Senior Research Associate Dr Nina Di Cara and Professor Claire Haworth in the School of Psychological Science, and Dr Oliver Davis, Associate Professor at Bristol Medical School, are investigating whether it is possible to gain information about mood by analysing the music people stream. The work is supported by the Jean Golding Institute and the Elizabeth Blackwell Institute, through the Wellcome Trust Institutional Strategic Support Fund grant and a generous donation in memory of Jo Richardson.
Dr Di Cara explains: “Music is widely used to match mood as well as to change it—participants in our advisory group gave examples of listening to music in their home language when they are living abroad and homesick; or listening to energetic music to help wake themselves up.”
Pilot study
Dr Di Cara began with a pilot study in 2020 to establish the feasibility of the project and to explore the ethical implications. She recruited ninety-eight students and asked them to fill out questionnaires on demographics, personality, listening preferences, depression, anxiety and wellbeing. The participants were also asked for a download of their Spotify streaming history.
The music they listened to could be linked to information provided openly by Spotify such as “valance”—how positive or negative the music is—and its tempo, as well as “speechiness” (how much speech is included) and “instrumentalness”. The researchers also looked at how many hours people spent listening to music, and how many tracks they skipped through.
“Most people listen to between one and two hours of music a day”, says Dr Di Cara. “In our pilot study the type of music was important in a variety of ways. In general, more negative, less “speechy” or intstrumental, and slower music was correlated with higher depression scores. Listening to a greater number of tracks was also correlated with increased depression and anxiety, and lower general wellbeing scores.”
Lockdown
“During lockdown in 2020 we could definitely see music listening changes in our subjects—particularly in instrumentalness, speechiness and valence. However, one of the challenges with our pilot study was that we only had a single occasion where we measured mood through the questionnaires. To understand if we can use music listening to understand change over time we needed multiple measures of mood to go alongside the music streaming data.”
The team’s next study in 2024 addressed this. Once again, the 171 participating students completed surveys at the beginning and at the end of the study. Crucially, though, they were also asked to complete short mood surveys four times a day, an approach known as Ecological Momentary Assessment, or EMA.
EMA?
This is a technique that involves repeated sampling of an individual’s thoughts and behaviours as they go about their normal lives, at—or close to—the time they usually carry out that behaviour. Such data is usually collected using an app on the participant’s phone.
Dr Di Cara developed a software framework to collect EMA data on mental health, and to link it to Spotify data collected from participants. Her software also collated useful information about each track in a listening history from Spotify.
“We asked the participants to request their Spotify data under GDPR,” explains Nina, “and we use secure servers at the University of Bristol store the data.”
“Techniques like EMA that allow us to track mood over time have many potential benefits”, she continues. “Researchers don’t need to rely on participants trying to remember how they felt weeks or months ago, and we can identify patterns in response to external events at a personal or population level. If these technologies are successful, in future people may be able to use them to track their own moods, and share recent patterns with mental health professionals.”
“Of course, with new technology like this it is always incredibly important to pair it with an in-depth understanding of people’s views on the acceptability of its development and the limitations of its use.”
ALSPAC
Digital footprint data like music streaming have great potential to enhance the information collected by longitudinal birth cohorts such as ALSPAC, the Avon Longitudinal Study of Parents and Children, also known as the Children of the Nineties, where the researchers have already linked data from the X social media platform (formerly known as Twitter).
“Birth cohorts can benefit greatly from getting permission from their participants to collect digital footprint data.”, says Dr Davis, “The huge amounts of real-world behavioural data can fill in the gaps between surveys and complement more traditional approaches by providing new types of insight.” However, he is quick to point out that digital footprint research has just as much to gain from birth cohorts: “Often digital footprint researchers know very little about the people who are taking part in their studies. Collecting these data in birth cohorts allows us to link digital footprints to a lifetime of information to validate new techniques, and, importantly, information about what parts of the population are taking part so that we can reduce bias in the ways we analyse the data.”
The researchers have made the EMA and Spotify integration software freely available under an Open Source license, and the team aim to publish open access articles shortly which will discuss the approach and the findings.