Music‐based interventions in community settings: Navigating the tension between rigor and ecological validity

Abstract

Empirical research of community‐based music interventions has advanced to investigate the individual, social, and educational implications of arts‐for‐wellbeing practices. Here, we present the motivations, aims, hypotheses, and implications of this complex field of inquiry. We describe examples of recent large‐scale investigations to reflect on the major methodological challenges. Community‐based music interventions strike a balance between the empirical rigor of clinical trials and the demands of ecological validity. We argue that this balance should be viewed as an asset rather than a mere pragmatic compromise. We also offer some perspectives on best‐practice models for effectively engaging in this type of work.

INTRODUCTION

Millions of individuals around the globe spend significant parts of their leisure time listening to music and going to concerts as well as learning to sing or play musical instruments. Some will invest years of deliberate practice to refine their musical abilities. While very few of these individuals go on to pursue careers as professional musicians, many remain engaged in music activity through formal and informal channels at home or in their communities. In so doing, music activity may become an important cornerstone of participation in cultural life. ¹ , ² Notably, this type of music activity has also become the subject of scientific inquiry through studies investigating the potential benefits for cognitive health, physical health, and social wellbeing. ³ , ⁴ There are a number of such music activities that are based in community settings and that have been subject to research inquiry both for their educational outcomes, for example, the Harmony project in Los Angeles (www.harmony‐project.org) or the Demos program in Paris (www.philharmoniedeparis.fr) as well as their therapeutic outcomes as in the example of a choir intervention for patients with Parkinson's disease (PD) at the John Hopkins University. ⁵ Here, we describe these programs collectively as “Music‐Based Interventions in Community Settings” where the use of music in a community setting is assessed for its influence on learning, education, or therapy outcomes.

Music activities in the community are intriguing subjects for psychological inquiry for several reasons. First, they often represent culturally and socioeconomically diverse groups, crossing boundaries between novice‐, amateur‐, and (semi‐)professional levels. Second, they tend to be driven by intrinsic motivation in both facilitators and participants. ⁶ Third, the population engaged in community music (e.g., choir singers, instrumentalists, or mixed ensembles) often spans several age groups and generations. Furthermore, individual differences may go beyond music proficiency to include education, culture of origin, spiritual and religious backgrounds. Target groups may also include individuals with special needs or mental and physical health issues. The spectrum of challenges to quality of life may extend to anxiety, depression, loneliness, and dire living conditions. In fact, there are community music programs that specifically address vulnerable groups encouraging participation in the presence of significant life challenges. Questions arise as to how these programs contribute to the betterment of individual cognition, health, social wellbeing, and quality of life? To what extent do they entail therapeutic value in the absence of formal therapists and in the presence of people who may consider musical activity primarily for aesthetic or recreational needs and do health and wellbeing exist as side effects? How do they promote education and learning in a manner which is complementary to that accomplished in the school context?

A growing body of empirical work has begun to address some of these questions and issues within a more general arts‐for‐health framework ⁷ but also with respect to specific target groups, such as people living with neurodegenerative diseases, ⁸ communication challenges, ⁹ mental health issues, ¹⁰ , ¹¹ and challenging socioeconomic barriers, despite the interest to learn to play a musical instrument. ¹² , ¹³ While some of this research has addressed or was inspired by existing programs, such as El Sistema, ¹⁴ other programs have developed community‐based music therapeutic interventions for specific vulnerable populations, such as singing groups for individuals living with communication challenges in SingWell (www.singwell.org). The overarching aim of these initiatives is to investigate the impact of these programs on the neural, perceptual, cognitive, and/or social and emotional wellbeing of their participants across the lifespan. In younger populations, specific attention has been given to assess the efficacy of music activities and their influence on neurocognitive development, ¹⁵ , ¹⁶ as well as aspects of psychological wellbeing, including a sense of agency ¹⁷ and emotion regulation. ¹⁸

We agree with the notion that randomized controlled trials are the gold standard of scientific inquiry where interventions are concerned. ¹⁹ However, we argue here that given the dynamic and complex nature of music learning and practice, the use of community‐based samples can be an appropriate alternative, particularly in the early stages of research prior to the development of a formal training protocol. Within the context of community‐based trials, there remain significant challenges concerning the design, development, and implementation of studies. A range of factors to consider include selection of appropriate control conditions, participant recruitment and retention difficulties, access to measures that can easily be implemented in a community setting, lack of ecologically valid, and standardized batteries across studies. ²⁰ Elements of community‐based music programs are particularly difficult to describe fully because of variation among community sites, selection of music based on the culture and preference of attendees, participants to practitioner ratio, and other factors. However, it is important to systematically note the exact elements of community music programs in order to provide evidence for their efficacy while at the same time consider their ecological fit and validity.

Recently, weighty concerns regarding assumptions and biases implicit in arts‐for‐health research, including a need for robust critique and higher rigor, have been proposed. ²¹ Although the myriad benefits of arts interventions for wellbeing and health are not disputed, the concern lies with a lack of transparency, frequent reporting errors, and ill‐defined research designs that can preclude the interpretation of studies that have been included in several systematic reviews. ²² While acknowledging this concern, we, however, do not see this as an inherent shortfall of community‐based interventions and note that such criticisms can be used to implement practices to enhance the validity and reliability of these trials.

The purpose of this paper is to consider the available evidence from correlational, cross‐sectional, and longitudinal studies on community‐based music programs targeted at populations across the lifespan both in the context of a parallel research study or an intervention. We first describe the design and implementation of four community‐based models of music interventions in children, adults, and older adults that we have individually conducted in four different locations. Using these studies as example models, we will discuss methodological and theoretical challenges and consider the limitations associated with this research area. We will then provide what we consider to be best practices for music intervention research in community settings in comparison to randomized controlled trials and identify promising directions for future investigations. Finally, we discuss the development and implementation of a comprehensive dissemination plan so as to close the circle, providing information and data to the community stakeholders and policymakers to further support the development and flourishing of the communities.

Study in a community setting

Our first example of a music‐intervention project launched in a community setting originates from Finland. In the Helsinki metropolitan area, some municipalities welcome music play schools and dance schools to give their lessons in the daycare centers. These activities are not a part of the regular daycare curriculum, instead, the parents need to enroll their children in the activities and, in most cases, pay an annual fee (around 400–500 euros/academic year) for these weekly group‐based sessions. The researchers of the University of Helsinki were given access to implement their longitudinal study on children's language development in this context. The study was conducted four times with over 80 children in 26 daycare centers for 2 academic years using neurocognitive behavioral tests and auditory event‐related potential recordings. All research was implemented in the daycare centers, meaning that the researcher and assistants visited the centers during four periods of data collection until the data collection was completed. In addition to participants of the music playschools, participants of dance schools were recruited as active control participants. Children in the passive control groups took part in the daily curriculum of the daycare centers but did not participate in group‐based music or dance sessions. So, regarding the active groups, the study was launched in a community setting with those children as participants whose parents had enrolled their children in the weekly music or dance lessons and whose parents gave their children permission to be part of the study.

The neuropsychological tests, conducted and analyzed for 66 children, indicated that phonological processing and vocabulary were more readily developed in the children involved in the music activities than in the dance activities or in the participants of the passive control group. ²³ This was indicated by a significant interaction between music playschool and time using a linear mixed model growth curve analysis in these language‐oriented tests. This suggests that even if both music and dance sessions include auditory information and music, rehearsal in sound production and singing might play a key role in facilitating both phoneme processing and vocabulary in music sessions. Interestingly, even if the mismatch negativity (MMN) evoked by acoustic changes in speech stimuli was associated with the performance in the phonemic awareness test in the follow‐up, the MMN was not modulated by the music or dance activities. ²⁴ Notably, it turned out that a subsection of the children had started the music or dance lessons before the follow‐up, dropped out of the groups during the follow‐up, or participated in such activities outside the daycare despite their original group status. Thus, the final analyses used the number of months in either activity as the predictor instead of the original group divisions. This enabled the inclusion of all children in the analyses despite the possible change in their group status. Notably, this study did not have a randomized control trial (RCT) design, as the research group was not able to provide music and dance lessons free of charge. However, the lessons were of reasonable price and each participating daycare center offered either the music playschool, dance lessons, or neither; thus, the parents were not asked to choose between the three options.

Systematic control interventions

Study 1

Our next example of music‐intervention project results from a collaboration between researchers at the University of Helsinki and Beijing Normal University. They launched a large longitudinal intervention study with over 120 participants in a Chinese elementary school. ²⁵ , ²⁶ The aim of the study was to investigate whether extracurricular music lessons facilitate the neurocognitive development of the children 7–11 years of age. Group‐based music lessons were given twice a week during the semesters for 1 academic year with some additional homework for the holidays. The impact of music lessons was compared to that of lessons in a foreign language (English). Additionally, a passive control group was included. Research methods included group‐based tests in academic skills, such as reading and math. Individual tests included cognitive tests (e.g., working memory) and auditory event‐related potential (MMN and P3a) recordings.

Various parts of the study outcome are still being analyzed and reported, but the following findings are already established. First, regarding the auditory working memory as indexed by the digit span test of the Wechsler Intelligence Scales for Children, it was found that in children involved in music lessons, their digit span was improved in the backward but not in the forward subtest. We observed a group difference for digital span backward tests (one‐way ANCOVA, F(2, 106) = 5.038, p = 0.008) that resulted from higher performance in the music group than in the language or control groups. No such group difference was observed for digit span forward scores (one‐way ANCOVA, F(2, 106) = 0.583, p = 0.560). ²⁴ This implies that music lessons selectively improved the central executive system of auditory working memory when compared with language lessons or no lessons. In other working memory functions in auditory or visual modalities, there were no effects of the lessons. Second, regarding the development of neural indices of sound discrimination, it was found that the lessons of the foreign language enhanced the MMN more than music lessons or no lessons. ²⁵ This was shown by a significant group × time interaction in the Melody transposition condition that resulted in a larger MMN amplitude in post‐ than pre‐training measurements in the English group but not in the other groups (p < 0.01). This somewhat surprising result was obtained by using melodic musical stimuli and was taken to highlight the reciprocal relationship between music and speech encoding in the developing brain. Despite these promising results, it should be noticed that the study is not without limitations. The original aim was to conduct the study as an RCT. However, after random group allocation of the children, 26 of them did not want to join the music or language lessons but, instead, became participants in the passive control group. Since 19 of these 26 were boys, this unavoidably caused an unbalanced gender distribution of the groups. In addition, while over 123 children participated in pre‐ and post‐program tests for working memory on the school premises, only 85 of them successfully completed the post‐program electroencephalography (EEG) tests in the laboratory of the university.

Study 2

In a similar vein, in Los Angeles, Habibi and colleagues intended to uncover the effects of music training on the developing brains and undertook a 5‐year longitudinal study of school‐aged children ages 6–7 years old. ²⁷ Given the scarcity of affordable music education programs in the United States in general and specifically in Los Angeles, they opted to partner with an existing and successful community‐based music training program. Youth Orchestra of Los Angeles at the Heart of Los Angeles program, known as YOLA at HOLA (www.laphil.com/learn/yola/youth‐orchestra‐los‐angeles), specifically catered to children from under‐resourced socioeconomic backgrounds. YOLA is a signature education program of the Los Angeles Philharmonic that is inspired by the Venezuelan approach to music studies known as “El Sistema.” ²⁸ It offers free group‐based music instruction 3–4 days a week to children from underserved communities of Los Angeles.

Two control groups were selected for comparison with the music training program. Children in the active control group were recruited from community‐based soccer or swimming programs. The selection of sports training allowed the research team to control for aspects of development that are shared between music and sport learning, namely, social engagement, discipline, motivation, and sustained effort. Sports training was also chosen because of its reliance on sensory‐motor learning, a component that is widely shared with music training. Children in the passive control group were recruited from public elementary schools in a similar Los Angeles area. At the time of recruitment, those children were not engaged in any organized and systematic after‐school programs. In order to monitor participation in extracurricular activities, the research team conducted a systematic interview with the children participants and their parents so as to better understand the frequency of participation in any enrichment program, including music and sports.

Over the span of 5 years, the influence of music training on the children's development was observed first as near transfer effects with changes in musical and auditory skills followed by far transfer changes in nonmusical skills, including cognitive abilities and socioemotional maturation. For example, children who received music training performed better than children in the comparison groups on tasks measuring rhythm discrimination (F (1, 46) = 3.01, p < 0.1) and were also better at perceiving tonal irregularities in musical phrases (F (2, 34) = 7.99, p = 0.001). ¹² , ²⁹ , ³⁰ They also showed an increased functional development of the auditory pathways as measured by changes in the amplitude of the early cortical auditory evoked potentials P1 and N1 to changes in the pitch of musical notes (F (2, 34) = 3.01, p = 0.05). ¹² In relation to cognitive abilities, children who received music training developed better inhibition control earlier than the comparison groups. For example, compared to their control counterparts, they were capable of rejecting a small reward in favor of larger and better rewards at a later time (F (2, 56) = 3.49, p < 0.05). ³¹ Music children also performed better in assessments requiring task‐switching skills, and they displayed stronger engagement of the brain's prefrontal network, including the bilateral pre‐supplementary motor area, precentral sulcus, insula, anterior cingulate, and the inferior frontal gyrus at an earlier age. ³² Importantly, they showed more robust connectivity—as measured by increased fractional anisotropy—in three segments of the white matter pathway (the corpus callosum) connecting the left and the right hemisphere (F (2, 40) = 4.19, p = 0.02). ³³ Despite the success of the trial and the robust and promising results, the study was not without challenges or limitations.

Among the challenges, the research team faced difficulties in recruiting and retaining participants. This was especially true given the participants were primarily from low socioeconomic communities with less stable housing, employment, and overall economic opportunities. The research team addressed issues related to retention and cross‐over by systematically monitoring participants’ daily activities and overall participation in enrichment programs using biannual interviews with them and their families. Also, given that participants first enrolled in their program of choice/interest and then recruited to the study, it was not possible to rule out pre‐existing biological tendencies for musicality, family's motivation, or different home environment with respect to music.

Study 3

Our last example comes from Toronto where Russo and his colleagues found their way into community‐based singing interventions rather serendipitously. In the late 2000s, they were studying spontaneous facial mimicry that occurs during audio‐visual perception of song. ³⁴ At around the same time, they had joined a research network led by Annabel Cohen, called Advancing Interdisciplinary Research in Singing (AIRS). The AIRS network sought to understand singing from a developmental, pedagogical, and health perspective. Other members of the network were studying the communication benefits of group singing. ³⁵ The confluence of these projects led the Toronto‐based team to assess whether they could use singing to support communication challenges faced by people living with PD, including but not limited to facial mimicry.

To do so, Russo and colleagues recruited participants living with PD who were associated with support groups affiliated with the Parkinson's Society of Canada. Their first study confirmed a deficit in facial mimicry but not a total absence. ³⁶ They also demonstrated a motor‐behavioral link by way of a negative correlation between zygomaticus activity in response to positive emotions and emotion identification response times. The team later asked these same participants to participate in a singing intervention they developed that involved intentional vocal and facial imitation. The intervention, which was carried out on an individual basis, was a success with respect to mimicry and vocal strength outcomes, ³⁷ but it became evident that this individualized approach was not scalable, nor was it highly enjoyable for participants. These realizations led the team to establish their first community‐based choir consisting of people living with PD.

This first community‐based choir had an emphasis on inclusivity and fun but not letting go of the pragmatic intentions regarding the rehabilitation of communication function. The choir flourished, eventually developing into a registered nonprofit charity (Singing with Parkinson's) with chapters in three other Canadian cities (www.singingwithparkinsons.com). The success of these community choirs prompted the team to establish the SingWell network, which considers the benefits of group singing for people living with various communication challenges. For example, in an RCT involving choir singing versus a do‐nothing control, the team found that 10 weeks of participation in a choir led to improvements in various aspects of auditory processing in older adults with hearing loss, including speech‐in‐noise perception. ³⁸ While the main measures in these early studies were communication outcomes, the most salient outcomes noted by the research team escaped measurement entirely. It became clear that participants were not only improving with respect to communication function, but they were also feeling good about themselves as individuals and as a community united in their ability‐focused activity.

Researchers in the SingWell network have documented regularities in the evolution of these new singing groups. Participants characteristically experience a mood boost following singing sessions. The absolute size of these boosts does seem to dwindle over time, but they continue to accrue. The team has found evidence that at least some of these mood‐boosting effects depend on group singing and cannot be obtained while singing individually. ³⁹ The team assumes that these group‐mediated benefits hinge on emphasizing fun and inclusivity while de‐emphasizing the pursuit of musical perfection. In keeping with these assumptions, the repertoire tends to focus on simple, familiar, and culturally appropriate songs.

The SingWell network now embraces a biopsychosocial model to capture the complete scope of their observations more fully. Specific outcomes typically include neurohormones, loneliness, stigma, and social connectedness, alongside communication outcomes (www.singwell.ca). The challenges encountered in conducting this type of research are manifold. How do you establish these singing groups while maintaining optional participation (i.e., without any coercion) in a parallel research study? What control group is appropriate? How do you effectively deal with demand characteristics? How do you maintain the joy of singing while ensuring treatment fidelity and adherence to protocols for testing and measurement? These challenges continue to present themselves as the SingWell project evolves and expands (now in its third cycle of project funding, 2021–2028, with partners across the globe).

The examples from Helsinki, Beijing, Los Angeles, and Toronto are necessarily diverse and situated in different cultural and research contexts. Yet, they converge in at least two respects. First, they reveal that detected patterns of effects are often robust across different research contexts. ⁴⁰ Second, and perhaps more importantly, they reveal that cognitive, neural, social, and health science approaches to investigating community‐based music interventions should be seen as two sides of the same coin.

Opportunities, challenges, and limitations

The transformation of musical activities as music interventions

Music is an attractive leisure activity, which rarely raises issues in terms of lack of compliance. Some motivations are intrinsic to music per se, such as experiencing pleasure and reward, or developing mastery, whereas other motivations can be extrinsic, such as socializing or playing music as a distraction from everyday routines. ⁴¹ However, it is also common that choral singers or amateur musicians report psychological, physical, and social benefits from engaging in music. ⁴² In brief, health and wellbeing outcomes can be understood as side effects that can play a major role in perceived quality of life. It is likely that they support the adherence and commitment to music engagement for far longer time spans than can be captured by research interventions. ⁴³ Therefore, high, ecological validity characterizes the investigation of the health and wellbeing implications of music interventions. Moreover, although they are not designed as clinical trials, they still can meet methodological criteria, such as randomization, blinding, or the installment of control groups. ⁴⁴ In these respects, music intervention research is similar to any other behavioral intervention, such as mild sports exercise, yoga, or meditation. ⁴⁵ All of these activities offer demand characteristics that partially overlap with one another, and, therefore, appear suitable for comparison and investigation of their pros and cons with respect to target groups as well as individual differences and preferences. Importantly, but not specific to music, the cultural background and context can be influential on the materials being selected for intervention. ⁴⁶

At first sight, the above differences may put community‐based approaches at a disadvantage as compared to clinical trials. However, community‐based approaches offer a number of features that render their findings as relevant to the quality of life, wellbeing, and health. ⁴⁷ First, research in music interventions has consistently shown the potential to reveal beneficial effects in all these respects while focusing on a wide range of target groups. Second, contrary to clinical trials, their value is not restricted to adjuvant therapy, but they also entail disease prevention and health promotion. Third, music interventions seem to work alongside different cultural activities, thus complementing a wider range of strategies rather than offering a specific form of psychoactive substances. In other words, the controversial notion of prescribing music for health purposes notwithstanding, engaging in music remains a matter of choice rather than prescription. Such observations urge the question of whether or not participation and compliance may or may not be considered as health‐related variables in their own right.

A more challenging side in the methodological approach to implementing music interventions in a research context comes with the understanding of the role of practitioners, such as teachers and leaders of choirs or ensembles—who are primarily experts in their musical domains. However, their roles certainly go beyond planning, organizing, and conducting music activities because of their relationship with the target group and the social dynamics that are typical for the interactions between singers, musicians, and their leaders. Therefore, it is essential to consider researchers’ and interventionists’ perspectives without presuming that they ought to be identical. For example, it is often overlooked that informed consent is required from participants only, as it appears that informed consent would be naturally assumed from music practitioners as long as they do not provide research data. Obviously, as the role of a music practitioner changes to become a research associate, it is not always clear what such a shift of roles implies or not implies. Given that music intervention studies entirely rest on the full compliance and mutual understanding between researchers and interventionists, it is not surprising that strategies, such as fidelity measures (see below), are being developed to address some of the potential issues and biases that surround the need for transparency and confidentiality in music intervention research.

Perhaps the most significant challenge in designing music intervention research is to install comparison conditions that match the target activity in terms of compliance, motivation, and other aspects. To give an example, Bullack and colleagues investigate the role of physical presence versus singing in presence on behavioral measures of self–other immersion and emotional affect, on the one hand, and physiological measures of stress in a mixed group of adults (N = 54; 44 females; mean age: 59.6 years, age range: 18–85 years). Findings suggest that singers showed enhanced self–other immersion and positive affect as compared to nonsingers, whereas physiological measures were unchanged. ⁴⁸ It could be argued, however, that not the singing per se, but rather the compromised motivation of the nonsingers drove the observed effects. Therefore, using wait‐list controls (i.e., having the control group engaged in the same activity as the target group) after an initial period of doing‐nothing could guard to an extent against demotivation effects. However, such strategies are obviously not appropriate in situations, such as palliative care. Additionally, they are suboptimal in educational settings when relatively long (2–3 years) training programs have been found to be necessary, causing differences in the degree of neuroplasticity and neural maturation between the children. ⁴⁹

In summary, the framing of everyday cultural activities as research interventions is subject to continued discussion and reflection. How to balance the need for methodological rigor while maintaining high levels of ecological validity remains a multifaceted challenge. Perhaps most importantly, it involves a practice‐research partnership that comes to live only by taking the positions, requests, and insights from either side seriously and at an equal level. In fact, scientific understanding only follows practical experience as it does not preclude or direct any activity in the field. Researchers must be aware that regular and frequent communication is crucial to avoid conflict that may arise from divergent expectations. Best‐practice models can only work if necessary, adaptations with respect to divergent personal, social, and cultural backgrounds are acknowledged. Mutual agreement between all participating parties should be reached in order to render a successful transformation of the activity in response to the intervention.

Motivation and compliance in the context of community‐setting music intervention

Psychologically, participants should be optimally challenged to retain their intrinsic motivation and pursue their own learning goals and expectations. ⁵⁰ Notably, in the opinion of many researchers, this agency to choose may invalidate a scientific study as it conflates motivation with treatment. The antidote here is an RCT consisting of all the proper blinding procedures. However, in some cases when RCT design has been used in the first place, ²⁶ there have been high numbers of dropout participants before and during the intervention, indicating that participants “vote with one's feet.” ⁵¹ , ⁵² So, we can argue that there is an important place for well‐designed studies that are not RCTs. Such studies would optimize intervention outcomes by building on the motivation of the participant rather than trying to suppress it. Furthermore, whether the protocol is based on randomized assignment or participant's choice, robust measures of fidelity that assess the degree of participants’ motivation (e.g., in the forms of questionnaires or brief interviews by researchers) are necessary. Such measures could become part of the statistical modeling to better understand the cause of observed outcomes.

Implementation challenges: Fundings, terminology, and timelines

This type of research often falls between conventional funding lines in its orientation making them less competitive for funding support compared to the traditional RCTs. However, their contribution to understanding human behavior is none the less critical. A key difference in this approach and something that emerges from a clinical trials perspective is that researchers will often work to cater a protocol to individual needs. So, for example, if a participant cannot stand for 1 h of singing, they may be allowed to sit, or if their vocal capacity does not allow for continuous singing, they may be encouraged to hum or mouth the words or simply take a time out. This is quite different from what is typical in an RCT or in the laboratory where we ask the participant to adhere to the experimental protocol as much as possible. In cases where the participant cannot adhere to the protocol, the participant is generally asked to inform the researchers so that their data may be screened from the final sample. Despite this flexibility, these studies often include methodological rigor that clearly draws inspiration from the natural sciences, where consistency and precision of measurement are paramount. So, for example, researchers may select to use EEG testing or a salivary assessment of a neurohormone to assess changes that may be occurring as a function of participation in the community music activity. These methods are often costly, and they require expertise, so they are carried out with strict adherence to protocol. Understanding the value of these tests may also require disciplinary expertise that add to the cost of the overall investigation. However, with the exception of special calls that span research councils, the dual orientation of these studies complexifies the interpretation and the right fit for finding support.

Another challenge in funding is providing support for the intervention program, which often is more costly in time and financial resources than typical laboratory‐based trials. In some cases, such as the study out of Los Angeles ²⁷ that was discussed above, researchers are interested in assessing interventions that may take up to 5 years with outcomes assessed at even longer durations. This is particularly true when the intended investigation is an assessment of possible benefits of music on nonmusical skills, such as language or executive functions in children and young adults that involve neuroplastic changes that typically emerge over longer intervention periods. For example, Holochwost et al. used a Go/no‐Go paradigm in children aged 7–13 participating in a collective musical training program and reported an increased inhibition performance after 2 and 3 years when compared to a passive control group. ⁵³ If a funding cycle spans only 2 or 3 years, it can be difficult to fit the entire paradigm into one proposal in a meaningful way. In addition, these longer interventions require more resources over time to support the practitioners and participants.

Further challenges are given by the increasing demands of the intervention protocols. The more we know about the efficacy of the music interventions, the less likely it is that the inclusion of traditional passive control groups is considered appropriate for ethical reasons, particularly in trials spanning more than a few weeks.

Best practices going forward for music intervention research in community settings

With the recognition that the gold standard for intervention is a double‐blind randomized controlled trial, we propose that community‐based trials are appropriate alternatives to RCTs when conducted following best practices of design, implementation, and dissemination. Developing and establishing best‐practice models for community‐based music intervention research need to combine elements from different worlds with seemingly opposing strategies. ⁵⁴ On the one hand, clinical trials often focus on a small set of outcomes within frameworks that must rely on strict inclusion and exclusion criteria. They rely on quantifiable measures to ensure the significance and impact of health strategies at group levels. On the other hand, community‐based interventions are often open to individuals with different needs and often have less strict inclusion and exclusion criteria, and also may entail larger degrees of freedom within intervention protocols. Therefore, even when target groups are considered, studies are associated with less rigorous methodologies and often include a broader range of outcomes.

Best‐practice models must rely on selecting optimal individual or group outcomes to mark their effectiveness in the domains for which they were designed. Moreover, they need to account for feasibility, pragmatism, and above all, cost‐effectiveness to increase their chances of becoming established in health policies. Both clinical trial and intervention research may entail musical activity as a central agent. This is remarkable and enough reason to consider their complementary value to inform health politics. Some of the similarities and differences between each of the research domains were noted, but a lot more remain to be explored. Given the complexity of community‐based music intervention, in the next section, we provide recommendations of best practices to enhance the quality of the research studies associated with such interventions. The recommendations are gathered from our individual and collective experiences in conducting music‐based trials and are meant to build on the existing guidelines. ⁵⁵

Design and participants

There is an emerging literature that addresses the methodological issues and challenges of arts in health and education research. ²² It seems clear that research designs and selections of outcome measures must accommodate the specific context and individual acculturation of study participants. Although there is no ideal control condition for music intervention, including a control group that is participating in a comparable activity (i.e., active control), it is still recommended (see https://bit.ly/3pvZQCC for a videocast of the National Center for Complementary and Integrative Health, March 31, 2021). We sympathize that adding an active control condition effectively doubles the cost and required resources of an intervention. However, given the necessary resources that go into designing and implementing community‐based trials, it would be a waste of resources to use inadequate control conditions. In cases where including an active control group is not feasible or the study question is in the beginning/pilot stages but a comparison with a passive control would yield important insights, we recommend that the passive control group be offered the possibility to join the intervention at a later time, such as in a wait‐list control design. Allocating adequate funding and resources for recruitment and retention of participants for both intervention and control groups should be considered during the design stage.

Partnership with community organizations, teachers, social workers, and parents

If the primary objective of the trial is to assess an intervention based on an existing community music program, it is recommended that the research group engages the stakeholders and leadership of the programs from the early stages of the development and design of the study. This will facilitate a better understanding of the study goals, the role and responsibilities of the community program, and the potential benefits to the program. ⁵⁶ , ⁵⁷ In our experience, engaged community partners are key to the recruitment, retention of participants, and overall success of trial. ³⁰ Planned regular and frequent communication and meetings with the program staff and leadership allow for early identification of possible issues with retention, dissatisfaction with the program, or general problems with the delivery and implementation of the intervention.

Another effective path to engage with the partner community organization is if planned assessments can be carried out at the community centers, schools, or Kindergartens. ⁵⁸ The use of community resources, which are familiar and held in familiar places, is cost‐effective and can separate the intervention from what participants might consider a medical or academic (university) setting leading to hesitation to participate.

There is a broad consensus that community‐based interventions are partly designed and developed to assess the potential benefits (or lack thereof) of community programs. Building a dissemination plan that engages the community and provides education about research findings to the community members is essential in the continuity of such trials. Presumably, community members would be more interested and invested in participation if they recognize that findings from the parallel research program can contribute to more investments and resources for the community (see Ref. 57, chapters 2 and 3, for details).

Treatment fidelity

Music intervention studies have been criticized in the past for lack of information related to the specific contents, modes of delivery, provider training, adherence to interventions, and other characteristics that could potentially influence outcomes. In brief, music interventions can be seen as “black boxes,” in which the levels and amount of engagement are often obscure, or at least they are at best partially reflected in methodological descriptions. ⁵⁵

Fidelity, in general, refers to the methodological strategies to systematically measure the degree to which an intervention is delivered as planned. The broad and diverse format of community‐based music interventions highlights the importance of treatment manuals and systematic measurements of fidelity. Music interventions provided in community settings often incorporate a more ecologically valid method of engaging with music over a rigid lab‐based treatment in which dosage and delivery might be easier to capture. This may be reflected in more flexibility of lesson plans, selection of repertoires, and inclusion of peripheral components that are not initially in the study design. Breitenstein and colleagues have identified several limitations in the implantation and maintenance of fidelity in the context of community settings. ⁵⁹ These include variations in training and competence of practitioners, personal adaptation to curriculum or lesson plans, limited resources for training and support, different level of adherence by participants, and competing demands for the practitioners' time in community settings that can impact their commitment.

The development of treatment manuals and fidelity measures has been introduced as one strategy to counteract the above biases, which eventually may even compromise outcome measures. ⁶⁰ The treatment manual is a user‐friendly scripted curriculum. When realized in community‐based research, the treatment manual needs to exhibit conditional flexibility to accommodate the needs of different communities. At least four main components of fidelity, namely, design, provider training, administration, and treatment receipt, need to be addressed in the context of music interventions. ⁶¹ Reporting fidelity thus has a dual purpose. First, it is thought to enhance the transparency and content of interventions and, in part, contributes to quantify the potential biases that could limit the interpretation of research findings. Second, such reporting may facilitate replication by providing the researchers with important guidelines on how to implement interventions in comparable settings. Finally, fidelity reporting does entail valuable research data that can be useful, for example, to address the feasibility or cost‐effectiveness of different kinds of interventions. In summary, implementing measures of fidelity can ensure that the music interventions are delivered with high quality and consistency across different sites, groups, and practitioners. They may also contribute positively to the validity and reliability of outcome measures.

Clearly, identifying and clarifying the importance of fidelity measures does not answer the question of how to approach such measures and what level of detail is required to assess their potential impact on outcome measures. For example, measuring the dosage of exposure to an intervention can be challenging, depending on its location (at home, in a school or Kindergarten, at a community center, or in a lab), the frequency of exposure, or individual differences.

With respect to the need for fidelity measures outlined above, it must be noted that they come with a cost. They may place an additional burden on research protocols as identifying biases must be performed with the same rigor as any other assessment, including the main and secondary outcome measures. They may also be perceived as somewhat intimidating by interventionists, who may not be used to being observed and evaluated as practitioners, and, as a consequence, could change their behaviors in response to fidelity measures. Moreover, the participants are not necessarily in favor of being recorded in video and/or audio during the progress of their training or therapy process. Alternative suggestions for fidelity measures are in the form of diaries (written or questionnaire‐based ones), numerical notes on the format of the intervention (frequency and duration of singing, playing, clapping, discussion, dancing, etc. during each session ⁴⁹ ), as well as adherence and content of homework assignments. In the case of child or patient participants, many of these measures imply the help of family members. For example, in the longitudinal study by Habibi et al., the research team conducted frequent interviews with the participants’ families to assess adherence to music practice at home. ³¹ Likewise, in the longitudinal study by Tervaniemi and colleagues, ²⁶ the attendance rate of the school children in the extracurricular lessons was counted and reported in their Table 1.

In summary, treatment fidelity comprises an attempt to unlock the “black box” that characterizes complex interventions, including music in social situations. From a research perspective, such unlocking appears necessary and appropriate. However, not all people feel comfortable or happy to be observed in social situations, in general, and while participating in cultural activities, in particular. Therefore, it is not just the protection of personal data that may be in conflict with a rigorous approach to treatment fidelity but also a systematic influence that can be brought into a quasi‐experimental situation by installing the full arsenal of fidelity measures. In other words, measuring treatment fidelity could interfere with the main goals of intervention research, and, in the worst case, undermine participant motivation. One solution could be to extend piloting phases to assess the provision and appropriateness of fidelity measures at each stage of the implementation or to limit fidelity measures to rather superficial measures, such as physical presence (rather than engagement and activity).

Economy, ecology, and ethics of community‐based music intervention research

Music interventions are complex and dynamic and as described above assigning participants who are not motivated or emotionally engaged with music poses the risks of increased dropouts or cross‐over to the control condition (e.g., sports). They are economical in the sense that activities at group levels require a minimum of physical resources. They are associated with usually high degrees of compliance, motivation, and engagement on the side of participants, irrespective of a large number of variables related to their sociocultural background and individual health status. ⁶² Community‐based music interventions also offer high ecological validity because they use resources provided by the participants (musicality, sociability) and exploit shared knowledge as most individuals have a musical biography of lifelong experiences. Finally, we consider such interventions as ethical because none of the existing large‐scale studies appear to have raised substantial ethical issues or queries that would question the appropriateness in health contexts. That does not preclude the continual monitoring of the ethical aspects, but for the time being, it must be assumed that music interventions are safe and provide opportunities for emotional, cognitive, and social reward, and, therefore, must be considered as viable strategies in promoting health and supporting education outcomes in communal settings.

CONCLUSION

In our view, the transition from a traditional lab‐based experimental setting to a more ecologically valid setting in the community reflects a more general trend in the studies of auditory cognitive neuroscience. Unavoidably, this transition sometimes implies compromises in the methodological choices, such as in participant recruitment. For example, RCT procedures could be compromised due to the high dropout rate in programs, which may take years to complete. Yet, in parallel, studies in which both intervention and data collection are conducted outside the academic context play a key role in improving the generalizability of our conclusions. As pointed out by Linnavalli and colleagues, ⁴⁰ by giving up on the need of the participants to come to the laboratory, be it in the hospital or university, we are highly likely to also involve those participants who would not have the resources or interest to join the lab‐based data collection. In other words, we increase the possibility of including diverse populations (ethnically, racially, socioeconomically, and age) in studies with a community‐based setting.

This paper has documented various experiences and reflections of researchers who have engaged in music‐intervention research that adheres to, and in some cases, steps outside the bounds of a traditional lab‐based study protocol. These experiences reveal an essential tension between aspirations for experimental rigor on the one hand and ecological validity on the other hand. We have argued that there is value in community‐based music interventions that engage stakeholders in the community and build on the personal motivation of participants. The benefits of such studies introduce risks that may be mitigated, at least to some extent, through the coding of variability in motivation and fidelity, a commitment to open science methods, and adherence to validated testing protocols. We argue that compromises are necessary when a clinical trial is not possible or is not desirable due to ethical considerations (e.g., palliative care) or the risk of dropout rates due to random assignment is especially high, for instance, in adolescence. These studies that embrace the variability and the humanity of community music have high ecological validity, are likely to scale easily, and are more likely to capture participants who might otherwise be averse to engaging in studies administered by universities or hospitals. Through the dissemination of such work outside of traditional journals, we also stand to have a societal impact that reaches far beyond academic journals. To conclude, we wish that our contribution highlights the value and contribution of community‐based music intervention projects in spite of their compromises, limitations, and challenges.

AUTHOR CONTRIBUTIONS

A.H. and M.T. conceived and devised the main conceptual idea and outline for the paper. G.K. and F.R. further formulated and critically revised the paper in keeping with the intellectual framework. All authors contributed to the writing of the final version of the manuscript and approved the final submission.

COMPETING INTERESTS

The authors confirm that there are no relevant financial or nonfinancial competing interests to report.

PEER REVIEW

The peer review history for this article is available at: https://publons.com/publon/10.1111/nyas.14908.

Habibi, A. , Kreutz, G. , Russo, F. , & Tervaniemi, M. (2022). Music‐based interventions in community settings: Navigating the tension between rigor and ecological validity. Ann NY Acad Sci., 1518, 47–57. 10.1111/nyas.14908