Sage Journals: Discover world-class research

Abstract

Advances in portable music listening technology have increased the extent to which music is integrated into everyday life, providing it with the ability to accompany listeners at any time, anywhere. Previous research has attempted to understand and parse the functions of music listening (FML). However, there is a tendency for subsequent models to be oriented towards purely cognitive domains (e.g., mood regulation), without considering the impact that contexts may have on listeners’ experiences. Rarely do such models provide a unified construct that captures the breadth of functionality more broadly (i.e., contextually-determined utility). In this study, we employed a mixed-methods exploratory approach to initially assess FML qualitatively through bibliometric analysis and a comparative experience sampling method (ESM) study, resulting in an exhaustive model of FML with 53 distinct functions. Following this, a list of 114 items intended to reflect the content of these functions was generated. This reflected the conceptual content of the qualitative model. These items were rated through an online survey, leading to dimension reduction through factor analyses. Exploratory factor analysis (EFA) implied a latent construct across five dimensions to explain the resulting underlying construct of FML (Identity and Social Bonding, Emotion Regulation, Focus and Concentration, Background and Accompaniment, and Physiological Arousal). This was subject to model constraints through confirmatory factor analysis (CFA) in which a measure containing 23 items based on the stabilized EFA was fit. The five-factor model was a good fit for the observed data, presenting a latent structure of utilitarian FML generated from the previously identified qualitative framework. This article concludes with suggestions regarding the potential re-application of this model to ecologically valid data to cross-validate this psychometric structure in future work.

Keywords

Functionality functions of music mixed-methods music in everyday life psychometric development

Introduction

Music listening is a ubiquitous mode of human behavior. It enables individuals to manage emotions, facilitate social interactions, and provide cognitive regulation, and researchers have taken varied approaches to uncover the functions that music serves in everyday life. These include qualitative literature reviews (e.g., Lamont et al., 2016), quantitative applications (e.g., Groarke and Hogan, 2018, Greb et al., 2019, Greb, 2018; Lonsdale and North, 2011), as well as summary approaches, consolidating both theoretical and empirical conclusions (Schäfer et al., 2013). These varied approaches have contributed significantly to our understanding of music and its functions, namely the diverse social, emotional, cognitive, behavioral, and physiological functionalities it serves, and it is to this area of study that this paper contributes.

Specifically, we aim to generate a consolidated utilitarian framework for the functions of music listening (that is, from the perspective of purposeful applications of music in everyday life); a perspective that has been thus far limited in its application to empirical research and measurement. Models of functionality typically refer to “affective” states of music listening, such as the attainment of cognitive goals over extended periods of use (e.g., Groarke and Hogan, 2018). Though important and valuable in their contribution, what such models seldom integrate is an awareness of (or applicability to) functions of music listening in situational or cross-sectional settings. We argue that an alternative framework is needed to gauge functionality, given that the functions of music listening are situationally determined (e.g., Greb et al., 2019). Therefore, the short-term nature of cross-sectional need is what drives this re-orientation.

This paper is structured as follows. We begin with an overview of the functions of music listening, before outlining the issue of consensus in the research literature. We then contextualize an exploratory mixed-methods approach to improving consensus in both qualitative and quantitative settings, which subsequently inform the respective studies conducted. In the first study, a qualitative bibliometric analysis and comparative experience sampling method (ESM) study were conducted to identify the content of listeners’ functional interactions with music in everyday life. This was applied to generate an overarching qualitative taxonomic framework of the functions of music listening, from the utilitarian perspective. In the second study, a quantitative exploration was undertaken to assess the underlying nature of the identified functions across a large item pool, which was subsequently reduced to uncover a latent variable model that may be applied in future work, subject to cross-validation.

Functions of Music Listening

Merriam (1964) initially proposed that music possesses, or exhibits, a degree of function or utility in everyday life by specifying that “function” refers to the “reasons for its employment” (i.e., the purpose music serves for the listener; p. 210). Alongside function, Merriam proposed “use” to refer to the situation in which music was utilized. Merriam also identified an initial set of functions, including “emotional expressions”, “aesthetic enjoyment”, “entertainment”, “communication”, “symbolic representation”, and “physical response” (pp. 222–223), and broader societal functions such as “validation of social institutions” and “contribution to the continuity and stability of culture” (pp. 224–225). The framing of these functions (particularly the latter functions) is Eurocentric and so requires caveats, but Merriam's overall contribution, particularly as pertains to the notion of definition, has been vital in sparking continued interest in the functions of music; an interest that has only gathered strength post-2000 with the increased flexibility and ubiquity (Kassabian, 2013, p. 9) of the “celestial jukebox” (Morris and Powers, 2015, p. 106) i.e., large-scale ubiquitous and (seemingly) exhaustive music streaming options.

Since Merriam, numerous publications and studies have attempted to parse both the meaning of “function” and present potential models of the functions of music. The functions of music have been approached from varying disciplines such as social functions (e.g., Boer, 2009; Boer and Fischer, 2012; Clayton, 2008; Hargreaves and North, 1999; North et al., 2004), physiological functions (e.g., Karageorghis and Terry, 2009; Laukka and Quick, 2013; Priest et al., 2004), psychological functions (e.g., Alea and Bluck, 2003; Clarke et al., 2010; Laiho, 2004; Schäfer et al., 2013), and emotional functions (e.g., Groarke and Hogan, 2016; Juslin and Sloboda, 2001; Lonsdale and North, 2011; Saarikallio, 2019; van Goethem and Sloboda, 2011). Further to this, several researchers have considered the difficulties that pervade the notion of “function” as a concept and offered alternative conceptualizations of the processes involved (e.g., Groarke and Hogan, 2018) with a focus placed on goal attainment. Some researchers have considered the role of listening devices in relation to function (e.g., Brown and Krause, 2020; Bull, 2000, 2005; Krause et al., 2015; Williams, 2004). Finally, exploratory work has been performed by Krause and North that considers the role of external variables and their relationship to functional listening, such as weather (e.g., Krause and North, 2017b) and location (e.g., Krause et al., 2016; Krause and North, 2017a).

Consensus in Functions Research

Given this diversity of perspectives and approaches, functions research suffers from a lack of consensus. Across the available publications, taxonomies, and discussions, three key issues are prevalent that give rise to disparity, and are responsible for confused and often contradictory conceptualizations of functions. The first issue is that of definition. Merriam's founding proposition as to how to define function is not applied consistently, with competing theoretical groundings being explored under the umbrella of function. For instance, Schäfer's (2016) functionality offers a compelling and well-founded variation on Merriam (particularly given the use of more contemporary terminology), but these debates are not considered in most literature, with several studies presenting no definition or aligning themselves with a particular conceptual perspective. Terminology further compounds this issue, with similar or related concepts referred to by a range of terms or neologisms (e.g., functions, uses, functionality, reasons, strategies, affordances, benefits etc.). Such terminological deviation does not appear to be related to disciplinary perspective. Secondly, a broad lack of systematization in our investigatory and epistemological paradigms within the field reduces possibilities for cross-study comparison and analysis. Several milestone publications within the functions literature present theoretical discussions and taxonomies based on non-empirical data (Gregory, 1997; Hargreaves and North, 1999; Merriam, 1964). With regard to empirical approaches, several designs have been utilized, with experience sampling methodology (ESM) being increasingly employed (Greb et al., 2019; Greasley and Lamont, 2011; Juslin et al., 2008; Krause et al., 2014), alongside questionnaires (Krause and North, 2014; Lonsdale and North, 2011) and mass observation data (Sloboda, 2005). However, as there is no systematized approach within the field, comparative analysis pertaining to the possible efficacy of approaches is moot. Such studies may apply both qualitative and quantitative methodologies, further partitioning research outputs in the extent to which findings hold unifying external validity. As a consequence, the presentation of findings is also not standardized. Many studies present itemized or factorized taxonomies of participant responses, whereas others purport to present conceptualizations of models or broader structural approaches to functions. Finally, the issue of cross-disciplinarity needs to be addressed. Whilst much functions research comes from music psychology and the study of music and emotions, other disciplines have also explored the phenomenon from vastly different disciplinary perspectives. Whilst multi-disciplinarity is likely a boon to this area of study, studying functions from an inter- or trans-disciplinary stance consistently is likely to reduce problems of cross-compatibility or discipline-specific concepts. When combined, these three issues present a field of research that is disparate, siloed between disciplines and lacking a unified theory or framework from which to work. As such, a consensus as to what functionality is and what the potential functions of music listening are, has not been achieved.

Rationale of a Sequential Approach

To consolidate these conceptual and practical issues, we consider a two-step approach in which, firstly, a qualitative framework from which to view the functions of music listening is identified. Secondly, a quantitative approach explores and evaluates the contents of that framework. As such, we aim to provide a practical framework to assess functionality in both qualitative and quantitative domains. To this end, we apply a mixed-methods approach.

As has been mentioned, previous research has employed a variety of methods to view and model the functions of music listening. Various taxonomies have been generated, with varying degrees of granularity concerning the functions that are identified. Previous models have largely focused on functions of music listening without considering particularly strong aspects of situational variables, rather considering adaptive approaches to goal attainment (e.g., Groarke and Hogan, 2018). What we hope is that by utilizing an exhaustive qualitative model of the functions of music listening from an emphasized utilitarian perspective (i.e., an emphasis on “use” as well as “function”), we may subsequently reduce this into a quantitative subset drawn from a qualitative assessment of situational listening episodes. Subsequent assessment may then be applied to quantify and reduce these in the framing of general listening habits and explore listeners’ functions of music listening from a utilitarian perspective. By viewing functions in this way, we believe that cross-sectional applications of music listening may be well suited to informing a latent model of the functions of music listening overall, which may in turn be reapplied in cross-sectional studies, hence reflecting “use” as well as “function”.

Study 1

The first study conducted addresses the issues concerning consensus, cross- or interdisciplinary research, and the variation of research methods across fields. Study 1 comprised two parts. First, a large-scale bibliometric analysis was performed to identify an exhaustive list of functions available in the extant literature (as of 2018). This data was thematically sorted and coded to construct a framework of functions of music that was as exhaustive as practicable from the data. The second part of the study explored the veracity of this bibliometrically sourced framework. Here, ESM data was collected and coded into a comparable framework. Finally, both frameworks were compared and combined to construct a framework of functions that represents an exhaustive framework of functions from the extant literature and ecologically valid “real-world” findings. This study was conducted in early 2019, and the following analysis employs literature published at that time.

Bibliometric Analysis

Beginning with publications identified in Schäfer et al.'s 2013 analysis of existing literature, a further search of academic databases was conducted using combinations of relevant terms (such as function, music, use, regulation, strategy, listening etc.). The search was conducted in 2019 and identified 52 relevant publications published between 1964 and 2018. The publications came from a range of disciplines such as sociology, sports science, music psychology, and various sub-disciplines such as music and emotion, and music in everyday life. Across the 52 publications that form the aggregate dataset, 72 distinct models or frameworks of the functions of music were identified, containing a total of 834 items. A full list of the sources used in this analysis is available in the Supplementary Material. Notably, here we attempt to address the first of the three issues outlined: uniformity of definition. All items within the bibliometric analysis were explored with respect to Merriam's initial founding definition of use (the situation or context of the music) and function (the reason for its employment). These items were passed through a filtering process whereby items that did not qualify as functional by Merriam's definition were excluded (e.g., such as descriptions “it's beautiful”). Thirty-two items were removed. As such, until a more appropriate or widely accepted definition is presented by the research community, we propose positioning Merriam's founding definition as the best available authoritative standard.

The remaining 802 items were cycled through a semantic and then iterative latent coding process to combine items into conceptually congruent groupings (see Braun and Clarke, 2019, Clarke and Braun, 2017 for further detail on thematic coding methodologies). Through this process, 45 distinct congruent function groups were identified. Two items, “physical discomfort” and “surveillance”, were only referenced once respectively in the aggregate dataset of publications. The “interaction and bonding” item was referenced most frequently, with 55 references within the aggregate dataset. The mean references per item were 18 (M = 17.83). A previous iteration of this analysis was published in Maloney (2017).

ESM Study

Several influential studies (e.g., Bailes, 2015; Greasley and Lamont, 2011; Juslin et al., 2008; Krause et al., 2016; Randall and Rickard, 2013) have employed ESM as a means to garner valuable real-world data regarding participants’ musical behaviors and exposure in a relatively non-intrusive manner. The data gathered in these studies may also be tailored to their requirements. Therefore, an ESM study provided the ideal methodology to address goal-oriented listening and the functions of music grounded in the experience of listeners in everyday life.

To address the issues surrounding consensus and parity within extant functions research, an ESM-based approach allows for rich, qualitative data to be gathered. Requesting participants to complete ESM questionnaires concerning their activity, thoughts, reflections, perceptions, feelings, and responses to everyday stimuli removes the issues so prevalent in the “pharmaceutical model” (Sloboda, 2005). Primarily, it reinforces ecological validity by allowing a naturalistic engagement with stimuli, rather than participants encountering stimuli within a laboratory or overly restrictive or non-naturalistic conditions (Hektner et al., 2007). ESM does bring with it some pitfalls, particularly concerning the subjective rating of experience. However, the accusations of distortions in ESM studies are generally restricted to smaller momentary variations in the participant (ibid.) and may occur in any context.

Methods

Participants

Participant recruitment for the first study was carried out in person and online. Participants were identified as “highly-engaged” listeners (see Greasley, 2008) through a mixture of in-person approaches (city center, UK), and self-selecting through online advertisements on music-focused podcasts. Participants were not remunerated for participation in this study. In total, 71 participants began the study (34 females, 33 males, 4 non-disclosed or non-binary; aged 18–54, M = 27.9 years; SD = 7.57). Forty-nine participants completed the final evaluation form. Attrition rates for the study were relatively low (12.7%); nine participants were removed for non-participation.

Materials and Procedure

Participants were sent SMS messages containing their credentials and instructions for participation before the study commenced. Further SMS messages were sent three times a day at pseudo-random intervals (morning, midday, and evening). The procedure lasted for seven days, with 20 SMS messages requesting participants fill in a short online form. The form contained open-text field responses to questions concerning listening material, context, activity, location, and driver for engagement (i.e., function), with 7-point Likert scale queries as to the efficacy, awareness, external influence, and preference. A final SMS message (day 7, evening) contained a link to a final evaluation form. Upon completion of the study, all data were anonymized, and participant contact details and entries were removed from all datasets.

The data obtained was analyzed thematically through a series of iterative coding phases. Initial inductive iterations were semantic, before moving through cycles of latent, more qualitatively nuanced iterations. This approach to thematic coding followed that performed in the bibliometric analysis.

Results of Study 1

Participants submitted 573 experience sampling forms for listening events. The qualitative data regarding functions was passed through a filtering process before a semantic and then an iterative latent coding process mirroring the previous bibliometric analysis. The coding identified 1,042 items referencing functions which were themed into 44 congruent groupings. Two functions were referenced only once within the ESM study data, “Aestheticisation” and “Personal Development”. The “Relaxation and Stress Relief” function was referenced most frequently, with 108 distinct references in the ESM dataset. The mean references per item were 23 (M = 23.45).

Consensus Functions Framework

The results of the bibliometric analysis and the ESM study were then combined into a final taxonomic framework of function. Both the bibliometric and ESM studies used a similar approach to construct the taxonomy. This methodology (primarily inductive semantic coding followed by iterations of latent coding) was replicated for several reasons. Firstly, we considered it to be the most appropriate way to deal with both datasets. Secondly, it would allow for stronger links to be drawn when comparing or combining the two datasets. Thirdly, it allowed for language across both phases to be maintained where possible. It was then possible to analyze and combine findings from the two studies and construct a refined final taxonomy of the functions of music.

Thirty-six functions were identified in both frameworks and included in the final combined framework. Nine functions were identified in the bibliometric analysis, but not in the ESM study data. The ESM study identified included three functions that had no evidence in the bibliography; however, it is believed that these could represent important novel functions of music listing that have been not previously reported. These are Earworm Fulfilment, whereby listeners engage with music to somehow “clear” an earworm, Habitual Listening, where listeners engage with music to adhere to an established routine, and the more unexpected Mimesis and Matching, which sees listeners selecting and engaging with music to somehow reflect their surroundings or environment. Participants provided a wealth of descriptions concerning how these functions operate, and they fulfil Merriam's criteria to be considered a function. Therefore, these have been included. One function (Health) was only identified in the bibliometric analysis but had little supporting description and evidence. However, more nuanced variations of this function were found in the ESM study and ascribed to more appropriate/accurate groupings (two references were attributed Therapy, three references were attributed Promote Autonomy & Agency, and one reference was attributed to Solace). Whilst whole fields of study (e.g., music therapy) are dedicated to the health benefits of music, the broader Health function identified in the bibliometric analysis was deemed too vague and is not included. As such, only the nuanced variations have been included here. Furthermore, functions were often found to be highly related variations or “subfunctions” of larger functional groupings. These have been combined to show conceptually congruent functions. The resulting sorting process identified 53 unique functions of music listening, summarized in Table 1.

Table 1.

Frequencies of functions in bibliometric and ESM evaluations.

Function	Bibliometric N	ESM N
1. Accompaniment	35	97
2. Activation, Arousal & Response	30	41
3. Aestheticisation & Filmic Listening	3	1
4. Approval & Cultural Capital	15	3
5. Background		32
6. – Create & Maintain Atmosphere	27	15
7. Boundary Demarcation	5	2
8. Communication	13	8
9. Company/Music As Proxy	29	72
10. – Silence Avoidance		14
11. Control & Conformity	14	2
12. Creative/Maintain Identity	12
13. Creativity	5	7
14. Dancing	12	3
15. Distraction	31	63
16. Earworm Fulfilment		5
17. Enhancing Activity or Ability	17	12
18. Entertainment & Hedonic Motive	28	99
19. Environmental Control	6	22
20. Escapism & Venting	19	3
21. Express Identity/Values	14
22. Focus & Concentration	26	83
23. Group Identity	17
24. Habitual Listening		12
25. Interaction & Bonding	55	11
26. Maintain/Express Cultural Values	4
27. Matching & Mimesis		20
28. Memories	30	13
29. Mental State		35
30. Motivation		37
31. Musicking	15	25
32. – Aesthetic Appreciation	26	4
33. – Listening Behaviors		31
34. – Musical & Lyric Analysis	16	8
35. Pacing & Movement	14	28
36. Personal Development & Understanding	21	1
37. Physical Discomfort	1	2
38. Promote Autonomy & Agency	12	2
39. Relaxation & Stress Relief	72	108
40. Reflection	10	6
41. Situational Relevance	9
42. Solace		2
Specific Emotional Strategies
43. – Accentuate Emotion/Mood	14	2
44. – Change Emotion/Mood	16	40
45. – Express Emotion/Mood	12	2
46. – Regulate Emotion/Mood	41	13
47. – Trigger Emotion/Mood	15	7
48. Structure Time	13	37
49. Surveillance	1
Symbolic Representation & Difference
50. – (Group Difference)	15
51. – (Individual Representation)	6
52. Therapy	6	13
53. Visual Imagery	7	2

To the best of our knowledge, this final Consensus Functions Framework (CFF) represents the most exhaustive list of functions in a consolidated framework to date. It attempts to express a series of bibliographically informed qualitative definitions of the functions of music listening. Whilst the list is extensive and informed by an ESM study, it has not been subjected to rigorous post-hoc testing or validation, for which one approach is applied in the following sections.

Study 2

In our earlier discussion regarding consensus in functions research (p. 2), we highlighted a broad lack of systematization concerning differences in theoretical approaches and methodological paradigms. Study 1 has sought to consolidate the lack of theoretical systematization by applying a utilitarian lens (i.e., Merriam, 1964), through which a theoretical baseline may be used for approaching the issue of functionality. What remains outstanding, however, is a subsequent approach that may operationalize this perspective in such a way that it may be applied in future work. In reflecting on this, we consider that a principled methodological paradigm that may extend the practical use of such a framework is that of psychometric development.

This approach is consistent with sequential designs (e.g., Fetters et al., 2013; Punch, 2014), as we consider the extended use and applicability of the applied framework to be potentially amenable to quantitative as well as qualitative paradigms. For example, the development of psychometric tools often stems from theoretical groundings (DeVellis, 2017), and holds varied uses that enable researchers to measure phenomena through latent constructs and assess the extent to which such models are fit in the population in new samples, cultures, and contexts. Greb et al. (2019) illustrate functions as being non-categorical and not bi-modal (i.e., they are continuous); and that they vary in their intensity. Furthermore, this mode of systemization has been applied elsewhere in literature evidencing latent structures underpinning the uses of music in everyday life (e.g., Groarke and Hogan, 2018; Lonsdale and North, 2011; Schäfer et al., 2013). Therefore, it was deemed reasonable to consider whether an underlying structure to functionality may be uncovered by extending the utilitarian framework identified during Study 1.

The aim of this second study, therefore, was to first generate an exhaustive list of items encapsulating the 53 identified functions of the CFF, and explore the ratings of those items for the presence of a latent structure. Analytically, we determined that a common factor model (e.g., Exploratory/Confirmatory Factor Analysis) would be well suited to identify this dimensionality. The value of this approach is two-fold. First, it maximizes the bandwidth of the theoretical framework from which a latent structure is generated, thus summarizing an extensive model grounded in both theoretical and empirical evidence. Secondly, an underlying structure generated in this way provides future opportunities to assess the utilitarian functions of music listening. As such, re-applications of an identified structure may well serve as a new measure of the functions of music listening (subject to cross-validation) that is derived from the CFF. Though this is one approach to systematization of the utilitarian framework generated, we consider this to be a useful step in bridging the theoretical and practical methodological implications of the framework we provide through Study 1.

Methods

Item Generation

To generate a set of items encapsulating the CFF, we reviewed the definitions of the 53 functions contained in the CFF, considering content and working definitions. In each case, a tentative set of items was generated to represent the content of each function. This was a collaborative process between the authors, initiated by the first author. Here, we considered the characteristic scope of each function and generated items we considered reflective of the characteristic nature of that function. Multiple items were generated for each function, with the items reflecting the aspects of each function as attributed (DeVellis, 2017). In most cases this yielded two items per function; however, when functions warranted additional complexity or nuance, further items were included to fully encapsulate the ascribed definition of each construct. Following this, we consulted to assess whether the item’s content was reflective of each function's scope and definition and, following relevant discussion and amendments, a finalized set of 114 items was generated which the authors agreed reflected the content of the 53 functions. This exhaustive set of items was consciously generated to avoid construct under-representation “which is when a scale does not capture important aspects of a construct because its focus is too narrow” (Boateng et al., 2018; p .6). Items were assessed on a 5-point Likert scale, ranging from 0 (Never) to 4 (Very Often*), which was framed to indicate the frequency with which each item was applied by respondents with regard to music listening. Literature has often viewed the functions of music listening as a continuum rather than as binary or categorical in nature, whereby the intensity of different functions may appear under the same conditions but to varying extents (e.g., Greb et al., 2019). In this sense, we considered rating items through an ordinal (Likert) response measure to be not only theoretically consistent by assessing the intensity or frequency of a given item in terms of prevalence but also methodologically, as this simultaneously provides the conditions necessary for factor analysis based on correlations between the rated items.

Materials and Procedure

To assess the dimensionality of the 114 items, an anonymous online survey was generated and distributed using Qualtrics. In this, participants were asked to read each item in turn and rate them accordingly. When applied to the online survey, items were presented in a series of matrix tables with the prompt: “On a scale from Never (being you do not recall ever using music for that purpose) to Very often (being you use music for that purpose very frequently), to what extent do you use music to…”. All items with relation to respective functions can be seen in the Supplementary Material of this article.

As per best practice, participants were given access to relevant study information, the right to withdraw, and anonymity. Participants were required to confirm being at least 18 years of age and provide informed consent to participate via checkboxes embedded in the survey.

Participants

Participant recruitment for this study was solely carried out online, partially gathered through Prolific* (in which verified participants are invited to participate in research with modest compensation), with additional recruitment taking place via other forms of internet distribution (e.g., social media, emailing lists). Participants taking part via the Prolific platform were compensated £1.35 for taking part, with those from other avenues taking part on a purely voluntary basis without compensation. In total, n = 327 complete responses were recorded, of which n = 208 were recruited via Prolific and n = 119 participants via other forms of internet distribution (51.4% female, 47.4% male, 0.9% non-binary/third gender, 0.3% prefer not to say). Age was recorded via coded bands, the frequencies of which are shown in Table 2.

Table 2.

Age distribution of Study 2 sample.

Age	n	%
18–24	78	23.9
25–34	106	32.4
35–44	61	18.7
45–54	42	12.8
55–64	34	10.4
65–74	3	0.9
75–84	3	0.9
Total	327	100

Results of Study 2

Initial Item Reduction

Factor analysis was deemed the most well-suited procedure to uncover an underlying structure of the items due to its prevalence in psychometric development (e.g., DeVellis, 2017; Field, 2018; Hinkin, 1998; Watkins, 2018; Worthington and Whittaker, 2006). First, we inspected the items to mitigate the presence of highly correlated items, as well as those plausibly deviating from the theoretical scope of the study. The inter-item correlation for items 97 and 99 (see Supplementary Material), measuring Company and Music as Proxy was high (r = .80, p < .001, n = 327). Following Field's (2018) suggestion that highly correlated items may be indicative of substantial cross-over, we reviewed the items in question and assessed there to be a large conceptual overlap between them. As such, item 99 was removed from subsequent analyses as we decided that item 97 would suffice in reflecting the content of both items.

Next, following Watkins (2018), we reviewed items further as variables submitted to factor analysis should adequately represent the domains thought relevant, and that unrelated variables from alternative domains should not be included (e.g., items relating to mathematical skills should not be included if a domain like reading is under consideration). We assessed that two further items should not be included in the factor analysis, namely items 104 and 105. These items relate to Musicking, in the sense of the creation or performance of music itself (such as through playing an instrument) which we considered to theoretically diverge from the intended scope and application of these analyses which are oriented towards functionality in music listening, rather than other forms of musical exposure like performance or playing.

It was assessed that the remaining 111 items (1) did not share excessively high inter-item correlations indicating excessive conceptual overlap, and (2) were relevant to the domain being assessed. Next, we proceeded to dimension reduction of the remaining items.

Exploratory Factor Analyses

EFA was carried out in JASP (version 0.16.1). For the initial iteration, we used the Kaiser–Meyer–Olkin measure of sampling adequacy, which indicated the data were suitable for dimension reduction (KMO = .95). Next, Bartlett's Test of Sphericity was used to ensure the correlation matrix of the data suitably diverged from the identity matrix (Watkins, 2018), the result of which was significant ( $χ^{2}$ (6,105) = 27,416.079, p < .001), further indicating the data were suitable for factor analysis.

This iteration identified the initial underlying structure of the submitted 111 items. Parallel Analysis (Ledesma and Valero-Mora, 2007; Lim and Jahng, 2019) was used in conjunction with an Oblique (oblimin) rotation (to allow for inter-factor correlation) and Maximum Likelihood (ML) estimation, as have been both recommended to generate valid constructs and applied elsewhere in the music psychology literature (e.g., Saarikallio et al., 2015; Tavakol and Wetzel, 2020).

In the initial extraction, a seven-factor solution was implied to fit the observed data best, explaining 51.5% of the variance, however, a substantial number of items (n = 61) failed to load onto any factor, and the removal of such items is recommended to generate stable solutions (Brown, 2006; Worthington and Whittaker, 2006). An iterative process was used to generate a stabilized solution, in which items were retained if they strongly loaded onto one factor at the ≥.50 level (Costello and Osborne, 2005), demonstrated adequate communalities (e.g., ≥.25; Eaton et al., 2019), did not cross-load onto more than one factor, and retained at least two items (Worthington and Whittaker, 2006). The iterative process of removing items failing to meet these criteria was conducted with theoretical considerations to generate a conceptually as well as statistically parsimonious model (e.g., Beavers et al., 2013), resulting in the removal of 70 items, retaining 41 in a stable structure across six latent factors. This six-factor solution ( $χ^{2}$ (589) = 901.358, p < .001, TLI = .948, RMSEA = .040) explained 61.1% of the variance in the observed data. Hinkin (1998) suggests that approximately 60% of the variance explained is appropriate in this regard, however, also notes that there are no strict guidelines as such for EFA (unlike with Principal Component Analysis, for instance). The scree plot was examined to help visualize the number of factors to extract, which in the case of Parallel Analysis can be visually assessed by observing at which point the observed and simulated data intersect. Although Parallel Analysis is generally considered more reliable than standalone inspections of a scree plot (Lim and Jahng, 2019), it can still be useful to visualize the simulated intersection of the data (e.g., DeVellis, 2017). In this case, the six-factor solution was consistent with our interpretation of the scree plot, shown in Figure 1, whilst the results of the stabilized solution can be seen in the Supplementary Material.

Figure 1.

Scree plot of stabilized EFA iteration.

Items loading onto Factor 1 (Identity and Social Bonding; α = .95) contained a broad set of 18 items relating to the use of music to bond, interact, and identify with others. These included items representing: Group Identity, Interaction and Bonding, Communication, Approval and Cultural Capital, Maintain and Express Cultural Values, Symbolic Difference, Express Identity and Values, Control and Conformity, Create and Maintain Identity, and Situational Relevance.

Factor 2 (Emotion Regulation; α = .94) also consisted of a fairly broad selection of 10 items relating to functions within the CFF: Escapism and Venting, Change or Shift Emotions, Trigger or Elicit Emotions, Therapy, Regulate and Maintain Emotions, and Relaxation and Stress Relief . These items were assessed, however, to consistently relate to mood management and/or regulation, as is consistent with other literature (e.g., Groarke and Hogan, 2018; Lonsdale and North, 2011).

Factor 3 (Focus and Concentration; α = .88) contained four items, of which three were intended to measure Focus and Concentration, whilst one item represents Mental State. The items were reflective of the aim of listeners to assist with focus and prevent distraction, not dissimilar to the Cognitive regulation factor presented by Groarke and Hogan (2018).

Factor 4 (Background and Accompaniment; α = .88) contained four items – three of which represented Company and Music as Proxy, and one item relating to Background, indicative of the use of music to mitigate feelings of loneliness and providing background noise.

Factor 5 (Physiological Regulation; α = .86) contained three items, with one relating to Motivation, Activation Arousal and Response, and Pacing and Movement respectively. These items pertain to physical activities, such as stimulation, the maintaining of pace, and motivation, thus aligning with the pursuit of physiological stimulation.

Finally, Factor 6 (Earworm Fulfilment; α = .86) contained two items relating to its namesake, which refers to listeners’ application of music to satisfy the need to listen to “songs stuck in their heads” (e.g., Williamson et al., 2012). These six factors are theoretically consistent functions of music listening and practically distinct when considered with the wider literature. However, upon inspection and interpretation of the 41 items across the six factors, we considered there was room for further reduction and simplification for the sake of brevity and reduced conceptual overlap. The following section expands on the rationale and methods by which we explored this.

Further Reduction of Items

Intending to reapply the identified construct in future work, we thought it would be pragmatic to further simplify the model if possible. The retained 41-item structure represented six dimensions of the functions of music listening, however, it was considered this structure would be more concise by limiting conceptual overlap and unnecessary length. The first two factors (Identity and Social Bonding and Emotion Regulation) retained a larger number of items than the others (18 and 10 respectively; see Supplementary Material). This theoretically runs the risk of exacerbating participant fatigue when responding to the items as part of future work, which may not be necessary given the large number of items for the two latent constructs. As such, we assessed whether a subset of the highest-scoring items from the stabilized EFA would suffice in representing the factors (e.g., Robinson, 2018). Moreover, we considered that given the larger number of items and strengths of factor loadings, the substantive interpretation of the model would not be adversely affected by reducing the number of indicators. Worthington and Whittaker (2006) note that researchers may adapt models on theoretical and practical grounds based on stable EFA solutions, and suggest retaining a subset of the highest-scoring items in factors with a larger than the desired number of items. They advise, however, that adapted structures are fit in additional EFA iteration to ensure the reduction has not impacted the substantive conclusions drawn from the construct, or resulted in remaining items cross-loading or failing to load onto any factor. Furthermore, we considered that, following this reduction, it would be beneficial to consequently inspect model fit and discriminant validity in a constrained setting using confirmatory factor analysis (CFA), for which it is preferential to have at least three items per factor (e.g., Brown, 2006; Schmitt et al., 2018). In the case of the sixth factor, Earworm Fulfilment, this is not met, and as such, we considered it pragmatic to also remove these two items as future cross-validation would require at least three items for each factor.

With these motivations, we conducted a further EFA iteration in which a subset of the six most strongly associated items was taken from the Identity and Social Bonding and Emotion Regulation factors, as these item subsets were assessed to maintain minimal conceptual overlap and suffice in representing the relevant constructs, with the two Earworm Fulfilment items also dropped. An EFA was run using the same method as before (i.e., Parallel Analysis with ML estimation and an oblimin rotation), allowing items to load freely, ensuring the reduction process had not affected the prior structure (Worthington and Whittaker, 2006). The reduced subset of 23 items generated the anticipated five-factor solution ( $χ^{2}$ (148) = 213.614, p < .001, TLI = .975, RMSEA = .037), albeit in a reduced subset, explaining 64.1% of the variance. This is shown in Table 3. The EFA with a reduced number of items held consistency with the previous iteration, with all items loading onto their prior factors at the > .50 level as in the initial stabilized solution and with no cross-loadings. Consequently, we proceeded to CFA as we wished to inspect the reduced model's fit to the data in a restrictive setting.

Table 3.

Reduced EFA solution.

Item	Factor 1	Factor 2	Factor 3	Factor 4	Factor 5	Communality
To identify with others through your shared values and/or culture	0.813	−0.076	0.003	0.008	0.051	.638
To help express your identities and values to others	0.750	0.079	0.078	0.022	−0.095	.632
To act as a topic of discussion with others and ease communication or interaction	0.736	−0.066	0.003	0.088	0.025	.549
To feel that certain artists, pieces, or genres of music are central to your social group's culture and sets you apart from others	0.733	0.075	−0.082	−0.029	−0.033	.540
To match a group's dynamic so you are able to bond with group members when listening with others	0.725	−0.016	0.046	−0.026	0.076	.562
To help you bond with others, and to subsequently feel a sense of belonging with those individuals	0.705	0.098	−0.024	0.035	0.026	.587
To distract yourself from negative or stressful situations	−0.102	0.822	0.017	0.100	0.045	.759
To relieve stress and negative emotions associated with negative events or situations	−0.042	0.743	0.046	0.090	−0.010	.645
To find meaning within music that allows you to reduce negative emotions or moods	0.206	0.705	0.035	−0.046	−0.032	.640
To manage emotions that you may be experiencing despite external influences, whether they are positive or negative	0.072	0.696	0.145	−0.048	−0.023	.610
To feel certain specific emotions, such as joy or sadness	0.073	0.687	−0.017	−0.064	0.088	.515
To help you reverse your emotions or moods	0.005	0.674	−0.025	0.117	0.041	.566
To help you focus or concentrate on tasks	−0.052	0.029	0.861	0.054	−0.014	.788
To help you “flow” when trying to concentrate on something	0.066	0.049	0.817	−0.055	0.031	.726
To stop external factors from distracting you when trying to concentrate on a task	0.010	−0.001	0.739	0.086	0.017	.637
To help you attain the necessary mindset to working on certain tasks	0.054	0.097	0.543	−0.001	0.200	.549
To avoid silence when you’re alone (e.g., playing music when nobody else is home)	0.010	−0.008	0.010	0.880	0.015	.789
To feel a sense of company in the absence of others (e.g., playing the radio when home alone)	0.039	0.036	0.022	0.782	−0.023	.672
To reduce feelings of being lonely when social interaction is not possible	0.047	0.228	−0.050	0.672	0.081	.717
To provide background noise and remove silence	0.020	−0.082	0.319	0.561	0.020	.545
To help physically stimulate you to carry out physical tasks, such as exercise or sports	−0.010	0.027	−0.022	0.008	0.867	.752
To help you maintain pacing during physical activities, such as yoga, walking or whilst in the gym	0.030	−0.036	−0.022	0.033	0.854	.721
To help you achieve goals by motivating you to further action (such as increased effort during exercise)	−0.001	0.056	0.159	−0.046	0.687	.615
Cronbach's α	.89	.90	.88	.88	.86
McDonald's $ω$	.89	.90	.89	.89	.87

Note. By default, JASP provides Uniqueness (Uniqueness = 1 − Communality). Due to the more common use of Communality in the literature, however, we choose to provide this measure instead (i.e., Communality = 1 − Uniqueness).

Confirmatory Factor Analysis

CFA can be applied to test a hypothesized or a priori structure, or once a model has been generated through EFA, to assess a model's ability to fit the data when restricted and also allow for comparability during future cross-validation (Matsunaga, 2010; Schmitt et al., 2018; Worthington and Whittaker, 2006). Our rationale in this instance is in these former uses of CFA, with the latter (i.e., cross-validation) only possible with a new sample. This is a caveat we explicitly acknowledge to make clear our intention with the application of CFA in this setting. Specifically, in the present study our interest lay in the model's fit to the data, items’ standardized loadings, average variance explained (AVE), and the discriminant validity of the overall construct of the retained set of 23 items. In this sense, we impose a restrictive model to assess these contributory determinants of model validity, which allows for future comparison with new data, as well as cross-validation of the structure's psychometric properties (Schmitt et al., 2018).

To this end, we fit a CFA based on the 23-item structure in R (version 4.1.0; R Core Team, 2021) using the lavaan package (Rosseel, 2012) to estimate the model, and the semTools package (Jorgensen et al., 2022) to calculate factors’ AVE. We used robust maximum-likelihood (MLR) and fixed factor variances to 1 as a means of scaling the latent variables (Brown, 2006; Kline, 2016). CFA indicated good overall model fit, according to both incremental and absolute measures of fit index ( $χ^{2}$ (220) = 367.506, p < .001, CFI = .963, TLI = .957, RMSEA = .049*). Following Hair et al. (2014), model convergent validity was deemed suitable as each item’s standardized parameter estimates (i.e., factor loadings; β) and factors’ AVE were > .50. In tandem with alpha and omega construct reliability scores (see Table 3), these provide evidence of the model's overall construct validity. Table 4 reports the unstandardized and standardized parameter estimates for each item concerning its latent factor, as well as the standard error and significance test (all of which were p < .001), and also factors’ AVE. Figure 2 illustrates the CFA structure and the covariances between factors.

Figure 2.

CFA of reduced item subset. Note. ISB = Identity and Social Bonding, ER = Emotion Regulation, FaC = Focus and Concentration, BaA = Background and Accompaniment, PA = Physiological Arousal. Bi-directional arrows indicate factor covariances and single-headed arrows indicate standardized parameter estimates. Refer to Item References in Table 4 to see items. This model was generated in R using the semPlot package (Epskamp, 2022).

Table 4.

CFA item loadings of reduced construct.

Factor	Item Reference	$λ$	$β$	SE	p	AVE
Identity and Social Bonding						.574
To identify with others through your shared values and/or culture	S1	0.898	.787	.054	<.001
To help express your identities and values to others	S2	0.908	.781	.051	<.001
To feel that certain artists, pieces, or genres of music are central to your social group's culture and sets you apart from others	S3	0.870	.716	.060	<.001
To help you bond with others, and to subsequently feel a sense of belonging with those individuals	S4	0.824	.771	.051	<.001
To act as a topic of discussion with others and ease communication or interaction	S5	0.830	.740	.053	<.001
To match a group's dynamic so you are able to bond with group members when listening with others	S6	0.842	.751	.051	<.001
Emotion Regulation						.609
To distract yourself from negative or stressful situations	E1	1.009	.859	.047	<.001
To relieve stress and negative emotions associated with negative events or situations	E2	0.922	.801	.053	<.001
To manage emotions that you may be experiencing despite external influences, whether they are positive or negative	E3	0.972	.775	.051	<.001
To find meaning within music that allows you to reduce negative emotions or moods	E4	0.964	.768	.053	<.001
To feel certain specific emotions, such as joy or sadness	E5	0.756	.702	.055	<.001
To help you reverse your emotions or moods	E6	0.878	.758	.052	<.001
Focus and Concentration						.666
To help you focus or concentrate on tasks	F1	1.019	.876	.048	<.001
To help you “flow” when trying to concentrate on something	F2	1.039	.843	.046	<.001
To stop external factors from distracting you when trying to concentrate on a task	F3	0.933	.797	.049	<.001
To help you attain the necessary mindset to working on certain tasks	F4	0.786	.726	.055	<.001
Background and Accompaniment						.663
To avoid silence when you’re alone (e.g., playing music when nobody else is home)	B1	1.027	.861	.051	<.001
To feel a sense of company in the absence of others (e.g., playing the radio when home alone)	B2	1.047	.818	.055	<.001
To reduce feelings of being lonely when social interaction is not possible	B3	1.100	.850	.053	<.001
To provide background noise and remove silence	B4	0.766	.696	.064	<.001
Physiological Arousal						.688
To help you maintain pacing during physical activities, such as yoga, walking or whilst in the gym	P1	1.090	.840	.056	<.001
To help physically stimulate you to carry out physical tasks, such as exercise or sports	P2	0.960	.859	.051	<.001
To help you achieve goals by motivating you to further action (such as increased effort during exercise)	P3	0.886	.786	.056	<.001

Note. Item references refer to labels in Figure 2. $λ$ = unstandardized path coefficient. $β$ = standardized path coefficient. SE = Standard error. AVE = Average variance explained.

Moderate factor covariances (Φ) were observed, with coefficients between factors ranging from .322 to .697. To ensure discriminant validity holds between latent variables (i.e., that they are distinct), we compared the AVE of each factor with the square of covariances between them (Φ²; Fornell and Larcker, 1981; Hair et al., 2014), shown in Table 5. Each squared coefficient between latent factors was lower than the AVE of the factors themselves, providing evidence of discriminant validity across the construct. In brief, this indicates that each factor explains more of the variance in its item measures than it shares with other factors (Hair et al., 2014). This provides evidence of discriminant validity between factors as well as construct validity within each factor.

Table 5.

Bivariate Φ² in comparison to AVE of latent factors.

Factors	$Φ$	$Φ^{2}$
Identity and Social Bonding (.574) & Emotion Regulation (.609)	.501***	.251
Identity and Social Bonding (.574) & Focus and Concentration (.666)	.322***	.104
Identity and Social Bonding (.574) & Background and Accompaniment (.663)	.359***	.129
Identity and Social Bonding (.574) & Physiological Arousal (.688)	.322***	.104
Emotion Regulation (.609) & Focus and Concentration (.666)	.642***	.412
Emotion Regulation (.609) & Background and Accompaniment (.663)	.697***	.486
Emotion Regulation (.609) & Physiological Arousal (.688)	.456***	.208
Focus and Concentration (.666) & Background and Accompaniment (.663)	.622***	.387
Focus and Concentration (.666) & Physiological Arousal (.688)	.554***	.307
Background and Accompaniment (.663) & Physiological Arousal (.688)	.409***	.167

Note. AVE of each factor is in brackets. (***p < .001).

Discussion

Study 2 sought to extend the applicability of the qualitative framework provided by Study 1. We have provided a (tentative) measure of the functions of music listening from a utilitarian perspective, by first generating an exhaustive list of 114 items encapsulating the content of the CFF, which were each rated by 327 participants. Through EFA, we reduced the dimensionality of the items to uncover a latent construct we hypothesized would represent core underlying functional domains of the functions of music listening. Once refined, we uncovered a six-factor structure containing 41 items we deemed to be theoretically consistent with commonly applied functions of music listening regarding prior research. We further reduced this, however, to mitigate conceptual overlap in the first two factors and model parsimony by dropping the sixth factor altogether, which only contained two items. This was conducted with the motivation to present a more concise model whose construct validity could be assessed under restrictive conditions using CFA. These results indicated the subsequent 23-item structure across five factors was a good fit for the data, with good measures of construct reliability (assessed using Cronbach's α and McDonald's $ω$ ), convergence (standardized factor loadings and factors’ AVE), and discriminant validity (factors’ AVEs in comparison to Φ²). Overall, this leads us to argue that the latent structure is both theoretically and statistically parsimonious. We do note, however, that our future work should seek to cross-validate this structure in a new sample. Moreover, we consider that it would be particularly interesting to inspect the model's applicability in ecologically valid settings, in which measurements are taken from real-time music use, as opposed to a retrospective or more general framing, improving ecological validity.

The strengths of the presented model are rooted in the sequential approach taken between the respective studies, and a focus on utility in functions of music listening. Regarding the former, Study 1 conducted a comparative bibliometric analysis in relation to ecologically valid data collated via ESM. This provided a rich qualitative corpus of functions of music listening, from which we were able to build an item pool we hypothesized may comprise an underlying structure, and through which there would be implications for quantitative work. Furthermore, this was motivated by a focus on utility in functionality. It has been noted that functionality in music listening is broader in scope than regulatory theories, and as such novel approaches are required to fully capture utilitarian listening potential (Maloney, 2019). It is for this reason that the perspective of music listening as a goal-oriented activity, for which Study 1 applied the described approach, is here argued to encapsulate the breadth of potential functionality.

Consequently, in leveraging this framework as a grounding from which to derive a psychometric construct, Study 2 provides a measure of functionality that extends the practicabilities of this contribution. The process of item generation and reduction outlined in Study 2, therefore, rates utilitarian function, providing an item subset that is reflective of the broader domain of utility, rather than narrow models focusing on regulatory strategies. For instance, models such as those of Saarikallio (2008) and Groarke and Hogan (2018) measure adaptive regulatory strategies of music listening, for which they are well suited. However, this alternative standpoint expands the domain of functionality and instead focuses on the utility of music listening in more generalized terms. Moreover, in being derived from the CFF, the construct actively integrates knowledge and observations about functionality in everyday life, since the framework is derived from data including ecologically valid observations. This utilitarian domain is also, therefore, informed by contexts of music listening, which other generalized measures in the past have not explicitly considered during construct development (e.g., Chin et al., 2018; Lonsdale and North, 2011), and is considered a greater driver in predicting functionality and music selection, rather than individual variables such as preferences or personality types (e.g., Greb et al., 2019). This is also where differences between the present and other structures may be noted. For example, the relevant module of the more generalist MUSE-BAQ (Chin et al., 2018) holds musical “transcendence” as the most important factor in terms of music use, which refers to eudaimonic well-being as a primary motivation for music listening. Though such factors may be of considerable interest depending on the research question, they are by nature less aligned with the short-term characterization of music “use” as laid out by Merriam. The present factor structure differs from such models in this way since the items are oriented to refer solely to short-term use due to the utilitarian perspective. Though we acknowledge this necessarily reduces the absolute scope of functionality, the measure extends the applicability of this perspective by providing a measurement tool that future work may seek to reapply. There are, however, limitations to the applied approach and model, which are now discussed.

Limitations

Regarding Study 2, limitations include a lack of reversed items. The inclusion of reversed items can offset such biases and provide a reference point to check participants’ attention, however, it has also been argued that reversed items can often be difficult to interpret and unnecessary, with theoretical consistency to the domain under assessment taking precedence (van Sonderen et al., 2013). In this case, we considered that reversed items would not make theoretical sense to the underlying theory, and that reversed items may be more confusing than practical given the large number of items. A possible side-effect of this, however, is that some participants may have overstated the frequency of the proposed functions with overly agreeable responses (e.g., Gove and Geerken, 1977; Porta, 2014). It may be the case that the effects of this did not have a drastic impact on results, however, this is difficult to assess.

Relatedly, however, the size of the item pool that was used may have affected responses due to fatigue. The 114 items were presented over 11 matrix tables in an online questionnaire that, whilst intended to be as manageable as possible, nevertheless required extended and close attention. Items were presented in a randomized order for each participant, however, which was intended to mitigate fatigue by distributing items from various domains to reduce repetitiveness. Nevertheless, the length may have additionally put further participants off from engaging or otherwise completing the study.

Finally, we note that the sample collected is likely non-representative of the total population when it comes to music listening. Data were collected in the UK, and whilst no specific demographic data was collected regarding nationality, it is likely the sample was generally representing Western listeners, not including minority or understudied cultures in particular. It is important, therefore, to consider that the underlying structure may not be broadly representative of listeners from different cultures, who may have nuanced engagement practices with music, further influencing functionality. Future work may therefore also assess whether it is possible to cross-validate the proposed structure cross-culturally.

This set of limitations should be considered when discussing the presented model. However, as has been discussed, the five-factor solution is largely consistent with literature relating to the functions of music listening. When coupled with the meeting of statistical criteria, there are grounds to note that, on balance, the present model sufficiently reduces the scope of the initial CFF item pool into a set of theoretically consistent factors.

Conclusion

This paper has sought to consolidate a utilitarian theoretical framework of the functions of music listening, derived from bibliometric and empirical phases. The paper identifies three core issues with functions research and proposes possible courses of action to address said issues whilst providing a possible method for conducting such functions research. Firstly, we propose that all functions research should explicitly adhere to Merriam's founding definition of function, or explore a systematic approach to the development of a new definition of function.

Secondly, we have sought to rationalize and present a principled approach to the systemization of functions of music listening through psychometric development, in this case extending the utility of the theoretical framework presented during Study 1. We followed a process of item generation and dimension reduction via factor analysis to identify a latent structure of items generated through the framework. This aligns with the view that functions of music listening are not discrete in nature, but can rather be gauged by intensity or magnitude during music listening, and that this can be assessed at the latent level (e.g., Greb et al., 2019). The subsequent model is necessarily reduced in scope when compared to the framework generated during Study 1, however, provides a practical means by which functions of music listening may be measured in future research (though this requires cross-validation, preferably in ecologically valid settings). Study 2 therefore illustrates one mode of systemization we consider to be both pragmatic and applicable to future research, providing a measure of functionality derived from the utilitarian perspective emphasized. We note, however, that this is nevertheless just one approach to systemization, and that other methods and perspectives could still plausibly be explored by others. Such research should, however, consider and discuss its approach to systemization also, such that discussions and considerations regarding this issue may be possible to help find more consensus in the future.

Finally, whilst we identified issues around cross-disciplinary study as a potentially confounding factor in the applicability of research, it is arguable that without the variation in disciplinary groundings presented in the initial bibliometric analysis, some functions would not have been identified. Yet without grounding this research in systematic musicology and providing a factor structure of utilitarian functionality, such disciplinarily siloed findings would remain disparate and unrelated. Thus we propose that, whilst cross- or interdisciplinary contexts provide areas and contexts (i.e., uses of music) that are fertile locations for functions research (e.g., sports science and ergogenic functions of music), such disciplines should participate in functions research as a matter of necessity, and the disciplinary grounding for the investigatory paradigm should remain firmly rooted in music psychology methods.

This process has been supplemented here by item generation (encapsulating the content of this framework) and dimension reduction to provide a latent construct of the functions of music listening from the utilitarian perspective. These latent structures are not reflective of all possible functionalities, but rather indicative of a core set of functions that explain large amounts of variance in the observed data whilst meeting reliability and validity criteria.

Subject to cross-validation, such a construct is amenable to quantitative research in particular, as it allows for reflective measurement of functionality, viewing it as a continuum rather than as discrete or categorical in nature; not unlike other psychometric structures of the functions of music listening (e.g., Greb et al., 2019; Groarke and Hogan, 2018). However, we argue that the emphasis of the derived structure on utility is particularly relevant in situations where it is the use of music that is of primary interest. Moreover, the presence of latent variables enables more complex modeling approaches through Structural Equation Modeling (SEM) for example, in which measurement error is more effectively accommodated than in more traditional approaches (e.g., multiple regression). Therefore, hypotheses relating to the functions of music listening as part of a system of variables (e.g., Greb et al., 2019) may be subject to hypothesis testing, for which this measure provides a utilitarian perspective relating to episodic music listening.

Future work could cross-validate this presented structure in ecologically valid settings and gauge how contextual variables influence these latent structures in the manner outlined above (e.g., Greb et al., 2019). This may provide useful insights into how utility in music listening is influenced by the contextual factors of the listening experience.

Copyright

Note that the data for Study 2 in this article relates to the data contained in the “Study 1” folder in the latter of these repositories.

*0 = Never, 1 = Rarely, 2 = Sometimes, 3 = Often, 4 = Very Often

*https://www.prolific.co/

*Scaled $χ^{2}$ and robust fit statistics reported.

Supplemental Material

sj-pdf-1-mns-10.1177_20592043241266972 - Supplemental material for A Mixed-Method Exploratory Approach to Identifying the Utilitarian Functions of Music Listening

Supplemental material, sj-pdf-1-mns-10.1177_20592043241266972 for A Mixed-Method Exploratory Approach to Identifying the Utilitarian Functions of Music Listening by Noah Henry, Liam Maloney and Hauke Egermann in Music & Science

Footnotes

Action Editor

Elaine King, University of Hull, School of Arts

Peer Review

Amanda Krause, James Cook University, Psychology, Thomas Schäfer, MSB Medical School Berlin GmbH, Department of Psychology

Contributorship

LM designed Study 1 and carried out the subsequent analysis. NH designed Study 2 and carried out subsequent statistical analysis, with the guidance and input of LM and HE. LM and HE additionally provided feedback and guidance on item generation for Study 2. All authors contributed to the writing, editing and reviewing of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Ethical Approval

Ethical approval for both studies was granted by the Arts and Humanities Ethics Committee (AHEC) at the University of York. All participants provided informed consent to take part in this research and were made aware that they were able to withdraw at any time.

Funding

Study 1 was supported by the University of York. Funding for the recruitment of participants in Study 2 via the Prolific surveying platform was also supported by the University of York.

ORCID iDs

Noah Henry

Liam Maloney

Hauke Egermann

Data Availability Statement

The datasets analyzed during this research are available on the Open Science Framework (OSF; see Maloney, 2024 and Henry, 2024) and can be found in the following repositories: Study 1) https://osf.io/fhtqs/; Study 2) . Note that the data for Study 2 is located in the “Study 1 data and syntax” folder of the respective repository.

Supplemental Material

Supplemental material for this article is available online.

References

Alea

Bluck

(2003). Why are you telling me that? A conceptual model of the social function of autobiographical memory. Memory (Hove, England), 11(2), 165–178. https://doi.org/10.1080/741938207

Bailes

(2015). Music in mind? An experience sampling study of what and when, towards an understanding of why. Psychomusicology: Music, Mind, and Brain, 25(1), 58–68. https://doi.org/10.1037/pmu0000078

Beavers

A. S.

Lounsbury

J. W.

Richards

J. K.

Huck

S. W.

Skolits

G. J.

Esquivel

S. L.

(2013). Practical considerations for using exploratory factor analysis in educational research. Practical Assessment, Research, and Evaluation, 18(1), 6. https://doi.org/10.7275/qv2q-rk76

Boateng

G. O.

Neilands

T. B.

Frongillo

E. A.

Melgar-Quiñonez

H. R.

Young

S. L.

(2018). Best practices for developing and validating scales for health, social, and behavioral research: A primer. Frontiers in Public Health, 6, 149. https://doi.org/10.3389/fpubh.2018.00149

Boer

(2009). Music makes the people come together: Social functions of music listening for young people . PhD thesis, Victoria University of Wellington.

Boer

Fischer

(2012). Towards a holistic model of functions of music listening across cultures: A culturally decentred qualitative approach. Psychology of Music, 40(2), 179–200. https://doi.org/10.1177/0305735610381885

Braun

Clarke

(2019). Reflecting on reflexive thematic analysis. Qualitative Research in Sport, Exercise and Health, 11(4), 589–597. https://doi.org/10.1080/2159676X.2019.1628806

Brown

T. A.

(2006). Confirmatory Factor Analysis for Applied Research. The Guildford Press.

Brown

S. C.

Krause

A. E.

(2020). Freedom of choice: Examining music listening as a function of favorite music format. Psychomusicology: Music, Mind, and Brain, 40(2), 88–102. https://doi.org/10.1037/pmu0000254

10.

Bull

(2000). Sounding Out the City: Personal Stereos and the Management of Everyday Life. Berg.

11.

Bull

(2005). No dead air! The iPod and the culture of mobile listening. Leisure Studies, 24(4), 343–355. ISBN: 0261-4367 https://doi.org/10.1080/0261436052000330447

12.

Chin

T.-C.

Coutinho

Scherer

K. R.

Rickard

N. S.

(2018). Musebaq: A modular tool for music research to assess musicianship, musical capacity, music preferences, and motivations for music use. Music Perception: An Interdisciplinary Journal, 35(3), 376–399. https://doi.org/10.1525/mp.2018.35.3.376

13.

Clarke

Braun

(2017). Thematic analysis. Journal of Positive Psychology, 12(3), 297–298. https://doi.org/10.1080/17439760.2016.1262613

14.

Clarke

Dibben

Pitts

(2010). Music and Mind in Everyday Life. Oxford University Press.

15.

Clayton

(2008). The social and personal functions of music in cross-cultural perspective. In Hallam

Cross

Thaut

(Eds.), The Oxford Handbook of Music Psychology (pp. 35–44). Oxford University Press.

16.

Costello

A. B.

Osborne

(2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assessment, Research, and Evaluation, 10(1), 7. https://doi.org/10.7275/jyj1-4868

17.

DeVellis

R. F.

(2017). Scale Development: Theory and Applications (4th ed.). Sage.

18.

Eaton

Frank

Johnson

Willoughby

(2019). Comparing exploratory factor models of the brief electricity and magnetism assessment and the conceptual survey of electricity and magnetism. Physical Review Physics Education Research, 15(2), 020133. https://doi.org/10.1103/PhysRevPhysEducRes.15.020133

19.

Epskamp

(2022). semPlot: Path Diagrams and Visual Analysis of Various SEM Packages’ Output. R package version 1.1.6.

20.

Fetters

M. D.

Curry

L. A.

Creswell

J. W.

(2013). Achieving integration in mixed methods designs—principles and practices. Health Services Research, 48(6pt2), 2134–2156. https://doi.org/10.1111/1475-6773.12117

21.

Field

(2018). Discovering Statistics Using IBM SPSS Statistics (5th ed.). Sage.

22.

Fornell

Larcker

D. F.

(1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18(1), 39–50. https://doi.org/10.1177/002224378101800104

23.

Gove

W. R.

Geerken

M. R.

(1977). Response bias in surveys of mental health: An empirical investigation. American Journal of Sociology, 82(6), 1289–1317. https://doi.org/10.1086/226466

24.

Greasley

A. E.

(2008). Engagement with music in everyday life: An in-depth study of adults’ musical preferences and listening behaviours , PhD thesis, University of Keele.

25.

Greasley

A. E.

Lamont

A. M.

(2011). Exploring engagement with music in everyday life using experience sampling methodology. Musicae Scientiae, 15(1), 45–71. https://doi.org/10.1177/1029864910393417

26.

Greb

(2018). Determinants of music-selection behavior: Development of a model. PhD thesis, Technische Universität Berlin.

27.

Greb

Steffens

Schlotz

(2019). Modeling music-selection behavior in everyday life: A multilevel statistical learning approach and mediation analysis of experience sampling data. Frontiers in Psychology, 10(390). https://doi.org/10.3389/fpsyg.2019.00390

28.

Gregory

A. H.

(1997). The roles of music in society: The enthnomusicological perspective. In Hargreaves

D. J.

North

A. C.

(Eds.), The Social Psychology of Music (pp. 123–140). Oxford University Press.

29.

Groarke

J. M.

Hogan

M. J.

(2016). Enhancing wellbeing: An emerging model of the adaptive functions of music listening. Psychology of Music, 44(4), 769–791. https://doi.org/10.1177/0305735615591844

30.

Groarke

J. M.

Hogan

M. J.

(2018). Development and psychometric evaluation of the adaptive functions of music listening scale. Frontiers in Psychology, 9(516), 1–19. https://doi.org/10.3389/fpsyg.2018.00516

31.

Hair

J. F.

Black

W. C.

Babin

B. J.

Anderson

R. E.

(2014). Multivariate Data Analysis (7th ed.). Pearson.

32.

Hargreaves

D. J.

North

A. C.

(1999). The functions of music in everyday life: Redefining the social in music psychology. Psychology of Music, 27(1), 71–83. https://doi.org/10.1177/0305735699271007

33.

Hektner

J. M.

Schmidt

J. A.

Csikszentmihalyi

(2007). Experience Sampling Method: Measuring the Quality of Everyday Life. Sage Publications, Newbury Park.

34.

Henry

(2024). Modelling music selection in everyday life with applications for psychology-informed music recommender systems: Data and analysis syntax. https://doi.org/10.17605/OSF.IO/UAT64

35.

Hinkin

T. R.

(1998). A brief tutorial on the development of measures for use in survey questionnaires. Organizational Research Methods, 1(1), 104–121. https://doi.org/10.1177/109442819800100106

36.

Jorgensen

T. D.

Pornprasertmanit

Schoemann

A. M.

Rosseel

(2022). semTools: Useful tools for structural equation modeling. R package version 0.5-6.

37.

Juslin

P. N.

Liljeström

Västfjäll

Barradas

Silva

(2008). An experience sampling study of emotional reactions to music: Listener, music, and situation. Emotion, 8(5), 668–683. https://doi.org/10.1037/a0013505

38.

Juslin

P. N.

Sloboda

J. A.

(2001). Music and Emotion: Theory and Research. Series Title: Series in affective science. Oxford University Press.

39.

Karageorghis

C. I.

Terry

P. C.

(2009). The psychological, psychophysical and ergogenic effects of music in sport: A review and synthesis. In Bateman

Bale

(Eds.), Sporting Sounds. Relationships Between Sport and Music (pp. 13–36). Routledge.

40.

Kassabian

(2013). Ubiquitous Listening: Affect, Attention, and Distributed Subjectivity. University of California Press.

41.

Kline

R. B.

(2016). Principles and Practice of Structural Equation Modeling (4th ed.). Guilford Publications.

42.

Krause

A. E.

North

A. C.

(2014). Contextualized music listening: Playlists and the Mehrabian and Russell model. Psychology of Well-Being, 4(1), 1–16. https://doi.org/10.1186/s13612-014-0022-7

43.

Krause

A. E.

North

A. C.

(2017a). How do location and control over the music influence listeners’ responses? Scandinavian Journal of Psychology, 58(2), 114–122. https://doi.org/10.1111/sjop.12352

44.

Krause

A. E.

North

A. C.

(2017b). Tis the season: Music-playlist preferences for the seasons. Psychology of Aesthetics, Creativity, and the Arts, 12(1), 89–95. https://doi.org/10.1037/aca0000104

45.

Krause

A. E.

North

A. C.

Hewitt

L. Y.

(2014). Music selection behaviors in everyday listening. Journal of Broadcasting & Electronic Media, 58(2), 306–323. https://doi.org/10.1080/08838151.2014.906437

46.

Krause

A. E.

North

A. C.

Hewitt

L. Y.

(2015). Music-listening in everyday life: Devices and choice. Psychology of Music, 43(2), 155–170. https://doi.org/10.1177/0305735613496860

47.

Krause

A. E.

North

A. C.

Hewitt

L. Y.

(2016). The role of location in everyday experiences of music. Psychology of Popular Media Culture, 5(3), 232–257. https://doi.org/10.1037/ppm0000059

48.

Laiho

(2004). The psychological functions of music in adolescence. Nordic Journal of Music Therapy, 13(1), 47–63. https://doi.org/10.1080/08098130409478097

49.

Lamont

Greasley

Sloboda

(2016). Choosing to hear music: Motivation, process, and effect. In The Oxford Handbook of Music Psychology. Oxford University Press.

50.

Laukka

Quick

(2013). Emotional and motivational uses of music in sports and exercise: A questionnaire study among athletes. Psychology of Music, 41(2), 198–215. https://doi.org/ISBN:10.1177/0305735611422507

51.

Ledesma

R. D.

Valero-Mora

(2007). Determining the number of factors to retain in EFA: An easy-to-use computer program for carrying out parallel analysis. Practical Assessment, Research, and Evaluation, 12(2), 1–11. https://doi.org/10.7275/wjnc-nm63

52.

Lim

Jahng

(2019). Determining the number of factors using parallel analysis and its recent variants. Psychological Methods, 24(4), 452. https://doi.org/10.1037/met0000230

53.

Lonsdale

A. J.

North

A. C.

(2011). Why do we listen to music? A uses and gratifications analysis. British Journal of Psychology (London, England: 1953), 102(1), 108–134. https://doi.org/10.1348/000712610X506831

54.

Maloney

L. T.

(2017). Music as water: The functions of music from a utilitarian perspective. AVANT. The Journal of the Philosophical-Interdisciplinary Vanguard, VIII, 57–67. https://doi.org/10.26913/80s02017.0111.0006

55.

Maloney

(2019). Music like water: Exploring the functions of music through thematic bibliometric analysis and comparative ESM study . PhD thesis, University of York.

56.

Maloney

. (2024). A mixed-method exploratory approach to identifying the utilitarian functions of music listening: Study 1 data and codebook. https://doi.org/10.17605/OSF.IO/FHTQS

57.

Matsunaga

(2010). How to factor-analyze your data right: Do’s, don’ts, and how-to’s. International Journal of Psychological Research, 3(1), 97–110. https://doi.org/10.21500/20112084.854

58.

Merriam

A. P.

(1964). The Anthropology of Music. Northwestern University Press.

59.

Morris

J. W.

Powers

(2015). Control, curation and musical experience in streaming music services. Creative Industries Journal, 8(2), 106–122. https://doi.org/10.1080/17510694.2015.1090222

60.

North

A. C.

Hargreaves

D. J.

Hargreaves

J. J.

(2004). Uses of music in everyday life. Music Perception, 22(1), 41–77. https://doi.org/10.1525/mp.2004.22.1.41

61.

Porta

(2014). A Dictionary of Epidemiology. Oxford University Press.

62.

Priest

D. L.

Karageorghis

C. I.

Sharp

N. C. C.

(2004). The characteristics and effects of motivational music in exercise settings: The possible influence of gender, age, frequency of attendance, and time of attendance. Journal of Sports Medicine and Physical Fitness, 44(1), 77–86. http://bura.brunel.ac.uk/handle/2438/2946

63.

Punch

(2014). Introduction to Social Research (3rd ed.). Sage.

64.

Randall

W. M.

Rickard

N. S.

(2013). Development and trial of a mobile experience sampling method (m-esm) for personal music listening. Music Perception, 31(2), 157–170. https://doi.org/10.1525/mp.2013.31.2.157

65.

R Core Team (2021). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.

66.

Robinson

M. A.

(2018). Using multi-item psychometric scales for research and practice in human resource management. Human Resource Management, 57(3), 739–750. https://doi.org/10.1002/hrm.21852

67.

Rosseel

(2012). Lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2), 1–36. https://doi.org/10.18637/jss.v048.i02

68.

Saarikallio

S. H.

(2008). Music in mood regulation: Initial scale development. Musicae Scientiae, 12(2), 291–309. https://doi.org/10.1177/102986490801200206

69.

Saarikallio

(2019). Access-Awareness-Agency (AAA) Model of Music-Based Social-Emotional Competence (MuSEC). Music & Science, 2, 1–16. https://doi.org/10.1177/2059204318815421

70.

Saarikallio

Gold

McFerran

(2015). Development and validation of the healthy-unhealthy music scale. Child and Adolescent Mental Health, 20(4), 210–217. https://doi.org/10.1111/camh.12109

71.

Schäfer

(2016). The goals and effects of music listening and their relationship to the strength of music preference. PLoS ONE, 11(3), 1–15. https://doi.org/10.1371/journal.pone.0151634

72.

Schäfer

Sedlmeier

Städtler

Huron

(2013). The psychological functions of music listening. Frontiers in Psychology, 4, 511. https://doi.org/10.3389/fpsyg.2013.00511

73.

Schmitt

T. A.

Sass

D. A.

Chappelle

Thompson

(2018). Selecting the “best” factor structure and moving measurement validation forward: An illustration. Journal of Personality Assessment, 100(4), 345–362. https://doi.org/10.1080/00223891.2018.1449116

74.

Sloboda

J. A.

(2005). Everyday Uses of Music Listening: A Preliminary Study (pp. 317–331). Oxford University Press.

75.

Tavakol

Wetzel

(2020). Factor analysis: A means for theory and instrument development in support of construct validity. International Journal of Medical Education, 11, 245–247. https://doi.org/10.5116/ijme.5f96.0f4a

76.

van Goethem

Sloboda

J. A.

(2011). The functions of music for affect regulation. Musicae Scientiae, 15(2), 208–228. https://doi.org/10.1177/1029864911401174

77.

van Sonderen

Sanderman

Coyne

J. C.

(2013). Ineffectiveness of reverse wording of questionnaire items: Let’s learn from cows in the rain. PloS one, 8(7), e68967. https://doi.org/10.1371/journal.pone.0068967

78.

Watkins

M. W.

(2018). Exploratory factor analysis: A guide to best practice. Journal of Black Psychology, 44(3), 219–246. https://doi.org/10.1177/0095798418771807

79.

Williams

A. P.

(2004). The functions of Walkman music . PhD thesis, The University of Adelaide.

80.

Williamson

V. J.

Jilka

S. R.

Fry

Finkel

Müllensiefen

Stewart

(2012). How do “earworms” start? Classifying the everyday circumstances of involuntary musical imagery. Psychology of Music, 40(3), 259–284. https://doi.org/10.1177/0305735611418553

81.

Worthington

R. L.

Whittaker

T. A.

(2006). Scale development research: A content analysis and recommendations for best practices. The Counseling Psychologist, 34(6), 806–838. https://doi.org/10.1177/0011000006288127

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.45 MB