Vous êtes sur la page 1sur 16

Journal of Personality and Social Psychology

Music Through the Ages: Trends in Musical Engagement


and Preferences From Adolescence Through Middle
Adulthood
Arielle Bonneville-Roussy, Peter J. Rentfrow, Man K. Xu, and Jeff Potter
Online First Publication, July 29, 2013. doi: 10.1037/a0033770

CITATION
Bonneville-Roussy, A., Rentfrow, P. J., Xu, M. K., & Potter, J. (2013, July 29). Music Through
the Ages: Trends in Musical Engagement and Preferences From Adolescence Through Middle
Adulthood. Journal of Personality and Social Psychology. Advance online publication. doi:
10.1037/a0033770
Journal of Personality and Social Psychology 2013 American Psychological Association
2013, Vol. 105, No. 9, 000 0022-3514/13/$12.00 DOI: 10.1037/a0033770

Music Through the Ages: Trends in Musical Engagement and Preferences


From Adolescence Through Middle Adulthood

Arielle Bonneville-Roussy, Peter J. Rentfrow, Jeff Potter


and Man K. Xu Atof Inc., Cambridge, Massachusetts
University of Cambridge

Are there developmental trends in how individuals experience and engage with music? Data from 2 large
cross-sectional studies involving more than a quarter of a million individuals were used to investigate age
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

differences in musical attitudes and preferences from adolescence through middle age. Study 1 investi-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

gated age trends in musical engagement. Results indicated that (a) the degree of importance attributed to
music declines with age but that adults still consider music important, (b) young people listen to music
significantly more often than do middle-aged adults, and (c) young people listen to music in a wide
variety of contexts, whereas adults listen to music primarily in private contexts. Study 2 examined age
trends in musical preferences. Results indicated that (a) musical preferences can be conceptualized in
terms of a 5-dimensional age-invariant model, (b) certain music-preference dimensions decrease with age
(e.g., Intense, Contemporary), whereas preferences for other music dimensions increase with age (e.g.,
Unpretentious, Sophisticated), and (c) age trends in musical preferences are closely associated with
personality. Normative age trends in musical preferences corresponded with developmental changes in
psychosocial development, personality, and auditory perception. Overall, the findings suggest that
musical preferences are subject to a variety of developmental influences throughout the life span.

Keywords: musical preferences, age differences, personality, life span development, exploratory
structural equation modeling

Are there age differences in the ways in which individuals Musical Engagement, Preferences, and Age
experience and engage with music? Anecdotally, it seems obvious
Considerable evidence indicates that musical engagement and
that young people invest more time and effort listening to music
preferences have a psychological basis. The theoretical framework
compared with adults. However, despite growing interest in the
guiding much of the research in this area is that music reflects and
social-personality psychology of music (Rentfrow, 2012), we
reinforces peoples psychological, biological, and social needs
know next to nothing about age trends in musical engagement or (Rentfrow, 2012). Because practically all of the research in this
preferences. The objective of the current project was to fill that area has relied only on young people, it is impossible to know
void by broadening our understanding of the role music plays whether music serves similar functions throughout the life span.
across the life span. To achieve that objective, we examined age Are there age differences in musical engagement and preferences?
differences in musical engagement and preferences in two large
cross-sectional samples covering adolescence through middle Musical Engagement
adulthood.
Music is a vehicle for self-discovery, self-regulation, and self-
expression for most young people. Studies consistently show that
young people place significantly more importance on music than
on clothing, films, books, magazines, computer games, TV, and
sports (e.g., Lonsdale & North, 2011; North, Hargreaves, & Har-
greaves, 2004; Rentfrow & Gosling, 2003). Young people report
Arielle Bonneville-Roussy, The Psychometrics Centre, Department of listening to music for arousal and emotion regulation, social net-
Psychology, University of Cambridge, Cambridge, United Kingdom; Peter working, and self-expression (North et al., 2004; Rentfrow &
J. Rentfrow, Department of Psychology, University of Cambridge; Man K. Gosling, 2003). They also report listening to music frequently and
Xu, Department of Psychiatry, University of Cambridge; Jeff Potter, Atof in various settings, including public, private, social, and solitary
Inc., Cambridge, Massachusetts. (North et al., 2004).
This research was supported by the Social Sciences and Humanities
It is not clear whether music continues to play an important role
Research Council of Canada and the Cambridge Commonwealth Trust for
Arielle Bonneville-Roussy. We are grateful to David Boyle for providing
throughout the life span. Only two studies have examined musical
access to the data used in Study 1. engagement in relation to age, and the results were not entirely
Correspondence concerning this article should be addressed to Peter J. consistent. Lonsdale and North (2011) reported that people over 30
Rentfrow, Department of Psychology, University of Cambridge, Downing regard music as less important compared with adolescents,
Street, Cambridge, CB2 3EB, United Kingdom. E-mail: pjr39@cam.ac.uk whereas Laukka (2007) observed that adults 65 considered music

1
2 BONNEVILLE-ROUSSY, RENTFROW, XU, AND POTTER

more important now than at any previous time in their lives. The best friends were significantly more similar than were preferences
results from both studies also indicated that middle-aged and older among random strangers.
adults use music for purposes of arousal and emotion regulation. Research in this area has established a firm psychological basis
In summary, research on musical attitudes and age clearly to understand musical preferences. However, most of the studies
indicates that music is very important during adolescence, but it is are only with young people, so we do not know whether the same
unclear whether it remains as important in adulthood. Thus, if we psychological characteristics are related to musical preferences or
are to develop a thorough and complete understanding of the role whether the magnitude of those associations generalizes to adults.
music plays throughout life, we need to investigate musical atti-
tudes and beliefs from adolescence through adulthood. Age
There have been very few investigations of age differences in
Musical Preferences musical preferences. Of the handful of studies conducted, there
The past decade has witnessed considerable interest in questions appears to be consensus for the hypothesis that musical prefer-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ences crystallize in early adulthood and do not change from that


concerning individual differences in musical preferences (e.g.,
This document is copyrighted by the American Psychological Association or one of its allied publishers.

point onward (Holbrook & Schindler, 1989; North & Hargreaves,


Boer et al., Lam, 2011; Colley, 2008; George, Stickle, Rachid, &
1995). The evidence offered as support for the hypothesis essen-
Wopnford, 2007; Rentfrow, Goldberg, & Levitin, 2011; Rentfrow
tially shows that people report stronger preferences for popular
& Gosling, 2003; Rentfrow & McDonald, 2010; Schfer &
music artists they listened to during young adulthood than they do
Sedlmeier, 2009; Zweigenhaft, 2008). The results from these in-
for popular artists who came before or after. However, the research
vestigations converge on two conclusions: (a) There is a latent
has relied on small samples, focused only on preferences for
structure underlying musical preferences and (b) preferences are
popular music, and is entirely cross-sectional. As a result, it is not
linked to various psychological characteristics. Studies on the
clear how preferences for other genres vary by age or whether the
structure of musical preferences strongly suggest that individual
variance in musical preferences is greater in adolescence than
differences in music-genre preferences can be conceptualized in
middle adulthood.
terms of approximately four to six music-preference dimensions.
An alternative perspective is that musical preferences vary
In a recent series of studies involving a wide range of musical
throughout adulthood. Indeed, evidence from numerous studies of
genres, Rentfrow, Goldberg, and Levitin (2011; Rentfrow et al.,
normative changes in personality clearly suggest that Neuroticism
2012) identified five robust preference dimensions, labeled Mel- is comparatively high during adolescence and decreases as people
low, Unpretentious, Sophisticated, Intense, and Contemporary age, whereas Agreeableness and Conscientiousness are low during
(MUSIC). Each dimension comprises multiple musical genres that adolescence and increase as individuals get older (e.g., Allemand,
share common musical (i.e., loudness, timbre) and psychological Zimprich, & Hertzog, 2007; Lucas & Donnellan, 2011; Roberts,
(i.e., complexity, affect) characteristics. Walton, & Viechtbauer, 2006; Soto, John, Gosling, & Potter,
Research on the psychological correlates of musical preferences 2011). This organization of traits converges with the storm and
indicates that preferences are associated with personality, values, stress period of adolescence when young people struggle to find
and cognitive abilities (e.g., Delsing, ter Bogt, Engels, & Meeus, their place (Arnett, 1999), and the changes in these traits during
2008; Rentfrow, Goldberg, & Zilca, 2011; Rentfrow & Gosling, young and middle adulthood seem to correspond with the periods
2003, 2006; Zweigenhaft, 2008). For example, preferences for of forming intimate relationships and pursuing careers. Thus, just
sophisticated musical styles (classical, opera, and jazz) are posi- as normative changes in personality occur throughout the life span,
tively related to Openness, imagination, liberal values, artistic it seems reasonable to expect normative changes might also occur
expression, and verbal ability. Preferences for intense music in musical preferences.
(heavy metal and punk) are positively related to Openness, sensa- Explanations for personality change emphasize the impact of
tion seeking, and impulsivity. Also, preferences for contemporary age-graded social roles, such as marriage, family, and work (Rob-
music (pop, rap, and dance) are positively associated with socia- erts, Wood, & Smith, 2005), as well as psychosocial developmen-
bility, status orientation, and physical attractiveness. tal challenges (Erikson, 1950, 1968). For example, Eriksons life
There is also evidence that musical preferences are influenced span model of development argues that individuals are faced with
by the social connotations associated with music (Tarrant, North, the challenge of balancing the demands of certain psychosocial
& Hargreaves, 2002), as people are drawn to musical styles with conflicts as they progress through different life stages. The chal-
social characteristics that reflect aspects of their identities. For lenge during adolescence is to develop a coherent identity and
example, studies on music-genre stereotypes indicate that fans of autonomy; from late adolescence through early adulthood, the
rock and heavy metal music are considered aggressive, disobedi- challenge is to develop intimate bonds of love and friendship; and
ent, and independent, whereas fans of classical music are consid- through middle adulthood the two major life challenges are to
ered wealthy, intelligent, and sophisticated (Rentfrow & Gosling, maintain intimate relationships and to pursue a profession. Given
2007; Rentfrow, McDonald, & Oldmeadow, 2009). In addition, that individuals use music for self-expression, arousal regulation,
studies of music preferences among adolescents suggest that indi- and social bonding (tasks that are more or less salient at different
viduals listen to particular styles of music to be popular among life stages), it is conceivable that musical preferences change in
peers and to enhance group affiliation (Bakagiannis & Tarrant, concert with the social roles and life stages encountered through-
2006; Boer et al., 2011; Tarrant et al., 2002). For example, Self- out life.
hout, Branje, ter Bogt, and Meeus (2009) found that the musical Taken together, research on the links between music preferences
preferences of individuals who mutually regarded each other as and personality, identity, and social bonding provide good reasons
AGE TRENDS IN MUSICAL ENGAGEMENT AND PREFERENCES 3

to believe that preferences might change throughout adulthood. Measures. Four items were used to assess music consump-
Indeed, adolescence through adulthood is marked by significant tion. Participants were asked to estimate separately for the average
changes that impact personality, identity, and social development. weekday and for the average weekend the number of hours they
Considering that people seek out styles of music that reinforce spent listening to music that you have chosen/bought and lis-
their personalities, reflect their social identities, and enhance their tening to music that may be on in the background. The sum of the
relationships, it stands to reason that the music they choose to four items was taken as a measure of music consumption (Cron-
listen to might change as well. bachs .78).
A single item was used to assess musical importance. Partici-
Overview of the Current Research pants were asked to choose one statement from a list of five that
best reflected their attitude toward music. In decreasing order of
The overarching aim of this work was to establish a foundation importance, the statements were as follows: Music means a lot to
on which to develop and test hypotheses about the development of me, and is a passion of mine; Music is important to me, but not
musical preferences. The theoretical framework guiding our inves- necessarily more important than other hobbies or interests; I like
tigation was informed by interactionist theories positing that indi-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

music, but it does not feature heavily in my life; Music is no


viduals seek out environments that meet their needs (Buss, 1987).
This document is copyrighted by the American Psychological Association or one of its allied publishers.

longer as important as it used to be to me; and Music has no


Along these lines, and consistent with previous work, we reasoned particular interest for me. Musical importance was examined by
that individuals use music to satisfy and reinforce basic psycho- investigating the proportion of participants who endorsed each
logical needs. Thus, we expected age differences in musical in- item.
volvement to correspond to normative changes in particular psy- To study the contexts of music-listening behavior, participants
chological processes. reported the frequency with which they listened to music in vari-
Using multiple samples, methods, and recruitment strategies, ous situations. Using a 4-point scale with endpoints at 1 (Never)
two large cross-sectional studies were conducted to investigate and 4 (Frequently), participants reported how often they listened to
trends in musical engagement and preferences from adolescence music while At home relaxing; At home relaxing with friends;
through middle adulthood. In Study 1, we examined age trends in Out with friends; At work; Doing the housework; Doing
musical engagement in a large nationally representative sample of other tasks at home, e.g., DIY; and In the car. Analyses of the
U.S. and U.K. residents, and in Study 2, we examined age trends associations among the items revealed a very large correlation
in musical preferences using data from a very large Internet between Doing the housework and Doing other tasks at home,
sample, and their relationships with gender and personality. e.g., DIY (r .75, p .001). Thus, we combined these two
variables to form a Doing housework or other tasks measure. All
Study 1: Age Trends in Musical Engagement in all, six music-listening contexts were examined.
Numerous studies have shown that young people listen to music
regularly and consider it very important (e.g., Rentfrow & Gosling, Results and Discussion
2003; Tekman & Hortasu, 2002), but little attention has been
given to the musical attitudes or behaviors of adults. The primary Preliminary analyses revealed no gender or country differences
objective of Study 1 was to examine age differences in how much in the three sets of musical engagement variables. For the sake of
importance people give to music, how often they listen to music, simplicity, we report results from analyses that did not include
and the contexts in which they listen to it. gender as a covariate and that included participants from both the
U.S. and U.K. samples.
Music consumption. Across all ages, the amount of time
Method
participants reported listening to music ranged from 0 to 96 hr per
Participants. Participants were members of nationally repre- week (M 16.21, SD 14.09). The correlation between the sum
sentative Internet panels recruited by IPSOS, a global market of the four music-listening items revealed a negative relationship
research company. Data on musical attitudes were available for between age and music consumption (r .20, p .001). Across
9,080 participants, of whom 4,009 (44.2%) were from the United ages, 18 year-olds reported listening to the most amount of music
Kingdom and 5,071 (55.8%) were from the United States. The in an average week (25 hr), whereas 58-year-olds reported listen-
sample consisted of 4,670 (51.4%) women and 4,410 (48.6%) ing to the least amount of music (12 hr).
men. The average age of participants was 40.53 years and ranged Musical importance. Descriptive statistics across all ages
from 13 to 65 years (SD 14.67). Of those who indicated, 223 indicated that 31% of participants felt passionately about music,
(2.5%) participants were Asian, 323 (3.6%) were Black or African, 38% felt that music was at least as important as other hobbies and
380 (4.2%) were Hispanic or Latino, 7,766 (85.7%) were White, interests, 26% reported that they liked music but that it did not
and 89 (1.0%) were of another ethnicity. feature heavily in their lives, 4% felt that music was less important
Procedure. From March through April 2009, U.S. and U.K. than it used to be, and 1% reported that they had no particular
members of the IPSOS Internet panel were invited to complete a interest in music. To determine whether there were significant age
survey about musical attitudes for EMI Music, a multinational differences in musical importance, a series of chi-square analyses
music company. The survey was designed to measure respondents were performed on the five musical importance categories from
attitudes about music as well as their knowledge and familiarity age 13 to 65. The results revealed significant effects of age on
with a variety of artists in the EMI music catalogue. In exchange musical importance, 2(208) 886, p .001. Post hoc compar-
for completing the survey, participants received value points that isons between groups revealed that the expected percentage of
could be used for purchasing a range of products. participants who were passionate about music decreased from 41%
4 BONNEVILLE-ROUSSY, RENTFROW, XU, AND POTTER

at age 13 to 15% at 65, whereas the proportion of participants who Effectively investigating age differences in musical preferences
thought that music was as important as other hobbies remained requires a robust framework for conceptualizing and measuring
fairly stable, from 38% at age 13 to 35% at 65. An increased preferences. The MUSIC model (Rentfrow, Goldberg, & Levitin,
proportion of participants stated that music did not feature heavily 2011) offers a potentially useful framework for conceptualizing
in their life, between 19% at age 13 and 42% at 65. Less than 10% musical preferences, but all the current evidence supporting the
of participants of all ages thought that music was not important or model has relied exclusively on young people. As a result, it is not
had no particular interest for them. Overall, these results indicate known whether the same music-preference dimensions generalize
that younger participants are more passionate about music com- across age. Thus, if we are to make meaningful comparisons of
pared with older participants but that music is generally quite musical preferences across age, it is crucial to first determine
important to people of all ages. whether the structure of musical preferences is age invariant.
Music-listening contexts. The relationships between age and
the frequency of music listening in different contexts were as- Study Overview
sessed using Pearsons correlations. All correlation coefficients
Aims. The first aim of the project was to determine whether
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

between age and contexts were significant at p .001. To avoid


the factor structure of musical preferences is invariant with age.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

interpreting spurious results, only correlation coefficients |.20|


The second aim was to examine age trends in musical preferences
are interpreted. Participants reported listening to music in the car
and to gauge the impact of personality on such trends. To address
more often than in any other context (M 3.59, SD .81), and
the aims of the project, we used exploratory structural equation
this trend remained constant with age (r .04). The second most
modeling (ESEM) to examine measurement invariance of the
common music-listening context was at home, relaxing alone
music preference structure across age. We then examined age
(M 3.19, SD .86); young participants reported listening to music
trends in preferences using regression analyses within SEM to
in this context slightly more than did older participants (r .13).
assess the links between music preferences, age, gender, and
Doing housework or other tasks at home was a context in which
personality.
music was listened to occasionally (M 3.00, SD .95) and was
Design. This study was designed to overcome several of the
slightly more common among younger than older participants (r
shortcomings of previous research in this area. First, a very large
.11). Participants reported listening to music at home, relaxing
cross-sectional sample of nearly a quarter of a million respondents
with friends less often than the other contexts (M 2.88, SD
was used in the current study. Such a large sample provides the
.98), and young people listened to music more in this context
statistical power necessary to reliably detect interactions and cur-
compared with older participants (r .22). Out with friends was
vilinear trends (McClelland & Judd, 1993). Second, samples with
a less popular music-listening context (M 2.51, SD 1.02), and
narrow age ranges were used most in previous studies, with many
it was negatively associated with age (r .34). Finally, the least
focusing only on adolescents and young adults or relying on small
frequent music-listening context was at work (M 2.32, SD
samples of middle-aged adults (Holbrook & Schindler, 1989;
1.19), which was most stable with age (r .07).
North & Hargreaves, 1995). In contrast, a much wider range of
Summary. Consistent with previous research, participants
participants between the ages of 12 and 65 was used in the current
over 30 considered music less important than did adolescents or
study. Our sample also included at least 164 participants at each
young adults, although middle-aged adults still listened to 12 hr or
individual year of age. Third, most of the studies that have exam-
more of music per week. The majority of participants across all
ined age trends in music preferences assessed preferences for pop
ages reported that music was either a passion or as important as
music only (Holbrook & Schindler, 1989; North & Hargreaves,
their other interests. Our analyses of music-listening contexts
1995). In contrast, the current study assessed individual differ-
indicated that adolescents listened to music in a wide variety of
ences in preferences for a wide assortment of music genres, which
public and private settings, whereas adults listened to music pri-
allowed us to investigate whether the structure of preferences is
marily in private. It is worth noting that although musics promi-
invariant across age and also whether trends in preferences vary by
nence declined after adolescence, in an absolute sense, music
musical style. Fourth, the current study included a measure of the
continued to play an important role throughout adulthood. Never-
Big Five domains, which enabled us to investigate the impact of
theless, the current findings are entirely cross-sectional, which
personality on musical preferences from adolescence through mid-
leaves open the possibility that the results reflect cohort differ-
dle adulthood. Given these design aims, we decided to rely on the
ences and not developmental trends. Longitudinal research is
Internet to collect data for the current study.
needed to examine intraindividual musical engagement.
A unique feature of the current project is that the data were
collected over a period of 8 years. This multiyear data collection
Study 2: Age Trends in Musical Preferences period produced variance in participants birth year that was not
shared with age at the time of participation. This aspect of the
The results from Study 1 indicate that individuals become less
research design enabled us to determine whether earlier born
passionate about music as they age, but the majority of people
versus later born participants music preferences differed system-
young and old alikestill place considerable importance on mu-
atically from each other, and thus whether birth-cohort effects
sic. This raises the question: Do musical preferences change, just
contributed to any of the current findings.
as musical engagement changes, or are preferences stable across
the life span? Do age differences in musical preferences corre-
Method
spond to normative changes in personality? The objective of Study
2 was to explore age trends in musical preferences and to examine Participants. Participants volunteered to complete a music-
the potential impact of personality on such trends. preference survey over the Internet. As in all studies that collect
AGE TRENDS IN MUSICAL ENGAGEMENT AND PREFERENCES 5

data from individuals over the Internet, there is the possibility that Personality was assessed using the Ten-Item Personality Inven-
respondents may complete a survey multiple times. Repeat re- tory (TIPI; Gosling, Rentfrow, & Swann, 2003). The TIPI assesses
sponding has the potential to produce unreliable and misleading each of the Big Five domains with only two items and has strong
results, so it was necessary to remove data from potential repeat convergent and discriminant validity with longer measures of the
responders. Several criteria were used to eliminate repeat respond- personality domains. In the current study, respondents were asked
ers. We removed participants who indicated that they had filled out to report the degree to which they agreed with each item using a
the questionnaire more than once. In addition, the questionnaires 7-point rating scale with endpoints at 1 (Disagree) and 7 (Agree).
that were answered from the same IP address within less than 1 hr Analytic strategy. The analyses were designed to address two
and those from the same IP address that matched on several goals. First, we wanted to determine whether the MUSIC model
demographic characteristics (gender, age, ethnicity) were also was invariant across age. Second, we wanted to map age trends in
eliminated. Finally, participants with spurious patterns of re- preferences and examine the impact of personality on those trends.
sponses were removed from the database (e.g., those who re- Those goals were accomplished within an SEM framework. All
sponded to all the survey items with 1s or 2s). analyses were conducted in Mplus 7.0 software (Muthn &
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Implementation of the aforementioned criteria resulted in data Muthn, 1998 2012).


This document is copyrighted by the American Psychological Association or one of its allied publishers.

for 254,825 participants. Of these participants, 147,589 (57.9%) Confirmatory factor analysis (CFA) has traditionally been used
were female, and 107,236 (42.1%) were male. The average age of to evaluate the factor structure of an instrument. However, CFA
participants was 23.52 years and ranged from 12 to 65 years involves strict assumptions that are not easily met in psychological
(SD 10.06). In terms of education, of those who reported, 76,692 research. For instance, CFA assumes highly restrictive, zero load-
(32.5%) participants reported having less than a high school degree, ings for items that are nontarget loadings for factors. This in turn
19,021 (8.1%) had a high school diploma, 73,554 (31.2%) attended could inflate the associations between factors and bias the rela-
some college, 31,810 (13.5%) had a college degree, and 34,606 tionship between the factors and external covariates. ESEM, which
(14.7%) had a postgraduate diploma. Among those who indicated combines features of CFA and exploratory factor analysis (EFA),
their socioeconomic status, 26,770 (21.9%) participants identified overcomes this restriction by allowing loadings of nontarget items
themselves as working class, 23,728 (19.4%) identified as lower- to be estimated and thus provides less biased factor correlations
middle class, 47,482 (38.9%) identified as middle class, 20,638 (Morin, Marsh, & Nagengast, 2013). The main advantages of
(16.9%) identified as upper-middle class, and 3,546 (2.9%) identified ESEM over CFA are to integrate the less restrictive assumptions of
as upper class. Of those who indicated, 23,150 (9.3%) participants EFA with the benefits of SEM, such as the presence of goodness-
were Asian, 7,899 (3.2%) were Black or African, 12,352 (4.9%) were of-fit indices, the possibility of performing multigroup invariance
Hispanic or Latino, 189,595 (75.9%) were White, and 16,742 (6.7%) analysis, and the combination of regression and structural equa-
reported another ethnicity. tions in the same model (see Marsh et al., 2009; Marsh, Nagengast,
Procedure. Data were collected as part of an ongoing study & Morin, 2012).
of music preferences involving volunteers assessed over the We took advantage of ESEM to accomplish the goals of this
Internet (www.outofservice.com/music-personality-test/). The study. Specifically, to determine whether the MUSIC model is
website where the data were collected is noncommercial and invariant across age, we first assessed the structural validity of a
advertisement-free, and contains a variety of psychology mea- five-factor model of music preferences. Next, measurement invari-
sures. Potential respondents could find out about the site through ance was examined with three age groups (age 1219, age 20 39,
several channels, including search engines or unsolicited links on and age 40 65) under the ESEM analytic framework. We then
other websites. The data reported in the current research were followed the measurement invariance steps suggested by Brown
collected between 2003 and 2010. Respondents volunteered to (2006) by first testing for configural invariance, then weak (metric)
participate in the study by clicking on the music test icon and invariance, and finally strong invariance. Configural invariance
were then presented with a series of questions about their music serves to confirm the equivalence of the factor structure across
preferences, personalities, and demographics. After responding to groups, weak invariance ensures that the items within each factor
each item and submitting their responses, participants were pre- have comparable meanings across groups, and strong invariance
sented with a customized personality evaluation based on their warrants mean-level comparisons across groups. To confirm mea-
responses to the music survey. surement invariance, we used Multiple Indicators Multiple Causes
Measures. Music preferences were assessed using a revised (MIMIC) within ESEM. Using MIMIC, the covariates can be
version of the Short Test of Music Preferences (STOMP-R; Rent- regressed on both the latent factor and the items, thus allowing
frow & Gosling, 2003). The STOMP-R comprises 23 music further tests of intercept invariance. As a measurement invariance
genres: alternative, bluegrass, blues, classical, country, electronica/ method, MIMIC is advantageous compared with multiple-group
dance, folk, funk, gospel, heavy metal, international/world, jazz, analysis in that it preserves the continuous nature of the covariate
new age, oldies, opera, pop, punk, rap, reggae, religious, rock, instead of dividing a continuous variable into categories (Morin et
soul/R&B, and soundtracks. Respondents were asked to report al., 2013). Finally, to examine age trends in musical preferences
their degree of liking for each genre using a 7-point rating scale and to gauge the impact of personality on those trends, we re-
with endpoints at 1 (Dislike) and 7 (Like). Two items, soundtracks gressed age and its polynomial increments, along with gender,
and oldies, encompass music of many different styles and from personality, and their interactions on the music-preference dimen-
different time periods, and it is very likely that these items might sions, using SEM.
be interpreted differently by people of different ages. We thus We followed the recommendations of Marsh et al. (2009, 2012)
removed these two items from further analyses and relied only on to evaluate the goodness of fit of the models tested. Accordingly,
the remaining 21 genres. a model with acceptable fit should have a comparative fit index
6 BONNEVILLE-ROUSSY, RENTFROW, XU, AND POTTER

(CFI) and Tucker-Lewis index (TLI) .90, and models with ized factor loadings of the ESEM, quartimin rotation, are reported
excellent fit should have fit statistics .95. Additionally, the in Table 1.
root-mean-square error of approximation (RMSEA) and standard- The next step in our analyses was to examine configural, weak,
ized root-mean-square residuals (SRMR) should be .08 for and strong invariance of the MUSIC factor structure between age
acceptable and .05 for excellent model fit. Standard evaluation groups. We first split the sample into three groups representing
of measurement invariance involves the comparison of chi-square Eriksons (1950) three primary life stages: adolescence (1219;
values (Brown, 2006), but chi-square measures are largely depen- n 118,605), young adulthood (20 39; n 112,343), and middle
dent on sample size and penalize large samples. Given the unchar- adulthood (40 65; n 23,877). We next performed a series of
acteristically large sample size used in this study, chi-square sta- increasingly restrictive models, starting with the configural model,
tistics are not presented. To compare the adequacy of a more then comparing the weak invariance model with the configural
restrictive model in measurement invariance, Chen (2007) recom- model, and then the strong invariance model with the weak invari-
mended that .010 in CFI or TLI, complemented by a change ance model.
of less than .015 in RMSEA or .030 in SRMR, reflects good model As can be seen in the first data row in Table 2, the assumption
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

fit. It is worth noting that because ESEM is a relatively new of configural invariance between age groups was met. Three of the
This document is copyrighted by the American Psychological Association or one of its allied publishers.

measurement approach, model-fit comparison for invariance four change indices of weak invariance were also within the
should be used as guidelines and not as strict thresholds. thresholds, indicating that metric invariance was also met, as
shown in the second data row. Strong invariance was then tested,
Results and Discussion but the differences between the models were above threshold, as
Age invariance of the MUSIC model. The first step in our shown in the third data row. Post hoc modification indices indi-
analyses was to examine the factor structure of musical prefer- cated that rap and funk music displayed small evidence of differ-
ences. Given previous evidence that individual differences in mu- ential item functioning across groups. When the intercepts of these
sical preferences can be characterized in terms of five factors, we two genres were freed, the general five-factor structure remained
performed an EFA using quartimin rotation within the ESEM unchanged. We found partial strong invariance when the intercepts
framework. We expected a priori that some residuals of the ob- of rap and punk were freely estimated, as shown in the last data
served music genres that share similar components would be row. To further assess intercept measurement invariance using
correlated (e.g., classical and opera, or jazz and blues). Failure to MIMIC, we regressed age, age quadratic (age2), and age cubic
take into account such correlated residuals can lead to bias in the (age3) on the MUSIC factors and the 21 items. This model fit the
factor correlations (Jreskog, 1979; Marsh & Hau, 1996), so using data well (CFI .950, TLI .912, RMSEA .048, SRMR
procedures consistent with other research (e.g., Marsh et al., 2013) .022). Post hoc examination of the modification index suggested
we allowed some residuals to correlate. The results with five latent that rap and punk music genres could be predicted by the age
factors provided satisfactory to excellent fits to our data (CFI covariates, indicating potential noninvariance in item intercepts.
.968, TLI .937, RMSEA .045, SRMR .018). The standard- The addition of direct paths between age, age2, and age3, and rap

Table 1
Exploratory Structural Equation Model With Quartimin Rotation of 21 Music Genres

Five quartimin-rotated factors


Genre Mellow Unpretentious Sophisticated Intense Contemporary

Electronica/Dance .54 .01 .16 .09 .24


World/International .50 .06 .31 .01 .00
New Age .43 .14 .04 .16 .01
Pop .14 .63 .16 .05 .22
Country .14 .52 .17 .03 .04
Religious .08 .38 .17 .12 .08
Blues .11 .02 .79 .06 .12
Jazz .15 .06 .64 .05 .11
Bluegrass .00 .10 .61 .06 .02
Folk .15 .12 .55 .06 .18
Classical .34 .04 .49 .08 .23
Gospel .00 .29 .41 .14 .11
Opera .32 .08 .38 .08 .16
Rock .13 .10 .10 .71 .10
Punk .02 .07 .07 .68 .08
Alternative .17 .05 .01 .66 .02
Heavy Metal .01 .24 .01 .40 .06
Rap .06 .09 .18 .00 .69
Soul/R&B .04 .26 .21 .14 .65
Funk .13 .12 .36 .15 .53
Reggae .09 .10 .38 .07 .40
Note. N 254,825. Coefficients are standardized. Primary factor loadings are in bold typeface.
AGE TRENDS IN MUSICAL ENGAGEMENT AND PREFERENCES 7

Table 2
Multiple-Group Invariance of Age

Model CFI TLI RMSEA SRMR CFI TLI RMSEA SRMR

Invariance M1 .965 .931 .047 .020


M2 .950 .935 .045 .029 .015 .004 .002 .009
M3 .935 .921 .050 .036 .015 .014 .005 .007
M3a .940 .926 .048 .034 .010 .009 .003 .005
Note. M1M3 multiple-group models. Age groups are 1219; 20 39; 40 65. M1 configural model; M2 weak (metric) invariance; M3 strong
(intercept) invariance. M3a partial strong invariance with intercepts of Rap and Funk freely estimated; Delta coefficients are the absolute differences
between the models. M2 is compared with M1; M3 and M3a are compared with M2. CFI comparative fit index; TLI Tucker-Lewis index; RMSEA
root-mean-square error of approximation; SRMR standardized root-mean-square residual.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

and funk did not lead to a significant increase in model fits (CFI sions on music preferences. All in all, 27 regression paths per
.957, TLI .921, RMSEA .046, SRMR .019). This indicated MUSIC dimension were examined (nine main effects and 18
that the items intercepts were invariant across age measured as a moderation effects), for a total of 135 paths.
continuous variable, providing further support for strong measure- For this series of analyses, factor scores for the MUSIC prefer-
ment invariance. ence dimensions were computed and then standardized to a mean
Taken together, tests of measurement invariance using multiple- of zero and a standard deviation of one to facilitate comparison,
group ESEM showed support for age invariance of musical pref- gender was contrast-coded (0 males, 1 females), and the
erences. The results from the analyses yield a factor structure that personality variables were standardized. The large sample size
resembles the MUSIC model reported in Rentfrow, Goldberg, and provided sufficient statistical power to detect small effects, but to
Levitin (2011), Rentfrow et al. (2012), and Rentfrow, Goldberg, avoid accepting regression effects that are unlikely to be replicated
and Zilca (2011). The factor loadings in the first data column of using smaller samples, we only focus on regression paths with
Table 1 show large loadings for electronica/dance, world, and new |.10|. These paths represent the unique contribution of the inde-
age, styles that are perceived as relaxing, unaggressive, and atmo- pendent variables on the MUSIC factors and are far more conser-
spheric, qualities consistent with the Mellow preference dimen- vative than Pearson correlations. To test the age-only model, all
sion. Genres with large loadings on the second factor were pop, 135 regression paths were specified, but only age and the higher
country, and religious, styles that typically have vocals and are order age polynomials were freely estimated; the other main ef-
perceived as uncomplicated, unaggressive, and simple, consistent fects and interactions were constrained to zero. The results indi-
with the Unpretentious dimension. The third factor included blues, cated that the age-only model did not fit the data well (CFI .735,
jazz, bluegrass, folk, classical, opera, and gospel, styles that gen- TLI .680, RMSEA .050, SRMR .026).
erally use acoustical instruments; are clear sounding; and are We next tested the main-effects model by allowing the age
perceived as intelligent, deep, inspiring, and complex, consistent variables, gender, and all five personality traits to be freely esti-
with the Sophisticated dimension. The fourth factor included rock, mated simultaneously. The results indicated that this model
punk, alternative, and heavy metal, genres characterized by electric, achieved an excellent fit (CFI .981, TLI .970, RMSEA
loud, and distorted instruments, and perceived as aggressive, tense, .015, SRMR .006).
and unromantic, consistent with the Intense dimension. And the fifth Finally, we tested the moderation effects model. Allowing the
factor included rap, soul/R&B, funk, and reggae, styles that are 135 paths to be freely estimated would yield a just-identified
percussive and electric and perceived as upbeat, danceable, and not model, with zero degrees of freedom. We relied on model modi-
sad, consistent with the Contemporary dimension. The current fication indices obtained from the main-effects model to assess the
factors clearly resemble the MUSIC model. Therefore, we labeled relevance of studying interaction effects on music preferences.
the factors in the current study accordingly. Given evidence for Comparatively large modification indices were found for the in-
partial strong invariance, the factor scores were saved and used to teractions between age and gender for the Intense music factor, so
examine age trends in musical preferences. we allowed the interaction terms of age, age2, and age3 with
Age trends in musical preferences and their relationships gender to be regressed on Intense. Including these terms, however,
with gender and personality. To investigate age trends in mu- did not yield fit statistics that were significantly better than the
sical preferences and to examine the impact of gender and person- main-effects model (CFI .985, TLI .976, RMSEA .014,
ality on those trends, we conducted a series of regression analyses SRMR .005). Furthermore, the expected path coefficients were
within the SEM framework by specifying regression paths be- below our regression coefficient benchmark (all s |.09|, SE
tween age, age2, age3, gender, and each of the Big Five personality .009).
traits, and their interactions. We first tested the age-only model to The aforementioned results provided confirmation that there are
see whether the age variables alone provide a good fit for age robust associations between musical preferences, age, gender, and
trends in the MUSIC factors. We next tested the main-effects personality. On the basis of these findings, and for the sake of
model to assess the simultaneous influence of age along with parsimony, we retained the main-effects model for all subsequent
gender and personality on music preferences. And finally, we analyses. Regression paths for the main-effects model with inter-
tested the moderation effects model to examine the interactive action terms fixed to zero are displayed in Table 3. Age trends in
effects of age and gender, and age and the five personality dimen- music preferences are graphed in Figure 1. The significant main
8 BONNEVILLE-ROUSSY, RENTFROW, XU, AND POTTER

Table 3 effects of personality with one standard deviation above and below
Standardized Coefficients of the Path Model of Age Trends, the mean for the MUSIC dimensions are plotted against age in
Gender, and Personality on MUSIC Preferences Figure 2.
Mellow. The results from our analyses supported a cubic
95% CI model for the Mellow music-preference factor. As can be seen in
Model Low High SE p Panel A of Figure 1, preferences for Mellow increased steeply
during adolescence. Early in young adulthood, preferences flat-
Mellow
Intercept .09 .10 .09 .003 .001
tened slightly and began to decrease from age 30 to the early 50s.
Age .16 .15 .16 .003 .001 Preferences for this dimension increased from late 50s through 65.
Age2 .44 .46 .43 .007 .001 Results presented in Table 3 and in Panel A of Figure 2 also
Age3 .31 .29 .32 .007 .001 indicated that Openness was positively related to preferences for
Gender .08 .07 .08 .002 .001 Mellow music, such that participants who were high in Openness
Extraversion .05 .06 .05 .002 .001
Agreeableness .03 .03 .03 .002 .001 scored approximately half of a standard deviation above those who
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Conscientiousness .03 .03 .02 .002 .001 were low in Openness. This difference remained constant as partici-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Emotional Stability .01 .00 .01 .002 .001 pants aged, with an effect size of d .57 (mean low Openness
Openness .19 .19 .20 .002 .001 .25, mean high Openness .23).
Unpretentious
Intercept .28 .29 .28 .003 .001
Unpretentious. The age trends in preferences for Unpreten-
Age .23 .22 .23 .003 .001 tious music are shown in Panel B of Figure 1, and fit a linear
Age2 .17 .18 .15 .007 .001 model. That is, preferences for Unpretentious music increased
Age3 .12 .11 .13 .007 .001 from adolescence to young adulthood and on through middle
Gender .24 .24 .25 .002 .001 adulthood. As shown in Table 3, women displayed stronger
Extraversion .10 .10 .11 .002 .001
Agreeableness .13 .12 .13 .002 .001 preferences for this factor compared with men (d .52; mean
Conscientiousness .11 .10 .11 .002 .001 men .24, mean women .17). Extraversion, Agreeable-
Emotional Stability .02 .02 .01 .002 .001 ness, and Conscientiousness were all positively related to pref-
Openness .09 .09 .08 .002 .001 erences for Unpretentious music, such that people high on those
Sophisticated
Intercept .05 .04 .05 .003 .001
traits scored approximately one third to one half a standard
Age .43 .42 .44 .003 .001 deviation higher in their preferences compared with people low
Age2 .43 .44 .41 .007 .001 on those traits. These personality differences are plotted in
Age3 .28 .27 .30 .007 .001 Panels B1B3 of Figure 2. They remained constant with age
Gender .04 .04 .04 .002 .001 (Extraversion d .32 [mean low Extraversion .15, mean
Extraversion .04 .05 .04 .002 .001
Agreeableness .04 .03 .04 .002 .001 high Extraversion .11], Agreeableness d .61 [mean low
Conscientiousness .05 .05 .05 .002 .001 Agreeableness .30, mean high Agreeableness .19], Con-
Emotional Stability .06 .05 .06 .002 .001 scientiousness d .53 [mean low Conscientiousness .20,
Openness .18 .17 .18 .002 .001 mean high Conscientiousness .23]).
Intense
Intercept .01 .01 .02 .003 .001
Sophisticated. As can be seen Panel C of Figure 1, the age
Age .12 .12 .11 .003 .001 trend for the Sophisticated factor was fit best by a quadratic
Age2 .02 .01 .04 .007 .003 model. Preferences for Sophisticated music increased sharply
Age3 .13 .14 .11 .007 .001 during adolescence to early young adulthood, where they con-
Gender .01 .01 .01 .002 .001 tinued to increase, but at a slower rate, until the 50s when they
Extraversion .02 .03 .02 .002 .001
Agreeableness .02 .03 .02 .002 .001 appeared to stabilize. As displayed in Table 3, and Panel C of
Conscientiousness .10 .11 .10 .002 .001 Figure 2, the results also revealed that preferences for Sophis-
Emotional Stability .02 .03 .02 .002 .001 ticated music were positively related to Openness, such that
Openness .15 .15 .16 .002 .001 participants who were high in Openness scored approximately
Contemporary
Intercept .08 .08 .07 .003 .001
one half of one standard deviation above those who scored low,
Age .10 .10 .11 .003 .001 with an effect size of d .49 (mean low Openness .25,
Age2 .22 .23 .20 .007 .001 mean high Openness .20).
Age3 .04 .03 .06 .007 .001 Intense. The trend plotted in Panel D of Figure 1 reveals that
Gender .06 .06 .07 .002 .001 preferences for the Intense music factor were fit best by a quadratic
Extraversion .19 .19 .20 .002 .001
Agreeableness .06 .06 .07 .002 .001 model. Preferences show rapid rise and then fall during adolescence,
Conscientiousness .02 .02 .02 .002 .001 with preferences for Intense music plateauing during young adulthood
Emotional Stability .01 .01 .02 .002 .001 and then decreasing sharply from late young adulthood through mid-
Openness .02 .01 .02 .002 .001 dle adulthood. As shown in Table 3, and Panels D1 and D2 of Figure
Note. N 254,825. Age, personality, and the five MUSIC factors are 2, the results also revealed main effects of Conscientiousness and
standardized. Gender is dummy-coded: male 0, female 1. MUSIC Openness, such that participants who scored high in Conscientious-
Mellow, Unpretentious, Sophisticated, Intense, and Contemporary; CI ness liked intense music less than individuals low in this trait (d
confidence interval; Age2 age quadratic; Age3 age cubic. Estimates in
bold typeface exceed the effect-size benchmark of |.10|. All cell entries are
.43; mean low Conscientiousness .16, mean high Conscientious-
from the main-effects path model. ness .22), and those who scored high in Openness consistently
liked Intense music more than did participants who scored low (d
.45; mean low Openness .22, mean high Openness .17).
AGE TRENDS IN MUSICAL ENGAGEMENT AND PREFERENCES 9
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 1. Mean Mellow, Unpretentious, Sophisticated, Intense, and Contemporary preference scores by age,
Study 2. Scores are standardized.

Contemporary. As displayed in Panel E in Figure 1, age trends a difference of just over half a standard deviation throughout adoles-
in preferences for Contemporary music were best fit by a quadratic cence and adulthood (d .63; mean low Extraversion .32, mean
model. The trend lines reveal a rise in preferences for Contemporary high Extraversion .23).
music during adolescence that stabilizes in early young adulthood, Effects of birth cohort on age trends in music preferences.
and then begins to decline steeply from late young adulthood through It is possible that birth-cohort effects are responsible for the age
middle adulthood. As can be seen at the bottom of Table 3, and Panel trends observed in the current study. Because the current study
E of Figure 2, a main effect of Extraversion was also observed, such included data collected over an 8-year period, we were able to
that participants who were more extraverted also displayed greater compare the preferences of participants who were the same age
preferences for the Contemporary dimension, which corresponded to (e.g., 35) when they originally completed the survey, but from
10 BONNEVILLE-ROUSSY, RENTFROW, XU, AND POTTER
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 2. Main effects of age and personality traits on the Mellow, Unpretentious, Sophisticated, Intense, and
Contemporary preference scores, Study 2. Scores represent individuals high and low scores on the personality
traits (1 SD above and below the mean).

different birth cohorts (e.g., 1968, 1969, 1970, 1971, 1972, General Discussion
1973, 1974, and 1975). To rule out possible cohort effects on
the relationships between age and music preferences, we re- The current research sought to fill a gap in the emerging
gressed age, age2, age3, and year of survey completion (i.e., literature on the social-personality psychology of music by
2003, 2004, 2005, 2006, 2007, 2008, 2009, and 2010) on music investigating age differences in how people experience and
preferences and fixed all of the other regression paths to zero. engage with music. Two independent studies involving more
The model provided a poor fit of the data (CFI .736, TLI than a quarter million participants were conducted to examine
.671, RMSEA .050, SRMR .026), and the path coefficients age trends in musical engagement and preferences. Findings
for year of survey completion failed to reach our regression from the studies provide a solid foundation on which to develop
coefficient benchmark ( .075, SE .002). These findings and test hypotheses about the role music plays throughout life.
strongly suggest that pronounced cohort effects did not drive Below we summarize the results and propose hypotheses for
the age trends observed in the current study. future research.
AGE TRENDS IN MUSICAL ENGAGEMENT AND PREFERENCES 11
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 2. (continued)

Summary of Results findings suggest that the music-preference dimensions general-


ize across age and offer additional support for the robustness of
The results from Study 1 shed light on age differences in
the MUSIC model. Results from the regression models revealed
musical engagement. The results show that although music
declines in importance after adolescence, young and middle- clear age trends in music preferences, such that preferences for
aged adults still consider music an important feature of their the Mellow, Unpretentious, and Sophisticated dimensions in-
lives. Findings from this study also suggest that young people creased with age, whereas preferences for Intense and Contem-
spend roughly 20% of their time listening to music, whereas porary declined. The regression models also revealed main
adults spend nearly 13% of their time listening to music (as- effects of personality for all five preference dimensions. The
suming the average person sleeps 8 hr per night). Furthermore, patterns of associations between Extraversion, Agreeableness,
the results indicate that young people listen to music in a variety Conscientiousness, and Openness and the music-preferences
of contexts, whereas adults typically listen to music in private. dimensions replicate previous research (e.g., Rentfrow & Gos-
Study 2 revealed age trends in musical preferences. Results ling, 2003; Zweigenhaft, 2008) and lend further support for the
from multigroup ESEM analyses indicated that the five- conclusion that musical preferences are manifestations of sim-
dimensional music-preference model is age invariant. These ilar psychological characteristics throughout the life span.
12 BONNEVILLE-ROUSSY, RENTFROW, XU, AND POTTER

Why Does Musical Engagement Decline With Age? porary music are most popular during young adulthood because
these styles of music reinforce desires for intimacy and also
Adolescence is a period marked by identity uncertainty and complement the settings where young people come together with
intense pressures to conform (Erikson, 1968; Harter, 1999; Hartup, the goal of establishing close relationships (i.e., house parties,
1989), and music plays an integral part in helping young people dance clubs, bars).
explore their identities and form relationships with peers (Delsing, By middle adulthood, the psychosocial stage that individuals
2008; ter Bogt, Keijsers, & Meeus, 2013). Indeed, adolescents use face is the challenge of developing a career, raising a family, and
music as an identity badge that serves as a symbolic representation maintaining friendships (Erikson, 1968; Hogan & Roberts, 2004).
of their values and beliefs (Frith, 1981). For example, by listening Preferences for Unpretentious and Sophisticated music were high-
to music with friends or in public, individuals are effectively est during this life stage. Both music dimensions are perceived as
making statements about their preferences, beliefs, and lifestyles positive and relaxing, but whereas Unpretentious is more simple
for others to hear (Tarrant et al., 2002). Although such statements and about themes of love and family, Sophisticated is more com-
are indirect, there is consistent evidence that music carries infor- plex and indicative of high culture (Rentfrow et al., 2012). Thus,
mation about the social and psychological characteristics of the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

the relaxing and familial themes of Unpretentious music may be


listeners (Rentfrow & Gosling, 2006, 2007; Rentfrow et al., 2009).
This document is copyrighted by the American Psychological Association or one of its allied publishers.

appealing for adults at a life stage where family life is the focus,
Yet, as individuals transition from adolescence to adulthood, and the aesthetic qualities of Sophisticated music may be appeal-
stable self-concepts are formed (Harter, 2003), the ability to resist ing among individuals preoccupied with the challenge of estab-
peer pressure increases (Steinberg & Monahan, 2007), and identity lishing social status and career success.
becomes invested in newly emerging social roles (e.g., spouse, Personality. Also stemming from a life span perspective of
parent, career professional). Raising a family and pursuing a career personality (McAdams & Olson, 2010), continuity and change in
provide adults with defining features of their identities. It seems particular traits appear to be expressed in musical preferences. Our
reasonable to suggest that the meaning derived from these roles results strongly suggest that there is a moderate association be-
diminishes the function that music serves in shaping identity and tween Extraversion and preferences for Contemporary music
offering fulfillment. The current findings suggest that adults may throughout the life span. This suggests that sociability, positive
not have as much psychologically invested in music as young affect, and enthusiasm are related to preferences for fun and
people, and instead listen to music for purposes of relaxation and sociable music throughout life. A comparatively small, but robust,
entertainment. In this way, adults may rely on music less as a association was observed between Agreeableness and preferences
means for identity development and more for purposes of emotion for Unpretentious. This is consistent with the hypothesis that
regulation or stimulation. individuals of all ages who possess traits associated with warmth
prefer music that is inoffensive and gentle sounding. Although the
Toward a Developmental Theory of Musical associations were comparatively small, Conscientiousness was
positively associated with preferences for Unpretentious but neg-
Preferences
atively related to Intense. Given that Intense has been associated
Although the current results are based entirely on cross- with disorderliness in past research (Rentfrow, Goldberg, & Levi-
sectional data, theory and research in developmental, personality, tin, 2011), it is not surprising that people of all ages who prefer
and biological psychology offers clues for understanding age dif- Intense music are comparatively low in conscientiousness. Finally,
ferences in musical preferences. We may thus draw on that re- Openness was moderately associated with Mellow, Sophisticated,
search and the current findings to lay a foundation for developing and Intense music from adolescence through to middle adulthood.
a theory of why musical preferences change with age. Research has shown that Sophisticated and Intense music are
Life span perspective. Eriksons (1950, 1968) psychosocial perceived as complex and that Mellow and Sophisticated are
stages of development provide a useful framework for understand- perceived as intelligent and thoughtful, psychological characteris-
ing age differences in musical preferences. The current results tics also associated with Openness (Rentfrow, Goldberg, & Levi-
indicated that preferences for the Intense music-preference dimen- tin, 2011; Rentfrow et al., 2012).
sion were highest among adolescents and then declined through In the same way that music reinforces dispositional personality
middle adulthood. This dimension is marked by loud and distorted traits, it is conceivable that normative changes in personality might
sounds and is perceived as aggressive, tense, and antiestablishment contribute to age differences in musical preferences. Several stud-
(Rentfrow, Goldberg, & Levitin, 2011; Rentfrow et al., 2012). The ies indicate that Neuroticism is high and that Agreeableness and
psychosocial stage during adolescence is the challenge of devel- Conscientiousness are low during adolescence (e.g., Allemand et
oping a coherent identity. It thus seems reasonable to argue that the al., 2007; Roberts et al., 2006; Soto et al., 2011)a period marked
rebellious connotations of Intense music might be what adoles- by significant cognitive and biological changes that influence
cents find appealing as they struggle to establish a sense of emotional stability, impulse control, and aggression (Keating,
independence and autonomy. 2004; Kuhn, 2006; Tremblay, 1998). It is conceivable that these
In early adulthood, once self-identities have formed, time and changes contribute to a greater preference for fast and edgy music,
effort are invested in developing intimate bonds of love as it might reinforce or complement adolescents psychological
(Carstensen, 1992; Erikson, 1968). Our results indicated that the and physiological states. The normative increases in Agreeable-
Mellow and Contemporary dimensions were most popular during ness and Conscientiousness correspond with declining preferences
young adulthood. These preference dimensions are considered for Intense music and increasing preferences for Unpretentious
romantic, emotionally positive, and danceable (Rentfrow et al., music, suggesting that as people become more caring and dutiful
2012). Taken together, it is conceivable that Mellow and Contem- with age, their degree of preference for warm and reflective music
AGE TRENDS IN MUSICAL ENGAGEMENT AND PREFERENCES 13

also increases. There is also evidence that Openness increases as music preferences and, at the same time, reveal the relative impact
people age, suggesting that people become more imaginative and of such factors throughout the life span.
aesthetic (Soto et al., 2011), which appears to be reflected in A limitation of both studies was the fact that they relied solely
greater preferences for reflective, imaginative, and unconventional on self-report data. Given the current evidence that music is more
music. The current results are cross-sectional and cannot determine important among adolescents than adults, it is conceivable that
causality. Longitudinal research tracking personality and musical showing passion for music is more socially desirable for young
preferences will be necessary for rigorously testing these hypoth- people. If so, social desirability might have inflated young partic-
eses. ipants reports of how important music was to them. Social desir-
Biological maturation. Another explanation for age differences ability might have also affected young peoples self-reports of
in musical preferences arises from developmental changes in auditory which musical styles they most prefer. Future research that gathers
perception. There is evidence of developmental changes in auditory objective information about musical engagement and preferences
threshold levels and in specific hearing impairments among normally longitudinally would go a very long way in establishing the gen-
aging individuals that could influence the degree to which particular eralizability of the current results.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

musical styles are enjoyed. In terms of normative auditory devel- A potential limitation of Study 2 is that the data were collected
This document is copyrighted by the American Psychological Association or one of its allied publishers.

opment, as people age, they lose the capacity to hear high-pitched on the Internet. It is conceivable that older middle-aged adults who
and soft sounds, and show an increased risk for developing hearing volunteer to complete a survey online may be comparatively high
impairments (Brant & Fozard, 1990). For instance, people with in Openness or savvy with new technologies. To address the
hearing recruitment tend to show higher sensitivity to sound in- possibility of a selection bias, we used a procedure described by
tensity (Buus & Florentine, 2002). Another hearing problem that Soto et al. (2011) to examine age variance in Openness across age
can affect music preferences is hyperacusis, that is, a decreased groups. If there was a selection bias for the older participants, we
tolerance to certain sound frequencies (Anari, Axelsson, Eliasson, should expect less variance in Openness among the older middle-
& Magnusson, 1999). As a result of such changes, perceptions of aged participants (because the majority of them should be similarly
loudness increase more quickly as a function of objective sound high on this trait) than among the younger participants (because
intensity for people with hyperacusis compared to people with they comprise a more representative sample). We tested that hy-
normal hearing function (Hood & Poole, 1966). pothesis by computing the standard deviation ratio of openness
Research has shown that the aforementioned changes in hearing between the three comparison groups (1219 vs. 20 39; 1219 vs.
are relatively frequent and develop over time, most often taking 40 65; and 20 39 vs. 40 65) and did not find evidence of a bias
decades to fully implement (Agrawal, Platz, & Niparko, 2008; in openness. Of course these results do not entirely rule out any
Gordon-Salant, 2005). Thus, it is easy to understand why middle- selection bias, yet the size of our sample and the evidence of small
aged adults were less likely to prefer loud and distorted music than variability between groups suggest there is not a pronounced bias.
adolescents and young adults. For middle-aged people, the Intense We assessed music preferences using the STOMP-R, which
and Contemporary music dimensions, with their mostly large comprises music genre items. Recent research on the structure of
changes in volume or their generally high intensity, might be music preferences indicates that assessing individual differences in
literally painful to hear and thus intolerable (Leek, Molis, Kubli, & music preferences by measuring affective reactions to audio ex-
Tufts, 2008). Consistent with that hypothesis, research by Smith cerpts of actual music could shed more light on age differences in
(1989) indicated that age was negatively related to preferences for preferences (Rentfrow, Goldberg, & Levitin, 2011; Rentfrow et al.,
listening to music at a high intensity. Although older adults dis- 2012). Measuring affective reactions to auditory excerpts instead
played lower auditory threshold levels, they tended to listen to of genre preferences overcomes the potential limitation that par-
music at a lower volume than did younger participants, suggesting ticipants might be unfamiliar with the genre classifications. An-
that loud music might be uncomfortable for older adults. other advantage of examining age differences in preferences for
musical clips is that it allows for studying age trends in preferences
for musical attributes, as well as the impact of auditory thresholds
Limitations and Future Directions
on musical preferences. Such research has the potential to deepen
A limitation with all studies based on cross-sectional designs is our understanding of why particular music dimensions are pre-
that it is never clear whether the age differences observed were ferred more at certain periods of life.
driven by age effects, birth-cohort effects, or period effects (Lucas
& Donnellan, 2011). We took advantage of the multiyear design of
Conclusion
the current research to assess the degree to which cohort effects
might be responsible for the age differences observed in Study 2. In the current work, we examined age differences in musical
We did not detect pronounced cohort effects, though the results engagement and preferences. Our results are the first to compre-
should be interpreted with care due to the cross-sectional nature of hensively document the ways in which individuals experience and
this study, as the 8-year period may not be sufficient for detecting engage with music from adolescence through middle adulthood.
cohort effects. It is possible that less pronounced cohort effects We identified important age differences in the role music plays at
drive the current results. Indeed, given that musical styles evolve various periods in life; we obtained compelling evidence that
and new genres emerge over time, it is only reasonable to expect musical preferences develop throughout adulthood; and we spotted
birth cohort to have some effect of musical preferences. Longitu- normative trends in the styles of music individuals prefer at dif-
dinal studies that gather information about preferences, personal- ferent life stages and how they are related to personality. These
ity, life experiences, and social networks will be invaluable for results replicate and greatly extend our knowledge of the social
teasing apart the social and psychological factors that influence psychology of music by providing a crucial life span perspective
14 BONNEVILLE-ROUSSY, RENTFROW, XU, AND POTTER

that, until now, has been absent. At the same time, the current personality factors of those who listen. Psychomusicology, 19, 3256.
research highlights the real-world relevance of mainstream social- doi:10.1037/h0094035
personality, developmental, and biological psychology by illustrat- Gordon-Salant, S. (2005). Hearing loss and aging: New research findings
ing how basic concepts and theories in these fields can inform our and clinical implications. Journal of Rehabilitation Research and De-
understanding of a facet of everyday life that is important to people velopment, 42(Suppl. 2), 924. doi:10.1682/JRRD.2005.01.0006
Gosling, S., Rentfrow, P., & Swann, W. (2003). A very brief measure of
of all ages. It is only by broadening our research foci and adopting
the Big-Five personality domains. Journal of Research in Personality,
a multidisciplinary perspective that we will develop a thorough
37, 504 528. doi:10.1016/S0092-6566(03)00046-1
and complete understanding of the role of music through the ages. Harter, S. (1999). The construction of the self: A developmental perspec-
tive. New York, NY: Guilford Press.
Harter, S. (2003). The development of self-representations during child-
References
hood and adolescence. In M. Leary & J. Tangney (Eds.), Handbook of
Agrawal, Y., Platz, E. A., & Niparko, J. K. (2008). Prevalence of hearing self and identity (pp. 610 642). New York, NY: Guilford Press.
loss and differences by demographic characteristics among US adults: Hartup, W. W. (1989). Social relationships and their developmental sig-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Data from the National Health and Nutrition Examination Survey, 1999- nificance. American Psychologist, 44, 120 126. doi:10.1037/0003-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

2004. Archives of Internal Medicine, 168, 15221530. doi:10.1001/ 066X.44.2.120


archinte.168.14.1522 Hogan, R., & Roberts, B. W. (2004). A socioanalytic model of maturity.
Allemand, M., Zimprich, D., & Hertzog, C. (2007). Cross-sectional age Journal of Career Assessment, 12, 207217. doi:1.1177/
differences and longitudinal age changes of personality in middle adult- 1069072703255882
hood and old age. Journal of Personality, 75, 323358. doi:1.1111/j Holbrook, M. B., & Schindler, R. M. (1989). Some exploratory findings on
.1467-6494.2006.00441.x the development of musical tastes. Journal of Consumer Research, 16,
Anari, M., Axelsson, A., Eliasson, A., & Magnusson, L. (1999). Hyper- 119 124. doi:10.1086/209200
sensitivity to sound. Scandinavian Audiology, 28, 219 230. doi: Hood, J. D., & Poole, J. P. (1966). Tolerable limit of loudness: Its clinical
10.1080/010503999424653 and physiological significance. Journal of the Acoustical Society of
Arnett, J. J. (1999). Adolescent storm and stress, reconsidered. American America, 40, 4754. doi:10.1121/1.1910062
Psychologist, 54, 317326. doi:10.1037/0003-066X.54.5.317 Jreskog, K. G. (1979). Statistical estimation of structural models in
Bakagiannis, S., & Tarrant, M. (2006). Can music bring people together? longitudinal investigations. In J. R. Nesselroade & B. Baltes (Eds.),
Effects of shared musical preference on intergroup bias in adolescence. Longitudinal research in the study of behavior and development (pp.
Scandinavian Journal of Psychology, 47, 129 136. doi:10.1111/j.1467- 303352). New York, NY: Academic Press.
9450.2006.00500.x Keating, D. (2004). Cognitive and brain development. In R. Lerner & L.
Boer, D., Fischer, R., Strack, M., Bond, M. H., Lo, E., & Lam, J. (2011). Steinberg (Eds.), Handbook of adolescent psychology (pp. 45 84). New
How shared preferences in music create bonds between people. Person- York, NY: Wiley.
ality and Social Psychology Bulletin, 37, 1159 1171. doi:1.1177/ Kuhn, D. (2006). Do cognitive changes accompany developments in the
0146167211407521 adolescent brain? Perspectives on Psychological Science, 1, 59 67.
Brant, L. J., & Fozard, L. J. (1990). Age changes in pure-tone hearing doi:10.1111/j.1745-6924.2006.t01-2-.x
thresholds in a longitudinal study of normal human aging. Journal of the Laukka, P. (2007). Uses of music and psychological well-being among the
Acoustical Society of America, 88, 813 820. doi:10.1121/1.399731
elderly. Journal of Happiness Studies, 8, 215241. doi:10.1007/s10902-
Brown, T. A. (2006). Confirmatory factor analysis for applied research.
006-9024-3
New York, NY: Guilford Press.
Leek, M. R., Molis, M. R., Kubli, L. R., & Tufts, J. B. (2008). Enjoyment
Buss, D. M. (1987). Selection, evocation, and manipulation. Journal of
of music by elderly hearing-impaired listeners. Journal of the American
Personality and Social Psychology, 53, 1214 1221. doi:10.1037/0022-
Academy of Audiology, 19, 519 526. doi:10.3766/jaaa.19.6.7
3514.53.6.1214
Lonsdale, A. J., & North, A. C. (2011). Why do we listen to music? A uses
Buus, S., & Florentine, M. (2002). Growth of loudness in listeners with
and gratifications analysis. British Journal of Psychology, 102, 108
cochlear hearing losses: Recruitment reconsidered. JARO-Journal of the
134. doi:1.1348/000712610x506831
Association for Research in Otolaryngology, 3, 120 139. doi:10.1007/
Lucas, R. E., & Donnellan, M. B. (2011). Personality development across
s101620010084
Carstensen, L. L. (1992). Social and emotional patterns in adulthood: the life span: Longitudinal analyses with a national sample from Ger-
Support for socioemotional selectivity theory. Psychology and Aging, 7, many. Journal of Personality and Social Psychology, 101, 847 861.
331338. doi:1.1037/0882-7974.7.3.331 doi:1.1037/a0024298
Chen, F. F. (2007). Sensitivity of goodness of fit indexes to lack of Marsh, H. W., & Hau, K.-T. (1996). Assessing goodness of fit: Is parsi-
measurement invariance. Structural Equation Modeling: A Multidisci- mony always desirable? Journal of Experimental Education, 64, 364
plinary Journal, 14, 464 504. doi:10.1080/10705510701301834 390.
Colley, A. (2008). Young peoples musical taste: Relationship with gender Marsh, H. W., Muthn, B., Asparouhov, T., Ldtke, O., Robitzsch, A.,
and gender related traits. Journal of Applied Social Psychology, 38, Morin, A. J. S., & Trautwein, U. (2009). Exploratory structural equation
2039 2055. doi:10.1111/j.1559-1816.2008.00379.x modeling, integrating CFA and EFA: Application to students evalua-
Delsing, M., ter Bogt, T., Engels, R., & Meeus, W. (2008). Adolescents tions of university teaching. Structural Equation Modeling, 16, 439
music preferences and personality characteristics. European Journal of 476. doi:10.1080/10705510903008220
Personality, 22, 109 130. doi:1.1002/per.665 Marsh, H. W., Nagengast, N., & Morin, A. J. S. (2012). Measurement
Erikson, E. H. (1950). Childhood and society. New York, NY: Norton. invariance of Big-Five factors over the life span: ESEM tests of gender,
Erikson, E. H. (1968). Identity: Youth and crisis. New York, NY: Norton. age, plasticity, maturity, and La Dolce Vita effects. Developmental
Frith, S. (1981). Sound effects. Youth leisure and the politics of rock n Psychology. Advance online publication. doi:10.1037/a0026913
roll. New York, NY: Pantheon. McAdams, D. P., & Olson, B. D. (2010). Personality development: Con-
George, D., Stickle, K., Rachid, F., & Wopnford, A. (2007). The associ- tinuity and change over the life course. Annual Review of Psychology,
ation between types of music enjoyed and cognitive, behavioral, and 61, 517542. doi:10.1146/annurev.psych.093008.100507
AGE TRENDS IN MUSICAL ENGAGEMENT AND PREFERENCES 15

McClelland, G. H., & Judd, C. M. (1993). Statistical difficulties of detect- Processes & Intergroup Relations, 12, 329 344. doi:10.1177/
ing interactions and moderator effects. Psychological Bulletin, 114, 1368430209102845
376 390. doi:10.1037/0033-2909.114.2.376 Roberts, B. W., Walton, K. E., & Viechtbauer, W. (2006). Patterns of
Morin, A. J. S., Marsh, H. W., & Nagengast, B. (2013). Exploratory mean-level change in personality traits across the life course: A meta-
structural equation modeling. In G. R. Hancock & R. O. Mueller (Eds.), analysis of longitudinal studies. Psychological Bulletin, 132, 125. doi:
Structural equation modeling: A second course (2nd ed., pp. 395 436). 1.1037/0033-2909.132.1.1
Charlotte, NC: Information Age Publishing. Roberts, B. W., Wood, D., & Smith, J. L. (2005). Evaluating Five Factor
Muthn, L. K., & Muthn, B. O. (1998 2012). Mplus users guide. Los Theory and social investment perspectives on personality trait develop-
Angeles, CA: Author. ment. Journal of Research in Personality, 39, 166 184. doi:10.1016/j
North, A. C., & Hargreaves, D. J. (1995). Eminence in pop music. Popular .jrp.2004.08.002
Music and Society, 19, 41 66. doi:1.1080/03007769508591606 Schfer, T., & Sedlmeier, P. (2009). From the functions of music to music
North, A. C., Hargreaves, D. J., & Hargreaves, J. J. (2004). Uses of music preference. Psychology of Music, 37, 279 300. doi:1.1177/
in everyday life. Music Perception, 22, 4177. doi:10.1525/mp.2004.22 0305735608097247
.1.41 Selfhout, M. H. W., Branje, S. J. T., ter Bogt, T. F. M., & Meeus, W. H. J.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Rentfrow, P. J. (2012). The role of music in everyday life: Current (2009). The role of music preferences in early adolescents friendship
formation and stability. Journal of Adolescence, 32, 95107. doi:
This document is copyrighted by the American Psychological Association or one of its allied publishers.

directions in the social psychology of music. Social and Personality


Psychology Compass, 6, 402 416. doi:10.1111/j.1751-9004.2012 10.1016/j.adolescence.2007.11.004
Smith, D. (1989). Preferences for differentiated frequency loudness levels
.00434.x
in older adult music listening. Journal of Music Therapy, 26, 18 29.
Rentfrow, P. J., Goldberg, L. R., & Levitin, D. J. (2011). The structure of
Soto, C. J., John, O. P., Gosling, S. D., & Potter, J. (2011). Age differences
musical preferences: A five-factor model. Journal of Personality and
in personality traits from 10 to 65: Big five domains and facets in a large
Social Psychology, 100, 1139 1157. doi:1.1037/a0022406
cross-sectional sample. Journal of Personality and Social Psychology,
Rentfrow, P. J., Goldberg, L. R., Stillwell, D. J., Kosinski, M., Gosling,
100, 330 348. doi:1.1037/a0021717
S. D., & Levitin, D. L. (2012). The song remains the same: A replication
Steinberg, L., & Monahan, K. C. (2007). Age differences in resistance to
and extension of the MUSIC model. Music Perception, 30, 161185.
peer influence. Developmental Psychology, 43, 15311543. doi:
doi:10.1525/mp.2012.30.2.161
10.1037/0012-1649.43.6.1531
Rentfrow, P. J., Goldberg, L. R., & Zilca, R. (2011). Listening, watching,
Tarrant, M., North, A. C., & Hargreaves, D. J. (2002). Youth identity and
and reading: The structure and correlates of entertainment preferences. music. In R. A. R. Macdonald, D. J. Hargreaves, & D. Miell (Eds.),
Journal of Personality, 79, 223258. doi:1.1111/j.1467-6494.201 Musical identities (pp. 134 150). New York, NY: Oxford University
.00662.x Press.
Rentfrow, P. J., & Gosling, S. D. (2003). The do re mis of everyday life: Tekman, H. G., & Hortasu, N. (2002). Music and social identity: Stylistic
The structure and personality correlates of music preferences. Journal of identification as a response to musical style. International Journal of
Personality and Social Psychology, 84, 1236 1256. doi:1.1037/0022- Psychology, 37, 277285. doi:1.1080/00207590244000043
3514.84.6.1236 ter Bogt, T. F. M., Keijsers, L., & Meeus, W. H. J. (2013). Early adolescent
Rentfrow, P. J., & Gosling, S. D. (2006). Message in a ballad: The role of music preferences and minor delinquency. Pediatrics, 131, e380 e389.
music preferences in interpersonal perception. Psychological Science, doi:10.1542/peds.2012-0708
17, 236 242. doi:1.1111/j.1467-928.2006.01691.x Tremblay, R. E. (1998). Testosterone, physical aggression, dominance, and
Rentfrow, P. J., & Gosling, S. D. (2007). The content and validity of physical development in early adolescence. International Journal of
music-genre stereotypes among college students. Psychology of Music, Behavioral Development, 22, 753777. doi:1.1080/016502598384153
35, 306 326. doi:10.1177/0305735607070382 Zweigenhaft, R. L. (2008). A do re mi encore: A closer look at the
Rentfrow, P. J., & McDonald, J. A. (2010). Preference, personality, and personality correlates of music preferences. Journal of Individual Dif-
emotions. In P. N. Juslin & J. Sloboda (Eds.), Handbook of music and ferences, 29, 4555. doi:1.1027/1614-0001.29.1.45
emotions: Theory, research, applications (pp. 669 695). New York,
NY: Oxford University Press. Received August 21, 2012
Rentfrow, P. J., McDonald, J. A., & Oldmeadow, J. A. (2009). You are Revision received May 13, 2013
what you listen to: Young peoples stereotypes about music fans. Group Accepted June 19, 2013

Vous aimerez peut-être aussi