Study: changing gender roles and attitudes to family formation
The qualitative study
An in-depth qualitative study was carried out in late 2008-early 2009 as a precursor to the quantitative study. The purpose of the qualitative study was to obtain in-depth qualitative information and insights into current attitudes to gender roles and gender relations and attitudes to family formation and childbearing. The results of the qualitative study played an essential role in identifying key issues for inclusion in the questionnaire to be administered in the quantitative study (Fine-Davis, 2011, 2016).
The sample used in the qualitative study consisted of 48 adults selected to mirror the types of respondents which were to be included in the main quantitative phase. The sample was stratified by gender, age (20-34 / 35-55 years), family status (single / cohabiting / married and living with spouse), socio-economic status and geographic location (Dublin / other towns and cities / rural areas).
Most of the questions were open-ended and designed specifically for this study. Interviews were carried out on a one-to-one basis in a variety of locations. They were conducted primarily in the respondent’s home, the interviewer’s home or in Trinity College Dublin. All interviews were tape recorded. Interviews lasted from 39 minutes to one hour and 58 minutes, with the average interview length being one hour. The full results of the qualitative study are contained in Fine-Davis (2009).
Pilot study
Following the development of an extensive questionnaire, a pilot study was carried out in the summer of 2010 on a stratified sample of 150 male and female respondents in the childbearing age group and in various marital statuses throughout the country. The results of the pilot were used in developing the questionnaire to be used in the main study. Factor analyses were carried out to identify underlying dimensions of attitudinal items and to provide an empirical basis for retaining the best items for inclusion in the main questionnaire.
Main study
Research design and sample
The population under investigation was adults of child-bearing age (20-49 years) in the Republic of Ireland. A stratified sampling design was employed. This was based on gender, family status (single, cohabiting, married), age (20-34 and 35-49 years), having one or more children or not, and rural vs. urban location. “Single” was defined as not living with a partner; it did not refer to marital status per se. “Married” was defined as married and living with one’s spouse. Presence of a child was defined as the respondent having given birth, fathered or adopted one or more children; it did not require that the child be resident with the respondent. This design was employed so that the sample would include people in all possible combinations of these characteristics, including married with children, married without children, single with and without children, etc., as this would allow for the use of analysis of variance - one of the primary multivariate statistical techniques used in the study. A major goal in creating this design was to allow for comparisons between single, cohabiting and married people, while controlling for the effects of gender, age and other key demographic variables.
A multi-staged quota controlled sampling design with randomly selected starting addresses from the Geo Directory was used. This was based on a total of 200 randomly selected sampling points throughout the country with six interviews conducted per point. A “random walk” from the random starting point was adopted within each sampling point. A total of 1,254 interviews were conducted during the main fieldwork phase, with the additional 150 pilot survey questionnaires boosting the final sample to 1,404. Because of the comparability of the questionnaires and the quality of responses elicited at the pilot stage, the pilot data were thus incorporated into the final dataset prior to analysis.
In order to obtain the final sample of 1,404 a total of 10,596 contacts were made. Of these, 38% were ineligible, 32% were not at home and 1% was derelict properties. Twenty-eight per cent of the total contacts were eligible, i.e., they fulfilled the characteristics of the quota sample design. Of these eligible respondents, there was a 51% response rate in urban areas and a 42% rate in rural areas, for an overall response rate of 47%.
Data collection procedures
All interviewing was conducted on a face-to-face basis in respondents’ homes by experienced and pre-briefed interviewers from Behaviour and Attitudes. Computer Aided Personal Interviewing (CAPI) was utilised. Some of the sections of the questionnaire were self-completed by respondents on the laptops - Computer Aided Self-Completion Interviewing (CASI). This protected the privacy of the respondent in providing answers to more sensitive questions and, accordingly, enhanced data quality. Fieldwork for the main survey took place from mid-August to early November 2010. The average interview time was 45 minutes.
Instrument of main study
The questionnaire used in the main study included, inter alia, the following sections: (1) demographics, (2) attitudes to relationships, (3) relationships and relationship history, (4) attitudes to having children, (5) attitudes to gender roles and (6) attitudes to work-life balance and related social policies.
Most of the items were developed on the basis of the qualitative study, while others were replications of items used in previous research by the author and others. Some of the items concerning gender-role attitudes were replicated from Fine-Davis (1983a, 1988a), while some items concerning work-life balance were replicated from Fine-Davis et al. (2004) and Fine-Davis, McCarthy, Edge, et al. (2005). Selected items measuring attitudes to childcare and other family policies were replicated from Fine-Davis (1983b, 1983c); Fine-Davis et al. (2004); and Fine-Davis, McCarthy, Edge, et al. (2005). Likert items were used extensively. These were presented on seven-point scales ranging from “strongly disagree” to “strongly agree.” Within each set, items were presented in a randomised order for each respondent so that there would be no ordering effects.
Weighting of the sample and comparison with population estimates
In line with best practice in all sample surveys, the completed sample was statistically adjusted or re-weighted to ensure that it reflected the socio-demographic structure of the relevant sub-group of the population under investigation. Fortyeight individual weighting variables were set which reflected the population breakdown of 20- to 49-year-olds in terms of gender, age, family status, presence of children and area. Additional educational and SES (occupational status) weights were applied. All population figures were derived from the 2006 Census of Population.
The completed sample, notwithstanding the quota controls, was slightly under-represented in terms of family composition by single and married respondents and over-represented in terms of cohabitees. This is due to the fact that certain groups, including cohabitees, were oversampled in order to have sufficient respondents in each cell of the design to make it possible to carry out analyses of variance. Other groups in the completed sample are largely in line with the population figures. Comparison of the re-weighted sample and the Census figures shows that the structure of the re-weighted sample is very close to the population figures. Thus, we may be confident that the re-weighted data is representative of the population in the childbearing age group.
In the final weighted sample there are 50% males and 50% females. There are 54% in the age group 20-34 and 46% in the age group 35-49. The final weighted sample consists of 45% single people, 44% married people and 11% cohabiting people.
The proportion of rural and urban respondents is 40% urban and 60% rural in the final re-weighted sample. Again, this was adjusted somewhat by the weighting to conform to the proportions of these groups in the population. The final re-weighted sample includes 46% with children and 54% people without children, reflecting a minor readjustment from the original sample.
