Couple Disagreement about Short-Term Fertility Desires in Austria: Effects on Intentions and Contraceptive Behaviour
Testa, Maria Rita, Demographic Research
Because of the dyadic nature of reproduction, the couple is the most suitable context for studying reproductive decision-making.
I investigate the effects of couple disagreement about short-term childbearing desires on the formulation and implementation of fertility intentions. Do men and women incorporate the perception of a disagreement with the partner about wanting a(nother) child now in their reports on short-term fertility intentions and contraceptive behaviour? Are there relevant differences by type of disagreement, parity, gender and gender equality within the couple?
Using individual-level data from the Austrian Generation and Gender Survey conducted in 2008, I regress respondent's short-term fertility intentions (ordinal regression models) and non-use of contraception (logistic regression model) on couple's short-term childbearing desires and a set of background variables.
The findings show that disagreement is shifted toward a pregnancy intention\pregnancy-seeking behaviour at parity zero and toward avoiding pregnancy and maintaining contraceptive use at higher parities. Childless women are less responsive to the perception of their partner's desires than childless men when they express their short-term childbearing intentions. Neither women nor men are likely to stop contraception if they perceive a disagreement with their partner about wanting a(nother) child. Moreover, if the man is actively involved in childcare duties the chance to resolve the couple conflict in favour of childbearing increases.
This paper calls for the collection of data from both members of each couple so that the analysis of the partner's actual desires can complement the analysis of the partner's perceived desires.
In this paper, I address the issue of fertility decision-making among couples using individual-level data. Following the approach proposed by Miller and Pasta (1996), a distinction is made between respondents who perceive a positive disagreement (with the partner having weaker desires than the respondent), and respondents who perceive a negative disagreement (with the partner having stronger desires than the respondent).
The importance of including both partners in the fertility analysis is well recognised (Ryder 1973). Several studies have adopted a couple-oriented approach (Fried and Udry 1979; Coombs and Chang 1981; Morgan 1985; Thomson et al. 1990; Corijn et al. 1996; Thomson 1997; Thomson and Hoem 1998; Jansen and Liefbroer 2006; Miller and Pasta 1996; Miller et al. 2004; Becker 1996), fertility research, however, has been primarily based on the female perspective. This choice has been justified by the high degree of homogamy within the couple, and the fact that women are the main actors and the most reliable reporters of childbearing events.
A major difficulty in the couple-level research lies in the need to have high quality data that collect information on both partners, possibly in repeated waves, which allow researchers to ascertain the differences between partners' reproductive goals, and to identify the contribution of each partner to the final childbearing outcome. If this is true for every country, the lack of adequate data is an even greater problem in European countries, where longitudinal household surveys have only rarely been conducted in recent decades.
Some surveys have tried to get around the difficulties related to the collection of couple data by asking individual respondents questions about their partner's childbearing desires. Empirical findings based on this kind of data have demonstrated that the respondent's reports about the couple's fertility intentions are influenced almost twice as strongly by the individual's own desire to have a child as by the perception of the partner's desire to have a child (Morgan 1985). Moreover, respondent's proxy reports of spouse's desired family size are subject to both systematic and random error (Williams and Thomson 1985).
Estimating the degree of egocentric biases contained in respondents' perceptions of their partner's fertility desires is beyond the scope of the current paper. My basic assumption is that the partner's perceived and actual childbearing desires are different measures, both contributing to creating the context in which individual fertility decisions are made (Miller et al. 2004).
The study addresses the following research questions: Do men and women incorporate the perception of a disagreement with the partner about wanting a(nother) child now in their reports on short-term fertility intentions and contraceptive behaviour? Are there relevant differences by type of disagreement, parity, gender and gender equality within the couple?
The empirical analysis is based on a sample of 3,402 individuals from the survey "Familienentwicklung in Österreich" conducted by the Austrian National Statistical Office (Statistik Austria) in 2008. Austria is a country with low gender equity and high incompatibility between childrearing and employment. In addition to lacking couple level data, which is the case for most European countries, Austria has just few micro-level studies on fertility intentions.
This article is organised as follows. I begin by reviewing the relevant literature on couple fertility decision-making (Section 2). This is followed by a statement of the research hypotheses (Section 3). I present the data and the methodology in Section 4, and describe the results of the statistical analyses in Section 5. In Section 6 I compare the main findings with those that could have been observed by using the actual partner's desires rather than the respondent's proxy report of them. The most relevant results are summarised and discussed in the concluding section.
2. The role of partner context in theories of fertility decision-making
There are two major theoretical frameworks suitable for studying fertility decision-making: the Theory of Planned Behaviour (TPB) and the Theory of Traits-Desires-Intentions-Behaviour (TDIB). I will review both with the intention of showing how they take into account the dyadic aspect of reproduction. Moreover, I will discuss an overview of decision rules used to resolve conflict.
The Theory of Planned Behaviour (Ajzen 1991) was first applied in the domain of fertility decision-making by Billari et al. 2009. It studies intentions as an immediate forerunner of the corresponding behaviour, and views intentions as being formulated under the immediate influence of three groups of factors: (a) personal positive and negative attitudes towards the behaviour, i.e., having a child; (b) subjective norms, i.e., perceived social pressure to engage or not to engage in the behaviour; and (c) perceived behavioural control, i.e., ability to perform the behaviour which may depend, for example, on the availability of housing, income, or other resources.
The partner's intentions are not explicitly considered in the theory, but it may be implicitly assumed that the perception of a disagreement with his/her partner may influence an individual's normative beliefs. An individual who wants to have a(nother) child, and perceives that his/her partner does not share this wish, is likely to form the belief that the partner does not want her/him to have a(nother) child. This perception may clearly influence the respondent's fertility intentions. In a recent conference on reproductive decision-making, Ajzen clarified that the close link between intentions and subsequent behaviour holds true only if the behaviour is specified in all of its four components: namely, the target, the action, the context and the time (Ajzen 2010). In the field of fertility, the target is a child, the action is giving birth, the context could be the current partner, and the time could be a short-term horizon, which may make the intentions more realistic. Consistent empirical evidence has been collected on the crucial importance of the partner's context for the construction of pregnancy intentions (Barret and Wellings 2002; Zabin et al. 2000).
The Theory of Planned Behaviour has been criticised because it does not explicitly consider the complexity of the dyadic nature of reproduction (Philipov 2011) nor does it describe the disagreement effects of couple's decisional conflicts (Miller and Pasta 1996).
Unlike the TPB theory, the traits-desires-intentions-behaviour theory (Miller 1986 and 1994) has been reformulated to explicitly consider the dyadic nature of reproduction (Miller and Pasta 1996; Miller et al. 2004). This theoretical framework sees the behaviour that determines whether or not a pregnancy occurs as the last step of a motivational sequence which encompasses four major stages. The first one refers to motivational traits, or the dispositions to feel, think, and behave in certain ways with respect to fertility; the second stage concerns the desires, emotional feelings or conscious wishes which do not lead directly to action; the third stage refers to intentions, desires constrained by reality, psychological states that represent conscious commitment to act in a certain way or to achieve a certain goal at some future time. Eventually, the reproductive behaviour is aimed at achieving (proceptive) or avoiding (contraceptive) a pregnancy. According to the TDIB intentions are assumed to incorporate the perception of desires of significant others, above all the partner, as well as other situational factors that may prevent individuals from simply doing what they want. This distinction is worth emphasising, given that the focus of my analysis is on the translation of desires into intentions and contraceptive behaviour. In the couple-based version of the theory, the interaction between partners is examined at each stage of the sequence, and the perception of the partner is considered an important element in the individual's construction and performance of the subsequent stages of the motivational structure (Miller et al. 2004).
Miller and Pasta (1996) have disentangled two main components of disagreement between partners which influence the individual's decision-making differently. The signed difference component describes an influence effect, i.e. which member of the couple has more or less influence on the behaviour. The absolute difference component describes a conflict effect, and it is independent of which outcome is desired by either partner. The latter causes a delay in fertility decision-making due to inertia, which tends to favour the partner who does not want to have a child in a context in which using contraception between births is standard practice (Davidson and Beach 1981; Beach et al. 1982) Indeed, research has shown that the behaviour of couples with conflicting childbearing desires is more similar to that of couples in which both partners want no (more) children than to that of couples with a shared desire for (more) children (Miller and Pasta 1994), which suggests the presence of a symmetric veto power model (Thomson 1997; Thomson and Hoem 1998; Voas 2003). The influence effect will depend on which of the partners has stronger and weaker childbearing desires, the level of gender equality between the partners, and the decision-making rules within the couple. Previous studies have demonstrated that, in low-fertility societies, the decision-making process tends to be dominated by the woman, who controls most methods of contraception (Beckman 1983). Moreover, a couple who disagree will tend to seek pregnancy if the wife's subjective expected utility favours pregnancy (Townes et al. 1980). Empirical evidence indicates that women have greater influence over the negotiation process in societies in which childbearing lies mainly in the women's sphere of interest (Fried et al. 1980; Beckman 1984; Rindfuss et al. 1988; Miller and Pasta 1994).
Jansen and Liefbroer (2006) offer an overview of the different decision rules used to resolve couple's conflict. According to the power rule, the partner who has greater access to socio-economic resources will prevail. As long as men have higher occupational and income levels than women they will predominate in the couple's negotiation process. Men will always exert a greater influence on fertility decisions in patriarchal cultures characterised by gendered power relations. A second heuristic is the sphere-of-interest rule, according to which the partner in whose sphere of interest a decision is located will have more influence over subsequent behaviour. As long as childbearing tends to lie in the female sphere of interest women will be more influential in the couple's fertility decision-making. A third heuristic is the social-drift rule, according to which the maintenance of the status quo will prevail by favouring the partner who does not want to have children if contraception between births is routine. Women who perceive their broader environment as being unpredictable develop a life style characterised by social drift and they respond to events like pregnancy as they happen rather than deliberately causing them to happen through their own efforts. Jansen and Liefbroer (2006) argued that in the Netherlands such an attitude controls the couples' reproductive choices, if neither of the partners has a clear intent to impose her or his own point of view on the other. Thomson (1997) and Thomson and Hoem (1998) demonstrated that in the US and Sweden the patterns of implementation of childbearing desires in childbearing intentions and birth were not different for couples with more or less traditional gender role attitudes.
3. Research hypotheses
I suppose that people incorporate the perceptions of their partner's childbearing desires into their own reports on fertility intentions and contraceptive behaviour but the extent to which they do so varies by the type of decision each partner wants to make, as well as by parity, gender and gender equality within the couple.
I expect that both women and men tend to emphasize their own desires over their partner's perceived desires in their reports of their own short-term fertility intentions. Hence, once disagreement is disaggregated into a positive and a negative component I hypothesise that men and women who want to have a child now and perceive that their partner does not want (positive disagreement) are more likely to report a child intention than men and women who do not want to have a child now but perceive that their partner wants (negative disagreement) (Hypothesis 1a, signed difference effects of disagreement on intentions.) Moreover, since contraceptive behaviour is a couple construct, I expect that the effect of disagreement with the partner about wanting a(nother) child now does not depend on whether respondent's own desires are stronger or weaker than the partner's perceived desires. Hence, I hypothesise that absolute and signed disagreement have the same effects on contraceptive behaviour (Hypothesis 1b, absolute difference effects of disagreement on contraception).
The lack of agreement between partners may favour a decision not to have a child, given that delaying fertility is a normatively acceptable life course strategy, while having a baby has immediate and permanent implications for individuals (Rindfuss et al. 1988; Miller and Pasta 2002). Hence, I hypothesise that men and women who perceive a disagreement with their partner about wanting a(nother) child now show intentions and contraceptive outcomes closer to that of respondents who perceive agreement on not having a(nother) child now than to that of respondents who perceive agreement on having a child now (Hypothesis 2, double-veto power effect).
Since the two-child family is the most prevalent normative model in Europe, I expect a shift of disagreement toward intending a child and stopping contraception at parity zero and one and a shift toward preventing a pregnancy intention and using contraception at parity two or above (Hypothesis 3, two-child norm effect).
Women generally have the primary responsibility for childrearing and face a number of barriers in combining work and family life in Austria (Prskawetz et al. 2008). It is therefore reasonable to assume that gender differences exist in the respondent's responsiveness to the partner's perceived desires. My third research hypothesis may be stated as follows: women are less responsive than men to the perception of a disagreement with their partner about wanting a(nother) child now when they express their own short-term fertility intentions (Hypothesis 4, gender effect).
Fertility is particularly low in those contexts where high levels of gender equity in education and labour market are combined with low levels of gender equity in the family (McDonald 2000). Hence, I expect that gender equality within the couple favours a decision to have another child. I hypothesise that men and women who perceive a disagreement with the partner about wanting a(nother) child now and who live in more gender-egalitarian relationships in terms of gender division of childcare tasks will be more prone to resolve the conflict in favour of childbearing (i.e., express an intention to have a child and stop contraception) than men and women who live in less gender egalitarian relationships (Hypothesis 5, gender equality effect).
In contemporary societies, the intra-household distribution of bargaining power has become an important factor driving the fertility decisions of spouses with different interests (Jansen and Liefbroer 2006). The partner with more socio-economic resources will be more empowered in the reproductive decision-making if the power rule is the predominant heuristic adopted by the couples in conflict. The last research hypothesis may be stated as follows: men and women with less bargaining power than their partner are more likely to reflect their partner's perceived desires into their own short-term fertility intentions and non-use of contraception, while respondents with more bargaining power than their partner are less likely to reflect their partner's perceived desires into their own short-term childbearing intentions and related behaviour (Hypothesis 6, bargaining power effect).
To test these hypotheses, I disaggregated the respondent's desires and the partner's perceived desires into positive and negative components, since the degree of responsiveness that people may have to the partner may vary with the type of the decision that has to be made. Miller and Pasta (1996) have proved the utility of such an approach by showing that each of these two components may vary independently of the other, and each of them may be differently associated with important demographic backgrounds.
4. Data and methods
4.1 Data sample
I used data from the survey Families in Austria (Familienentwicklung in Österreich), conducted by the Austrian National Statistical Office in 2008 in the framework of the international Generations and Gender Programme. The whole sample size included 5,000 individuals, men and women, aged 18-45.
I restricted the analysis to 3,402 heterosexual couples in which the female partner was neither pregnant nor older than age 50 at the time of the interview. Furthermore, I selected only those unions in which neither partner had been surgically sterilised, and the respondent answered both questions about his or her own and the partner's fertility desires.
4.2 Short-term fertility desires, short-term fertility intentions and non-use of contraception
Short-term fertility desires were measured by asking individuals not only about their own fertility desires (Do you yourself want to have a/another child now?), but also about their partner's fertility desires (Does your partner want to have a/another child now?), if they had a partner. The later question was introduced by the following statement: "Couples do not always agree on when and how many children they want". Respondents were given a choice between the following answers: yes, no, or don't know; and in the question about the partner's wishes, they were offered an additional option: partner is not sure. 12 respondents answered "don't know" to the question on their own fertility desires and 15 respondents did so for the question on the partner's desires. Four individuals reported a "don't know" answer to both questions.
Short-term fertility intentions were measured by asking respondents the following question: "Do you intend to have a/another child during the next three years?" The interviewees were asked to indicate on a four-point scale how certain they were about their intentions. The specific options were: "Certainly not", "Probably no", "Probably yes" and "Certainly yes". 22 respondents answered "don't know" to this question.
"Don't know" answers may be informative of particular desires and intentions (Morgan 1981). However, since the uncertain individuals were very few in the Austrian GGS survey, I could not consider them as a distinctive category in the bivariate and multivariate statistical analysis.
The literature on fertility decision-making generally uses the term 'want' to capture wishes, or emotional feelings that do not incorporate any situational constraints, and do not lead directly to action, and the expression 'intent' to measure goal-related actions (Miller 1994). Since, however, the interviewers who administered the questionnaire were not trained to first explain to the respondents the difference between desires and intentions, the extent to which the interviewees may have perceived the distinction between the concepts remains unclear.
Moreover, it is worth noting that the two questions on fertility desires and intentions refer to periods of time of differing lengths: namely, "now" for the desires and "three years" for the intentions. This may have introduced an additional source of ambiguity.
Non-use of contraception was measured by asking respondents the following question: "Do you currently use contraception?" The interviewees could choose between two options: yes or no. 78 respondents answered "don't know" (2% of the analytic sample).
I excluded from the multivariate analyses the 122 individuals reporting a "don't know" answer on any of the items on fertility desires, intentions, or contraceptive behaviour. This eliminated 122 respondents and resulted in a sample size of 3,280 individuals.
4.3 Respondent's perception of agreement/disagreement with the partner about wanting a(nother) child now
I defined the agreement/disagreement variable from a respondent's individual perspective by disaggregating the entire sample into four categories, outlined in Scheme 1: 1) respondents who do not want to have a child and perceive that their partner does not want one either (agreement on no); 2) respondents who do want to have a(nother) child, and perceive that their partner wants to have one as well (agreement on yes); 3) respondents who do not want a(nother) child, but perceive that their partner wants to have a(nother) child or is unsure (respondent's perception of negative disagreement); 4) respondents who want to have a(nother) child, but perceive that their partner does not want a(nother) child or is unsure (respondent's perception of positive disagreement).
Since I made the distinction between the positive and negative types of disagreement from the respondent's perspective, the adjective 'negative' describes the type of disagreement in which the respondent does not want to have a(nother) child and the adjective 'positive' describes the type of disagreement in which respondent does want to have a(nother) child.
I considered disagreement to be any discrepancy between the respondent's and the partner's perceived desire, irrespective of the direction of disagreement (Scheme 1).
In the empirical analysis I labeled the four categories as follows: R yes, believes partner yes (agreement on wanting), R yes, believes P no or unsure (positive disagreement), R no, believes P yes or unsure (negative disagreement), R no, believes P no (agreement on not wanting). R stands for respondent and P for partner.
I used ordinal regression models to analyse the predictors of the respondent's short-term fertility intentions, and logistic regression models for examining the determinants of the respondent's report on non-use of contraception. In the ordinal regression models, the response variable is the intention to have a child in the next three years, which has the following four categories: certainly not, probably not, probably yes, certainly yes. They are assumed to be ordinal, and are coded one, two, three and four. The key covariate of the model is the respondent's perception of agreement/disagreement with the partner about wanting a(nother) child now. This is included as a three-category variable (Model 1), and as a four-category variable (Model 2), depending on whether the difference between the respondent's and the partner's desires is considered as a whole, or is specified as a positive or a negative component. The ordinal regression models assume that the ratio of the odds of two individuals does not depend on the category (the proportional odds property). This assumption, which is known as the parallel regression assumption, is convenient for the parsimony of interpretation, and can be checked using, for instance, the test developed by Brant (1990). Since just a few covariates in each model violated such an assumption, and they did so only slightly, I used the proportional odds models for the sake of simplicity. Estimates of the partial proportional odds models may be provided by the author upon request.
In the logistic regression models, the response variable is a dummy which takes a value equal to one if the respondent says that she/he does not currently use any contraceptive method, and zero otherwise. The key covariate of the model is the respondent's perception of agreement/disagreement with the partner about wanting a(nother) child now. All of the models are stratified by parity, which is defined as the number of children born in the current union. Accordingly, separate models are run on childless persons and parent respondents, i.e., respondents who have at least one common child with the current partner. This method choice is based on the assumption that the predictors of childbearing intentions are different in the two selected groups, which is in line with a conditional-sequential fertility decision-making process (Namboodiri 1972; Bulatao 1981). I test the assumption, in the context of the test about normatively appropriate desires for one, two, or more children.
4.5 Background variables
The same set of explanatory variables was included in all the models: the respondent's gender, the type of union, the duration of the partnership, the man's and the woman's education, the man's and the woman's employment status, the man's and the woman's age, the man's and the woman's stepchildren (i.e., children from a previous union), the number of common children, the number of years that have elapsed since the birth of the last child, the difference in education levels between the partners, men's involvement in childcare duties. The percentage distribution of the variables used in the multivariate analysis is given in Table 1.
I measured men's involvement in childcare tasks with a dummy variable equal to one if men take part in at least six of the following childcare duties: dressing the children, putting them to bed, staying home with them when they are sick, playing with them or spending leisure time with them, helping them with their homework and taking them to school or to kindergarten; and zero otherwise.
Previous studies have shown that partners often differ in their reports on the division of household and childcare tasks (Kamo 2000). These findings seem to be supported by the Austrian GGS data. However, since the data at hand provide information from just one of the partners, I was not able to rigorously test for the possibility of discrepancies in the partners' reports.
I measured the intra-household distribution of bargaining power with the difference in education levels between partners (Lundberg and Pollak 1996). The variable has three categories: the man is better educated than the woman, the man and the woman have the same levels of education, and the woman is better educated than the man. One of the two partners is considered to be better educated if the achieved level of education is higher than that of the partner's, regardless of the size of the gap between the two education levels. A refinement of the variable that takes the size of this gap into account has not been introduced because the sample sizes are too small.
5.1 Descriptive analysis
In Table 2, I looked at the percentages of respondents who were in perceived agreement/disagreement with their partner about wanting a(nother) child now, following the classification described in the methodological section. The share of respondents who perceived that their partner agreed with them was high: 87%. The share who agreed about wanting a(nother) child, 76%, was considerably higher than the share who agreed about not wanting a(nother) child, 11% (Table 2). When interpreting these results, it should be noted that the desires refer to a short time frame, and do not imply any long-term decision. Of the 13% of the respondents who perceived that their partners disagreed with their own fertility desires, they were almost evenly divided between those who wanted a child when their partners were perceived as not wanting a child or being unsure, and vice versa (6% and 7%, respectively). Parent respondents reported negative disagreement more frequently (8%) than positive disagreement (5%) (Table 2).
In Figure 1 I examined the share of interviewees who reported the intent to have a child over a three-year period for each combination of the respondent's own and the partner's perceived desires. In this analysis, individuals who intended to have a child are those who answered yes, either certainly or probably, to the question on fertility intention. The proportion of men and women who formulated a short-term childbearing intention gradually decreased, going from perceived agreement on yes (97% and 91% among childless persons and parent respondents, respectively) to perceived agreement on no (33% and 10%, respectively). The decline was steeper for parents than for childless respondents. The percentage of men and women expressing a childbearing intention in the disagreement categories was between that of respondents who agreed on wanting a child and that of those who agreed on not wanting a child. When disagreement is distinguished by the respondent's and partner's desire for a child now, the respondent's intentions are much closer to those of respondents who perceive partner agreement, whether on having or not having a child. That is, the respondent's desire appears to have a stronger influence on her/his intentions than perception of the partner's desire.
In Figure 2 I looked at the proportion of interviewees who reported non-use of contraception for each combination of the respondent's own and the partner's perceived desires. The share was high if agreement on yes was perceived: 68% and 66%, respectively. It was substantially smaller, between 10% and 34%, in all other combinations of the respondent's own and the partner's perceived desires. Regardless of the direction of the disagreement, contraceptive non-use is substantially reduced in comparison to respondents who perceive agreement on wanting a child. For parents, it is not significantly different from the levels reported by respondents who perceive agreement on not wanting a child.
5.2 Multivariate analysis
In this section I present the findings of the multivariate analysis in two different sub-sections. First, I discuss the estimates related to the socio-demographic determinants of respondents' own short-term fertility intentions and contraceptive behaviour (Tables 3 and 4). Next, I discuss the effect of the disagreement by describing the results of the goodness of fit tests used to accept/reject each of my research hypotheses (Tables 5 and 6). The evidence provided by these tests justifies keeping the signed difference disagreement variables in the final models for intention and the absolute disagreement variable in the final models on contraceptive non-use.
5.2.1 Socio-demographic determinants of the respondent's own short-term fertility intentions and contraceptive behaviour
In Table 3, all the estimates of the ordinal regression models for the respondent's short-term fertility intentions are reported. Short-term fertility intentions were negatively correlated with the man's or the woman's enrolment in education, non-active status and stepchildren if the respondents were childless (1st column of Table 3). They were negatively associated with the length of the relationship, the number of shared children and the years elapsed from the birth of the last child if the respondent already had children at the time of the survey (2nd column of Table 3). In this latter sub-group, the intention to have a child in the short-term was positively associated with the cohabiting status.
In Table 4, all the estimates of the logistic regression models for the respondent's report on the non-use of contraception are provided. Models run on childless and parent groups yielded different results.
In the sub-sample of childless respondents, non-use of contraception was negatively associated with cohabiting status, with the man's high level of education, and the man's stepchildren. It was positively associated with the woman's non-active status (1st column of Table 4). In the sub-sample of parents, the non-use of contraception was negatively associated with the cohabiting status, and positively associated with the man's non-active status (2nd column of Table 4).
As expected, the chance of reporting non-use of contraception was higher if the respondent expressed an intention to have a child in the next three years (Table 4). The fertility intention's coefficients were smaller than those related to the perception of agreement with the partner about wanting a(nother) child now in both models on childless couples and parents. This evidence suggests that the latter may play a more relevant role than the former in predicting fertility-related behaviour.
5.2.2 Effects of disagreement: Testing the research hypotheses
Hypothesis 1a: Signed difference effect of disagreement on intention. To test this hypothesis I compared the fit of a model in which perceived positive and negative disagreements are collapsed in one single category with that of a model in which they are kept separated. As shown in Table 5a, the latter fit the data much better than the former in the case of fertility intentions. Hence, one can argue that, for childbearing intentions, the conflict effect (absolute disagreement) is different from the influence effect (signed disagreement). More specifically, perceived negative disagreement had stronger inhibiting effects on childbearing intentions than perceived positive disagreement: the coefficients for intending a child were 1.06 and 2.40, respectively, for childless persons and 0.93 and 2.50 for parents (Table 5a).
Hypothesis 1b: Absolute effect of disagreement on non-use of contraceptives. The hypothesis that both types of disagreement have the same effect on non-use of contraception could not be rejected, and the combined coefficient was not significantly different from zero (Table 5b). For this reason, the remaining hypotheses were tested using the four-category specification of desires for respondent's fertility intentions, the three-category specification (both types of disagreement combined) for contraceptive non-use. The four-category specification produced essentially the same results for contraceptive non-use but is less parsimonious.
Hypothesis 2: Double-veto power effect. To test this hypothesis, I compared the fit of a model with a linear specification of both partners' combined desires in which disagreement had a score midway between agreement on yes and agreement on no with that of a model in which a categorical variable (N-1 dummies) reflected the possible combinations of the respondent's own and the partner's perceived desires. The linear variable for the partners' combined desires took the value 0 for agreement on no, 1 for disagreement and 2 for agreement on yes. Because equal effects for the two types of disagreement on fertility intentions were rejected above, we would expect this model to be rejected for fertility intentions, and it is. A second model that allows the two forms of disagreement to differ in their effects, but maintains a 'linear' character was also tested: 0=agreement on no, 1=negative disagreement, 2=positive disagreement and 3=agreement on yes (Table 6a). This model was not rejected in comparison to the categorical model and therefore signals that the four categories are not only ordinal but also equidistant. Although there is a 'bias' toward reporting intentions that are more similar to one's own than to one's partner's desires, the bias is symmetrical - it doesn't matter whether one wants or does not want a child.
The category specification of partners' combined desires also fitted the data better than the 0-1-2 specification of the combined partners' desires in the models on non-use of contraception. That is, disagreeing couples were not in the middle of couples who agreed on wanting or not wanting a child. From the magnitude of the absolute disagreement coefficients (beta=0.73 and 0.23 for childless persons and parents, respectively, in Table 6b), one can argue that the perception of a disagreement with the partner about wanting a(nother) child now produced a contraceptive outcome closer to that of an agreement on not wanting a child (beta=0 for both childless persons and parents) than to that of an agreement on wanting a child (beta=2.46 and 1.92 for childless persons and parents, respectively, Table 6b). Remarkably, in the models run on the parents' sub-sample, the effect of disagreement was not statistically significantly different from that of perceived agreement on no, which provides very strong evidence for a double-veto effect.
Hypothesis 3: Two-child norm effect. To test this hypothesis, I compared the fit of a model with an interaction term between a dichotomous variable, indicating whether or not respondents have at least one child, and the partners' combined short-term fertility desires with that of a model in which the interaction for parity zero, one and two or above are contemplated. Since the latter model did not fit better, I reject the hypothesis of norms regarding the two-child family and I support the hypothesis of parenthood as a norm (results not shown but available upon request). This result provides good justification for stratifying the analysis by childless respondents and parents.
Hypothesis 4: Gender effect. To test this hypothesis I compared the fit of two models with and without an interaction term between gender and partners' combined short-term fertility desires. The likelihood ratio test provided some evidence in favour of the full models which included gender effects for the childless sub-sample. In respect to respondents agreeing on not wanting a first child now, childless women were more likely to report a pregnancy intention if they themselves wanted a child but perceived that their partner did not want one (beta=2.89, Table 3). This was also true for men, but to a lesser extent, as evidenced by the magnitude of the related coefficients (beta=2.89-1.01=1.88, first column of Table 3). I could find only weak evidence for a similar gender effect in the models on non-use of contraception among childless respondents. Although the gender interaction term for absolute disagreement had a statistically significant effect on contraception, it did not improve the fit of the model, therefore I did not keep it in the final model (Table 4). In Figure 3 I show the predicted probabilities of short-term childbearing intentions stemming from the full model run on childless respondents. Men were more likely than women to report negative or uncertain short-term childbearing intentions if they wanted to have a child but perceived that their partner did not (Figure 3).
Hypothesis 5: Gender equality effect. To test this hypothesis, I compared the fit of a model in which the effect of the perception of partner's agreement/disagreement was constrained to be the same for couples in which the man participates in child care tasks and for couples in which he does not with that of a model where this effect was allowed to be different in the two types of couples. The inclusion of such an interaction term significantly improved the fit of the model on non-use of contraception. From the sign of the coefficient, one can argue that when men were involved in childcare duties a positive disagreement tended to be shifted toward stopping contraception. The coefficient was 0.16 for couples in which the male partner did not participate in childcare and 1.64 (0.16+1.48) for couples in which the male partner participated in childcare (Table 4). In Figure 4 I show the predicted probabilities of contraceptive non-use from the models run on parents' sub-sample. The man's involvement in childcare tasks increased the chance to report non-use of contraception if disagreement was perceived (Figure 4).
Hypothesis 6: Bargaining power effect. To test this hypothesis, I compared the fit of a model in which the effects of the combined partners' desires was constrained to be the same for couples in which either the woman or the man was more educated with that of a model in which these effects were allowed to differ between the two types of couples. The empirical evidence was always in favour of the restricted models, meaning that the inclusion of the interaction term between the partner's perceived desires and a difference in partners' educational levels did not significantly improve the fit of the model. Results are not shown but available upon request.
6. Accuracy of the perception of the partner's short-term fertility desires
In the Austrian GGS, a subset of the interview questions were administered to the partners of respondents living in a union, with the goal of keeping the partners busy during the interviews with the respondents. Questions on fertility desires and intentions were included in the partner's questionnaire. Of the 3,892 partners who were asked to participate, 1,115 returned the questionnaire. The response rate was, at 29%, very low. Hence, the couple-level data could not be used for any robust statistical analysis.
Since there were 796 individuals of my selected analytic sample for whom a comparison between perceived and actual partner's short-term fertility desires could be performed, I looked at this small sub-sample to assess the accuracy level of the respondents' perceptions of their partner's desires. It is worth pointing out that the share of perceived disagreement did not significantly differ in the main sample and in the small sub-sample of couple data.
In Table 7 the distribution of the partners' combined short-term fertility desires is reported by using either the proxy or the partner's actual desires. The full distribution of proxy by actual report is not shown because of the small sample size. The share of disagreement differed depending on whether it was based on respondent's proxy report for partner's desires or on each partner's own report of desires. The differences were statistically significant here only for childless respondents. As expected, the respondent's perceptions of their partner's desire tended to overestimate the actual level of agreement and to underestimate the actual level of disagreement between the partners. Parents' proxy reports were more correct than childless respondents' ones, the consistency levels here were 90% and 84%, respectively (Table 7). In each of the sub-groups considered, the highest level of error was observed in the category 'respondent no, partner yes'. This was the most difficult case to be ascertained by the respondent in his/her proxy report. The other category of disagreement 'respondent yes, partner no' registered a higher proportion of consistent reports, even higher than that of the agreement category 'respondent yes, partner yes'.
I ran on the sub-sample of respondents with couple data the same models used for the analysis of respondents who had just a proxy report of their partners' desires (Tables 8 and 9). I included in these couple models the variables associated with partners' participation in the survey in order to control to some extent for partners' response bias. This is because in the sub-sample of respondents with an interviewed partner, couples who were married, couples with children and couples in which the male partner was the main respondent tended to be over-represented.
I tested all the research hypotheses by using the same procedure adopted in the main analysis. Some of the results obtained in the earlier analysis were confirmed. In the models for intentions run on childless or parent respondents, the effects of signed differences were different from those of absolute disagreement and the disagreement produced an outcome which was lying midway between that of an agreement on yes and that of an agreement on no (Table 8). Moreover, men reported a childbearing intention less frequently than women if they wanted a child but the partner did not want one (beta coefficient for men=1.9 (3.67-1.77) (Table 8). Disagreeing couples, whether childless or not, reported non-use of contraception almost as often as those who agreed on not having a child (Table 9). This result held, independent of whether the disagreement was signed or absolute, and did not depend on whether the woman or the man was higher in their short-term fertility desires.
Some other findings obtained in the main analysis were not confirmed. The effect of the partner's disagreement on the respondent's intentions and contraceptive behaviour did not depend on the gender equality within the couple. Moreover, the parity-specific effect of disagreement was not consistent with the one observed in the main sample: disagreement had the strongest inhibiting effects at parity zero and two or above. This result, however, has to be interpreted with great caution, given that the sub-sample of respondents with an interviewed partner was strongly selected by parity.
7. Summary and concluding remarks
In this paper, I examined how couple disagreement about short-term childbearing desires affects two aspects of reproductive decision-making: the formation of short-term fertility intentions, and their implementation, as measured by the prediction of non-use of contraception. I disaggregated differences between the respondent's own and the partner's perceived desires into negative and positive components depending on whether the respondent's desires were weaker or stronger than the partner's perceived ones, and examined those components separately.
Results showed that the perception of the partner's childbearing desires is an important element in the formation and implementation of one's own fertility intention, as theoretically discussed by Miller and Pasta (1996), Miller et al. (2004), and recently emphasised by Ajzen (2010). The perception of a lack of agreement with the partner about wanting a(nother) child now discourages the intention to have a child in the short-term and inhibits the non-use of contraceptives. There are, however, important differences by type of disagreement (positive versus negative), gender, parity and gender equality within the couple.
Absolute and signed disagreements produced different effects on the respondent's own short-term childbearing intention. Positive disagreement was closer to agreement on yes and negative disagreement was closer to agreement on no while absolute disagreement had an effect lying between the two indicated signed differences. This result may reflect the circumstance that individuals are not willing to be discouraged by their partner if they want to have a child. It is also consistent with the acceleration effect found by Miller and Pasta (1996) in their longitudinal research according to which the member of the couple who intended to have a child sooner had more influence on the timing of subsequent proceptive behaviour. However, it may also signal the predominance of the respondent's desires over his/her partner's desires and it is consistent with the bias in the perception of partners' desires which is usually driven by the respondents' own desires (Thomson 1997; Miller et al. 2004).
The double-veto model worked only for contraceptive behaviour but not for respondents' own short-term childbearing intentions. Respondents who perceived a difference between their own and their partner's desires reported an intention outcome in between that of respondents who perceived an agreement on having a child and that of respondents who perceived an agreement on not having a child. This evidence suggests that the partner context is not always adequately incorporated in the respondent's report of his/her own short-term fertility intentions.
The disagreement tended to be shifted toward a pregnancy intention/pregnancy-seeking behaviour at parity zero and toward avoiding pregnancy and maintaining contraceptive use at higher parities. This signals that below replacement family size ideals are the norm in Austria, as evidenced in previous studies (Goldstein et al. 2003).
Moreover, at parity zero, women were less responsive than men to their partner's desires if they wanted to have a child now but perceived that the partner did not. This finding supports the women's greater influence on childbearing decision-making outlined in my fourth research hypothesis. It is also in line with a recent qualitative study according to which Austrian men think it is fair to leave the final say on having a child to their female partner (Rille-Pfeiffer 2009). A woman's prevalence in reproductive decision-making was also found in other countries, such as Italy, where childbearing lies mainly in woman's domain (Testa et al. 2011).
In the sub-sample of parents, the man's active involvement in childcare duties shifted disagreement more in favour of childbearing rather than contraception (see interaction effects between man's participation in childcare tasks and disagreement, in the Model for the prediction of non-use of contraceptives, sub-sample of parents, Table 4). The result provides evidence for my fifth research hypothesis. It is also consistent with the positive association between an equal gender division of childcare duties within the couple and the second-birth intentions observed in a previous Austrian empirical study (Buber-Ennser 2003).
The data reveal that the respondent's perception of agreement with the partner about wanting a(nother) child now predicts more strongly the non-use of contraception than the respondent's own fertility intentions. This result indicates that fertility desires may bypass fertility intentions and act directly on behaviour (Miller 2011). It is worth pointing out that fertility desires have a closer time referent than fertility intentions in the Austrian GGS. This discrepancy is in many respects a limitation, but the use of the word 'now' is important in identifying conflicts between the partners, which may vary over time, given that desires are frequently reassessed over the individual's life course (Morgan 1985). Moreover, it reduces the area of potential agreement because the partners may disagree about whether they want to have the next birth at the current time (now), or whether they want to postpone it to a later point in time.
The remarkably high proportion of fertility desires converted into fertility intentions in the case of a perceived agreement between the partners implies that fertility desires could be the most suitable context for investigating the couple's interaction. Indeed, intentions might conceal the conflict between the partners because they reflect the resolution of it, as has been suggested by previous research (Ajzen 1991; Miller 1994; Barret and Wellings 2002). This would explain why in some studies looking at partners' differences in fertility intentions, a high degree of homogeneity in partners' responses has been found (Berrington 2004).
One shortcoming of this study concerns the use of the respondent's report of their partner's fertility desires. Ideally, research into couples' fertility decision-making processes should be based on surveys in which the two partners are interviewed independently (Thomson 1997; Thomson and Hoem 1998). The ideal situation would be to have a double source of information on the partner's desires-i.e., as perceived by the respondent and as reported by the partner-as this may ease the interpretation of conflicting desires (Becker 1996). My analyses of the restricted couple sample suggest that the bias introduced by using partners' perceived fertility desires as proxies for their actual fertility desires was not particularly high. This means that real disagreement, not just perceived disagreement, is the source of continuing to use contraception among conflicting couples. This assessment, however, could only be based on a valid sub-sample affected by a very low response rate and high selectivity.
Because of the dyadic nature of reproduction, the couple is the most suitable context for investigating fertility decision-making. In this study I attempted to take a step forward in the analysis of couples' childbearing decisions. My findings reveal that respondents who perceive that their partner disagrees with them about wanting another child now do not always refrain from expressing a short-term child intention but are not likely to stop using contraception. This sub-sample of people probably will not actively try to conceive until an agreement has been reached within the couple. Whether they will eventually have a baby in the short-term future and whether this will happen more often to couples in which only the woman wanted to have a child while the man did not want one might be clarified in further investigations using a longitudinal approach, once the data from the second wave of the Austrian GGS become available.
I am grateful to Caroline Berghammer and Isabella Buber for helping me become acquainted with the Austrian GGS data, and to Dimiter Philipov and Leonardo Grilli for their beneficial comments on a first draft of the paper. I am indebted to Elizabeth Thomson and two anonymous reviewers who helped me improve an earlier version of the paper.
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes 50(2): 179-211. doi:10.1016/0749-5978(91)90020-T.
Ajzen, I. (2010). Fertility intentions and the theory of planned behaviour. Paper presented at the International Conference on Reproductive Decision-making in a Macro-Micro Perspective, Vienna, December 2-3, 2010.
Barret, G. and Wellings, K. (2002). What is a 'planned' pregnancy? Empirical data from a British study. Social Science & Medicine 55(4): 545-557. doi:10.1016/ S0277-9536(01)00187-3.
Beach, L.R., Hope, B.D., and Campbell, F.L. (1982). The expectation-threshold model of reproductive decision making. Population and Environment 5(2): 95-108. doi:10.1007/BF01367489.
Becker, S. (1996). Couples and reproductive health: A review of couple studies. Studies in Family Planning 27(6): 291-306. doi:10.2307/2138025.
Beckman, L.J. (1983). Communication, power, and the influence of social network in couple decisions on fertility. In: Bulatao, R.A. and Lee, R.D. (eds.). Determinants of fertility in developing countries, Vol. 2. New York: Academic Press: 415-443.
Beckman, L.J. (1984). Husbands' and wives' relative influence on fertility decisions and outcomes. Population and Environment: Behavioral and Social Issues 7(3): 182-197. doi:10.1007/BF01255488.
Berrington, A. (2004). Perpetual postponers? Women's, men's and couples' fertility intentions and subsequent fertility behaviour. Applications and Policy Working Paper A04/09. Southampton: Southampton Statistical Sciences Research Institute.
Billari, F., Philipov, D., and Testa, M.R. (2009). Attitudes, norms and perceived behavioural control: Explaining fertility intentions in Bulgaria. European Journal of Population 25(4): 439-465. doi:10.1007/s10680-009-9187-9.
Brant, R. (1990). Assessing proportionality in the proportional odds model for ordinal logistic regression. Biometrics 46(4): 1171-1178. doi:10.2307/2532457.
Buber-Ennser, I. (2003). The influence of the distribution of household and childrearing tasks between men and women on childbearing intentions in Austria. Vienna Yearbook of Population Research 1: 165-180. doi:10.1553/populationyearbook 2003s165.
Bulatao, R.A. (1981). Values and disvalues of children in successive childbearing decisions. Demography 18(1): 1-25. doi:10.2307/2061046.
Coombs, L.C. and Chang, M.C. (1981). Do husbands and wives agree? Fertility attitudes and later behavior. Population and Environment 4(2): 109-127. doi:10.1007/BF01255801.
Corijn, M., Liefbroer, A.C., and De Jong Gierveld, J. (1996). It takes two to tango, doesn't it? The influence of couple characteristics on the timing of the birth of the first child. Journal of Marriage and the Family 58(1): 117-126. doi:10.2307/353381.
Davidson, A.R. and Beach, L.R. (1981). Error patterns in the prediction of fertility behaviour. Journal of Applied Social Psychology 11(6): 475-488. doi:10.1111/ j.1559-1816.1981.tb00837.x.
Fried, E.S., Hofferth, S.L., and Udry, J.R. (1980). Parity-specific and two-sex utility models of reproductive intentions. Demography 17(1): 1-11. doi:10.2307/ 2060959.
Fried, E.S. and Udry, J.R. (1979). Wives' and husbands' expected costs and benefits of childbearing predictors of pregnancy. Biodemography and Social Biology 26(4): 265-274. doi:10.1080/19485565.1979.9988385.
Goldstein, J., Lutz, W., and Testa, M.R. (2003). The emergence of sub-replacement family size ideals in Europe. Population Research and Policy Review 22(5-6): 479-496. doi:10.1023/B:POPU.0000020962.80895.4a.
Jansen, M. and Liefbroer, A.C. (2006). Couple's attitudes, childbirth, and the division of labor. Journal of Family Issues 27(11): 1487-1511. doi:10.1177/ 0192513X06291038.
Kamo, Y. (2000). "He said, she said": Assessing discrepancies in husbands' and wives' reports on the division of household labor. Social Science Research 29(4): 459-476. doi:10.1006/ssre.2000.0674.
Lundberg, S. and Pollak, R.A. (1996). Bargaining and distribution in marriage. Journal of Economic Perspectives 10(4): 139-158. doi:10.1257/jep.10.4.139.
McDonald, P. (2000). Gender equity in theories of fertility transition. Population and Development Review 26(3): 427-439. doi:10.1257/jep.10.4.139.
Miller, W.B. (1986). Proception: An important fertility behavior. Demography 23(4): 579-594. doi:10.2307/2061352.
Miller, W.B. (1994). Childbearing motivations, desires, and intentions: A theoretical framework. Genetic, Social, and General Psychology Monographs 120(2): 223-258.
Miller, W.B. (2011). Differences between fertility desires and intentions: Implications for theory, research, and policy. Vienna Yearbook of Population Research 9: 75-98.
Miller, W.B. and Pasta, D.J. (1994). The psychology of child timing: A measurement instrument and a model. Journal of Applied Social Psychology 24(3): 218-250. doi:10.1111/j.1559-1816.1994.tb00580.x.
Miller, W.B. and Pasta, D.J. (1996). Couple disagreement: Effects on the formation and implementation of fertility decisions. Personal Relationships 3(3): 307-336. doi:10.1111/j.1475-6811.1996.tb00119.x.
Miller, W.B. and Pasta, D.J. (2002). The motivational substrate of unintended and unwanted pregnancy. Journal of Applied Biobehavioral Research 7(1): 1-29. doi:10.1111/j.1751-9861.2002.tb00073.x.
Miller, W.B., Severy, L.J., and Pasta, D.J. (2004). A framework for modelling fertility motivation in couples. Population Studies 58(2): 193-205. doi:10.1080/ 0032472042000213712.
Morgan, S.P. (1981). Intention and uncertainty at later stages of childbearing: The United States 1965 and 1970. Demography 18(3): 267-285. doi:10.2307/ 2060997.
Morgan, S.P. (1985). Individual and couple intentions for more children: A research note. Demography 22(1): 125-132. doi:10.2307/2060991.
Namboodiri, N.K. (1972). Some observations on the economic framework for fertility analysis. Population Studies 26(2): 185-206. doi:10.2307/2173577.
Philipov, D. (2011). Theories on fertility intentions: A demographer's perspective. Vienna Yearbook of Population Research 9: 37-45.
Prskawetz, A., Sobotka, T., Buber-Ennser, I., Engelhardt, H., and Gisser, R. (2008). Austria: Persistent low fertility since the mid-1980s. Demographic Research 19(12): 293-360. doi:10.4054/DemRes.2008.19.12.
Rille-Pfeiffer, C. (2009). Kinder - jetzt, später oder nie? Generatives Verhalten und Kinderwunsch in Österreich, Schweden und Spanien. Schriftenreihe des Österreichischen Instituts für Familienforschung (ÖIF) 21. Vienna: Budrich UniPress.
Rindfuss, R.R., Morgan, S.P., and Swicegood, G. (1988). First births in America: Changes in the timing of parenthood. Berkeley: University of California Press.
Ryder, N.B. (1973). A critique of the National Fertility Study. Demography 10(4): 495-506. doi:10.2307/2060877.
Testa, M.R., Cavalli, L., and Rosina, A. (2011). Couples' childbearing behaviour in Italy: Which of the partners is leading it? Vienna Yearbook of Population Research 9: 157-178.
Thomson, E. (1997). Couple childbearing desires, intentions, and births. Demography 34(3): 343-354. doi:10.2307/3038288.
Thomson, E. and Hoem, J.M. (1998). Couple childbearing plans and births in Sweden. Demography 35(3): 315-322. doi:10.2307/3004039.
Thomson, E., McDonald, E., and Bumpass, L.L. (1990). Fertility desires and fertility: Hers, his, and theirs. Demography 27(4): 579-588. doi:10.2307/2061571.
Townes, B.D., Beach, L.R., Campbell, F.L., and Wood, R.L. (1980). Family building: A social psychological study of fertility decisions. Population and Environment 3(3-4): 210-220. doi:10.1007/BF01255339.
Voas, D. (2003). Conflicting preferences: A reason fertility tends to be too high or too low. Population and Development Review 29(4): 627-646. doi:10.1111/j.1728-4457.2003.00627.x.
Williams, R. and Thomson, E. (1985). Can spouses be trusted? A look at husband/wife proxy reports. Demography 22(1): 115-123. doi:10.2307/2060990.
Zabin, L.S., Huggins, G.R., Emerson, M.R., and Cullins, V.E. (2000). Partner effects on a woman's intention to conceive: 'Not with this partner'. Family Planning Perspectives 32(1): 39-45. doi:10.2307/2648147.
Maria Rita Testa1
1 Vienna Institute of Demography, Austrian Academy of Sciences. Wittgenstein Centre for Demography and Global Human Capital, Vienna, Austria. E-mail: firstname.lastname@example.org.…
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: Couple Disagreement about Short-Term Fertility Desires in Austria: Effects on Intentions and Contraceptive Behaviour. Contributors: Testa, Maria Rita - Author. Journal title: Demographic Research. Volume: 26. Publication date: January-June 2012. Page number: I+. © Max Planck Institut für Demografische Forschung Jan-Jun 2007. Provided by ProQuest LLC. All Rights Reserved.
This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.