Viral tunes: changes in musical behaviours and interest in coronomusic predict socio-emotional coping during COVID-19 lockdown

Pandemics have recurred throughout human history and, while containment measures reduce public health risks, they can create economic, societal, and political co-crises that affect socio-emotional wellbeing. During COVID-19, people faced fear of the virus, elevated stress due to physical distancing, reduced income or unemployment, added childcare burdens, uncertainty, and social isolation, with accumulating evidence for increased anxiety, depression, stress, loneliness, and impaired sleep. Maintaining mental wellbeing affects compliance with containment measures and helps prevent social unrest, stigma, and violence. Behavioral data can illuminate coping strategies that strengthen resilience. This study investigates music engagement—a widespread, accessible, lockdown-compatible behavior—as a coping strategy. Music can alleviate stress and anxiety, enhance positive feelings, and serve as a social surrogate, offering belonging and perceived company. However, it is unclear whether the effectiveness of musical engagement as a socio-emotional strategy generalizes to a broad collective crises context like COVID-19. It is also unclear what roles specific listening and practice behaviors play, and how situation-specific factors (e.g., access to music vs other coping resources, perceived severity) and individual variables (musical attitudes/behaviors, stress, pandemic realities) mediate use of music to cope. The study explores these questions using large-scale, cross-national survey data collected during the first COVID-19 lockdown.

Design and questionnaire: A custom multilingual (English, German, French, Italian) online questionnaire was developed to assess changes in musical engagement during COVID-19 lockdown. It included seven sections: (1) demographics; (2) living situation, employment, and importance of leisure activities/choirs during lockdown; (3) changes in listening formats and services, music-making forms, and social activities; (4) changes in the functions of music making (e.g., emotional regulation, social surrogacy) based on established scales; (5) analogous questions for music listening; (6) open-ended questions (not reported); (7) musicianship (OMSI), importance of music, musical participation, personality (BFI-10), and self-reported changes in health and wellbeing. Sections 3–5 were shown only to participants who reported listening to or making music in section 2. Most items assessed changes relative to pre-lockdown using 7-point Likert scales from “significantly less” to “significantly more,” with options “never do this” and “prefer not to say” for some items. Sampling and data collection: Data were collected mid-April to mid-May 2020 from adults experiencing lockdown in France, Germany, India, Italy, USA (New York State only), and the UK. Samples (≈700–1000 per country) were recruited via Prolific, MTurk, and IPSOS-MORI to approximate representativeness by age, gender, and education. The Indian sample, limited to English speakers, is not representative of the full population. Data were collected via Qualtrics (median completion time 17 min). Inclusion: age ≥18; quotas enforced for demographics. Ethics: exemption from University College Dublin HREC (REARN: HS-20-626). Pre-processing: Analysis in Python 3 with publicly available code/data (https://github.com/lkfnk/CMO and https://github.com/liknfk/CMQ). Participants in the lowest 10th percentile of completion time (conditional on sections completed) were removed. Participants with no response variability across large matrices (activities/functions) aside from “no change/never/prefer not to say” were removed. Multi-select items were one-hot encoded. “I never do this” responses were recoded as 4 (no change) for change questions. For country-level rankings, responses were centered by subtracting 4, medians computed per item within country, and items ranked by mean; ties received average ranks. Exploratory Factor Analysis (EFA): To capture individual differences related to lockdown experiences, an EFA was conducted on 51 variables spanning living situation, employment, COVID-19 contact, demographics, changes in health/wellbeing (e.g., loneliness, stress, positive/negative affect), and personality traits. Missing values were imputed with the country-specific median; continuous variables were centered/scaled. Diagnostics supported factorability (Bartlett’s test: χ2=678.790, p<.001; KMO=0.548). Parallel/eigenvalue inspection suggested six factors; EFA used six factors with quartimax rotation. Factor labels: (1) Negative Emotions (depression, stress, anxiety, loneliness, negative affect); (2) Age (age, living with parent, student, retired); (3) Positive Emotions (positive affect, energetic arousal, general health); (4) Living Situation (with partner/child/alone); (5) Employment Situation (full-time, retired, working from home during pandemic); (6) City Type (suburban/urban/rural). To describe profiles, participants in top vs bottom quartiles per factor were compared descriptively on means/SDs across variables; no inferential tests were conducted to avoid multiple-comparison inflation. Socio-emotional coping outcome and modeling: The primary individual-level outcome was a holistic socio-emotional coping score for music listening and for music making, computed as the average agreement with three items: (a) induced joy/helped in present situation, (b) made them feel connected to others, (c) served as a replacement for social interaction. Participants missing these items were excluded. Owing to multicollinearity and violated linear model assumptions (e.g., smallest eigenvalues near 0; many VIF>5; non-normal residuals: Jarque–Bera listening=270.535, p<.01, skew −0.491, kurtosis 3.663; making=502.022, p<.01, skew −0.620, kurtosis 4.287; heteroscedasticity via Breusch–Pagan: listening LM=434.319, p<.01; making LM=256.417, p<.01), non-linear models were used. Machine learning: For each domain (listening and making), a Light Gradient-Boosted Machine (LGBM) regressor predicted the coping score from demographics, COVID situation, activities, functions, situations, and music selection changes. Data were split 80/20 for train/test; grid search with 3-fold CV tuned hyperparameters. Best model example (listening): LGBMRegressor(learning_rate=0.5, max_depth=10, min_child_samples=10, num_leaves=40, random_state=123, subsample_for_bin=1000). Feature importance was assessed using SHAP values to quantify each feature’s marginal contribution to predictions. Open materials include analysis code and cleaned data.

- More than half of respondents reported using music to cope during the first COVID-19 lockdown (April–May 2020). - Between 34% and 57% reported adapting musical behaviors; 57% reported moderate to extreme interest in coronomusic (COVID-related music content). - Music listening ranked among the most increased-in-importance daily activities during lockdown, behind staying connected/informed and household chores; among leisure behaviors, only watching movies ranked higher. - People experiencing increased negative emotions used music primarily for solitary emotional regulation (e.g., reduce stress/anxiety, provide comfort, support, solace, distraction, venting), whereas those experiencing increased positive emotions used music as a proxy for social interaction (feeling connected, part of a group) and to seek aesthetic/spiritual experiences. - Changes in listening and making situations reflected these patterns: negative-emotion groups listened more while alone or during other activities; positive-emotion groups listened more with others and in attentive contexts (evenings, travel) and engaged more in social/dance-related music making. - LGBM models identified predominantly music-related behavioral adaptations as top predictors of coping via both listening and making. Shared key predictors included: interest in coronavirus-related music (coronomusic), selecting different music than before, social-surrogacy functions (feeling like having company/part of a group), the general importance of music, personality trait neuroticism, and the confirmed number of COVID-19 infections. - For listening, additional strong predictors included social-surrogacy (reduces loneliness, sharing one’s experience) and self-directed regulation functions (creating personal space, comfort and support, feeling good about self, new perspectives, reduced stress), increased importance of music listening, and listening during other activities. Older age predicted higher coping; lower agreeableness, neuroticism, and extraversion related to lower coping via listening; increased anxiety changes during lockdown were also predictive. - For making music, predictors were more centered on generating positive affect, emotion expression, arousal regulation, relaxation, entertainment, strong sensation, interpersonal creativity/communication, and socially shared music making (performing for others, posting on social media). - Tabled descriptive contrasts (top vs bottom quartiles) showed larger mean differences for: Negative Emotions factor—listening functions such as distract (0.90), reduce stress (0.85), comfort (0.78), support in bad mood (0.78), feel less lonely (0.77), vent negative emotions (0.75); Making functions such as feel less lonely (0.94), comfort (0.92), reduce stress (0.92), daydream (0.82), vent negative emotions (0.80), distract (0.79). Positive Emotions factor—listening functions such as spiritual (0.93), understanding others (0.88), aesthetic (0.88), connected to culture (0.85), identify with artist (0.84), sharing experience (0.82); Making functions such as spiritual (1.23), identify with artist (1.15), dwell on worries (1.15), aesthetic (1.13), new perspectives (1.09), part of a bigger group (1.09). Situational differences mirrored social vs solitary use. - Overall, interest in coronomusic emerged as the single strongest predictor of socio-emotional coping.

The study asked whether musical engagement served as an effective socio-emotional coping strategy during the COVID-19 lockdown and which factors predicted its use. Findings show that more than half of adults used music to cope, and that coping was tied less to who people are (stable traits/demographics) than to how they adapted their music-related behaviors (functions, situations, and selections). Individuals reporting increased negative emotions primarily leveraged music for solace-oriented, solitary regulation (comfort, stress reduction, companionship by proxy). Those reporting increased positive emotions engaged with music more socially, using it as a stand-in for social interaction and to seek aesthetic/spiritual experiences. Predictive modeling confirmed that adaptive changes in music functions/situations and an interest in pandemic-specific musical content (coronomusic) were the most important determinants of perceived coping benefits, whereas personality and demographics played smaller roles. These results address the research question by demonstrating that musical engagement is an effective, context-sensitive coping strategy during societal crises and that tailoring behaviors and selections—particularly engaging with topical, socially resonant repertoires—enhances socio-emotional benefits. Differences between listening and making suggest complementary coping potentials: listening aligns more with solace and reduction of negative affect, whereas making music supports agency, positive affect generation, and social connection.

This study contributes large-scale, cross-national evidence that during the first COVID-19 lockdown, people adapted their music-related behaviors to meet socio-emotional needs, with more than half using music to cope. The most influential predictors of perceived coping via music were changes in functional use and situations of engagement and, strikingly, interest in coronomusic. Listening and making provided complementary coping potentials—listening more aligned with solace and mitigation of negative states, making more with agency, expression, and social connection. Future research should analyze textual/musical/visual features of coronomusic to determine what makes content resonate and support coping at scale, and examine causal mechanisms and longitudinal effects of music-based coping. Policy implications include recognizing the public value of musical practices and repertoires during crises and considering support for musicians (e.g., commissioning pandemic-related works, funding participatory musical activities, compensation schemes) to bolster collective resilience.

- Sampling and representativeness: Although quotas aimed for demographic representativeness, the Indian sample comprised English-literate participants and is not representative of the full population; the U.S. sample was restricted to New York State. Online-panel recruitment may introduce selection biases. - Cross-sectional, self-report design: Data rely on retrospective comparisons (before vs during lockdown) and subjective reports; causal inferences cannot be drawn. - Measurement and preprocessing choices: Recoding “I never do this” as no change, centering/ranking within countries, and imputation by country medians may affect estimates; acquiescence/response biases (noted especially for India) remain possible despite rank-based summaries. - Factor-analytic descriptives: High/low quartile comparisons were descriptive without inferential testing to avoid multiple-comparison inflation, limiting statistical conclusions about group differences. - Modeling constraints: Many correlated predictors and violated linear-model assumptions necessitated tree-boosting models; while SHAP aids interpretability, feature importance does not establish causality. Some variables were removed due to survey errors (e.g., certain listening mode items).