Musical taste,in-group favoritism,and social identity theory: Re-testing the predictions of the self-esteem hypothesis

Method

Results

Main analysis

A repeated measures t-test showed that participants rated fans of their favorite musical style (M = 15.29, SD = 19.82) significantly more favorably than fans of their least favorite musical style (M = −7.72, SD = 13.99), t(150) = 8.86, p < .001, d = .72. Pearson’s r correlations investigated the extent to which the in-group bias exhibited by participants toward fans of their favorite musical style was related to their (personal and collective) self-esteem, their in-group identification, the importance of music to their everyday life, and how often they listened to music. To achieve this, a new in-group bias score was calculated.

The overall score for fans of a participant’s least favorite musical style was subtracted from the overall score for fans of his/her favorite musical style (Lonsdale & North, 2009). In this context, participants with high overall in-group bias scores would be considered to have exhibited in-group bias to greater extent that those with lower scores. ² This measure of in-group bias was then correlated with measures of self-esteem, in-group identification and musical interest/consumption. Table 1 shows that all of the variables under investigation, except for public CSE, correlated significantly with the in-group bias that participants exhibited toward fans of their favorite musical style.

OPEN IN VIEWER

Table 1.

Summary of in-group bias correlations.

	Personal self-esteem	Membership self-esteem	Private CSE	Public CSE	Importance to identity	In-group identification	Importance of music	How often do you listen?
In-group bias score	.24**	.52***	.53***	.15	.20*	.57***	.51***	.31***

CSE: collective self-esteem.

*

p < .05. ** p < .01. ***p < .001.

N = 151 in all cases.

To assess their relative importance, each of these eight variables was entered into a multiple regression. Together these eight predictor variables predicted 41% of the variation (adjusted r²) in the in-group bias shown to the fans of participants’ favorite musical style, F(8, 142) = 14.16, p < .001. Upon closer inspection, it was evident that participants’ membership self-esteem (β = .21, p = .039), their private CSE (β = .26, p = .011), and the extent to which they identified with the typical fans of their favorite music style (β = .41, p < .001) were only three significant predictors of their in-group bias score.

When asked to indicate their favorite musical style, it was clear that participants liked a variety of different music styles. ³ For this reason, a one-way between-group analysis of variance (ANOVA) was used to test the possibility that fans of different musical styles might exhibit in-group bias to differing extents (e.g., do R&B fans exhibit in-group bias to a greater extent than jazz music fans?). No significant differences were found between the eight sub-groups, F(7, 143) = .47, p = .85. ⁴

Discussion

The findings of the present study are in keeping with the predictions of social identity theory and, most notably for the purposes of this article, the self-esteem hypothesis. As expected, participants rated the fans of their favorite musical style more favorably than fans of their least favorite musical style. This apparent in-group favoritism is consistent with previous studies (e.g., Lonsdale & North, 2009; North & Hargreaves, 1999; Tekman & Hortaçsu, 2002, 2003) and offers further empirical evidence for the predicted relationship between musical taste, group membership, and social identity (Frith, 1981; North & Hargreaves, 1999). The present findings also suggest this tendency to regard those who share our musical tastes more favorably than those who do not is one that is common to most (if not all) music fans and does not appear to be limited only to musical styles that are closely associated with contemporary youth subcultures (e.g., chart pop, hip-hop, and dance music).

Social identity theory asserts that in-group favoritism is motivated by the need to achieve, maintain, or enhance a positive social identity and self-esteem; this idea is widely referred to as the “self-esteem hypothesis.” The findings of the present study offer, for the first time, evidence of a significant link between an individual’s self-esteem and the in-group favoritism shown to those who share their musical tastes. With the exception of public CSE, significant positive correlations were found with all of the measures of PSE and CSE under investigation. These findings are in keeping with the predictions of social identity theory and suggest that people’s willingness to positively discriminate between those who do and do not share their musical tastes might be linked, in some way, to their self-esteem.

The significant positive correlations with self-esteem suggest that participants who feel good about themselves both as individuals and as fans of their favorite musical style were most likely to exhibit in-group favoritism toward those who share their musical tastes. Nonetheless, the correlational research design does not permit us to draw conclusions concerning the causal relationship between self-esteem and the in-group bias shown to fans of participants’ favorite musical style. This issue of causality is, however, nothing new; the evidence for the role played by self-esteem in intergroup discrimination is far from clear cut (Abrams & Hogg, 1988; Brown, 2000; Rubin & Hewstone, 1998). Indeed, Abrams and Hogg (1988) set out two competing hypotheses to explain this relationship: (1) successful intergroup discrimination leads to an increase in self-esteem (Corollary 1) or (2) low or threatened self-esteem motivates increased intergroup discrimination (Corollary 2). It therefore seems imperative that future studies approach this question experimentally to determine whether self-esteem is the cause or effect of the in-group bias shown to those who share our musical tastes.

The present findings highlight the importance of using measures of self-esteem adapted to the particular in-group under investigation, in this case fans of participants’ favorite musical style. Lonsdale and North (2009) used a global measure of collective esteem that asked non-specific questions about the social groups that participants belonged to rather than ones specifically tailored to assess their feelings as fans of their favorite musical style. This oversight may explain why Lonsdale and North (2009) failed to find a significant link with self-esteem, while the present study did. Research suggests that specific measures of self-esteem are often more powerful predictors of behavior than global measures (e.g., Crocker, Luhtanen, Blaine, & Broadnax, 1994; Dutton & Brown, 1997). Future investigations on this topic should bear this in mind and ensure measures of group-specific social self-esteem are used wherever possible.

The present findings are the first to show that the in-group favoritism shown to those who share an individual’s musical tastes might be linked to their self-esteem; this link was found using two well-established measures of PSE and social self-esteem (Luhtanen & Crocker, 1992; Rosenberg, 1965). Although both scales are psychometrically robust (Gray-Little, Williams, & Hancock, 1997; Greenberger, Chen, Dmitrieva, & Farruggia, 2003; Luhtanen & Crocker, 1992) and have been used extensively, there are countless other ways future researchers might go about assessing self-esteem.

Rubin and Hewstone (1998) suggest that there are at least three different types of esteem: (1) global or specific self-esteem, (2) trait or state self-esteem, and (3) PSE or social self-esteem. In most cases, self-esteem scales have traditionally tended to regard self-esteem as a global, personal trait, concerned with an individual’s overall evaluation of themselves as a person (rather than as a group member) that is unlikely to change over time or in the light of new experiences. This conception of self-esteem seems somewhat limited and ignores evidence that our sense of self-worth is often unstable and subject to fluctuations over time (e.g., Kernis, Cornell, Sun, Berry, & Harlow, 1993). Future studies might therefore consider employing state measures of self-esteem (e.g., Heatherton & Polivy, 1991) to explore the links with in-group bias and to detect any short-lived or temporary effects of intergroup discrimination at the time of questioning.

It is also interesting to note the study’s non-significant findings. While two measures of CSE (i.e., membership self-esteem and private CSE) significantly predicted participants’ in-group bias score, PSE did not. These results arguably add further weight to the argument that people’s musical tastes, group membership, and social identity are linked. Put simply, these findings suggest that the extent to which participants exhibited in-group bias toward those who shared their musical tastes was related to how positively participants regarded themselves as group members rather than how they felt about themselves as individuals.

The absence of a significant correlation with public CSE also highlights the possibility that a subtle distinction in how people evaluate themselves as group members might have differing consequences for in-group bias. Specifically, the present findings seem to imply that if participants personally hold fans of their favorite musical style in high esteem (i.e., high private CSE), they would be more inclined to exhibit an in-group bias. Whereas, in contrast, there was no such association between in-group bias and how participants think other people judge fans of their favorite musical style. This distinction warrants further exploration. However, this non-significant correlation should be regarded with caution given the disappointing internal consistency of the 4-item public CSE sub-scale (α = .58) and further investigation will be required to improve the reliability of this adapted sub-scale before any firm conclusions are drawn. The present findings, however, indicate that self-esteem is unlikely to be the sole factor responsible for the in-group favoritism shown to those who share our musical tastes.

A significant positive correlation with a measure of in-group identification suggested that participants who identified strongly with their fellow music fans were most likely to exhibit in-group favoritism toward those who share their musical tastes. In-group bias scores also correlated significantly with the self-rated importance of music and how often participants listened to music. Participants who ascribed greater importance to music may have been more inclined to regard musical taste as a salient criterion for social categorization and social comparison, and as such, this may explain why they were more likely to exhibit in-group favoritism toward fans of their favorite musical style. Subsequent analysis indicated that in-group identification was the most powerful of the three variables found to significantly predict participants’ in-group bias score (i.e., membership self-esteem and private CSE). This positive relationship is consistent with previous research (e.g., Brown, 2000; Brown et al., 1992; Hinkle & Brown, 1990; Perreault & Bourhis, 1998) and leads us to question the relative importance of self-esteem in explaining the in-group bias observed here.

It is possible that the relationship between in-group bias and self-esteem is moderated by the degree to which people identify with the in-group (De Cremer, 2001). If this were the case, we would expect the link between self-esteem and the in-group bias shown to those who share our musical tastes to be stronger among individuals who identify closely with fans of their favorite musical style than among those who do not. In any case, the present findings suggest that the self-esteem hypothesis offers only a partial explanation of the in-group bias observed, and it is clear that the interactions between in-group identification, self-esteem, and in-group bias warrant further investigation.

Although the present findings are the first to demonstrate a link between self-esteem and the in-group favoritism shown to those who share an individual’s musical tastes, it is important to acknowledge the limitations of the study. First, it is unclear if the measure of in-group bias used in the present study offers any meaningful insights into how participants regard music fans in everyday interactions. Indeed, it remains to be seen if the in-group bias shown to hypothetical/imagined music fans when completing an online questionnaire corresponds with how participants behave toward “real” music fans encountered in more naturalistic, real-life settings. Future studies should therefore look to explore alternative methods for assessing in-group bias that are more realistic and less obtrusive than rating the “typical” fan of several different musical styles.

Second, the eight musical styles under investigation arguably offered a more comprehensive account of participants’ musical tastes than previous studies (Lonsdale & North, 2009; North & Hargreaves, 1999; Tekman & Hortaçsu, 2002, 2003), but this list was far from exhaustive. It cannot therefore be assumed that participants’ favorite (or least favorite) musical style was among this pre-determined list of eight musical genres; it is unclear how this restricted range of choices may have affected the present results, but this draws into question the accuracy of the present findings. Future research on this topic will need to weigh up the benefits of investigating an ever-broader range of musical styles alongside the disadvantages of asking participants to complete an overly long questionnaire (e.g., Galesic & Bosnjak, 2009; Herzog & Bachman, 1981; Roszkowski & Bean, 1990). One possible solution might be to allow participants to report their musical taste for themselves or to select their favorite/least favorite musical style from a comprehensive list of options (e.g., North, 2010) and then program the remainder of the questionnaire to reflect these choices. Nonetheless, an approach like this would have limitations of its own; participants would immediately understand that the two musical styles were the focus of the study and this understanding is likely to bias/polarize how they rate their typical music fans.

Third, the present study used a single-item measure to assess the extent to which participants identified with fans of their favorite musical style. While this approach is far from unprecedented (e.g., Brown & Williams, 1984; Perreault & Bourhis, 1998; Sachdev & Bourhis, 1985), some argue that researchers should avoid using single-item measures because they are more prone to random error, and they often lack content validity and the sensitivity needed to discern subtle differences between individuals on complex, heterogeneous constructs (e.g., Churchill, 1979; Loo, 2002; Nunnally, 1978). Research over the past 40 years suggests that in-group identification is likely to be a multidimensional construct, including cognitive, affective, and evaluative components (e.g., Ellemers, Kortekaas, & Ouwerkerk, 1999; Hinkle, Taylor, Fox, Cardamone, & Crook, 1989; Leach et al., 2008; Roccas, Sagiv, Schwartz, Halevy, & Eidelson, 2008). Future research on this topic might therefore benefit from using measures that reflect the complex, multidimensional nature of in-group identification.

Finally, respondents taking part in the present study were predominantly young adults, as were those who took part in earlier studies (Lonsdale & North, 2009; Tekman & Hortaçsu, 2002, 2003). There is evidence to suggest that music is particularly important for adolescents and young adults (16–29 years) and that people aged 30 years and above become less and less likely to report listening to music as a means to define and express their identity as they grow older (Lonsdale & North, 2011). Moreover, research that has consistently shown that self-esteem tends to increase steadily as individuals grow older (see Orth & Robins, 2014 for an overview of the longitudinal evidence). It is, therefore, entirely possible that the in-group bias shown to those who share our musical tastes is limited to adolescents and young adults, who may be more inclined to regard musical tastes as a salient dimension for intergroup comparison, more likely to identify with fans of their favorite musical style, and have greater self-esteem concerns than their middle-aged counterparts. Future studies should look to determine whether the willingness to positively discriminate between those who do and do not share our musical tastes is common to people of all ages, or if this is a phenomenon uniquely associated with young people.

In conclusion, the present findings lend further empirical support to the idea that musical taste might function as a socially symbolic “badge” of group membership (Frith, 1981; North & Hargreaves, 1999) that contributes to an individual’s sense of social identity. Participants rated fans of their favorite musical style more favorably than fans of their least favorite musical style; this apparent in-group bias was found irrespective of participants own musical tastes and using a broader range of musical styles than has been used previously. In keeping with the predictions of the self-esteem hypothesis, the present study also found that the extent to which participants exhibited in-group favoritism to those who share their musical tastes was positively linked with their self-esteem. More broadly, these findings also suggest that musical taste might offer researchers a means to study the predictions of social identity theory in a way that is far more naturalistic and less obtrusive than traditional minimal group experiments.

Supplemental Material