• Skip to content
• Contact AIFS
•  
• Search box

Research report no.11 2004

'It's not for lack of wanting kids...'
A report on the Fertility Decision Making Project

by Ruth Weston, Lixia Qu, Robyn Parker and Michael Alexander

2. Method and sample

In this chapter the process by which the sampling and fieldwork were conducted is set out, and the rationale for defining the parameters of the population from which the Fertility Decision Making Project sample was drawn. The major steps involved in the process of designing the interview schedule and the recruitment and interviewing of the participants are described. Response rates are presented and the demographic characteristics of the sample are then compared with the population.

2.1 Development of the interview schedule

The interview schedule was developed over a period of several months, at first in consultation with officers from the Office for Women and subsequently in collaboration with both the Office for Women and the Wallis Consulting Group, the fieldwork company awarded (through a competitive tendering process) the contract to undertake the fieldwork for the Fertility Decision Making Project.

Research themes and questions relevant to the investigation of issues related to fertility decision making were generated by both the Office for Women and the Australian Institute of Family Studies. Institute researchers then collated a range of instruments designed to measure the concepts and issues of concern to each agency. Through an extensive period of consultation, the amount and nature of the information required by each agency in relation to particular domains of fertility behaviour were discussed, clarified and finalised.

Australian Institute of Family Studies researchers then worked with the fieldwork company to refine the content of the interview to allow maximum information to be gathered in the time available. The fieldwork company recommended the interview be structured to be no longer than 30 minutes. Within this constraint, extensive revision of the content of the interview was undertaken. The content areas covered by the interview were:

• Relationship and childbearing status and history;
• Other sociodemographic characteristics such as age, education, labour force participation, income, and religion;
• Family background, health and wellbeing, quality of relationships;
• Fertility aspirations and expectations of respondents and their partners;
• Employment conditions, work values, job-related perceptions and stress; and
• Level of importance attached to various issues that may influence fertility decision making behaviour.

2.2 Overview of the sampling procedure and fieldwork

Two samples were selected: a primary sample comprising those who were initially contacted for interview, and a secondary sample formed by recruiting the partners of those primary respondents in committed relationships who provided contact information.

2.2.1 Sampling

The primary sample was a national simple random sample of people between the ages of 20 and 39 years who lived in a private dwelling with a telephone. Thus the sample could include boarders and those in group households but not those residing in shared accommodation such as university Halls of Residence where there was no private phone access. Where several members of a household were eligible to participate in the survey, the sample member with the next birthday was chosen. This approach improves the chances of the sample resembling the target population in terms of gender and age (Salmon and Nichols, 1983).

Much of the fertility literature is biased to the extent that researchers have tended to restrict their attention to the views of women, as though they alone make the decisions (Greene & Biddlecom 2000). There is also a tendency to gather data from only one member of a couple, yet it appears that men's fertility preferences are also important in shaping couples' decisions about having children (Barber & Axinn 1998; Miller & Pasta 1995; Thomson 1997a, 1997b). Yet, as Greene and Biddlecom (2000) point out, the individual reproductive histories of the partners and the level of agreement between partners regarding fertility preferences and decisions can have important effects on reproductive behaviour (Greene & Biddlecom 2000; Stewart 2002; Thomson 1997a).

The Fertility Decision Making Project has addressed this bias by recruiting a primary sample of both men and women so that the similarities and differences in factors that influence men's and women's fertility preferences and expectations can be examined. An additional feature of this project is its capacity to explore some of the dynamics of fertility-related decisions in couples via the collection of data from the respondent regarding their partner's fertility histories and preferences. However, there are three issues to consider with respect to obtaining information about their partner's fertility preferences and intentions from the respondent:

• respondents may not be prepared to indicate their partner's views,
• the information provided by respondents about their partners may not be accurate, and
• evidence suggests that the views of each partner are independently important in shaping reproductive behaviour (Miller and Pasta, 1995).

For these reasons, primary respondents in committed relationships (married, de facto, or committed but not living together) were asked to provide contact details for their partners. These partners formed a secondary sample, which was subsequently mailed a self-completion questionnaire asking for information about their personal fertility aspirations and expectations, as well their general demographic characteristics. The data provided by this secondary sample enables the assessment of the extent to which couples share similar aspirations and expectations regarding having children, and the relative influence of each partner's aspirations in shaping their fertility expectations.

2.2.2 Target population

The population from which the sample was drawn was restricted to those aged between 20 and 39 years - the years in which women are most likely to have children. Current fertility rates (expressed as the number of babies per 1000 women) indicate that relatively few women under 20 and over 40 have children. In 2002, the fertility rate for women under 20 was 17.1 and for women aged 40-44 it was 9.7. By contrast, the fertility rates for women aged in their early and late twenties were 55.5 and 104.2 respectively, while those for women aged in their early and late thirties are 111.2 (30-34 year olds) and 52.2 (35-39 year olds). While there could be merit in exploring the fertility intentions of younger people, given the implications of their views for future fertility intentions and patterns, their intentions may be too fluid to yield reliable information about likely future fertility patterns.

Surveying people aged 20-39 also allows us to gain an understanding of ways in which fertility expectations differ across the primary childbearing years. This aspect of the study design was based on the proposition that fertility preferences and expectations can change, often in response to changes in life circumstances and the accumulation of life experiences (McDonald 2000c, 2000e). Many preferences and expectations are likely to be contingent rather than absolute, being re-framed in the light of job or career opportunities, partnership formation and quality, experiences with an earlier child, and so forth.

2.2.3 Fieldwork

Computer-Assisted Telephone Interviewing (CATI) was used to conduct interviews with respondents enlisted via random digit dialling (RDD). In random digit dialling, telephone numbers to be contacted are based on the first five digits of the series of telephone numbers known to be allocated to a given postcode in the various metropolitan and regional areas of the country. The last three digits are unknown and are generated randomly by the CATI system. Interviews lasted on average 34.3 minutes and required an average of 4.4 calls to the residence. Almost ten per cent of interviews were achieved only after ten calls to the dwelling or the selected resident.

Fieldwork began on 10 December 2003 and concluded on 22 March 2004. The final primary sample of 3,201 respondents comprised 1,250 men (39%) and 1,951 (61%) women. Male respondents were unexpectedly difficult to recruit, particularly in NSW/ACT and Victoria, even when the "last birthday" method of selection was abandoned. Up to 10 callbacks were made to secure an interview. Initial non-contacts were called back one hour later and if contact still not established a further callback was scheduled for the following day. If four of these attempts were unsuccessful, callbacks were re-scheduled to take place in a different time frame - for example, a weekend rather than an evening. In this way successive attempts to enlist a respondent or make contact with the selected household member were spread across daytime and evening periods, weekdays and weekends.

Piloting of the CATI interview on a sample of 70 primary respondents was conducted by the fieldwork company to identify any inadequacies of the questions, clarify interviewer protocols, and to check the routing of respondents into correct subgroups.

2.3 Response rates

2.3.1 Response rates - Primary sample

Table 2.1 provides a summary of outcomes of the random digit dialling process used to derive the sample. In total, calls were made to 140,547 separate phone numbers, with 12 per cent turning out to be business numbers. In addition, 37 per cent failed to result in a contact. Where a private dwelling was contacted, the nature of the study was described to the person answering the telephone, and this person was then asked whether there was anyone in the household aged 20 to 39 years. In 76 per cent of cases, the person to whom the fieldwork company spoke indicated that there was no one in the household aged 20 to 39 years (here called ineligible households). Some of these people may have falsely claimed that their household was ineligible as a polite way of refusing contact. It appears that around 60 per cent of private households in most capital cities and around 65 per cent in the rest of the States/Territories would have been "out of scope".⁶

Where it was known that an eligible person lived in the private dwelling, 45 per cent agreed to participate, 21 per cent refused, and the remaining 34 per cent were not available for interview during the survey period. Some of these people may, in fact, have not wished to participate, while others are likely to have been genuinely absent at the time of the survey. In total, 68 per cent of interviews were achieved where the eligible person was contacted. However, if it is assumed that 40 per cent of refusers at the household level contained an eligible person, then only 26 per cent of interviews were achieved of the total sample of assumed refusers, known refusers, eligible respondents who were unavailable, and survey participants.

In other words, the best scenario is a response rate of 68 per cent (based on the sample of refusers and participants). The second best scenario is a response rate of 45 per cent (based on the sample of known eligibles who refused or were not contacted, and those who participated). It is more likely, however, that the response rate based on eligible households was closer to the lower bound estimate of 26 per cent.

2.3.2 Response rates - secondary (partner) sample

At the end of each telephone interview, respondents in committed relationships (married, de facto, and exclusive but non-cohabiting; n = 2129) were asked whether they thought their spouse or partner would also be willing to participate in the study. Partner contact details were obtained from almost 50 per cent of these, and questionnaires subsequently mailed to the partners of 1,057 respondents. Of these, 64 were returned marked "return to sender". Completed questionnaires were received from 313 partners of primary respondents, a response rate of 30 per cent. The secondary sample of 313 partners comprised 173 men (55%) and 140 (45%) women.

2.4 Comparison of sample and population characteristics - Primary sample

In order to gain insight into the extent to which the sample represents the Australian population aged 20-39 years, several sociodemographic characteristics were compared with the ABS one per cent sample from the 2001 Census. These included age, social and registered marital status, the number of children, labour force participation, education and religion.

Age:

The proportions of Fertility Decision Making Project respondents in the age ranges used in this project compare favourably to those found in the general population. However the sample does comprise more women than men (61% vs 39%), which will potentially bias any analysis in which gender is not taken into account.

Social Marital Status:

In the Fertility Decision Making Project sample, the proportions of currently married and unmarried women are roughly similar to those in the general population. However, de facto relationships and currently unmarried men are slightly over-represented in the Fertility Decision Making Project sample, and married men slightly under-represented.

Registered Marital Status:

There are more never married and fewer married men in the Fertility Decision Making Project sample compared to the Australian population, however the proportions of the other categories for men and all categories for women are very similar.

Number of children ever born:

Overall, the sample of women represented the population well in terms of the number of children they already had. Compared to the general population, slightly more women in the Fertility Decision Making Project sample already had one child, while the proportions in the sample without children and with two, three, four or more children were similar to those in the population. The 2001 Census did not ask women how many children they had, thus the 1996 Census is used here. The Census only asked this question of women.

Labour force participation:

Compared to the general population there were slightly fewer employed men and women in the Fertility Decision Making Project sample, and more men and women who are not in the labour force at all. Proportions of unemployed men and women in the sample and in the general population were similar.

Family Type:

Compared to men in the general population, approximately half as many male Fertility Decision Making Project sample respondents were living with a partner and children. Women with partners and children were also underrepresented in the sample, but not to the same extent. Slightly more men in the Fertility Decision Making Project sample were living with a partner and no children, and there were fewer single fathers in this sample than in the general population.

Highest Level of Education Completed:

The Fertility Decision Making Project sample is, overall, more highly educated than the general population. A lower proportion of participants in the study had less than Year 12 schooling while there was an over-representation of respondents at most other levels of education, most notably graduates and postgraduates.

Religion:

Although the proportions of Catholics were similar both between the sample and the Australian population, and between men and women, respondents in the Fertility Decision Making Project sample was much more likely to report subscribing to a non-Christian religion or to having no religion at all. In the Australian population these proportions are affected by those who refused to answer or provided answers that could not be classified, making comparison difficult.

Region:

The distribution of female Fertility Decision Making Project respondents across geographic regions was very similar to that found in the general population. However, there tended to be a slightly smaller proportion of male respondents in Sydney and Melbourne, and a slightly larger proportion from Adelaide.

2.5 Description of secondary (partner) sample.

The characteristics of the sample of partners are largely similar to those of the respondent sample.

Partner's age:

Since the partner sample was not constrained by screening procedures, the distribution of partner ages extends beyond that of the primary sample. As women tend not to form relationships with younger men and men tend to form relationships with younger women, there are no men aged less than 20 years and very few women aged over 40 years in the partner sample. In Australia married men are, on average, approximately two years older than their partners (ABS 2003). In the Fertility Decision Making Project, female respondents were 2.5 years younger, and male respondents approximately 9 months older than their partners. The proportion of men and women in their early 20s in the partner sample is lower than that in both the general population and the Fertility Decision Making Project primary sample, probably because of the later age at which they tend to form committed relationships. The comparatively small number of women partners aged 35 to 39 years is likely to be strongly influenced by the absence of men aged in their forties in the primary sample.

Social marital status:

Completed questionnaires were received from a slightly higher percentage of the married male partners and a lower percentage of the men in de facto relationships than would be expected based on the proportions of each in the primary sample. Thus, analyses of couple data were representative of both married and de facto couples.

Registered marital status:

Due to the nature of the secondary sample, far fewer partners had never been married, and far more were currently married, than in the primary sample. Only a very few of the partners had been divorced.

Number of children:

As would be expected in a sample consisting of partnered respondents, more male partners had one, two, three, four or more children than had male primary respondents and less than one third were childless. Women partners with two or three children were slightly more common in the secondary than the primary sample, although fewer of them had four or more children.

Highest level of education completed:

Postgraduate qualifications were more common in the partner than the respondent sample, although there were fewer women with Bachelor degrees. When categories of tertiary qualifications were combined, however, the proportions of university-educated respondents and partners were very similar. There were more tradesmen and fewer men who had completed Year 12 in the partner sample.

Religion:

The overall proportions of non-religious and Christian men and women (whether Catholic, Anglican or unspecified) were similar for primary respondents and their partners, although Anglican men and women were over-represented in the partner sample. There were also fewer male partners subscribing to a non-Christian religion, and fewer women partners whose religion was an unspecified Christian faith compared to primary sample.

2.6 Summary

Apart from being comprised of more women (61%) than men (39%), the Fertility Decision Making Project sample was reasonably representative of the Australian population. The use of CATI technology to conduct the interviews may be part of the reason there are fewer employed respondents (Steel, Vella & Harrington 1996; Trewin & Lee 1988), but it is likely that other deviations are due to the restriction of the Fertility Decision Making Project sample to those aged between 20 and 39 years. It would be reasonable to expect there to be fewer married and more unmarried (including de facto) respondents in a sample aged under 40 years, and with more young Australians completing secondary and university education it should not be surprising that the Fertility Decision Making Project sample is more highly educated than the general population. The restricted age range may also explain why the proportion of respondents indicating "no religion" was much higher for the Fertility Decision Making Project sample than the general population.

Response rates for the primary sample, while difficult to determine exactly, probably ranged somewhere between 30 and 40 per cent, which is on a par with many other telephone surveys of this kind. The fieldwork company noted that potential male respondents were especially reluctant to discuss the subject matter, reflected in the lower percentage of interviews achieved for men, particularly in NSW/ACT (36.3%), Victoria (36.5%) and Western Australia (39.7%). A further contributing factor may have been the length of the interview. Upon contacting a potential respondent the interviewer explained that the interview would take approximately 30 minutes to complete (the average was 34.3 minutes), which may have been sufficient to deter respondents. With respect to the partner sample, the response rate of 30 per cent is within the range expected for mailout surveys (Neuman 1997).

6. These estimates were provided by various fieldwork companies, including the Wallis Consulting Group, which conducted the FDMP fieldwork. [back]

Research Report 11: Contents | Previous | Next

• Web feedback
• Privacy
• Disclaimer
• Accessibility
• Site map
• © Commonwealth of Australia 2010