Participants were contacted by phone using random-digit dialling in order to constitute a national representative sample of adults of reproductive age living in France in 2010. To take part in the Fecond survey (for detailed information on the survey, see Legleye et al., 2013), participants had to be aged between 15 and 49 years old. Participants were randomly divided into two ballots (J1 and J2), each one having specific questions to limit the length of the interview to 40 min. Half the participants (n = 4,261) answered to the J1 ballot that included questions on the parenthood norm. Five percent of these participants (n = 194 out of 4,261) were excluded because they failed to respond to at least one of the questions that were used in the present study. The results are based on a sample of 4,067 participants.
Table 1 presents descriptive statistics for the sample. First, we selected three variables that reflected how close the participant was to parenthood: age at interview, being in a relationship, and parental status. Age at interview is considered an indicator of proximity to parenthood because in France the majority of people have children between the ages of 25 and 34 years (Breton et al., 2017). In our study population, nearly half (47%) of the participants were aged 35 years and older, 26% were aged 25–34 years and 27% were younger than 25 years. The majority stated that they had a partner (68%). Nearly half of the participants (49%) had no children at the time of the survey. Second, we used two variables to measure the social class of the participant: education and occupation. In our study population, 20% had a bachelor’s degree, 50% had a lower level of education and 30% a higher degree. Lastly, we took into account the importance of religion in the participant’s life (whatever the religion) as this may reflect some of their life values (Koropeckyj-Cox & Pendell, 2007a; Merz & Liefbroer, 2012; Rijken & Merz, 2014; van de Kaa, 1987). For half of the participants (49%), religion was not important in their life or they had no religion at all, whereas 6% of participants declared that religion was very important in their life.
In the Fecond survey, participants were interviewed about their sexual and reproductive life (Bajos et al., 2014; Moreau & Bohet, 2016; Moreau et al., 2014). The interview included 12 sections in the following order: sociodemographic profile, socialization, fertility intentions, reproductive biography, infertility care, contraceptive biography, focus on recent contraceptive use, first sexual intercourse, sexuality, norms, health, and health care.
Participants indicated their gender in binary terms with the following question “are you … a male? A female?” Participants who defined themselves as female were classified as “female participants” and those who defined themselves as male were classified as “male participants.”
The parenthood norm was investigated using a gendered question in order to distinguish the parenthood norm for women and men. The parenthood norm for women was measured using the following question: “In your opinion, can a woman have a fulfilled life without having children?” The parenthood norm for men was measured with the same question adapted for men: “In your opinion, can a man have a fulfilled life without having children?” [translation of the French version: à votre avis, une femme/un homme peut-elle/peut-il réussir sa vie sans avoir d’enfant ?]. The concept of a “fulfilled life” was not defined so that participants would be free to define it according to their own life values and purpose. Similarly, the concept of “having children” was not defined so that participants were free to define it in as a biological notion only (to conceive a child) or also as a sociological notion (to parent a child, including stepchildren, adopted children, etc.).
Participants selected one of the following response options: “Yes, easily,” “With difficulty,” or “No” [translation of the French version: Oui facilement, Difficilement, Non]. Participants were classified as endorsing the parenthood norm when they answered that a woman/man cannot, or can only with difficulty, have a fulfilled life without children. Participants were classified as not endorsing the parenthood norm when they answered that a woman/man can easily have a fulfilled life without having children.
Consistent with recommendations for investigating double standards (Crawford & Popp, 2003), the Fecond Survey used a within-subject design: each participant was interviewed on the parenthood norm for women and men. Survey guidelines recommend handling question-order effects by randomly splitting the participants into two groups (Moore, 2002; Perreault, 1975): one group answers question A and then question B (first ballot) and the second group answers question B and then question A (second ballot). This is called a split-ballot design. This methodology was applied in the Fecond survey: half of our study population were first asked about the parenthood norm for women and then immediately asked about the norm for men (first ballot), whereas the other half were first asked about the parenthood norm for men and then immediately asked about the norm for women (second ballot). Participants are responding within a personal background context for the first question (i.e. based on personal history, values and environment) whereas they are responding within a social relational context for the second question (i.e. in comparison to the first question).
All analyses were performed using Stata 13.0 software (StataCorp). Proportions were compared using chi-square tests. We expected that double standards would be activated only in the social relational context through the question-order effect. A first preliminary analysis was conducted to determine whether there was a question-order effect (Hypothesis 1), i.e. if responses to the parenthood norm for women and for men among female and male respondents varied between the two response contexts, the personal background context and the social relational context. Then, we examined whether gender double standards emerged only in the social relational context and not in the personal background context among female and male respondents (Hypothesis 2). First, in the social relational context (i.e., responses to the second question), we examined whether gender double standards emerged in line with our hypothesis of a stronger endorsement of the parenthood norm for women than for men among female and male participants. Next, we examined the responses from the personal background context (i.e., responses to the first question) to determine if a different pattern emerged and if so, what was the logic of this pattern.
Finally, stratified multivariate analyses were conducted to determine whether the univariate analysis was affected by confounding variables. The dichotomous dependent variable was endorsement of the parenthood norm. Based on the literature (Merz & Liefbroer, 2012), we included six potential confounding variables that might affect endorsement of the parenthood norm: family profile (age, having a partner, having children), socioeconomic status (education, occupation) and life values (importance of religion for the participant). Results of the multivariate models give the odds ratio (OR), a measure of the association between a factor and an outcome (i.e., the parenthood norm). If the 95% confidence interval for the odds ratio does not include 1, then the association is statistically significant.