Hypothesis 1: Perceived happiness can predict songs valence

Introduction

The demand for music has grown to become an important aspect of many peoples’ everyday life (Fuentes et al., 2019). Especially, the mode of music listening has shifted from only actively listening to music from time to time, to the soundtracking of daily activities. People accompany various activities with music, and it becomes an affective-practical resource. Songs have shown to change and strengthen peoples’ emotions, help patients’ with their anxiety and improve workers’ concentration (Sloboda et al., 2001; Mok & Wong, 2003; Lesiuk, 2005). A high demand for music automatically leads to a high demand for research in music listening behaviour. Several studies have focused on Music Emotion Recognition (MER) (Aljanaki et al. 2014). It focuses mostly on two dimensions: “valence (positive vs. negative) and arousal (quiet vs. energetic). However, the perfect MER model has not yet been found.

Research Question

This study will try to understand the measurement of valence and whether subjective and objective valence differs as much in our sample. Additionally, many studies have shown that musical sophistication has an effect on the relationship between objective and subjective musical perception (Castro & Lima, 2014). We will examine whether the level of expertise in music will impact musical perception in the context of valence, leading to the following research question:

To what extent does objective valence judgement predict subjective happiness ratings and to what extent is this relationship moderated by music sophistication?

Playlist

Data cleaning

The original data set contained 103 responses, but some had to be dropped. 21 participants did not finish the questionnaire and were thus removed from the data set. Furthermore, four participants took less than four minutes to complete the questionnaire. This was considered unreasonably low and so these responses were also removed from the data set. After this cleaning process, the data set contained 78 responses.

Method

In accordance with the hypotheses, the measured variables were participant’s happiness ratings per song (DV), Spotify’s valence ratings per song (IV) and the participants’ respective levels of musical sophistication (IV / Moderator). A convenience sample was gathered by distributing the questionnaire to friends, family, colleagues and other students. The valence ratings were sourced from the web-version of the Spotify API. The remaining, participant-specific data was gathered through Qualtrics using a mixed design. After signing the informed consent, participants were first asked to judge the happiness of 15 songs which were selected from the Dutch Charts playlist on Spotify. The songs were selected in a manner so that the spectrum from very low to very high valence (Spotify API) was completely covered. Following this, the participants were asked to fill out the Goldsmith Musical Sophistication Index (Gold-MSI) questionnaire, specifically the 18 items measuring the general musical sophistication subscale. For each item, participants are asked to indicate their agreement to the given statement on a scale from “Completely Disagree” to “Completely Agree”. Example items include “I can sing and play music from my memory” and “I would not consider myself a musician” (reverse-coded).

Hypothesis: Perceived happiness can predict songs valence

Valence is a valid proxy for happiness, however not perfect

In the existing literature, Spotify’s valence measure has often been used as a way to operationalize the happiness of a track or playlist. Given that our study found valence to significantly predict subjective happiness ratings, this practice seems to be valid. It should be noted however, that not all of the happiness’ variance was explained by valence (R²=0.12). Future research may therefore combine several of Spotify’s other features such as valence but also danceability or mode, assign weights to them and use this compound of features as a more accurate representation of happiness in music.

Distribution of valences for songs rated at different happiness level

The musical sophistication measured by the Gold-MSI does not seem to include the ability to judge happiness in music.

According to its authors, the Goldsmith-MSI is an instrument measuring “self-reported musical skills and behaviours” (Müllensiefen et al., 2014). In our study, a participant’s musical sophistication was not found to significantly strengthen the relationship between a song’s valence and the subjective happiness perceived by people listening to it. Thus, judging happiness in music is either not included in the skills and behaviours captured by the Gold-MSI or self-reported musical sophistication does not match actual performance, in our case judging the happiness of songs. In the above-mentioned original study presenting the Goldsmith-MSI, the authors validate it by demonstrating a positive association between musical sophistication and performances on melodic memory and beat perception tasks respectively. One common feature of those two tasks is that the measured aspect, whether melody or beat, is objective and precisely definable. This is not the case with happiness, which is a quality of a track that is more intangible and lies a lot more in the eyes (or rather ears) of the beholder. Thus, the Goldsmith-MSI seems to be more adequate when measuring people’s abilities to judge objective aspects of music, rather than intangibles such as happiness.

Valence distributions at each level of happiness by sophistication levels

People with a higher musical sophistication do not ‘agree more’ in their judgement of music happiness.

In Hypothesis 3 we tested to what extent the more sophisticated participants showed less spreaded answers (i.e. if the variance in their response was smaller). If this was the case, it pointed at more sophisticated participants ‘agreeing more’ in their answers. So even if they did not agree with the Spotify valence more than less sophisticated participants, they did agree among themselves. This prediction stems from the fact that more sophisticated participants indeed ‘agree more’ than less sophisticated ones when, for instance, judging the pitch of sound. Nevertheless, we found no sign of greater agreement in any of the 15 songs. This could have been explained by our sample lacking a good representation of more musically sophisticated people. It could have been the case that, out of chance or due to sampling bias, our participants were all rather unsophisticated in absolute terms. Nonetheless, after comparing our distribution to that from the original paper that introduced the Goldsmith-MSI questionnaire, this explanation is ruled out. In sum, our data clearly points at sophisticated people not showing a higher agreement in their happiness judgment.

Happiness perception by musical sophistication