce because these costs represent the most salient difficulties that parents face in raising their children after divorce. Mothers and fathers are, however, assumed to differ in the kind of costs they encounter. Given that mothers predominantly obtain physical custody (Seltzer, 1990; Fox & Kelly, 1995; Cancian & Meyer, 1998), the most salient costs for fathers are parenting costs, whereas mothers are more likely to face economic costs (Kalmijn, 1999). Non-resident fathers’ parenting costs arise because they have less contact with their children after divorce (Seltzer, 1991, 2000; Stephens, 1996; Manning & Smock, 1999). Custodial mothers experience relatively large financial declines after divorce due to their role as primary caretakers (Smock, 1994; Bianchi, Subaiya & Kahn, 1999). Assuming that couples are less likely to divorce when the perceived costs are higher (Levinger, 1979; Becker, 1981), children lower the risk of divorce because mothers expect economic costs and fathers expect parenting costs after divorce.
So far, most research has merely assessed the relation between children and the risk of divorce, thereby implicitly assuming that children deter divorce due to the higher costs. A few studies have explicitly addressed the role of parents’ expected costs for divorce, but they have used indirect and objective measures for expected costs and have focused on either economic costs for mothers or parenting costs for fathers. For example, Dechter (1992) measures expected economic costs by estimating the predicted postdivorce economic well-being for mothers on the basis of statistical modeling and subsequently estimates its effect on the divorce risk. Kalmijn (1999) measures expected parenting costs by fathers’ involvement in childrearing, assuming that highly involved fathers face higher parenting costs, and assesses the effect of fathers’ childrearing investments on perceived marital stability. A more general line of research on the perceived “barriers to divorce” has used direct and subjective measures of expected costs, such as respondents’ reports on the factors that keep their marriage together (Knoester & Booth, 2000; Previti & Amato, 2003) or their perceptions of post-divorce life (Heaton & Albrecht, 1991). Although these subjective evaluations usually include both anticipated economic and parenting costs, these studies have not explicitly focused on parents but combined parents and childless persons, and used reports from one spouse only. This study aims to gain more insight in the role of parents’ expectations about economic and parenting costs for their chances to divorce. We extend upon prior work in four respects. First, we focus only on parents, which may lead to stronger evidence for the role of anticipated costs than studies combining parents and non-parents, which have found only weak support for the relevance of expected economic and parenting costs (Heaton & Albrecht, 1991; Knoester & Booth, 2000). Parents are likely to attach greater
weight to the possible costs of divorce when they decide to split up, because they are not only concerned about their own well-being but also their children’s and have invested more in the relationship than childless couples (Dechter, 1992). Research indeed shows that expected economic costs are only relevant for mothers’ risk of divorce, not for childless women (Dechter, 1992). Second, we take into account parents’ reports about both types of costs, which allows us to examine whether mothers and fathers actually expect different costs and asses which kind of costs weigh more heavily in the decision to divorce. Although the common assumption is that children deter divorce because mothers expect economic costs and fathers parenting costs, it is likely that both parents face economic and parental costs. Due to recent changes in divorce custody law (see Buehler & Gerard, 1995), mother-sole custody has decreased and shared custody and father-sole custody have increased (Meyer & Garaski, 1993; Garasky & Meyer, 1996; Cancian & Meyer, 1998). Hence, mothers are increasingly unsure of gaining custody before hand, and the costs of losing contact with the children might have become increasingly relevant. Also, fathers may now obtain custody and are therefore more likely to face the associated economic costs. More importantly, even in the (still) most common case of mother-sole custody, mothers are likely to face parenting costs and fathers economic costs. The non-custodial parent may expect economic costs for example because of future alimony payments or the loss of economies of scale of a larger household. The custodial parent may face parenting costs due to increased responsibilities and pressures in case of sole-parenting or worries about the possible adverse effects of divorce for the children (Waite & Lillard,
1991). Concerns about children’s well-being are in fact parenting costs that are likely to be shared by both parents. Third, we use the reports of both parents from the same household. We can therefore assess the extent to which the mother and father agree about the expected costs, and examine whose expected costs matter more in the divorce decision. Divorce usually involves a longer decision making process during which partners are likely to evaluate and discuss the costs and benefits of divorce. Particularly in case of children, parents are likely not to take a lighthearted decision and will discuss each others’ concerns before they split up. When the parents’ expectations about costs do not coincide, the issue is whose considerations matter most in the decision making process. On the basis of the literature about the gendered division of emotional labor in relationships (Bernard, 1976; Thompson & Walker, 1989; Amato & Rogers, 1997), the mother’s expected costs may be argued to weigh more heavily than the father’s. Prior research indeed offers suggestive evidence. Fathers’ higher parenting costs do not account for the positive relation between their childrearing investments and marital stability; instead it’s all about mothers being more content with marriage when fathers invest a lot (Kalmijn, 1999). Also the finding by Heaton and Albrecht (1991) that perceived marital stability is more strongly related to expected parental costs for women than men suggests that the mother’s concerns matter more, although their analyses pertain to the person instead of household level. Finally, we distinguish between happy and unhappy couples to examine whether expected costs deter divorce more when parents are unhappily married. People’s concerns about the post-divorce situation have been argued to become relevant only when the option to divorce has become more than hypothetical, as is the case when a couple is no
longer happily married. Such worries are presumed to play no role in case of a happy marriage, when people are not actually considering divorce (Levinger, 1965; Heaton & Albrecht, 1991; Knoester & Booth, 2000; Previti & Amato, 2003). Because there is more at stake for couples with children, the moderating role of marital unhappiness might be particularly strong in this case. Prior findings, though not exclusively on parents, are mixed; some studies do find a stronger effect of expected costs on marital stability for unhappy couples compared to happy ones (Heaton & Albrecht, 1991), whereas others do not (Koester & Booth, 2000). Using two waves of the National Survey of Families and Households, we examine the role of parents’ anticipated economic and parenting costs by using both mothers’ and fathers’ subjective reports on these costs. The first wave contains information on their expectations about how their standard of living and how being a parent would be different after divorce. This information is used to examine whether mothers and fathers expect different costs and the extent to which they agree about costs. The second wave has information about whether the couple subsequently divorces. Combining this information with data from the first wave allows us to assess the effect of parents’ expected economic and parenting costs on the risk of divorce and in particular, which and whose costs matter most. Also, these data enable us to examine the extent to which the relation between expected costs and divorce is moderated by marital happiness.
Parents’ expected costs and divorce How do parents’ expected economic and parenting costs affect the chance of divorce? Following micro-economic and social-psychological theories on divorce (Levinger, 1979; Becker, 1981), we start off with the basic premise that, all else equal, people are less likely to divorce when the associated costs are higher. It can therefore be expected that parents anticipating greater difficulties in their post-divorce standard of living or parenting role are less likely to divorce than those expecting fewer difficulties. Obviously, a divorce entails different costs and benefits and an individual’s willingness to divorce is based on weighing all these pros and cons. In case of children, a divorce most notably involves economic and parental costs. Mothers and fathers may differ as to how important these respective costs weigh in considering divorce. It is difficult to predict which costs matter more for mothers and fathers. On the one hand, financial declines are most salient for mothers and parenting costs for fathers, given the prevalence of mother-sole custody. To the extent that the most salient costs are also the strongest deterrents of divorce, expected economic costs should matter more than parental costs for mothers, whereas the reverse holds for fathers. On the other hand, one could argue that given the special bond between parents and children, both mothers and fathers attach greater weight to considerations referring to their parental role than to economic difficulties; the latter do complicate the parent-child relationship after divorce but rather operate as secondary strains for child rearing. This line of reasoning thus suggests that both mothers’ and fathers’ expected parental costs have a stronger effect on the risk of divorce than their respective expected economic costs. These conflicting
predictions lead us to refrain from formulating a hypothesis about which costs matter most. Whether or not a couple divorces depends on both spouses’ expectations about the costs of divorce. Parents are likely to discuss the option of divorce on the basis of their expected costs. The outcome of the decision making process is more easily arrived at when both parents agree about the costs of divorce, because (all else equal) their objections to divorce are alike; when both expect high costs they will be less likely to divorce than when they both agree that the costs are low. However, when parents’ expectations about costs do not coincide, one of them has greater objections to divorce than the other. In case of such conflicting interests the question is whether the mother’s or father’s considerations weigh more heavily in the decision making process. Arguments about the gendered division of emotional labor in families (Bernard, 1976; Thompson & Walker, 1989; Amato & Rogers, 1997) suggest that the mother’s concerns about postdivorce life override the father’s. Women tend to be more relationship-minded than men, and are generally more oriented towards domestic and family life. Because relationships and family live figure more prominently in the minds of women, they are not only more likely to sensor any problems but are also more often the ones who express their relational concerns (see the review by Thompson & Walker, 1989). Not only are women more sensitive to and more ready to speak out on relational problems, they are also more likely to act upon it, as is reflected in the higher percentage of women initiating divorce (Pettit & Bloom, 1984; Kitson, 1992). If women are the ones who have the upper hand in relational issues, it is likely that the mother’s expected costs are more important than the father’s in the decision making process. In case of couples with children, this may all the
more be true because mothers, as primary care takers, are generally more involved with the children than fathers; anticipated difficulties in post-divorce child rearing may therefore be more salient and important deterrents to divorce for mothers. This leads us to hypothesize that when the mother expects greater economic or parental costs than the father, divorce is less likely than in the opposite case where the father expects greater costs than the mother. The extent to which expected costs actually deter divorce not only depends upon whose costs are considered, but may also depend upon the stage in which couples are in the decision making process. Expected costs have been argued to become increasingly important as deterrents to divorce, the further a couple is in the process leading up to divorce (Levinger, 1965; Heaton & Albrecht, 1991; Knoester & Booth, 2000; Previti & Amato, 2003). In fact, people’s concerns about the post-divorce situation are argued to be of little importance when people are not considering divorce, as is the case when couples are happy together; for happy couples the potential costs of divorce are only hypothetical. As soon as the relationship becomes to be considered as problematic, expected costs may delay or even prevent divorce because people more carefully consider the possible consequences of divorce, particularly when there are children involved (ibid). This leads us to formulate an hypothesis about interaction effects: the expected parental and economic costs are stronger deterrents of divorce in an unhappy marriage than in a happy one. Stated differently, mothers’ and fathers’ expected parental and economic costs have a stronger effect on the risk of divorce in unhappy marriages than in happy ones.
Data We use two waves of the National Survey of Households (NSFH1 and NSFH2), a panel survey of adult members of households in the United States. During the first wave in 1987/1988, 13007 randomly selected primary respondents per household were interviewed face-to-face and received a self-administered questionnaire (Sweet, Bumpass & Call, 1988). In addition, the spouse or cohabiting partner was given a less detailed selfadministered questionnaire. The sample of primary respondents consists of an oversample of minority families, one-parent and reconstituted families, cohabiting couples and recently married persons (Sweet et al., 1988).Weights have been constructed by the NSFH –team to correct for this oversampling. The response rate for main respondents of the NSFH1 was about 74 percent. From 1992 to 1994 members of the original sample were re-interviewed in the follow-up survey (Sweet & Bumpass, 1996). Next to follow-up interviews with the main respondent, the NSFH2 also includes more or less similar interviews with the current spouse or cohabiting partner at wave 2 and the original spouse or cohabiting partner of wave 1 if the relationship had ended. The response rate for main respondents in the second wave was approximately 82 percent. Given our research purposes we use several selection criteria for the sample. For substantive reasons, the sample is limited to respondents who: (a) were married at wave 1 and were either still married or divorced/separated at wave 2 (N=5354), (b) have at least one biological child who is under 18 at the time of the first wave (N down to 2706). 1 Given our interest in the effect of expected economic and parenting costs, having children over the age of 18 are of less interest because they are making the transition to adulthood at the time of divorce or separation. The sample has been further reduced to include only
couples for which we have valid information about expected costs from both parents. 2 Given the centrality of expected costs in the analyses, we decided not to resort to the practice of missing substitution. The main reason for missing data on expected costs is the absence of the spouse’s questionnaire. For about 15 percent (N=420) of our sample the spouse’s questionnaire is missing and another 13 percent (N=351) did not give valid answers to the questions about expected costs (partial missing data). Exclusion of these cases leaves us with 1935 couples.
Measures of and descriptive findings on parents’ expected costs The NSFH1 contains information about parents’ expected economic and parenting costs. The self-administered questionnaire for primary respondents and their spouses contained the following question:
Even though it may be very unlikely, think for a moment about how various areas of your life might be different if you separated. For each of the following areas, how do you think things would change?
Six areas of life were asked for: ‘standard of living’, ‘social life’, ‘career opportunities’, ‘overall happiness’, ‘sex life’ and ‘being a parent’. Respondents could choose from the following answering categories: (1) much better, (2) better, (3) same, (4) worse, and (5) much worse. To measure parents’ expected economic and parenting costs we use respondents’ answers about ‘standard of living’ and ‘being a parent’, respectively. Although the data thus also contain expected costs on other areas than parenting and
standard of living, these costs are not our main interest, because they are less likely to interfere with childrearing after divorce. 3 To examine whether mothers and fathers expect different costs, we present the distribution and descriptive statistics of mothers’ and fathers’ answers concerning their expected economic and parenting costs in Table 1.
[Here Table 1]
The upper panel of Table 1 shows the expected costs with respect to their standard of living. Only a minority of mothers and fathers think that their standard of living would improve if they were to split up; about 4 percent of the mothers and 10 percent of fathers thinks that their standard of living would be (much) better. In contrast, over half of the fathers and almost three quarters of the mothers thinks their standard of living would be (much) worse after marital dissolution. Mothers expect higher economic costs than fathers. The means and the percentages expecting a worsening of their standard of living differ significantly between mothers and fathers. Although this gender difference was to be expected on the basis of the common practice of mother-sole custody and the wellknown gender differences in the economic consequences of divorce (e.g. Smock, 1994), it is noteworthy that a considerable number of fathers also expect a decline in their standard of living. For parenting costs (lower panel in Table 1), we see that mothers and fathers who think parenting would improve after dissolution are an exception; less than 5 percent thinks so. The majority thinks parenting would be (much) worse; about 60 percent of the mothers and 70 percent of the fathers. Now we see that fathers expect
higher parenting costs than mothers, which is as would be expected on the basis of the prevalence of mother-sole custody. Although parenting costs are significantly higher for fathers compared to mothers, the difference is smaller than in case of economic costs and also most mothers expect that parenting would be worse, even though they are likely to gain custody in case of divorce. These results suggest that mothers and fathers do expect different costs, but only to some degree; most mothers and fathers expect high economic as well as parenting costs after divorce. Table 2 shows the extent to which the mother and father from the same household agree about the expected costs. To this end, we dichotomized parents’ answers about expected costs resulting in two categories for both types of costs: (much) worse versus (much) better or the same (i.e., not worse). Cross-classifying the answers of the mother and the father from the same household, leads to the four possible situations presented in Table 2. With respect to expected economic costs, in little over half of the families the mother and father agree about the costs (about 54 %); in about 14 percent of the families both parents expect that their standard of living will not be worse, and in 40 percent they agree that their economic situation will be worse. The other 46 percent of the families disagree about the costs, with most instances pertaining to the situation that the mother expects that her standard of living will be worse and the father does not (32 %) instead of the other way around (14 %).
[Here Table 2]
For parenting costs, we observe that in the majority of families parents agree about the costs (about 62 %); about 16 percent of the families consists of parents agreeing that parenting will not be worse and in 46 percent of the families the mother and father agree that parenting will be worse. Disagreement is prevalent in almost 40 percent of the families; in about 15 percent of the families the mother thinks parenting will be worse and the father does not, and in 23 percent of the families the reverse is the case. Overall, agreement between the mother and father shows considerable variation. For both economic and parenting costs we find that a substantial number of families consists of parents whose expectations about the costs following divorce do not coincide, which points at conflicting interests in case of deciding whether or not to divorce.
Analytical strategy and measures for multivariate analyses To examine the relation between parents’ expected costs and the risk of divorce and the moderating role of marital happiness, we combine information from the NSFH1 with that from the NSFH2. The NSFH2 contains information about our dependent variable, marital dissolution. The NSFH1 not only has information on the parents’ expected costs (see previous section), but also about marital happiness and other well-known determinants of marital dissolution, which are controlled for in the analyses. Marital dissolution is measured as a dichotomy indicating whether the spouses separated or divorced (coded 1) or remained married (coded 0) between the first and the second interview. The majority of couples continued to stay married (N=1677) and about
2 percent (N=50) of the couples separated due to marital problems and 11 percent (N=208) actually got divorced. Parents’ expected economic and parenting costs are, for the purposes of the multivariate analyses, measured in two ways. First, we simply constructed four dichotomous variables to measure the mother’s and father’s economic and parenting costs. These variables indicate whether they expected their standard of living or parenting role to be (much) worse (coded 1) or (much) better/the same (coded 0). Descriptive information has already been presented in Table 1 (under Percent worse or much worse). Second, we use the variables measuring the mother’s costs vis-à-vis the father’s from the same family, as presented in Table 2. For both standard of living and parenting, we constructed three dichotomous variables indicating whether (a) both expect it will be worse, (b) mother expects worse, father does not, and (c) father expects worse, mother does not. The reference group consists of couples in which both expect it will not be worse, which are assumed to be most likely to divorce since both spouses have no objections to divorce. The other independent variables are listed in Table 3. Besides the obvious control variable indicating the duration between wave 1 and wave 2, they broadly refer to parents’ social-demographic characteristics, characteristics of the children, parents’ socioeconomic characteristics and marital characteristics. Information about precise measurement and descriptive statistics can be found in Table 3.
[Here Table 3]
Parents’ social-demographic characteristics pertain to their ethnicity, age at marriage, religiosity, prior cohabitation, parental divorce and prior divorces. Studies have shown that non-white persons, those who marry at a younger age, non-religious persons, those who did not previously cohabit, persons from broken families and previously divorced persons have higher chances of divorce (e.g., South & Spitze, 1986; Bumpass, Martin & Sweet, 1991; Call & Heaton, 1997; Wolfinger, 1999). Although these characteristics were measured for both spouses, high correlations between the mother’s and father’s measures for some characteristics (i.e. ethnicity, age at marriage, religiosity and prior cohabitation) led us to include only the mother’s. When correlations were not too high (i.e., parental divorce and previous divorces), we tested whether the effects for the mother and father differed. Since they did not differ significantly, we constructed composite variables for reasons of a more parsimonious model. Children’s characteristics refer to the number of minor biological children, the presence of stepchildren, age of the youngest biological child and sex-composition of the minor biological children. Prior findings, though sometimes mixed, indicate that having fewer children, stepchildren, older children and solely daughters increases the risk of divorce (e.g., White & Booth, 1985; Morgan, Lye & Condran, 1988; Heaton, 1990; Waite & Lillard, 1991). We focus on characteristics of children, who are biological to both spouses, but take into account whether there are stepchildren in the household. 4 Furthermore, the children have to be under the age of 18, because otherwise they have already reached adulthood at the time of marital dissolution. The socio-economic characteristics refer to parents’ educational level, employment statuses and wage rates, and family income. Previous studies most
commonly start from the assumption that the wife’s greater socioeconomic resources increase the risk of divorce, whereas the husband’s decrease the risk of divorce. However, the findings are mixed and depend upon the measures used, which is no surprise given the variety of theoretical arguments for the effects of spouses’ socioeconomic resources, some being economic in nature (e.g., Becker, 1981) and others sociological (e.g., Parsons, 1949). The results are most consistent when looking at the husband’s and wife’s employment statuses and least consistent in case of income measures (e.g., Cherlin, 1979; Spitze & South, 1985; Greenstein, 1990; Hofmann & Duncan, 1995; Ono, 1998; Sayer & Bianchi, 2000). We therefore consider several indicators for the mother’s and father’s socioeconomic position simultaneously: employment status, wage rate and education. As to parents’ education, it should be noted that education not only reflects someone’s socioeconomic resources but also socialcultural resources, such as better communication skills, implying a smaller chance of divorce (e.g., Ono, 1998) or modern attitudes, implying a higher chance of divorce, at least for women (e.g., Greenstein 1995; Kaufman, 2000; Sayer & Bianchi, 2000). We also control for family income, because a higher family income, irrespective of the spouses’ separate socioeconomic resources, is argued to lead to a lower chance of divorce because of the fewer financial problems couples encounter (e.g., Cherlin, 1979). Marital characteristics refer to the duration of the marriage and marital happiness. Although the divorce risk is sometimes found to increase during the first years of marriage and to decline at later marital durations (e.g., Morgan & Rindfuss, 1985; Ono, 1998), preliminary analyses showed that a linear specification of the duration effect fitted the data best. This is probably because most couples in our sample are already married
for quite some years (mean length is 13 years) at wave 1 and already ‘survived’ the first unstable years of marriage. Marital happiness is consistently found to decrease the risk of divorce (e.g., Booth et al., 1985; White & Booth, 1991; Sayer & Bianchi, 2000) and is also central to our analysis when we assess whether the effect of expected costs varies by marital happiness. In light of the question about the moderating role of marital happiness we constructed a composite variable indicating whether either spouse thought the marriage is unhappy, instead of using both spouses’ reports separately; it is likely that only one spouse’s unhappiness with the marriage is sufficient to prompt discussions about divorce, leading to greater salience of the expected costs. Empirically, the effects of the mother’s and father’s reports of marital happiness did not differ significantly either. In the multivariate analyses to examine the role of parents’ expected costs for marital dissolution, we use weighted logistic regression of marital dissolution on our measures of expected costs and the control variables listed in Table 3. As a preliminary step we also estimated a baseline model in which only the control variables are included. The results are shown in the last column of Table 3, and they are generally in line with findings from prior studies. The only exceptions are the positive effects of the number of children and family income on the chance to divorce. Perhaps the reason for these counterintuitive findings is that our sample is limited to parents, whereas most other studies combine parents and non-parents. With respect to the number of children, additional analyses (not shown) for example show that the destabilizing effect particularly refers to couples who have four or more children; parents of large families are significantly more likely to divorce than those from smaller families (see also Heaton, 1990).
The analyses to come consist of two steps. In the first step, we examine the relation between parents’ expected costs and marital dissolution by estimating two models. The first model contains the four separate measures for mothers’ and fathers’ expected economic and parenting costs. This model not only shows whether expected costs matter, but also which costs weigh more heavily for mothers and fathers in the decision to divorce (and thus have a stronger effect on the risk of divorce). We also test whether the effect of economic costs is equal to the effect of parenting costs for mothers and fathers. In the second model, we include the measures for the mother’s vis-à-vis the father’s expected economic and parenting costs, which indicate whether and how they agree or disagree about the costs. This model allows us to examine whether the mother’s expected costs weigh more heavily than the father’s in the decision to divorce, in case they disagree. We also test whether the effect of the mother expecting higher costs is equal to the effect of the father expecting higher costs. In the second step, we examine whether expected costs deter divorce more for unhappy couples by making a distinction between happy and unhappy couples. Both models are estimated for unhappy and happy couples separately, and to assess whether the effects of expected costs differ significantly we estimated interaction effects as well.
Multivariate results Table 4 presents the results of the weighted logistic regression models of marital dissolution on parents’ expected economic and parenting costs. The results for Model 1, including separate measures for mothers’ and fathers’ costs, show that both mothers’ and
fathers’ anticipated economic costs do not deter divorce. When mothers expect that their post-divorce standard of living will be worse, couples are not significantly more likely to split up than when mothers do not expect such a decline, and the same is found for fathers’ expected economic costs. When it comes to parenting, however, mothers’ and fathers’ expected costs do matter. When mothers expect that parenting will be worse, couples have a 41 percent ([1-e-.533]*100%) lower odds of divorce than when the mother does not expect such costs, and this figure is 36 percent for fathers’ expected parenting costs. Although these results suggest that parenting costs weigh more heavily in the decision to divorce than economic costs, the differences in the effects of economic and parenting costs for mothers and fathers are not large enough to be significant. The Wald tests of whether the effects of economic and parenting costs are equal are insignificant, and such equality can therefore not be statistically rejected.
[Here Table 4]
Model 2 includes the measures for agreement about costs between parents from the same household. In case of expected economic costs, all coefficients have a negative sign, which is not surprising given that the reference group consists of families in which parents’ agree that there are no economic costs implying that neither sees barriers to divorce in this respect. We further observe that only one of the estimates is significant. The significant estimate surprisingly does not pertain to families where both parents expect high costs, which would be expected to be the least divorce prone because both parents perceive barriers to divorce. Rather, the effect is significant for families where the
mother expects higher costs than the father; compared to families where both parents perceive no economic costs, couples where the mother expects her standard of living will be worse but the father does not have a 41 percent lower odds to divorce. Because a smaller and insignificant effect is found for the reverse case of disagreement (i.e., where the father expects higher economic costs), the mother’s expected costs seem, as hypothesized, to matter more for divorce than the father’s in case they disagree. However, the difference in effects is small and the results for the Wald-test confirm that there is no significant difference between the effects of the mother expecting higher economic costs and of the father expecting higher costs. The results for parenting costs in Model 2, again show that high parenting costs strongly reduce the risk of marital dissolution. Compared to families where both expect no costs, all other types of families — where at least one parent expects parenting to be worse — are significantly less likely to divorce. The effect is strongest for families in which both parents expect costs and least strong in case the father expects higher parenting costs than the mother, with the effect of the mother expecting higher costs than the father being intermediate. In case of disagreement between parents, the odds of divorce are 51 percent lower in case the mother expects higher parenting costs than the father and 42 percent lower in case the father perceives of higher costs compared to families in which both expect no costs. Although this does seem to support our hypothesis that the mother’s parental concerns matter more than the father’s, the Wald-test shows that this difference is not significant.
[Here Table 5]
Table 5 presents the results of the analyses to examine whether expected costs actually deter divorce more when a couple is unhappily married. A comparison of the effects of expected costs between unhappy and happy couples shows whether this is the case. The interaction effects, indicating the size of the differences in effects, shows whether unhappy couples differ significantly from happy ones in this respect. Although the results for Model 1 in Table 4 showed that economic costs do not deter divorce in the overall sample, they may still matter when only unhappily married couples are considered since expected costs have been argued to be particularly relevant in this case. However, the results for Model 1 in Table 5 show that parents’ expected economic costs do not deter divorce in either happy or unhappy couples, and the effects do not differ significantly. For parenting costs, a different pattern emerges. In case couples are happily married, mothers’ expected parenting costs do not affect the chance of divorce, whereas fathers’ costs do. For unhappy couples we observe the reverse; mothers’ expected parenting costs strongly deter divorce, whereas fathers’ do not. When it comes to mothers’ anticipated difficulties in parenting, the effects also differ significantly between happy and unhappy couples. So for mothers’ expected parenting costs, these results are in line with our hypothesis that expected costs are stronger deterrents of divorce in unhappy than in happy marriages. As to fathers’ expected parenting costs, the stronger effect for happy marriages compared to unhappy marriages is opposite to what we expected, but the difference in effects is not significant. Interestingly, the Wald-tests now suggest that in unhappy couples, mothers’ expected parenting costs weigh more heavily in the decision to divorce than mothers’ expected economic costs.
In Model 2, the role of anticipated economic difficulties is again limited. Even for unhappily married couples, we only see a marginally significant lower odds of divorce for couples where only the mother expects her standard of living will be worse compared to those where both parents expect no costs. Furthermore, none of the effects differs significantly between happy and unhappy couples, which does not support our hypothesis. The results for parenting costs are more in line with expectations. Whereas expected parenting costs are not related to the chances of divorce for happily married couples, they strongly affect the risk of divorce when the couple is unhappy. In unhappy couples, it is sufficient that at least one of the parents expects parenting will be worse to deter divorce. Particularly, when both parents expect high costs or when the mother expects higher parenting costs than the father divorce is less likely compared to couples where both expect no costs. The estimates for these types of families also differ significantly between happy and unhappy couples, which lends support for our hypothesis that costs matter more for unhappy couples. Finally, the Wald-tests point at differences in the importance of the mother’s and the father’s expected costs for divorce. Although we observe a significant difference in effects for happy couples, the coefficients for parenting costs were not significant to begin with. In case of unhappy couples, where we do find significant effects for parenting costs, the Wald-test shows that the mother’s concerns about parenting matter more for divorce than the father’s, which is in line with our hypothesis in this respect.
Conclusion and discussion In this study we aimed at gaining more insight into the role of parents’ anticipated economic and parenting costs for their chance to divorce. Previous studies have argued that children lower the risk of divorce because both parents anticipate difficulties in postdivorce childrearing, but mothers mainly expect high economic costs and fathers parenting costs (e.g., Kalmijn, 1999). We elaborated upon this argument, by arguing that both mothers and fathers may expect economic as well as parenting costs and specifying the conditions under which the mother’s and father’s costs are more or less important for a couple’s chance to divorce. On the basis of previous literature (e.g., Levinger, 1965; Bernard, 1976), we argued that the mother’s concerns about the post-divorce situation may override the father’s when their perceptions about the costs do not coincide, and that anticipated costs become all the more important when couples are actually in the phase that they consider a divorce. The descriptive analyses showed that mothers and fathers do expect different costs. Mothers generally expect greater economic difficulties than fathers, whereas the reverse holds for expected parenting costs after divorce. Although this gender difference reflects the predominant practice of mother-sole custody, we also found that the majority of mothers expects high parenting costs as well, just as about half of the fathers expect considerable economic costs. Most mothers and fathers therefore seem to anticipate a worsening in both their economic situation and parenting role. When the reported costs of parents in the same household are compared to each other, parents were found to only partly agree about the costs and in this case, most parents agree that their standard of living and parenting role is going to be worse. In a considerable number of cases, parents
were found to disagree about the expected costs, which points at the possibility of conflicting interests when they have to decide whether or not to divorce. The descriptive findings raise questions about which costs matter more for divorce, economic or parenting costs, and whose concerns, the mother’s or father’s, are more important in case of conflicting interests. Although we did not explicate beforehand which costs would matter more, the multivariate analyses showed that expected parental costs tend to be more important than economic ones. Whereas mothers’ and fathers’ expected economic costs were found to be of little importance, parents’ anticipated parenting costs strongly reduced the divorce risk — although the differences did not always reach significance. Apparently, the special bond between parents and their children leads parents to attach greater weight to their concerns about their parenting role than to the possible adverse economic consequences, which more indirectly affect the relation with their children and their children’s well-being. As to the question whose concerns matter more, the findings indicate that the mother’s expected costs weigh more heavily in the decision to divorce than the father’s in case they disagree. We consistently found that when the mother expects greater costs than the father, divorce is less likely than in the opposite case where the father expects greater costs than the mother. This particularly holds for parenting costs, for which we also found a significant difference between these two situations. The finding that the mother’s concerns are more important role is in line with results from earlier studies on gender differences in emotional labor in marriage, which indicate that women are more affected by relational problems, more likely to express their concerns and act upon problems by initiating divorce (e.g., Pettit & Bloom, 1984; Thompson & Walker, 1989; Amato &
Rogers, 1997). This is not to say that the father’s concerns do not matter — in fact, their parental concerns do — , but in case of conflicting interests the mother’s bigger concerns are more likely to delay or prevent divorce than when the father expects higher costs. Finally, we found that the stage at which couples are in the decision making process matters as well. At least for parental costs, it was found that anticipated difficulties deter divorce more when the couple is unhappily married than when they are happy together. The stronger effect of parental costs in unhappy marriages, confirms the previously stated idea that people’s perceptions of the post-divorce situation only become salient and important deterrents to divorce when the option to divorce is more than hypothetical (e.g. Levinger, 1965; Knoester & Booth, 2000; Amato & Previtti, 2003). Although earlier findings on this issue were inconsistent (Heaton & White, 1991; Knoester & Booth, 2000), our supportive findings may be due to our focus on parents, for whom more is at stake than for childless couples. All in all, our findings show that parents’ expected costs do deter divorce, but that the relation between costs and divorce is conditioned by which and whose costs are considered, as well as the stage of the divorce process. We advanced upon earlier studies in this area in that we focused on parents and used both parents’ subjective reports on economic as well as parents costs. As such, we were able to show that, contrary to the common assumption that children deter divorce because mothers expect economic and fathers parenting costs, it is mainly a matter of parents’ parenting instead of economic costs. Furthermore, the use of both parents’ reports from the same household, yielded new evidence for the idea that women do most of the emotional labor in families — that is, also in the decision making process leading up to divorce. And lastly, our focus on
parents, has probably led to stronger evidence for the importance of expected costs and the moderating role of marital happiness, because particularly parents may be concerned with the post-divorce situation. Future research, using couple instead of person data, may examine whether the conditions we found to be important for the relation between costs and divorce, also apply to non-parents as well as other types of costs.
Of the original 5405 cases married at t1 and still married or divorced/separated at t2, we
excluded cases (N=51) in which (former) spouses’ reports on marital status at t2 did not coincide because these cases raise questions about whether or not a divorce or separation actually occurred. 2.
Respondents were asked about the perceived consequences of divorce with respect to six areas
in life. To be included in our sample, they had to have valid answers on all six items about expected costs of divorce — not only parenting and economic costs — , so we can be more sure that the resulting sample has at least made an effort to think about the possible consequences of divorce. 3.
Besides this substantive argument, there are also empirical reasons why we limit our analyses to
economic and parenting costs. First, particularly the expected costs with respect to overall happiness correlate strongly with expected costs in other areas for mothers and fathers (max. r=0.65), which is no surprise considering that overall happiness is likely to incorporate all other aspects. Second, additional multivariate analyses of the relation between expected costs and marital dissolution, in which the other expected costs are included as well (extension of Model 1 in Table 4), do not yield substantially different results. Even when the costs as to overall happiness are included, mothers’ and fathers’ parenting costs remain significant, although the coefficients become smaller. Moreover, the other costs are not significantly related to divorce, except for mothers’ expected changes in overall happiness. 4.
Of all couples having also non-biological children in the household (N=247), the great majority
consists of couples having stepchildren (87%) and due to the complexities of stepfamilies, stepchildren are probably also more likely to affect divorce than adoptive or foster children.
References Amato, P. R., & Rogers, S. J. (1997). A Longitudinal Study of Marital Problems and Subsequent Divorce. Journal of Marriage and the Family, 59: 612-24. Becker, G. S. (1981). A Treatise on the Family. Cambridge: Harvard University Press. Becker, G. S., Landes, E. M., & Michael, R. T. (1977). An Economic Analysis of Marital Instability. Journal of Political Economy, 85: 1141-87. Bernard, J. (1976). The Future of Marriage. Toronto: Bantam Books. Bianchi, S. M., Subaiya, L., & Kahn, J. R. (1999). The Gender Gap in the Economic Well-Being of Nonresident Fathers and Custodial Mothers. Demography, 36: 195-203. Booth, A., Johnson, D. R., White, L. K., & Edwards, J. N. (1985). Predicting Divorce and Permanent Separation. Journal of Family Issues, 6: 331-46. Buehler, C., & Gerard, J. M. (1995). Divorce Law in the United States: A Focus on Child Custody. Family Relations, 44: 439-58. Bumpass, L. L., Martin, T. C., & Sweet, J. A. (1991). The Impact of Family Background and Early Marital Factors on Marital Disruption. Journal of Family Issues, 12: 2242. Call, V. R. A., & Heaton, T. B.(1997). Religious Influence on Marital Stability. Journal for the Scientific Study of Religion, 36: 382-92. Cancian, M. & Meyer, D. R. (1998). Who Gets Custody? Demography, 35: 147-57. Cherlin, A. J. (1979). Work Life and Marital Dissolution. Pp. 151-66 in G. Levinger, & O. C. Moles (eds.), Divorce and Separation: Context, Causes and Consequences. New York: Basic Books.
Dechter, A. R. (1992). The Effect of Women’s Economic Independence on Union Dissolution. CDE Working Paper no 92-28. Center for Demography and Ecology, University of Wisconsin, Madison. Fox, G. R., & Kelly, R. F. (1995). Determinants of Child Custody Arrangements at Divorce. Journal of Marriage and the Family, 57: 693-708. Garasky, S., & Meyer, D. R. (1996). Reconsidering the Increase in Father-only Families. Demography, 33: 385-93. Greenstein, T. N. (1990). Marital Disruption and the Employment of Married Women. Journal of Marriage and the Family, 52: 657-76. Greenstein, T. N. (1995). Gender Ideology, Marital Disruption, and the Employment of Married Women. Journal of Marriage and the Family, 57: 31-42. Heaton, T. B., & Albrecht, S. L. (1991). Stable Unhappy Marriages. Journal of Marriage and the Family, 53: 747-58. Heaton, T. B. (1990). Marital Stability Throughout the Child-Rearing Years. Demography, 27: 55-63. Hoffman, S. D., & Duncan, G. J. (1995). The Effect of Income, Wages, and AFDC Benefits on Marital Disruption. Journal of Human Resources, 30: 19-41. Kalmijn, M. (1999). Father Involvement in Childrearing and the Perceived Stability of Marriage. Journal of Marriage and the Family, 61: 409-21. Kaufman, G. (2000). Do Gender Role Attitudes Matter? Journal of Family Issues, 21: 128-44. Kitson, G. C. (1992). Portrait of Divorce: Adjustment to Marital Breakdown. New York: Guilford.
Knoester, C., & Booth, A. (2000). Barriers to Divorce. When Are They Effective? When Are They Not? Journal of Family Issues, 21: 78-99 Levinger, G. (1965). Marital Cohesiveness and Dissolution: An Integrative Review. Journal of Marriage and the Family, 27: 19-28. Levinger, G. (1979). A Social Psychological Perspective on Marital Dissolution. Pp. 3760 in G. Levinger & O. C. Moles (eds.), Divorce and Separation: Context, Causes and Consequences. New York: Basic Books. Manning, W. D., & Smock, P. J. (1999). New Families and Non-resident Father-Child Visitation. Social Forces, 78: 87-116. Meyer, D. R., & Garasky, S. (1993). Custodial Fathers: Myths, Realities and Child Support Policy. Journal of Marriage and the Family, 55:73-89. Morgan, S. P., Lye, D. N., & Condran, G. A. (1988). Sons, Daughters, and the Risk of Marital Disruption. American Journal of Sociology, 94: 110-29. Morgan, S. P. & Rindfuss, R. R. (1985). Marital Disruption: Structural and Temporal Dimensions. American Journal of Sociology, 90: 1055-77. Ono, H. (1998). Husband’s and Wife’s Resources and Marital Dissolution. Journal of Marriage and the Family, 60: 674-89. Parsons, T. (1949). The Social Structure of the Family. Pp. 173-201 in R. N. Anshen (ed.), The family: Its function and destiny. New York: Harper and Brothers. Pettit, E. J., & Bloom, B. L. (1984). Whose Decision Was It? The Effects of Initiator Status on Adjustment to Marital Disruption. Journal of Marriage and the Family, 46: 587-95.
Previti, D., & Amato, P. R. (2003). Why Stay Married? Rewards, Barriers, and Marital Stability. Journal of Marriage and Family, 65: 561-73 Remez, L. (1990). Marital Dissolution Risk Affected by Number and Age of Couple’s Children. Family Planning Perspectives, 22: 238. Sayer, L. C., & Bianchi, S. M. (2000). Women’s Economic Independence and the Probability of Divorce. Journal of Family Issues, 21: 906-43. Seltzer, J. A. (1990). Legal and Physical Custody Arrangements in Recent Divorces. Social Science Quarterly, 71: 250-66. Seltzer, J. A. (1991). Relationships Between Fathers and Children Who Live Apart: The Father’s Role After Separation. Journal of Marriage and the Family, 53: 79-101. Seltzer, J. A. (2000). Families Formed Outside of Marriage. Journal of Marriage and the Family, 2000, 62: 1247-68 Smock, P. J. (1994). Gender and the Short-Run Economic Consequences of Marital Disruption. Social Forces, 73: 243-62. South, S. J., & Spitze, G. (1986). Determinants of Divorce over the Marital Life Course. American Sociological Review, 51: 583-90. Spitze, G. & South, S. J. (1985). Women’s Employment, Time Expenditure and Divorce. Journal of Family Issues, 6: 307-29. Stephens, L. S. (1996). Will Johnny See Daddy This Week? An Empirical Test of Three Theoretical Perspectives of Post-divorce Contact. Journal of Family issues, 17: 466-494.
Sweet, J. A., & Bumpass, L. L. (1996). The National Survey of Families and Households - Waves 1 and 2: Data Description and Documentation. Center for Demography and Ecology, University of Wisconsin, Madison Sweet, J., Bumpass, L. L., & Call, V. (1988). The Design and Content of the National Survey of Families and Households. NSFH Working Paper No. 1. Center for Demography and Ecology. University of Wisconsin, Madison. Thompson, L.,
& Walker, A. J. (1989). Gender in Families: Women and Men in
Marriage, Work, and Parenthood. Journal of Marriage and the Family, 51: 845-71. Waite, L. J., & Lillard, L.A. (1991). Children and Marital Disruption. American Journal of Sociology, 96: 930-53. White, L. K., & Booth, A. (1985). The Quality and Stability of Remarriages: The Role of Stepchildren. American Sociological Review, 50: 689-98. White, L. K., & Booth, A. (1991). Divorce over the Life Course. The Role of Marital Happiness. Journal of Family Issues, 12: 5-21. Wolfinger, N. H. (1999). Trends in the Intergenerational Transmission of Divorce. Demography, 36: 415-20.
Table 1. Married Mothers’ and Fathers’ Expectations about Standard of Living and Parenting If They Were to Divorce Expectation Standard of Living
Much better (%)
Percent worse or much worse
Much better (%)
Percent worse or much worse
Notes: NSFH-data. Sample is parents living with at least one biological child under age 18 at first interview. Percentages may not equal 100 due to rounding. Data are weighted. Unweighted N = 1935. a Difference between mothers and fathers statistically significant p ≤ .01.
Table 2. Agreement between Married Mothers’ and Fathers’ Expectations about Standard of Living and Parenting If They Were to Divorce Agreement
Standard of Living (%) 13.8
Both expect it will not be worse
Parenting (%) 16.4
Both expect it will be worse
Mother expects worse, father does not
Father expects worse, mother does not
Notes: NSFH-data. Sample is parents living with at least one biological child under age 18 at first interview. Data are weighted. Unweighted N = 1935.
Table 3. Means and Standard Errors of the Control Variables and Logistic Regression of Marital Dissolution on the Control Variables Control variables Duration between wave 1 and 2 (in years)
Mother non-white (whether she describes herself as Hispanic, Black or other non-white ethnicity)
Mother’s age at marriage (in years)
Baseline model 0.634** 0.206
Mother at least once a week to church
Mother ever cohabited (with current spouse and/or other partners)
Parents of either divorced (parents (adoptive or biological) of either spouse divorced before he/she turned 19 or left the parental home)
Either divorced (either spouse divorced or separated prior to current marriage)
Number of biological children under 18 years old (biological to both spouses)
Stepchildren in household
Youngest biological child under five years old
All biological children under 18 are sons
All biological children under 18 are daughters
Mother high school or less
Father high school or less
Father works b
Family income (in 10,000; of all family members)
Missing indicator for family income
Mother’s hourly wage rate (average if not working) Father’s hourly wage rate (average if not working) b
Duration marriage (in years)
Couple unhappy (whether either spouse described the marriage as less than (very) happy)
Notes: NSFH-data.Sample is parents living with at least one biological child under age 18 at first interview. Data are weighted. Unweighted N=1935. a Not for dichotomous variables; b Assigning the average implies that the effect of the wage rate only applies to working persons and that the effect of employment status indicates the difference between nonworking persons and persons with an average wage rate. ~ p ≤ 0.10 ; * p ≤ 0.05; ** p ≤ 0.01
Table 4: Logistic Regression of Marital Dissolution on Parents’ Expected Economic and Parenting costs Expected costs Standard of living
Mother expects worse
Father expects worse
Both expect worse (versus both not worse)
Mother expects worse, father not
(versus both not worse) Father expects worse, mother not
(versus both not worse) Being a parent Mother expects worse
Father expects worse
Both expect worse (versus both not worse)
Mother expects worse, father not
(versus both not worse) Father expects worse, mother not
(versus both not worse) Wald tests a Mother economic worse=mother parenting worse
Father economic worse=father parenting worse
Mother expects economic worse, father not = father expects economic worse, mother not
Mother expects parenting worse, father not=father expects parenting worse, mother not
Notes: NSFH-data. Sample is parents living with at least one biological child under age 18 at first interview. Models control for the control variables of Table 3. Data are weighted. Unweighted N is 1935. a Coefficient for interaction term with marital happiness. ~ p ≤ 0.10 ; * p ≤ 0.05; ** p ≤ 0.01
Table 5. Logistic Regression of Marital Dissolution on Parents’ Expected Economic and Parenting Costs by Whether Couples Have a Happy Marriage Expected costs
Interaction effect a
Mother expects worse
Father expects worse
Unhappy Interaction effect a
Both expect worse
Mother expects worse, father not
Father expects worse, mother not
Both expect worse
Mother expects worse, father not
Father expects worse, mother not
Standard of living
Being a parent Mother expects worse
Father expects worse
Wald tests b Mother economic worse=mother parenting worse
Father economic worse=father parenting worse
Mother economic worse, father not = father economic worse, mother not
Mother parenting worse, father not=father parenting worse, mother not
Notes: NSFH-data. Sample is parents living with at least one biological child under age 18 at first interview. Models control for the control variables of Table 3. Data are weighted. Unweighted N for happy marriages is 743, and for unhappy ones 1192. a Coefficient for interaction term with marital happiness in a model in which all variables are interacted with marital happiness. b F (1, n-1) statistic displayed. ~ p ≤ 0.10 ; * p ≤ 0.05; ** p ≤ 0.01