2.2.1 Sampling method

The researchers used the snowballing technique to obtain the sample. They identified a set of initial contacts, and asked them whether they knew other people who met the criteria and might be willing to participate in the research. The researchers then followed up on these contacts.

The researchers chose this technique because it can be an effective and efficient way to generate a sample of participants from a range of backgrounds. They successfully recruited participants who are often hard to reach through other means, such as through advertising, organisational databases or cold calling.

To enable them to answer the research questions adequately, the researchers developed a sampling matrix which included targets for minimum numbers within particular sub-sample groups, for example, people with children with special needs. In the few cases where couples shared caring responsibilities fairly evenly, the team interviewed the person with the demographic characteristics they needed for the sample.

The researchers also sought to include participants in different situations: people who were in paid work, both full time and part time, or not in paid work at the time of the study; men who had parenting and/or other caring responsibilities; participants who were Maori and Pacific people; and people living in urban and rural areas.

2.3.1 Recruitment procedures

The researchers contacted potential participants by email or telephone to explain the research and determine their willingness to participate. Those who agreed to take part received a written information sheet and consent form (Appendix I)^[3]. The information sheet discussed the purpose of the research, the areas the questions would cover, use and confidentiality of the information collected, and gave the researchers' contact details. A researcher then confirmed an interview time and place with the participant. All of the people contacted agreed to participate.

2.3.2 The interviews

The interviews took place during February and March 2006. Most were conducted in the participant's home. The researcher began the interview by going through the project again and giving the participant a copy of the consent form. If the participant had no questions and was happy to continue, s/he signed the form.

Many interviews were tape recorded and transcribed with the participant's permission. In some cases it was not practical to tape the interview, for example, due to background noise. In all cases researchers took detailed notes. The interviews took between one and one and three quarter hours.

Two researchers were present at all but three interviews, where it was impractical for two people to attend. One researcher asked the questions while the other took notes, and on occasion prompted for further responses or asked additional questions. This practice allowed the interviewers to probe areas in more depth and minimised the risk that they missed important points. It also enabled the researchers to take thorough notes without compromising their rapport with participants. A further benefit was that the researchers could discuss each interview after it was completed, and make sense of the data they collected.

The interviewing team comprised two Pakeha women, a Pakeha man, and for one interview, a Samoan woman.

At the completion of the interview, researchers gave participants a gratuity of $50 cash in recognition of their time.

2.3.3 Interview guide

The researchers developed a semi-structured interview guide in conjunction with the project advisory group.^[4] The guide (Appendix I) was based on the research questions and covered the following areas:

• nature and extent of caring responsibilities
• paid work - current and historical
• decision making related to current situation - factors influencing decisions and decision making processes
• demographic information.

The interview guide ensured that researchers collected information pertinent to the research questions and asked key questions consistently. Using a guide, rather than a questionnaire, also allowed researchers to explore themes and issues that arose from a participant's circumstances and experiences, omit questions that were not appropriate, and modify the wording of questions to suit a participant's situation.

The researchers reviewed the interview guide after the first four interviews, and made some minor alterations including adding some new questions and prompts.

2.3.4 Data analysis

The team analysed the data in several stages, beginning by coding transcripts according to research questions, labelling each transcript with participants' personal characteristics, and entering the data onto an NVivo database.

Researchers then looked at each research question according to the 'paid work' category participants were in, and within that according to participants' personal characteristics. The researchers also paid attention to sub-groups of interest, such as men, and people with children with special needs.

For the third stage, the researchers reviewed the research questions for all the sub groups of interest, looking, for example, at ethnicity and type of caring responsibility. They examined data from all participants to capture the dynamic of people's patterns of participation in paid work.

Finally, the researchers amalgamated reports from these analyses and examined the commonalities and full range of exceptions. This allowed for consistent 'within case' and 'between case' analysis, and ensured validity and reliability in interpreting findings.

2.3.5 Limitations of the research

This study sought to answer a number of broad research questions. The report provides a picture of some of the factors that influence the decisions of people with caring responsibilities, but given the broad scope of the study, it was not possible to meet all of the objectives in great depth.

The snowballing technique used to recruit participants has a number of advantages and some limitations. It does not aim to provide a representative sample, and can be biased. There is also a risk that participants will be too similar to the people who selected them - that is, they will have the same or similar characteristics and beliefs as the researchers. This can mean that only a narrow range of perspectives and experiences is represented in the findings. The researchers sought to overcome this problem by drawing on their networks with people with a range of characteristics and beliefs, and by using a sampling matrix which identified particular groups they needed to recruit. Although they attempted to include a range of participants, it was not possible to include every variant on caring responsibilities or other circumstances. Some groups, such as single parents of children with special needs and ethnic groups other than European, Maori and Pacific people, could be included in further research.

With the snowball technique, participants' confidentiality is potentially more at risk. Some of the initial contact people will know others who took part in the study, and may be able to identify them through discussion of their circumstances in the report. The researchers attempted to mitigate this risk by reducing references to specific characteristics of participants.