The fact that stepfamilies play a considerable role in the Austrian society has also been demonstrated in the study `Kindsein in Österreich' by
[Wilk 1998]. Among a representative sample of 2,745 children aged 10 who were interviewed in 1993 on various topics related to their family, school, spare-time, etc., about 6 to 8 per cent reported that they lived in a stepfamily.
Note that we use the respondent's order of the union as the defining characteristic of stepfamilies as opposed to the common definition of stepfamilies that relies on the presence of pre-union children of either the respondent or the partner. Our definition of stepfamilies therefore excludes possible first unions of respondents where pre-union children are already present and, on the other hand, we include second unions where neither the respondent nor the partner has any pre-union children. The latter assumption is used so as to compare stepfamily fertility, i.e., if at least one partner has pre-union children, with fertility if no pre-union children are present. The selection criteria of considering second unions only will be discussed in section 2.
In the Austrian Family and Family Survey approximately 49% of all male and 51% of all female respondents indicated two children as the ideal norm for a family
The beginning of a union is defined as the point in time the couple starts living together and sharing a common household.
16.3 per cent of all respondents recorded a second or even higher-order union.
This number differs when we distinguish between male and female respondents. It is 76.0 per cent for male and 83.6 per cent for female respondents.
Pregnancies that did not lead to a live birth or an abortion were not considered in our analyses since the data available do not include such information.
In Austria a third of all children of parity one are conceived before marriage and born after marriage formation, the pregnancy being the incentive for the marriage. Although the first common child in a second union can be of parity two or higher and despite of the fact that the above percentage does not directly refer to second unions considered in our paper, one should keep in mind a couple's tendency to marry when the woman is pregnant.
This restriction allows sufficient time for childbearing in a second union for those desiring a child.
The exclusion of those who had a birth shortly before the union formation is similar as in
[Griffith et al. 1985]. Children born more than 11 month before the formation of the second union are considered as pre-union children.
Our data set includes five records where the first child of the second union was conceived within the second union but born after the end of the second union. In none of the five cases was a third union recorded to which the child might have been assigned if the third union had been formed soon after the end of the second union.
We thank Elizabeth Thomson for the suggestion that we should conduct methodological studies to verify whether men and women are equally good reporters of their own and their partner's children. In U.S. data Elisabeth Thomson found that men's reports of union and birth dates have more inconsistencies than do corresponding reports of women. Among those 38 respondents who underreported stepchildren 17 were women, and almost the same number (21 persons) were men, which corresponds to 3.4 per cent among female and 10.6 per cent among male respondents. We thereby confirm Thomson's findings for our Austrian data as well.
We are grateful to Hans-Peter Kohler for suggesting this alternative viewpoint to us.
Only in the case that the couple was not married at union formation the respondent was asked whether the partner was single, married, widowed, divorced or separated at the time of union formation. If the couple was married when they moved into a joint household, this question was not posed and we have incomplete information on the partner's previous union. This is why we had to run gender-specific regressions.
Our sample includes only one occurrence where the man brought two children into the common household. The corresponding coefficient is significant and very small (0.11). To see whether our model is robust, we excluded this one case. The coefficient for "one child of the man living in the household" then changed from 3.18 to 3.13, which is still significant. In a second step we formed two categories for the man's pre-union children living in the household at the time of the formation of the second union, namely "none" and "one or more". In this case the covariate "man's pre-union children living in the household" is no longer significant, indicating that the opposite and significant effects for "one" (3.18) and "two or more" (0.11) "neutralise" to 1.27, which is no longer significant. We therefore retain three levels of the covariate "man's pre-union children living in the household".
Note that we have combined various levels of the three-way interactions, in particular we have reduced the duration variable to at most 4 levels.
This argument is supported by the following observation: the percentage of childless female partners in second unions of male respondents is nearly twice as high (81.41 per cent) that of childless female respondents (46.17 per cent). If we assume that women with pre-union children have also had previous unions to a higher extent, our argument follows.
While about half of all respondents reported that they have formerly been married, widowed respondents constitute a very distinct group: four per cent of female and two per cent of male respondents were widowed.
The interaction between woman's age and man's age is not significant, nor is a three-way interaction between woman's, man's age, and duration.