Music Preference, Social Identity, and Self-Esteem

Author(s): Daniel Shepherd and Nicola Sigg

Source: Music Perception: An Interdisciplinary Journal , Vol. 32, No. 5 (June 2015), pp. 507-514

Published by: University of California Press

Stable URL: https://www.jstor.org/stable/10.1525/mp.2015.32.5.507

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms

University of California Press is collaborating with JSTOR to digitize, preserve and extend access to Music Perception: An Interdisciplinary Journal

This content downloaded from �����������68.230.212.70 on Wed, 31 May 2023 01:01:32 +00:00������������

All use subject to https://about.jstor.org/terms

MUS IC PREFERENCE, SO CIA L IDENTIT Y, AND SE L F-ESTEEM

DA NI E L SHEPHERD & NICOLA SIGG

Auckland University of Technology, Auckland,New Zealand.

SOCIAL IDENTITY THEORY POSITS THAT MEMBERSHIP

to social groups serves to enhance and maintain self-esteem. In young people music plays a prominent role indefining social identity, and so a relationship betweenmusic preference and self-esteem is expected, but is asyet unconfirmed by the literature. The objective of thisstudy was to further examine the association betweenmusic preference and the self-esteem, and to applysocial identity theory to differences in music prefer-ences and self-esteem. The present study measuredself-esteem from university students (n ¼ 199) usingRosenberg’s (1965) self-esteem scale, and employedconfirmatory factor analysis to derive a representativemodel of the self-esteem data. Music preference scoresfor clusters of music genres were found to significantlycorrelate with self-esteem. Furthermore, some mea-sures of group differentiation based on music prefer-ence were significantly associated with self-esteem, butthe relationships differed depending on gender. Over-all, the results provided both support and challengesfor social identity theory.

Received: April 11, 2014, accepted October 27, 2014.

Key words: music preference, self-esteem, music genres,social identity theory, Rosenberg self-esteem scale

M USIC PREFERENCE IS A WAY OF SIGNALING

social identity (Tarrant, North, & Hargreaves,2001) and constitutes a window to an indivi-

dual’s identity (Steele & Brown, 1995). Self-esteem,a commonly measured construct in psychologicalresearch, can be influenced by social processes. Socialidentity theory (Tajfel, 1978) proposes that group mem-bership endows individuals with social identity, and thatgroup identification (i.e., ingroup membership)strengthens and maintains self-esteem through ongoingpositive evaluations of the ingroup. Furthermore, indi-viduals compare their ingroups to other groups (i.e.,outgroups), and use the outcome of these comparisonsto maintain positive social identity and self-esteem

through in-group favoritism, outgroup derogation, andpositive distinction from the outgroup (Tarrant et al.,2001). For adolescents, and young adults especially,there is often genuine and meaningful categorizationof ingroups and outgroups along the lines of musicpreference (North, Hargreaves, & O’Neill, 2000). Studies(e.g., Tarrant, 1999) and commentaries (e.g., Zillmann &Gan, 1997) have reinforced the notion that music pref-erence forms the dominant analysis when adolescents aremaking group comparisons, and hence music preferencemanifests a prominent dimension of adolescents’ socialidentity (Tarrant et al., 2001).

Though music has been coupled with well-being sinceancient times, there is very little research on the topic(Laukka, 2007). This is regrettable, as many young peo-ple in modern society consider music to be essential totheir well-being (Steele & Brown, 1995), and there isa growing concern that a preference for some musicgenres can negatively impact social behavior (Greite-meyer, 2009) and self-esteem (Baker & Bor, 2008).Some studies have touched on the association betweenself-esteem and music (e.g., North & Hargreaves, 1999;Tarrant et al., 2001), but few have directly examinedthe relationship between them. Tarrant and colleaguesdemonstrated that, for a sample of male adolescents,intergroup discrimination on the basis of music prefer-ence was related to self-esteem. They tested the self-esteem hypothesis (Abrams & Hogg, 1988), which wasdeduced from the central tenants of social identity the-ory, and suggests that intergroup discrimination andself-evaluation are intimately linked. The self-esteemhypothesis, which must be tested using a within-subjects design (Abrams & Hogg, 1988), consists of twodirectional hypotheses. First, that successful intergroupdiscrimination may lead to an increase in self-esteemand, second, low or threatened self-esteem may moti-vate increased intergroup discrimination. In support ofthe self-esteem hypothesis, Tarrant et al. (2001) reportedthat greater levels of discrimination and outgroup der-ogation were negatively associated with self-esteem, andthat ingroup favoritism was positively associated withself-esteem. Furthermore, they found that ingroupswere associated with positively evaluated music, andoutgroups with negatively evaluated music. Therefore,social identity theory presents a useful frameworkfor examining the relationship between self-esteem and

Music Perception, VOLUME 32, ISSUE 5, PP. 507–514, ISSN 0730-7829, ELECTRONIC ISSN 1533-8312. © 2015 BY THE REGENTS OF THE UNIVERSIT Y OF CALIFORNIA ALL

RIGHTS RESERVED. PLEASE DIRECT ALL REQUESTS FOR PERMISSION TO PHOTOCOPY OR REPRODUC E ARTICLE CONTENT THROUGH THE UNIVERSIT Y OF CALIFORNIA PRESS’S

RIGHTS AND PERMISSIONS WEBSITE, HT TP://WWW.UCPRESSJOURNALS.COM/REPRINTINFO.ASP. DOI: 10.1525/MP.2015.32.5.507

Music Preference, Social Identity, and Self-Esteem 507

This content downloaded from �����������68.230.212.70 on Wed, 31 May 2023 01:01:32 +00:00������������

All use subject to https://about.jstor.org/terms

music preference, the latter of which can be considereda component of an individual’s social identity (Tekman &Hortaçsu, 2002).

Indirect evidence that music is related to self-esteemcomes from Tekman and Hortaçsu (2002), describinga music style (‘‘Arabesk’’) whose listeners were depictedby the study’s participants as alienated, pessimistic,defeated, and ‘‘ . . . without hope or power to controltheir lives.’’ (p. 283). In the most direct examinationof the association between music preference and self-esteem, Rentfrow and Gosling (2003) failed to uncoversignificant correlations, implying that perceived self-worth has no effect on music preference, or vice versa.They used the Rosenberg Self-Esteem Scale (RSES;Rosenberg, 1965), a scale purporting to measure a uni-dimensional conceptualization of self-esteem con-structed from more specific facets of self-evaluation.The factor structure of the RSES has come under muchscrutiny (e.g., Baranik et al., 2008; Gray-Little, Williams,& Hancock, 1997), and commonly, a two-factor struc-ture is found. Depending on the sample, the two factorsmight reflect either positively (e.g., I take a positive atti-tude towards myself) and negatively (e.g., I certainly feeluseless at times) worded items. Others suggest that thetwo factors manifest a fundamental dichotomy of self-competence/self-assessment (e.g., I am able to do thingsas well as most other people), reflecting an objective formof self-evaluation based on instrumental value, and self-liking/self-acceptance (e.g., At times I think I am no goodat all), reflecting a more subjective form of self-evaluationbased on intrinsic value (Tafarodi & Milne, 2002).

In addition to using the RSES, Rentfrow and Gosling(2003) also measured self-view, and reported significantcorrelations between self-view and music preference.Their self-view measure assessed participant’s self-reports of athletic prowess, intelligence, and physicalattractiveness, all of which can be considered a form ofeither self-competence or self-liking. If self-competenceand self-liking are dimensions of self-esteem (Tafarodi &Swann, 2001), then one would also expect a relationshipbetween music preference and the RSES. That Rentfrowand Gosling failed to report a significant relationshipbetween music preference and the RSES indicates thatthe operationalization of self-esteem should be carefullyconsidered, and perhaps treated as a multidimensionalconstruct such as that uncovered by Tafarodi and Swann(2001). Additionally, the importance of gender as a deter-minant of music preference has yet to be resolved in theliterature. While some have emphasized substantial gen-der differences in music preference (Schwartz & Fouts,2003), others (e.g., Tekman & Hortaçsu, 2002) havereported no effect of gender. For example, Rentfrow and

Gosling (2003) reported no differences between malesand females when originally developing their four-factor music preference scale but, when updating theirscale to a five-factor model, reported significant genderdifferences in preferences to specific pieces of music(Rentfrow, Goldberg, & Levitin, 2011).

The objectives of the present study were to examinethe structure of the RSES, and then to use the RSES totest predictions of social identity theory. To this end theRSES will be factor-analyzed in order to derive a validrepresentation of self-esteem, where the emergence ofself-liking and self-competency factors would supportthe analysis of Tafarodi and colleagues (2001, 2002).The most psychometrically sound representation of theRSES will then be related to music preference, and fur-thermore, to biases towards specific music genres. Socialidentity theory attempts to account for the associationbetween self-esteem and intergroup discrimination byhypothesizing that stronger intergroup discriminationincreases self-esteem by strengthening one’s sense ofsocial identity. Specifically, when social identity isstrongly coupled to a preferred social category (or sub-group), individuals tend to behave in a way that mini-mizes within-group differences (i.e., converge upon thegroup prototype) which, by association, induces positiveself-evaluation. Consistent with social identity theory,individuals preferring particular music styles over otherswould be expected to identify with a particular subgroupbased on music preference (i.e., higher ingroup bias andoutgroup discrimination), and thus evaluate themselvesmore positively (i.e., have higher self-esteem). Further-more, because the effects of gender on the relationshipbetween music preference and self-esteem have yet to besufficiently examined, the data from males and femaleswill be scrutinized independently.

Method

PARTICIPANTS

The participants were 199 New Zealand university stu-dents. The sample consisted of 199 students: 42 males(M¼ 18.7 years, SD¼ 3.16) and 157 females (M¼ 19.24years, SD ¼ 3.61).

MEASURES

Two surveys were distributed, the RSES (Rosenberg,1965), and the Short Test of Music Preference (STOMP:Rentfrow & Gosling, 2003). The RSES is a self-reportinventory containing ten statements, each rated ona four-point Likert-type scale (0 ¼ ‘‘strongly disagree’’to 3 ¼ ‘‘strongly agree’’). The ten ratings are thensummed to provide a total score of self-esteem, ranging

508 Daniel Shepherd & Nicola Sigg

This content downloaded from �����������68.230.212.70 on Wed, 31 May 2023 01:01:32 +00:00������������

All use subject to https://about.jstor.org/terms

from 0 to 30, with a score of 15 or below indicatinga low level of self-esteem (Rosenberg, 1965). Whenresponding to the STOMP scale, participants wererequired to indicate, on a seven-point Likert-type scale,how much they enjoyed a particular music genre (1 ¼‘‘not at all’’ to 7 ¼ ‘‘very much’’). The genres covered bythe STOMP are classical, jazz, blues, rock, heavy metal,country, pop, folk, alternative, religious, soundtracks,rap/hip-hop, soul/funk, and electric/dance.

PROCEDURE

The surveys were distributed during lecture time oncethe respondents were informed that participation wasvoluntary and anonymous, and that ethical approvalhad been sought and granted by the University’s EthicsCommittee.

Results

Item responses were entered into SPSS, with negativelyworded items being recoded in Microsoft Excel prior toanalysis. All analyses were conducted in SPSS, with theexception of the confirmatory factor analyses, whichwere undertaken in LISREL 8.8.

SELF-ESTEEM (RSES)

The mean total RSES score was 19.38 (SD¼ 4.78, min¼10, max ¼ 30), and a Cronbach’s alpha (ac) of 0.86 wascalculated. The average male (M ¼ 16.83, SD ¼ 4.38)and female (M¼ 17.81, SD¼ 4.24) scores did not differsignificantly, t(197) ¼ �1.32, p ¼ .19. The RSES totalscores for this study are not remarkable as norms typ-ically fall between 15 and 25 (Gazzaniga & Heatherton,2006).

The dimensionality of the RSES was examined usingconfirmatory factor analysis (CFA). Three models werespecified a priori: Model I) a one-factor model contain-ing all ten items of the RSES; Model II) a related two-factor model consisting of positively (items 1, 2, 4, 6, 7)and negatively (items 3, 5, 8, 9, 10) worded questions,and; Model III) a related two-factor model comprisingself-competence (items 1, 2, 3, 4, 5) and self-liking(items 6, 7, 8, 9, 10) questions. Maximum likelihoodestimation methods were employed to compare the fitof the models. Two goodness-of-fit indices, chi-square(�2) and the root-mean-square of approximation(RMSEA), are reported in Table 1 for the three models.The greater the �2 statistic and RMSEA value, thepoorer the match between data and model. Of the threemodels, the related two-factor model dividing the RSESinto self-liking and self-competence affords the bestrepresentation of factor structure, though both the

two-factor models provide a better fit than the one-factor model. A model with a greater number of para-meters may provide a superior fit over a model withfewer parameters if its functional form is representativeof the process generating the empirical data or, of lessutility, if the extra parameters are soaking up residualnoise. If both models provide fits to the data that are notsignificantly different, then the simpler model should beretained in accordance with the principle of Occam’srazor. Because Model I is essentially a singly constrainedversion of Models II and III (Tafarodi & Milne, 2002),its goodness-of-fit can be compared to them using �2

difference tests. With an alpha level of .05, Model Iprovided a significantly worse fit than both Model II,�2

diff (1) ¼ 69.07, p < .001, and Model III, �2diff (1) ¼

89.13, p < .001. Models II and III are not hierarchicallyrelated and so a �2 difference test cannot be performedto compare their fits. However, the �2 statistics andRMSEA values presented in Table 1 provide strong evi-dence that Model III better accounts for the covariationbetween the ten RSES items than Models I and II, andwhat is more, concurs with previous findings (Tafarodi& Swann, 2001) and is more theoretically interpretable.

MUSIC PREFERENCE (STOMP)

With reference to Rentfrow and Gosling’s (2003) genres,the same four factor solution was extracted using a prin-cipal components analysis. The correlation coefficientmatrix for music genres was assessed for factorability byfirst examining the matrix for adequate associationsbetween genres (Pearson’s r) and then by conductinga Kaiser-Meyer-Olkin (KMO) test (KMO ¼ .74) anda Bartlett’s test of sphericity, �2(153) ¼ 1389.50, p <.001. Satisfied that the matrix was factorable, a principalcomponents analysis was performed to ascertainwhether or not the 14 genres of music could be reducedinto a smaller number of music domains. An initialunrotated solution indicated that four eigenvectors hadeigenvalues greater than one, explaining 66.4% of thetotal variance. However, the extracted communalityvalue for the ‘‘Religious’’ genre was deemed too low(¼ .13) to justify its inclusion in further analyses, andwas removed. To increase the interpretably of the

TABLE 1. Goodness-of-fit Estimates for Three Models Subjected toa CFA

MODEL �2 df p-value RMSEA

Model I 219.33 35 < .001 .16Model II 288.40 34 < .001 .16Model III 130.20 34 < .001 .12

Music Preference, Social Identity, and Self-Esteem 509

This content downloaded from �����������68.230.212.70 on Wed, 31 May 2023 01:01:32 +00:00������������

All use subject to https://about.jstor.org/terms

solution a direct oblimin rotation was undertaken and,since the correlations between the four extracted com-ponents were all less than .30, a varimax rotation wasperformed to achieve simple structure (see Table 2).

The four extracted factors were labelled as follows: 1)Energetic/Rhythmic (M¼ 4.03, SD ¼ 1.62, ac¼ .83); 2)Reflective/Complex (M ¼ 3.52, SD ¼ 1.34, ac ¼ .74); 3)Intense/Rebellious (M ¼ 3.76, SD ¼ 1.51, ac ¼ .81),and; 4) Upbeat/Conventional (M ¼ 4.18, SD ¼ 1.38,ac ¼ .68). A repeated-measures ANCOVA with the fourmusic preference factors as the within-subjects factor,gender as the between-subjects factor, and age asa covariate, was performed. There were no main effectsof preference, F(3, 597)¼ 1.62, p ¼ .19, or gender, F(1,197) ¼ 0.22, p ¼ .64, no significant effect of age, F(1,197) ¼ 2.26, p ¼ .14, but a significant interaction termbetween the music preference factors and gender wasnoted, F(3, 597) ¼ 3.65, p ¼ .013. Figure 1 displaysmean preference score as a function of the music pref-erence factors for males and females. Subsequent posthoc analysis of simple effects showed significant differ-ences in mean scores across gender for both the Intense/Rebellious (p¼ .008) and Upbeat/Conventional (p¼ .03)factors.

MUSIC PREFERENCE AND SELF-ESTEEM

Simple correlation analysis was performed to test forassociations between the four STOMP music factorsand summed items from the RSES: 1) all ten items; 2)five items corresponding to self-competence, and; 3)five items corresponding to self-liking. Zero-order cor-relations (Pearson’s r) were calculated for group, maleand female data, and a number of small but significant

correlations were uncovered. For males, there was a neg-ative correlation between the self-liking subscale and theReflective/Complex component, r(41) ¼ �.22, p ¼ .03.Females likewise exhibited small negative correlationswith the self-liking subscale and both the Energetic/Rhythmic component, r(156) ¼ �.20, p ¼ .01, and theUpbeat/Conventional component, r(156)¼�.19, p¼ .02.No significant correlations were noted for the group data.

TABLE 2. Means, Standard Deviations, and Component Loadings for the 13 Music Genres Subjected to a Principal Components Analysis

Descriptive Statistics Principal Component

Genre M SD 1 2 3 4

Dance 4.31 1.82 .90 �.12 .03 .05Rap 3.47 1.93 .85 �.002 .09 �.11Funk 4.31 1.86 .79 .07 �.15 .32Blues 3.97 1.72 .13 .84 �.02 .22Jazz 3.77 1.75 �.03 .83 �.001 .10Classical 3.21 1.86 �.13 .59 .07 .27Folk 4.50 1.89 .47 .49 .12 �.28Alternative 3.66 1.92 .003 .15 .87 �.12Metal 2.99 1.88 .06 .003 .85 �.03Rock 4.47 1.81 �.02 �.08 .78 .23Pop 4.41 1.64 .14 .007 .15 .76Country 3.12 1.79 .05 .35 �.13 .71Sound tracks 4.43 1.84 �.13 .36 .02 .59

Note. All loadings > |.40| are underlined. The highest factor loadings for each dimension are presented in bold.

FIGURE 1. Mean preference ratings as a function of music preference

factor for both males and females.

510 Daniel Shepherd & Nicola Sigg

This content downloaded from �����������68.230.212.70 on Wed, 31 May 2023 01:01:32 +00:00������������

All use subject to https://about.jstor.org/terms

MUSIC PREFERENCE AND SOCIAL IDENTITY THEORY

The four music preference factors were further used totest the predictions of social identity theory, whichentailed the construction of a measure of intergroupdifferences. To create a measure of intergroup differ-ences, six new variables were created by calculating thedifferences in preference ratings between each factorpairing. Table 3 displays mean difference scores, withpositive and negative symbols affording further scrutinyof gender differences. Each difference variable was thencorrelated with the RSES and the self-liking/self-competence subscales as described in the preceding sec-tion. For example, the absolute difference between thefirst (i.e., Energetic/Rhythmic) and second (i.e., Reflec-tive/Complex) factors were computed and then thesedifferences, ranging from 0 (no preference between thetwo factors) to 6 (a strong preference of one factor overthe other) were correlated to the self-esteem measures(see Table 4). For females, negative correlations werefound between the difference scores calculated from theReflective/Complex and Energetic/Rhythmic factorsand measures of self-esteem. The difference scoresbetween these two factors correlated with both theten-item RSES, r(198) ¼ �.15, p ¼ .03, and the self-competency subscale, r(198)¼�.21, p¼ .01. For males,positive correlations were noted between the difference

scores calculated from the Intense/Rebellious andReflective/Complex scores and both the self-liking sub-scale, r(41) ¼ .27, p ¼ .05, and the full ten item scale,r(41)¼ .32, p¼ .04. Also for males, positive correlationswere noted between the Intense/Rebellious and Upbeat/Conventional difference scores and both the self-likingsubscale, r(41) ¼ .32, p ¼ .04, and the ten-item RSES,r(41) ¼ .34, p ¼ .03. Fishers r-to-z transformations(two-tailed) revealed no significant differences in corre-lation coefficients across gender (p > .05).

Discussion

Our results reflect a number of interesting findingsrelating to the relationship between music preferenceand self-esteem, the factor structure of the RSES, andthe predictions of social identity theory. Structural anal-ysis of the RSES data presented here did not extracta single self-esteem dimension. Instead, of the threecompeting models, the unidimensional model contain-ing all ten RSES items returned the poorest fit to thedata. Thus further evidence has been marshalled in sup-port of the position that two distinct dimensions under-lie the RSES, namely, the two substantive dimensions ofself-competency and self-liking proposed by Tafarodiand colleagues (2001, 2002).

TABLE 3. Mean Difference Scores (M) and Standard Deviations (SD) for Group, Male, and Female Data

Group (n ¼ 199) Males (n ¼ 42) Women (n ¼ 157)

Difference Scores M SD M SD M SD

Intense/Rebellious vs. Energetic/Rhythmic 1.66 1.43 – 1.88 1.44 þ 1.61 1.43 –Intense/Rebellious vs. Reflective/Complex 1.58 1.29 þ 1.64 1.21 þ 1.56 1.31 þIntense/Rebellious vs. Upbeat/Conventional 1.55 1.28 – 1.43 1.26 þ 1.59 1.29 –Energetic/Rhythmic vs. Reflective/Complex 1.67 1.30 þ 1.71 1.23 þ 1.65 1.32 þEnergetic/Rhythmic vs. Upbeat/Conventional 1.66 1.18 – 1.59 1.15 þ 1.68 1.20 –Reflective/Complex vs. Upbeat/Conventional 1.23 0.89 – 0.96 0.78 – 1.31 0.91 –

Note: Positive and negative symbols indicate the direction of the difference.

TABLE 4. Correlation Coefficients (Pearson’s r) Estimating the Association Between Difference Scores and Self-esteem Measures for Group,Male, and Female Data

Group (n ¼ 199) Males (n ¼ 42) Women (n ¼ 157)

Difference Scores RSES Compt. Liking RSES Compt. Liking RSES Compt. Liking

Intense/Rebellious vs. Energetic/Rhythmic .04 .05 .02 .15 .13 .17 .02 .03 �.007Intense/Rebellious vs. Reflective/Complex .02 .01 .02 .17 .01 .27* -.15* -.21* �.04Intense/Rebellious vs. Upbeat/Conventional .06 .03 .08 .34* .23 .32* �.02 �.02 .001Energetic/Rhythmic vs. Reflective/Complex �.11 �.10 �.10 �.06 �.08 �.02 �.11 �.10 �.12Energetic/Rhythmic vs. Upbeat/Conventional .03 �.02 .08 .05 .06 .01 .02 �.04 .09Reflective/Complex vs. Upbeat/Conventional .04 �.02 .12 �.03 �.76 .12 .04 �.03 .10

Note. All tests were two-tailed. * p < .05. RSES ¼ 10-item RSES. Compt. ¼ 5-item self-competence scale. Liking ¼ 5-item self-liking scale.

Music Preference, Social Identity, and Self-Esteem 511

This content downloaded from �����������68.230.212.70 on Wed, 31 May 2023 01:01:32 +00:00������������

All use subject to https://about.jstor.org/terms

The research reported here found no evidence sup-porting a relationship between the ten-item RSES andany of the four music preference dimensions derivedfrom the STOMP. This result concurs with the findingsreported by Rentfrow and Gosling (2003), who alsofound no significant correlations between the ten-itemRSES and the four factors of music preference extractedfrom their music preference scale. However, when wedivided the RSES scale into subscales based on self-competence and self-liking items, then small but signif-icant correlations emerged between three of the fourmusic-preference dimensions and RSES items tappinginto self-liking. Specifically, negative correlations werefound between self-liking and the following two factors:Reflective/Complex (males only) and for females Ener-getic/Rhythmic and Upbeat/Conventional. Here, theimplications of the findings are two-fold. First, it is theself-liking component of self-esteem that may be asso-ciated with music preference, and second, gender mod-erates the relationship between music preference andself-esteem.

Social identity theory predicts an association betweengroup identification and self-esteem, which our resultspartially supported. Using measures of group differ-ences based on music preference, we uncovered smallcorrelations (�.14 to .34), consistent in size with thosereported in the literature (e.g., Tarrant et al., 2001). Thesmall magnitude of the statistical relationships betweengroup differences and music preference dimensionsmay be explained by a number of considerations. First,it may be that certain musical genres lend themselves togreater attachment and hence will be more stronglyassociated with social identity (Tekman & Hortaçsu,2002). For example, those enjoying classical music maynot feel motivated to wear clothes and regalia expressingtheir preference, nor play the music loudly in public,while those who identify with the rock or rap genresmay do. Second, tolerance of different genres peak atdifferent times across the lifespan (North & Hargreaves,1999), and so the sample of young adults we tested mayno longer invest their identities into a single genre whiledebasing others.

Two previous studies have examined social identitytheory and music preference (Tarrant et al., 2001; Tek-man & Hortaçsu, 2002), though recruited predomi-nantly young males. Their findings are extended hereby the inclusion of a greater proportion of females in thesample. Coley (2008) reported gender differences inrelation to music preference, and gender differences inself-esteem are well documented (Baranik et al., 2008),though did not reach significance in the current study.For females we noted a negative association between

self-competency and the difference scores calculatedfrom the Reflective/Complex and the Energetic/Rhyth-mic factors. Thus a high preference score on one of thetwo factors and a low score on the other is associatedwith lower self-esteem. This finding is not immediatelyexplained by social identity theory, and the study designdoes not afford analyses testing the self-esteem hypoth-esis, which argues that increased intergroup discrimina-tion may be predicted by low or threatened self-esteem.For males, positive associations were noted betweenself-liking and difference scores calculated using theIntense/Rebellious factor and either of the Reflective/Complex or Upbeat/Conventional factors. The maledata provides some support for the predictions of socialidentity theory, namely that increased intergroup dis-crimination leads to increased self-esteem. Thus thedata once again indicate the importance of gender asa moderating factor. However, while the magnitudes ofthe significant correlations were greater for males thanfemales, they were not significantly so.

These gender differences are difficult to reconcile onthe basis of current theory, though conjecture can beformed around gender differentials in music preferenceand use. For males, we note that intergroup discrimina-tion, according to social identity theory, is a strategy forachieving self-esteem via social competition aimed atincreasing the positive distinctiveness of one’s owngroup (Lemyre & Smith, 1985). Thus, it may be themore competitive nature of males that produced bothstronger and positive correlations relative to females.For males, outgroup discrimination along music dimen-sions were between aggressive styles of music, reflectingdominance, and ‘‘lighter’’ music types manifestingsocialization themes of emotional expressiveness andrelationships (Schwartz & Fouts, 2003). For females, thenegative correlation involving the Reflective/Complex(blues, folk, jazz, classical) and Energetic/Rhythmic(dance, rap, funk) difference scores is more problematic.The lack of positive correlations could be becausefemales do not use music socially like males, and insteaduse music emotionally (North et al., 2000). Within-group processes might also explain the differencesbetween the genders, as to some extent the processesoccurring within one’s group, such as acceptance fromother members, will determine group identification andhence the ability to maximize between-group differences(Tekman & Hortaçsu, 2002). Because individuals alsomake ingroup comparisons (i.e., relatively privileged ordeprived), the degree of group membership and sense ofbelonging may vary, and with it social identity. It maybethat the negative correlation found with the female dataarose because while intergroup discrimination was high,

512 Daniel Shepherd & Nicola Sigg

This content downloaded from �����������68.230.212.70 on Wed, 31 May 2023 01:01:32 +00:00������������

All use subject to https://about.jstor.org/terms

within group processes were serving to attenuate self-esteem.

There were several limitations to this exploratorystudy, all of which motivate further research. First, thesample was fairly homogenous, as all participants wereuniversity students. Second, grouping the music intodiscrete, unidimensional genres constitutes an addi-tional limitation, since many musicians fit into severalgenres. Third, the indirect measure of intergroup differ-ences represents the differences between the preferencesfor various music domains, but further research isrequired to elucidate the psychological meaning of thesedifferences. Lastly, the modest sample size precluded theuse of multivariate techniques that better guard again anincrease in the Type I error rate, and for the small malesample, avoid potential Type II errors. Future researchin this area is needed to address these limitations, and toadvance more comprehensive models that include thedeterminants of music preference.

In summary, previous research has failed to convinc-ingly uncover a relationship between music preferenceand self-esteem (Rentfrow & Gosling, 2003), thoughmethodological considerations may partly explain thenull finding. This study demonstrated that measuresof self-esteem need to be employed with care, and whenso used can uncover an association between music pref-erence and self-esteem, and can test theories of socialprocesses. Furthermore, the moderating effect of gendershould be estimated when reporting music preferencedata.

Author Note

Correspondence concerning this article should beaddressed to Daniel Shepherd, Department of Psychol-ogy, P. O. Box 92006, Auckland University of Technol-ogy, Auckland, New Zealand. E-mail: [email protected]

References

ABRAMS D., & HOGG, M.A. (1988). Comments on the motiva-tional status of self-esteem in social identity and intergroupdiscrimination. European Journal of Social Psychology, 18,317-338.

BAKER, F., & BOR, W. (2008). Can music preference indicatemental health status in young people? Australasian Psychiatry,16, 284-288.

BARANIK, L. E., MEADE, A., LAKEY, C. E., LANCE, C. E., HU, C.,HUA, W., & MICHALOS, A. (2008). Examining differential itemfunctioning of the Rosenberg Self-Esteem Scale across eightcountries. Journal of Applied Social Psychology, 38, 1867-1904.

COLEY, A. (2008). Young people’s musical taste, Relationshipwith gender and gender-related traits. Journal of Applied SocialPsychology, 38, 2039-2055.

GAZZANIGA, M. S., & HEATHERTON, T. F. (2006). Psychologicalscience, mind, brain, and behaviour. New York: W.W.Norton & Co.

GRAY-LITTLE, B., WILLAIMS, V. S. L., & HANCOCK, T. D. (1997).An item response theory analysis of the Rosenberg self-esteemscale. Personality and Social Psychology Bulletin, 23, 443-451.

GREITEMEYER, T. (2009). Effects of songs with prosocial lyrics onprosocial behavior: Further evidence and a mediating mecha-nism. Personality and Social Psychology Bulletin, 35, 1500-1511.

LAUKKA, P. (2007). Uses of music and psychological well-beingamong the elderly. Journal of Happiness Studies, 8, 215-241.

LEMYRE, L., & SMITH, P. M. (1985). Intergroup discriminationand self-esteem in the minimal group paradigm. Journal ofPersonality and Social Psychology, 49, 660-670.

NORTH, A. C., & HARGREAVES, D. J. (1999). Music and adoles-cent identity. Music Education Research, 1, 75-92.

NORTH, A. C., HARGREAVES, D. J., & O’NEILL, S. A. (2000). Theimportance of music to adolescents. British Journal ofEducational Psychology, 70, 255-272.

RENTFROW, P. J., & GOSLING, S. D. (2003). The do re mi’s ofeveryday life: Examing the structure and personality correlatesof music preferences. Journal of Personality and SocialPsychology, 84, 1236-1256.

RENTFROW, P. J., GOLDBERG, L. R., & LEVITIN, D. J. (2011).The structure of musical preferences: A five-factor model.Journal of Personality and Social Psychology, 100,1139-1157.

ROSENBERG, M. (1965). Society and the adolescent self-image.Princeton, NJ: Princeton University Press.

SCHWARTZ, K. D., & FOUTS, G. T. (2003). Music preferences,personality style, and developmental issues of adolescents.Journal of Youth and Adolescence, 32, 205–213.

STEELE, J. R., & BROWN, J. D. (1995). Adolescent room culture:Studying media in the context of everyday life. Journal of Youthand Adolescence, 24, 551-576.

TAFARODI, R. W., & MILNE, A. B. (2002). Decomposing self-esteem. Journal of Personality, 70, 443-483.

TAFARODI, R. W., & SWANN, W. B. (2001). Two-dimensional self-esteem: Theory and measurement. Personality and IndividualDifferences, 31, 653-673.

TAJFEL, H. (1978). Differentiation between social groups. London,UK: Academic Press.

Music Preference, Social Identity, and Self-Esteem 513

This content downloaded from �����������68.230.212.70 on Wed, 31 May 2023 01:01:32 +00:00������������

All use subject to https://about.jstor.org/terms

TARRANT, M. (1999). Musical and social development in adoles-cence (Unpublished doctoral dissertation). University ofLeicester, Leicester, United Kingdom.

TARRANT, M., NORTH, A. C., & HARGREAVES, D. J. (2001). Socialcategorization, self-esteem, and the estimated musical prefer-ence of male adolescents. The Journal of Social Psychology, 141,565-581.

TEKMAN, H. G., & HORTAÇSU, N. (2002). Music andsocial identity: Stylistic identification as a response tomusical style. International Journal of Psychology, 37,277-285.

ZILLMANN, D., & GAN, S. (1997). Musical taste in adolescence.In D. J. Hargreaves & A. C. North (Eds.), The social psychologyof music (pp. 161-187). Oxford, UK: Oxford University Press.

514 Daniel Shepherd & Nicola Sigg

This content downloaded from �����������68.230.212.70 on Wed, 31 May 2023 01:01:32 +00:00������������

All use subject to https://about.jstor.org/terms