Cohen and colleagues followed older adults in a professionally led chorale and a comparison group in the United States and reported better self-rated health and morale among the singers (Cohen et al., 2006). A randomised trial in England found that community singing groups improved mental health-related quality of life in older people (Coulton et al., 2015). The scoping review prepared for the World Health Organization places these studies within a wider body of evidence on the arts and health and notes the variable quality of that evidence (Fancourt and Finn, 2019).

Studies closer to the present question look at the social mechanism. Pearce and colleagues compared adult education classes in singing with classes in other subjects and found that singers felt close to their group sooner, which the authors called an ice-breaker effect (Pearce et al., 2015). Creech and colleagues, studying active music-making among older people in England, linked it to a sense of purpose, control and social affirmation (Creech et al., 2013). Qualitative work with a choir for adults living with disadvantage showed how membership can supply a valued identity (Dingle et al., 2013). None of these studies was set in the Irish countryside.

Rural Ireland has an older age profile than its cities, a scattered pattern of housing, and heavy dependence on the car. Post offices, creameries, shops and public houses that once served as meeting places have closed in many districts, while the parish, the Gaelic Athletic Association club and voluntary associations remain strong. The framework of old-age exclusion developed by Walsh and colleagues identifies social relations, services and mobility, and neighbourhood and community among its domains (Walsh et al., 2017). A choir touches each of them, and a study of its social life has to look outward to transport and venue as well as inward to the rehearsal.

Two ideas help to frame what to look for. Turino distinguishes participatory performance, where the aim is to involve everyone present, from presentational performance, prepared by one group for another (Turino, 2008). A choir of older adults usually does both, and the balance between them shapes who feels welcome. Putnam's distinction between bonding ties among people who are alike and bridging ties across social divisions (Putnam, 2000) raises a second question: whether the choir deepens existing parish networks or brings in people who stood outside them, such as recent arrivals, returned emigrants and the recently bereaved.

The case is one choir, followed for one full choir year from the autumn return to the summer break, so that the Christmas season, any festival or competition, funerals and the annual meeting all fall within the fieldwork. Figure 1 shows the case as a set of nested levels, from the individual member out to the policies and services that fund or house the group. The unit of analysis is the choir; members, rehearsals and the surrounding parish are embedded units within it.

A suitable choir is one that describes itself as open to older adults without audition, meets regularly, and is based in a village or open countryside and not in a commuter town. Access should be negotiated in stages: first with the conductor and committee, then with the whole membership at a rehearsal, with time allowed for the group to discuss the request in the researcher's absence. The researcher should expect to be asked to sing, and should decide beforehand how to record the effects of doing so.

Case study method depends on several sources that can be checked against one another. Table 1 lists the six proposed here. Participant observation of rehearsals, tea breaks, performances and outings is the core, written up the same day as fieldnotes that separate description from interpretation (Emerson et al., 2011). Attention goes to arrivals and departures, seating, who helps whom with music or mobility, how mistakes and absences are treated, how the conductor addresses the group, and what is talked about when the singing stops.

Semi-structured interviews follow once the researcher is known to the group. Members are chosen for variety in age, sex, length of membership, distance travelled and living situation, and the conductor, accompanist and committee officers are interviewed about the running of the choir. The design also includes a few people who have left or who decided against joining, since a study that speaks only to current members will overstate how open the group is. Documents such as concert programmes, minutes and attendance books give a history against which memories can be set.

Two short instruments are suggested for describing the membership: the three-item loneliness scale developed for large surveys (Hughes et al., 2004) and the six-item version of the Lubben Social Network Scale (Lubben et al., 2006). They are offered to members once, on a voluntary basis, and are used only to characterise the group alongside age and household details. With one choir and no comparison group they cannot show that singing changed anything, and the design does not ask them to.

Fieldnotes, transcripts and documents are analysed together using thematic analysis as described by Braun and Clarke: familiarisation, coding, the search for themes, review, definition and writing (Braun and Clarke, 2006). Coding begins inductively and is then set against the ideas introduced earlier, participatory and presentational performance and bonding and bridging ties, to see where they fit the material and where they fail. Analysis runs alongside fieldwork so that early readings can be tested in later observation and interviews.

Credibility rests on the measures set out by Lincoln and Guba: prolonged engagement, triangulation across sources, deliberate search for cases that contradict the emerging account, an audit trail of analytic decisions, and checking of interpretations with participants (Lincoln and Guba, 1985). A reflexive journal records the researcher's position, which will differ sharply for a local and an outsider, a singer and a non-singer. Reporting should follow the consolidated criteria for qualitative research so that readers can judge how the study was done (Tong et al., 2007).