Key issues and challenges of FGDs

It is the ‘group effect’ that really sets FGDs apart from other qualitative and quantitative social research methods. Morgan (Reference Morgan1996) compared the focus-group method with other social research data-collection methods, such as surveys and in-depth interviews, and considered their strengths and weaknesses. Focus groups are very useful in exploring personal and attitudinal subjects, as the group setting often releases inhibitions in talking about ‘taboo’ or ‘sensitive’ topics. Illiterate persons and those with minimal years of schooling can participate in focus groups as they are comfortable in sharing their perceptions and experiences with their peers. In Asian communities, where through either poverty or lack of school provision, many in the current cohort of elders have been deprived of educational opportunities, the focus-group method is a more appropriate approach to gathering information than self-completion questionnaires. FGDs also provide insights into the sources of complex behaviours and motivations (Morgan and Krueger Reference Morgan, Krueger and Morgan1993). The synergy that emerges through group interaction needs to be carefully managed by the moderator, who has to be trained in group dynamics (for more discussion seeMoen et al. Reference Moen, Antonov, Illson and Ring2010). This is probably the most important single influence on the quality of the data collected through focus groups. The second key issue is the recruitment and screening of group participants. Usually purposive sampling is applied, but some level of stratification may also be involved, especially if a particular ethnic culture or gender or economic class is to be studied. Thirdly, the ethics of conducting group discussions must be assiduously considered and respected. The participants should be informed of the risks (if any) and of the topics (particularly if personal data are to be gathered), and the confidentiality of the information and opinions that they share should be assured. In relation to the latter, all the participants must be informed about the confidentiality rules and agree to abide by them. Fourthly, methodological rigour in collecting and analysing the data should not be compromised in any way, as the analysis of complex textual or discourse data requires dedication and patience, as applies to most qualitative research. Indeed, many of the issues relating to other qualitative methodologies apply to FGDs, such as the investigator needing to be aware of their subjectivity, which among other things leads to personal interpretations and blind-spots in analysing one's own culture. Other shared methodological features include the likelihood of sample saturation, and the need for flexibility in implementing the focus-group questionnaire or topic guide (seeBerg Reference Berg1998).

Goss (Reference Goss1996) classified focus groups as ‘alternative research’ due to the reflexive and liminal experiences that participants experienced in his projects. He organised focus groups with trans-migrants (people from another country on their way to a third) in Indonesia to collect in-depth data on their socio-economic adjustment strategies. He believed that his participants found the experience to be enriching and empowering. The process that took place during the FGDs was ‘dialogic’ leading to a ‘polyvocal production’, i.e. ‘a multiplicity of voices speaking from a variety of subject positions’ (Goss Reference Goss1996: 118). It is this synergy of dialogues that sets focus groups apart as a research methodology. Focus groups have established an important role in social research during the last decade. The rise of this method is partly due to the expansion of international and cross-cultural research. Specialised guides or ‘tool kits’ have been produced and are helpful for researchers who wish to apply the method (Morgan, Krueger and King Reference Morgan, Krueger and King1998).

Focus groups and gerontology research

This section draws on some of the lessons learned from the author's empirical research with older people in Singapore over the last 20 years. Having been trained in the social work discipline, the author is very familiar with group dynamics and group processes, hence running focus groups as a social research tool was in certain respects a variant of familiar situations. After the first initiation into the FGD methodology in the 1990s as a senior researcher in an international project spearheaded by Albert Hermalin of the University of Michigan, the author was convinced of the effectiveness of FGDs in researching older populations. The project, titled Comparative Study of the Elderly in Four Asian Countries, studied older people in Thailand, the Philippines, Taiwan and Singapore. The huge dataset that was generated by the focus groups was analysed using Ethnograph (version 4) and a number of publications resulted (Knodel Reference Knodel1995; Mehta Reference Mehta1997, Reference Mehta1999; Mehta, Lee and Osman Reference Mehta, Lee and Osman1995). Knodel (Reference Knodel1995) summarises the research team's experience of using focus groups.

One of the key issues in the above-mentioned cross-cultural (and cross-national) research was to ensure the cultural equivalence and language equivalence of the semi-structured questionnaire that was used as a standardised instrument across the four countries. To this end, each country had a local senior researcher who had to ensure that the translation of the questionnaire passed rigorous ‘checks’ such as back translation and inter-rater validity. Pilot focus groups were invaluable for indicating whether the questions were correctly understood and relevant to the local cultural context. A point worth noting is that in many Asian communities, the older members have difficulty in answering general questions such as, ‘What are the perceptions of people of your age in your community about re-marriage?’ Such questions might be thought relevant and simple to understand, but many Asian elders were reluctant to speak on behalf of other fellow seniors. An effective way to deal with this issue is to ask, ‘What are the views of people you know in your age group about re-marriage?’

My involvement with other studies involving focus groups with seniors and/or their family members has included: (a) a study of older people's service needs and views about policies in Singapore, which involved six focus groups and 42 participants; (b) a study of the concerns about and preparation for retirement among people aged 50–59 years, involving 12 focus groups and 76 participants; (c) a study of the views of older Singaporeans (aged 60 or more years) on primary health care for seniors, involving six focus groups and 34 participants; and (d) a study of family care-givers' experiences and stresses in caring for an older parent or parent-in-law, with eight focus groups and 49 participants.

In all four studies, the author conducted some if not all of the focus groups, and was directly involved in the coding and drawing of the overview grid that helps to compare across the groups. While Ethnograph version 4 was used for the first two projects, NVivo version 7 was used for the last research. NVivo is a very creative tool that aids deeper analysis; the mapping tool allows the researcher to create visual diagrams to illustrate the inter-connectedness of the various variables. As a focus-group transcript can be more than 30 pages, computer software helps in the coding, analysis and retrieval processes. In the coding process, when suitable quotes are identified, their retrieval for academic writing purposes is made easy and convenient. From the author's experience, while the use of computer software can be frustrating, at the same time it is invaluable, especially when it comes to writing reports and publications.

Some recurring issues in conducting focus groups with older people are: participants (or moderators) not turning up as a result of unforeseen circumstances or lack of motivation; difficulties with the participants' access to the venue, such as the lack of elevators and ramps for wheelchair-bound participants; the need for washrooms to be within walking distance; setting the times, which should bear in mind the participants' routines of meals and naps; and being aware of a gender effect, that women are uncomfortable discussing certain issues in the presence of men. When this issue surfaced in one project, an additional small group exclusively of women was conducted to clarify and elicit the female seniors' views. In addition, researchers who intend to conduct focus groups with older adults should be prepared for the possibility of cognitive impairment and dementia, particularly among those aged more than 80 years. Therefore careful screening of the potential participants is crucial. If such a possibility is overlooked, not only will the group dynamics be affected, but the participant with mental health problems can be stigmatised. As older people have long biographies and many experiences that they like to share, the moderator should be prepared for ‘detours’ in the discussion. The moderator needs to be skilled in focusing the discussion, yet have the flexibility to allow insights to be explored that may open up new lines for understanding of issues and problems.

An incident occurred during the first listed research project that may have useful lessons for the reader. We scheduled the FGD in a university seminar room. It lasted from 7 pm to 9.30 pm. After a very animated discussion, we escorted the eight group members to the lobby and on the way had to walk over uneven corridor floors that were being repaired. The moderator assiduously warned the participants to take care, but in vain. One participant tripped on a loose floor tile and almost fell – it was fortunate that the moderator was holding her hand and managed to prevent a fall. This incident is permanently etched in the author's memory as a warning that when conducting research with older people, you must obtain an indemnity certificate, just as an Informed Consent form is essential. Also, you need to be prepared for emergencies and to act in the best interest of the respondent. Contact numbers of family/next of kin should be easily available so that they can be notified in an emergency.

Researchers may be asked to make alternative arrangements for those who are cared for by FGD participants. In one research project, for example, we asked a university student to ‘entertain’ a few grandchildren of the respondents. This facilitated the participation in the FGD of the older people (who were grandparents providing care to grandchildren). In another research project, care-givers of older people with dementia asked whether the day-care staff could help supervise their care recipients while the FGD session was in progress. As the venue of the FGD was a day-care facility, this was easily arranged.

In general, for the data and analysis to be rigorous in qualitative research, much time and patience is required. Ethnographers spend months living within a community collecting data on a daily basis and these have to be verified to ensure their validity and reliability. Those who adopt the case study method, in-depth interviews, or other data-collection modalities such as narrative discourse, when they come to coding and interpreting the data usually find themselves scouring through numerous pages of transcripts. The process has become less tedious with computer software, but the required rigour and methodical procedures are still time-consuming and laborious. In the research projects mentioned earlier, research assistants trained in the application and software were hired to expedite the analysis. The author, after training the research assistant for the particular project, closely examined the coding categories, and the links between the categories. To compare across groups conceptually, manual overview grids were found useful, as the software may not pick up implicit themes. To clarify, an overview grid captures the themes that arise in each group in a table format, thus enabling the researcher visually to analyse the contents of all focus groups at a glance.

Inductive as well as deductive inferences help build up the within-category integration, the across-category integration and finally the hypothesis refinement (if there are hypotheses). In the second listed research project, it was concluded from the FGD data that it is not only employees that experience retirement, so do housewives. During the FGD with housewives and pre-retirees, the housewives made a strong case for re-conceptualising the retirement concept to include ‘home-makers/housewives’ as they felt that their work within the home, though unpaid, was still productive work and they looked forward to ‘retiring’ from their home-maker role. One Asian Indian lady, who was a housewife, said that she would like to start a Match-Making Bureau for young people in her retirement, as she felt that career-minded young Asians had little time to socialise.

Apart from the practical issues, the conceptual and mechanical tasks of employing computer-aided qualitative software raise many challenges. As Kelle explained, ‘conceptual tasks are those activities by which the researcher groups and labels data, then generates inferences from the data to move to a broader understanding of the phenomena under study’, and mechanical tasks are ‘those activities by which the researcher stores, organizes and retrieves the data’ (1995: 115). While computing environments can facilitate the mechanical tasks, the conceptual tasks have to be executed by the researcher. If a research assistant is hired, the challenge lies in supervising his or her work while simultaneously doing the conceptual tasks. The author has learnt through experience that an inept research assistant can delay a research project through being incapable of using or inexperience with computer software. Is this any different from hiring a research assistant in a quantitative study? In my opinion it is because qualitative methodology and analysis is a highly intuitive process which requires ‘tuning into’ the subject to a much higher level than quantitative analysis. Where focus-group analysis is concerned, the context (physical, social, cultural, linguistic) has to be deeply understood so that the statements made are analysed accordingly. Barbour and Kitzinger (Reference Barbour and Kitzinger2001: 32–5) recognised that the researcher's understanding of the context influenced the kinds of data that were produced, which in turn affects the results.

Lessons learnt from FGDs with diverse older Asian people

As mentioned earlier, the author has been involved in international research both in Singapore (with various cultural groups including Chinese, Malays and Indians) as well as in cross-national and cross-cultural research projects. This article synthesises much of what was learned about focus-group methodology and implementation from the author's experiences in these diverse projects. The participants in almost all the projects were aged 50 or more years, except for the study of family carers who looked after their elderly parents or parents-in-law. While the focus-group setting enables the participants to interact and share their experiences with peers, and there is a therapeutic effect, there is the risk of confidentiality being breached if the members do not adhere to the rule of respecting each other's privacy. While this point applies to all applications, it seems that the loneliness and ‘conversation deprivation’ of some older Asians leads to incautious self-disclosures of personal information. This process is called ‘mirror reaction’ (Fern Reference Fern2001: 100), whereby commonalities among the group participants emerge and there is a sense of comfort in sharing. While some Asians are introverts and reluctant to share their life story, in our research we found that after the first 30 minutes most participants were comfortable and contributed to the group discussion. To ensure that the need for confidentiality is well understood, at the beginning of the session the moderator must firmly remind all the participants about the importance of this issue.

Given the diverse languages of the participants, a common lingua franca has to be agreed at the planning stage. In some of our sessions, bilingual exchanges were appropriate, as in English and Mandarin, because the participants were fluent in both languages. Another issue is the ‘standardisation’ or steering of the discussion by the moderators. If different cultural groups are involved, there could be two or three moderators, and so standardisation requires close attention. The depth to which certain topics were discussed varied in different groups, even though the same standard semi-structured questionnaire or topic guide was used. Hence, the moderators have to discuss their strategies, and be committed to debrief in detail the dynamics of each session with the team. Consistent and faithful translation is another challenge. In one research project, we had to decide whether to use a two-step or one-step procedure for the translation and transcription. In the former, the translator first transcribes in the speaker's language and then translates; in the latter, the transcription and translation into the language for analysis are simultaneous. As is obvious, the latter saves time and resources, but the former enables the contributions to be reviewed by the participants (if literate) and by researchers of the same language. Eventually we agreed on direct translation into English, but archived the original tapes for use by ‘indigenous’ researchers.

In cross-cultural and inter-disciplinary research, sensitive topics are difficult to avoid. In gerontology research, they include sexual intimacy, elder abuse, emotional blackmail and family feuds (cf. Culley, Hudson and Rapport Reference Culley, Hudson and Rapport2007). Even if the guideline does not include any anticipated sensitive topic, a group discussion sometimes takes an unexpected direction and touches on cultural or religious sensitivities. The moderator has at all times to feel the pulse of the group discussion and, if necessary, step in and take control. It is not uncommon for a dominant member to steer the group discussion in a particular direction or emotional level for his or her own satisfaction. In such situations, the moderator has adroitly to bring the discussion back to ‘safe water’. The following example of an exchange between a woman participant (W) and a moderator (M) exemplifies the skill. It is from a focus group for the primary health care project and was held at a Senior Day Care Centre:

It emerged that some of the participants had financial problems. The experience was particularly painful for this lady who had recently lost her husband. The moderator had to comfort her and assure her that she would be able to cope with the help of her friends and the agency. As a follow up, after gaining the participant's consent, the manager of the agency was informed of this information so that appropriate action could be taken.

The relationship of the moderator with the group participants is a key influence on the character and quality of the data that are generated. It is not always a good idea to have a moderator who knows the group participants. In many studies, particularly in evaluative research, the moderator may be a member of staff of the service provider organisation, and the FGD participants are clients. In such circumstances, the participants may voice views with considerable response bias, in that they may focus on messages for the staff and management but with self-censoring. Being an academic, I was usually not known to the older people recruited from social service agencies, and this was advantageous as a moderator. In the focus groups moderated by an outsider, clients of social and health services gave frank opinions of government policies and elder-care services. A similar consideration applies in cross-national studies. A researcher that sets out to collect data from another country may be inclined to ask a ‘local’ person to moderate on the grounds of cultural and linguistic commonality, but a ‘local’ person may inadvertently generate discussions that have strong response bias (for an example seeNaylor et al. Reference Naylor, Wharf-Higgins, Blair, Green and O'Connor2002).

‘Ethical issues are relevant to all stages of focus group design, implementation and presentation’ (Barbour and Kitzinger Reference Barbour and Kitzinger2001: 19). While some common ethical issues were discussed earlier in this article such as informed consent, confidentiality and the treatment of cultural sensitivities, another major issue is that focus-group research can be used for ‘political’ purposes. In the field of gerontology, it would not be difficult to imagine the methodology being employed to justify government policies to the less educated, or to learn that findings are ‘massaged’ to support a proposal for radical change. The paramount ethical principle is to be scrupulously honest in reporting the findings, and to state clearly the limitations or weaknesses of a design. Researchers must work hard at every turn to maintain the integrity of their work, and the final presentations (such as reports or conference papers) should take a neutral stance. Any conflict of interests arising from the research should be clearly stated.

Finally, focus-group data are best used in triangulation with evidence collected by other research methods, for example qualitative interviews or surveys. The complementary strengths of different research methods can lead to richer and more authoritative evidence and deeper understanding. Surveys are very useful instruments for capturing the breadth of circumstances and experiences of a population, but qualitative instruments such as focus groups and case studies allow the researchers to gain in-depth understanding of what, why and under what conditions certain processes or phenomena occur. In my own research projects, focus groups have sometimes been the only research mode, and at other times they have been combined with other forms of data collection. My conclusion is that the triangulation of different methods leads to a more comprehensive research outcome. The focus-group literature indicates the value of conducting focus groups to collect information that helps formulate the questions for a later survey. Alternatively, focus groups have been conducted after a survey to deepen the understanding of some of the complex or unexplained findings. A mixed methodology may not be possible in cross-national and inter-disciplinary studies because of the high costs, but even for projects with a small budget, the triangulation of FGD data with case studies is a viable approach.