Predicting Music Consumption

Music consumption is a tricky thing to try to predict. We are all individuals with unique tastes and lives, into which music is interwoven to varying degrees. Yet millions of pounds is spent every year by music, radio, advertising and leisure industries (to name a few) to try to better understand the factors that predict our listening and music consumption behaviour.  The question on their agendas is, are there any universals that might at least loosely predict who will listen to what and how? In other words, how do we tailor music for ‘maximum impact’?

Psychologists interest in individual differences in music preference goes back a long way and one of the most consistently used experimental paradigms in this field is to measure a large group of peoples’ listening habits and then determine if there are any relationships between their habits and aspects of the individuals’ underlying personality traits

One of the problems with this approach is that it can be rather hit and miss. Filling in personality questionnaires takes time  so you can’t possibly test for everything. And you shouldn’t actually just go testing for everything – that is the scientific equivalent of having a good poke around in the dark! Basic research design courses will tell you (rightly) that you should go into an experiment with a workable hypothesis that you can test. Unfortunately, when it comes to music consumption there are lots of candidate personality traits that would be interesting to test and therefore lots of different hypotheses, because many factors have been shown to influence musical behaviours (e.g. musicianship, music enjoyment and musical development). So there is a danger that psychologists in this field end up chasing a multitude of different hypotheses, probably in individual experiments due to time constraints, without ever drawing together a coherent piece of research.

A recent study by a psychologist in my department at Goldsmiths, Tomas Chamorro-Premuzic, and colleagues has done a very good job of balancing the need for multiple, grounded hypotheses testing on the same population. Tomas and his team were interested in further exploring the Uses of Music Inventory (UMI), a scale published first by Tomas and Adrian Furnham in 2007. The UMI is a simple 15 point scale which assesses three distinct motives for using music:

1) Emotional (using music to induce moods and change an emotional experience)

2) Cognitive (listening in an intellectual and rationale manner)

3) Background (while working, studying, socialising or performing other tasks).

An individual can rate as high or low across all three scales, thereby establishing a profile of their patterns of music listening. The first aim of the present study was to replicate previous work on the scale with a larger sample (see Tomas’ page for references)

Another aim of the study was to extend previous work that related musical preferences and listening behaviours to personality traits as measured by the Big Five Scale (extraversion, neuroticism, openness to experience, agreeableness and conscientiousness) and a related but different personality construct of Emotional Intelligence (EI). Because the Big 5 and EI are related, an important question was whether the EI predicted anything unique in musical consumption.

The hypotheses:

  • All 3 uses of music subscales should be positively associated with music consumption: that is, the more you use music for any reason the more music you should consume
  • Neuroticism would be positively associated and trait EI negatively correlated with emotional use of music
  • Extraversion would be positively correlated with background use of music
  • Openness would be positively correlated with cognitive use of music

They tested their predictions with structural equation modelling (SEM) which is a fancy statistical way of looking at the variance in music consumption, building a model to show how each of the background measures relates to these patterns, and then determining how much of the original variance is uniquely explained by each of the measures. So you can work out a) which factors explain patterns of music consumption and b) their relative importance.

The Music Consumption Scale they used included 10 factors like ‘ Purchase music from online stores’, ‘ Download music’, ‘Watch TV or films about musicians’, ‘ Attend music concerts’, ‘ Play a musical instrument’ and ‘Visit music shops with the idea of buying music’. As a participant you simply rate each statement on a 5 point scale from 1 = very rarely to 5 = Very often (3 = sometimes).

They tested 535 people in the present study, which is a good size for a personality study. In individual differences test you need HUGE numbers before the statistical modelling makes sense, and before you can really be sure that your conclusions have any real relationship to what is happening in the real world (what psychologists call ‘ecological validity’).

Their final model suggested that individual differences were associated with different uses of music and that music uses did predict a significant degree of music consumption (36%), although there were some unexpected associations, as well as lack of association.

A summary of the findings:

  • Men were more likely to use music for cognitive reasons, whereas women were more likely to use music for emotional reasons. You can insert your own ideas about response biases here, but all being said these results do support previous findings.
  • Older participants were less likely to use music for background. This may reflect decreasing appreciation of music with age or a difference in uses of music between generations.
  • All 3 uses of music subscale factors had significant effects on music consumption, which was not affected by the other personality traits or EI.
  • Neuroticism positively predicted emotional use of music. The authors say that this is in line with the theory that neurotics are more emotionally sensitive to music when compared to their emotionally stable counterparts.
  • Openness to Experience predicted cognitive uses of music. This supports research suggesting that open individuals have an interest in more sophisticated forms of music.
  • The study failed to find the predicted association between Extraversion and background use of music. This finding adds to growing evidence that the relations between Extraversion and music use are less stable than other factors and that as yet unexplained confounding variables are confusing matters in this area.
  • Significant correlations between EI and emotional use of music were found, but in other studies this effect becomes negligible when you throw in other personality traits that better explain the variance. Also a weird finding was that people who are low on EI tended to listen more to background music, which is counterintuitive with predictions about Extraversion. The authors conclude that EI is so linked with traits within the Big 5 that there is no unique variance that it can explain on its own.  They (and I) suspect the hand of confounding variables is at work here too.

Overall this paper does a very good job of throwing a good number of pre-determined, scientifically valid hypotheses about individual differences into the mixer and determining which, if any, predict how we consume music. It seems that participant demographics such as gender and personality can make limited predictions regarding music use (although EI looks a non-starter) but that overall the best predictor of our music consumption is the Uses of Music Inventory, with its 3 subscales.

This finding emphasises the need to take into account our everyday patterns of music use when trying to explain our levels and patterns of music consumption. It has been, perhaps, an overlooked aspect of a chain which has a degree of independence between the components: personality affects how we use music, which in turn affects our music consumption behaviour.

A final point made by the authors is that the inventory in its current state may be far from a finished scale: Now that the UMI effects have been replicated and their utility in music research psychology research has been demonstrated, a next step would be to include aspects of music listening that the current short 15 item scale does not encompass. These factors include personal identity formation, social identity formation (see North and Hargreaves still un-rivalled book on the subject for more information) and interpersonal exchange/communication.

Paper: Chomorro-Premuzic, T., Swami, V., & Cermakova, B. (2011 – in press) Inidividual differences in music consumption are predicted by uses of music and age rather than emoational intelligence, neuroticism, extraversion or openness. Psychology of Music, DOI: 10.1177/0305735610381591

One thought on “Predicting Music Consumption

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>