volume 8
july 2005

The four dimensions of popular music

Mapping the continental drift of pop and rock music preferences

by Ger Tillekens and Juul Mulder

Over the past year Ger Tillekens and Juul Mulder have tried to get a grasp on the shifting music prefences of secondary and primary school pupils. In the summer of 2005, they presented a paper on this subject on the 13th biennial conference "Making Music, Making Meaning" of the International Association for the Study of Popular Music (IASPM) in Rome. The paper shows the results of their analyses of some large datasets covering the 1980s and 1990s up to 2001.

Looking behind the charts. As the hit charts so clearly suggest, the preferences of the pop audience seem to be in constant flux. The history of pop artists and their songs, indeed, has been written many times with the hit charts serving as an indication of the twists and turns in their careers — see for instance Anderson a.o. (1980). By now, popular music research has also produced many studies on specific groups of fans, ranging from Simon Frith's study of the Keighly teenagers (1978) and Paul Willis' analysis of the British Motor-bikers and Hippies (1978) to Jeffrey Arnett's penetrating discussion of Heavy Metal fans (1996) with many other interesting studies in-between. Hit charts and fan analyses, though, will tell us little about the changing preferences of the pop audience at large (Parker, 1991; Hakanen, 1998). Moreover, while theorists like Lawrence Grossberg (1992) have provided some deep theoretical reflections about the course of pop and rock music over time, there still are few empirical investigations into the changing preferences of the pop audience spanning more than a few years.

In this paper, we try to fill this gap somewhat by taking an empirical look at the overall popular music audience and its internal divisions over a span of about fifteen years. We will do so by assembling the outcomes on music preferences of three Dutch data sets in a cross-sectional overview. The first one concerns the Groningen data set: a small cohort of 4th grade secondary school pupils dating back to yearend 1986 (n = 712; mean age: 16.2). The second one is the HBSC data set: a large school-based survey of 12-to-18-year old pupils (n = 5,730; mean age: 13.9) collected in the school year of 2001/2, also in secondary education. The third one is the Prima-4 data set: a still larger set of data consisting of a cohort of fourth grade primary school pupils (n = 4,263; mean age: 10.0) and a cohort of sixth grade primary school pupils (n = 14,594; mean age: 12.0) collected a year before. [*] By comparing the outcomes of these data sets, we will try to forge an impression of the main shifts within the field of popular music over the last decades. But, first — to constitute a starting point — we take a tentative look at the changes taking place in the field of popular music since the 1970's, when Graham Murdock and Guy Phelps (1972; 1973) tried their hand at locating the main coordinates of the field of rock music preferences.

Diverging tastes. At the time of the mid-1960's' beat-explosion, new forms of popular music were developing into what we now know as rock music. Appropriated by a rising tide of middle-class youth, at that time this music clearly constituted a cultural unity. The music itself as well as its practices, at least momentarily, seemed able to bridge the existing social divisions of class, gender and ethnicity (Tillekens, 1998). Though these divisions, of course, did not disappear, they did not crystallize in clear subgenres. Rather, they showed themselves in more subtle preferences for specific artists, groups and their songs, like the fan-bases of the Beatles and the Rolling Stones. The 1970's in turn are for most analysts special by themselves for the budding of new genres within the field of rock music itself. Rock turned into, as Peter Christenson and Donald Roberts (1998: 73 ff.) say, "fragmented rock." Underground or Psychedelics on the one hand and Hard Rock and Heavy Metal on the other were the first genres to be labelled as subgenres of rock music, followed by a middle field of dance-oriented genres like Disco, Soul and Funk. Later on in the decade Punk arose, expelling Underground. It would take some time for these genres, spawned by what David Matza (1961) once called "subterranean traditions of youth," to emerge and reach the larger audience.

The first traces of this process of internal differentiation can be found in the British landmark study of Graham Murdock and Guy Phelps. Judging by the chart-oriented kids' favourite record title — Marc Bolan's T. Rex' hit single "Ride A White Swan" — Murdock and Phelps (1973: 159) must have collected their data shortly before or after the year-end of 1970. By that time new genres as Underground, Reggae and, what the authors tentatively call "Heavy Rock" had made themselves known, though these were not yet all fully labelled. In this respect it is a telling fact, that the authors, next to song fragments, used group names and song titles for their research. Already by that time, though, the differentiation of rock music made itself be felt. In their study, Murdock and Phelps use ample space to discuss the adoption of Underground music by the more intellectual youngsters and the rise of Reggae, which they perceive as a reaction among the lower class Skinheads. Murdock and Phelps (1973: 9) even motivate their study with this observation: "The average reggae record is as far away from many of the songs of Bob Dylan as it is from the poetry of Keats."

The stepwise analysis of Murdock and Phelps reveals all of the usual difficulties of handling data on music preferences. For one thing, the differences in musical taste between social categories of young people prove to be not that big. Evidently, in modern societies rock music first and for all constitutes a field of shared preferences and meanings. Rock and pop songs are easily accessible to all and only a small percentage of the audience exclusively adheres to one single genre. It is as difficult to find the system of coordinates that span the field of musical taste. To this end Murdock and Phelps performed three subsequent analyses. The first step involved a questionnaire relating to sound fragments of twelve selected hit songs. These sound bytes were rated according to the principles of the "semantic differential" by a small sample of 211 boys and girls in the fourth grade of secondary education. The analysis resulted in a two dimensions indicating the preferred type of pop music, labelled "activity" and "intelligibility." The second step consisted of an explorative factor analysis on a small sub-sample of 55 of these respondents. The outcomes of this analysis urged Murdock and Phelps to comprise and broaden both dimensions into a single one — the opposition of intelligibility and potency — and to add another one: the opposition between mainstream music and a broadly defined category of Underground, covering the likes of Leonard Cohen as well as the heavy, blues-oriented rock of the Rolling Stones and Johnny Winter.

Figure 1: A reconstruction of the dimensions of Murdock and Phelps (1972; 1973); data collected at year-end 1970

The third step involved a larger sample of secondary school pupils of grade 2 and 4 to investigate the impact of relevant background variables. This analysis added yet another category of more conventional or non-pop-oriented youth, inclined towards more traditional styles of music or to actual chart hits. Taking these groups together along the lines resulting out of the subsequent analytical steps, the pattern emerged that is being presented in Figure 1. Here we find the field of music preferences delineated by two dimensions. The first one differentiates a predilection for more traditional music (Clusters A and D) from the preference for more modern and new forms of music (Clusters B and C); the second one contrasts intellectual, highbrow forms of music (Clusters A and B), now including pop and rock, with lowbrow ones (Clusters C and D). The middle field is constituted by a cluster (Cluster E) of "black" dance music — or as we now would say Soul — represented at that time by the songs of Aretha Franklin and the Jackson Five. This cluster seemed to be more oriented towards shopping and partying, following fashion and exuberantly buying and showing the attributes of their taste as a leisure activity. Because of their sociable manners and outgoing behaviour, Murdock and Phelps label this category as "activity," though in fact this is more often an attitude than real conduct, as the authors themselves conclude. This cluster mainly recruited middle-class youth, so Murdock and Phelps conjecture, who contrary to their lower class counterparts didn't had easy access to the existing street-culture and therefore adopted pop culture to achieve the same end.

The data, moreover, showed a shift along the line of traditional, non-pop-oriented towards modern or at least actual pop-oriented song types from grade 2 to grade 4, which according to Murdock and Phelps testified to the rise to cultural dominance of the middle classes. A dominant opposition, dubbed by the authors as their most important outcome, was found to exist between early school-leavers with a middle-class background and their lower class counterparts — represented by a diagonal line from the upper right (Cluster B) to the bottom left (Cluster D) in Figure 1 in the criss-cross pattern constituted by the model. The former category proved to be more deviant from their parent cultures and also more oriented towards the Underground style of rock music, while the latter category did deviate less from their parent culture and kept to the Top-Ten. This polarization, by the way, seemed to manifest itself mostly in lower class schools, thereby showing the growing importance of school as a social divider.

The turn towards the 1980's. Geographically the Netherlands are not that far removed from the British Isles. Economic and social conditions being rather similar, in time the cultural developments on the youth front tended to converge. At the time the Beatles were playing Hamburg, Dutch bands were successfully competing them in the clubs at the Reeperbahn (Mutsaers, 1990). As soon as Beatlemania had arrived, countless Dutch bands in the major cities appropriated the sound and conquered the country. At the end of the 1960's Amsterdam even became known as the magical centre of the counter culture. Even so, due to the global impact of rock music, by and large, one should expect a pattern similar to one found by Murdock and Phelps. Just like in Britain and almost everywhere, the situation changed in the 1980's. As anticipated by the findings of Murdock and Phelps, Hard Rock and Underground developed into fully-fledged genres within the corpus of rock music. Next to that, several new genres made their presence known, notably Punk, Funk, and Disco. The turn of the 1970's towards the 1980's even may be seen to present an important marking point, because now a youth generation that itself had fully grown up with pop and rock music, entered the scene. For them pop music was not something radically new, but a factual matter of everyday life.

Already by 1976, the book of Murdock and Phelps had been translated and published in Dutch. It had to wait, though, for seven long years for a follow-up. In 1983 Remko van Bork and Jan Jacobs (1986) took to this job. By that time the fragmentation of rock music had steadily been progressing and in their analysis a whole range of new genres show up. There even had been a significant rise in Dutch Pop, especially in the age group of 12 to 14 years-old, mainly due to the successes of a single band by the name of "Doe Maar." However, the genre preferences, collected during April-March, 1983 (n = 302; age: 12-23 years), could be allocated to five clusters conforming, as Van Bork and Jacobs (1986: 147) themselves say, with those of Murdock and Phelps. Their first cluster of genres, comprising MOR (Middle of the Road), Country and Western, Rock and Roll and Dutch language Schlagers, coincides conceptually with Cluster D. Their second category largely conforms to Cluster E and comprises Disco, Funk and Reggae. The third one, covering Blues and Symphonic Rock, coincides with Cluster A, while the fourth one, uniting Ska, Punk and New Wave, conforms to Cluster B. Finally, Hard Rock, coinciding with Cluster C, formed a separate and exclusive genre, with a relatively small following of youngsters that admitted to not liking or even knowing any of the other genres (Van Bork and Jacobs, 1986: 88). This clearly signals the early start of, what by now we can call, the enduring Heavy Metal isolation (Bryson, 1996).

Next to the crystallisation of genres within the domain of pop music itself, the most remarkable facts are the new additions of Punk and Disco (Lopes, 1992: 65). Disco is clearly present but still rather elusive, while Punk instead of Underground now is serving as the great divider on the diagonal line B-D. By 1983, so this shows, Punk and its softer variant New Wave had successfully driven out Underground and Psychedelics. Van Bork and Jacobs add some information confirming the identity match of their findings with those of Murdock and Phelps. Moreover, the young people preferring the left-hand clusters A and D consume music mainly passively on radio and television and use it as a discussion topic with their local peers with whom they share their preferences, while those on the right-hand clusters B and C are visiting concerts, making friends at these occasions and even making pop music themselves. The inhabitants of the centre, Cluster E, on average counting more girls than boys so the data of Van Bork and Jacobs indicate, cultivate a more active attitude: they choose their own records to listen to, are out-going and not only sit down to listen but also like to dance to the music of their choice. Confirming the findings of Murdock and Phelps, this aura of "activity" first and for all is a kind of attitude and does not result in excessive outgoing or partying behaviour. The vertical dimension appears to be related to the level of education. Symphonic rock and Blues represent highbrow tastes, while Dutch Schlager, Rock and Roll, Country and Western, MOR and Hard Rock stand for lowbrow tastes. Just like Murdock and Phelps, Van Bork and Jacobs (1986: 90) mark the diagonal opposition between Cluster B and Cluster D as the dominant opposition, that according to them is highly reducible to differences between levels of education.

Not all results of this study correspond with those of Murdock and Phelps. The street culture to which Murdock and Phelps lay so much weight, for example, does not seem to play such an important role. Music consumption instead is strongly mediatized as most youngsters enjoy their music by listening to the radio or playing records. This may be partly due to national differences, as Van Bork and Jacobs (1986: 96) themselves argue. On the other hand, it can also be attributed to the growing acceptation of pop and rock by parents, many of which by now had lived actively through rock's formative years. This does not implicate that parents and their children by then shared an appreciation of the same songs and genres. On the contrary, the available data instead show a growing divergence, to which the newly arriving genres of Punk and Disco may have paid their share. During the 1980's, as Koen van Eijck a.o. (2002: 161) recently showed, the older generations — those born before 1954 — relatively turned more towards conventional patterns, while the younger ones more fully and exclusively adopted popular culture.

This shift, by itself, proves not to be an age effect, but is first and for all due to differences between birth cohorts. The authors argue that the parental acceptance of children listening to their own music within the confines of the home is related to emergence of a new permissive socialization regime, mainly due to the social upward mobile post-war generations that, during the 1950's and more so during the 1960's, parted from conventional culture. Which, we may add, by itself also was an effect of the successful cultural revolution of pop music in the 1960's. This growing acceptance by parents of their children's deviating music tastes is also emerging from Van Bork and Jacobs' (1986: 108/9) data. By 1983, almost 60% of the Dutch parents themselves had quite different tastes, but accepted that their children cultivated their own taste of music, be it not played too loud. Mainstream pop music, by then, was forming a middle ground between parents and children (De Meyer, 1980: 69).

Mapping the 1980's. After Van Bork and Jacobs had set the example, more studies followed in their wake. In the mean time, the first generation of social scientists that had fully grown up with rock music had acquired academic positions, and they now stood ready to apply the tools of their trade to the object of their liking. This even caused a modest Big Bang in Dutch popular music studies (Sikkema, 1988; Van Wel and Van der Gouwe, 1990; Janssen and Prins, 1991; Tillekens, 1993). These studies, by and large, produced almost identical results. The results even conform strongly to the outcomes of comparable studies in other countries like the one of Keith Roe (1992) among Swedish pupils (n = 1,334; age 15-16; data collected in 1984), thereby confirming the international character of pop music. By the mid-1980's popular music, indeed, had become a global phenomenon, even in its internal differentiations.

The Groningen data set, collected at year-end 1986 (n = 712; mean age: 16.2), offers a good example (Tillekens, 1993). This research project looked into the preferences of a sample of 16-year-old youngsters for nineteen different musical genres, encompassing some older and some new ones, which could be reduced into five categories by means of a factor analysis explaining 56.8% of variance (see Table 1). For the total variance to be explained each and every genre would have to be taken as a separate factor. Luckily many genres converge in clusters, proving that many people prefer combinations of related or adjacent genres. The five factors found in the Groningen data set, are quite similar to those found by Van Bork and Jacobs and thus can be equated with the clusters of Murdock and Phelps. The first factor can be identified as their conventional highbrow Cluster A, the second one with their centre Cluster E, the third one with their pop-oriented highbrow Cluster B, the fourth one with their conventional lowbrow Cluster D, and the fifth one with their lowbrow Cluster C. Again, outcomes like these are surprisingly stable. Most subsequent analyses result in the same number of four to five dimensions with analogous interpretations (see for instance Dheil a.o., 1983; Fink a.o., 1985; Hakanen and Wells, 1993; Mizell a.o., 2003; Rentfrow and Gosling, 2003; and even the outcomes of the two-factor solution of Finnas, 1987; and the replication of the latter by Schwartz and Fouts, 2003).

Table 1: Five factors emerging from the Groningen data set (data collected at year-end 1986; n= 712; mean age: 16.2); rotated factor matrix (Varimax; iterations: 17; explained variance: 56.8%)

The orthogonal character of the factor solution indicates that the respondents seem to treat these factors as independent dimensions along which to measure their likings. So the final outcome can be a mixture of quite unlikely partners. Indeed, as revealed by a closer investigation into the individual ratings, there clearly were groups of respondents who combined likings for genres usually thought to be incompatible, as for instance highbrow Classical Music with the lowbrow Dutch Schlager. In the same vein, individual combinations were found to exist of strong preferences for Classical Music and Hard Rock. This fact, though, will not surprise those who know that many a Hard Rock and Heavy Metal guitarist in those days was inspired by Classical Music, just like highbrow New Wave groups like Blondie rediscovered Rock and Roll and the Girl Group sound. Ostensibly, also their fans sometimes seemed to share this predilection. This mutual fascination of musical opposites, indeed, may be an important driving force in the history of popular music. Another element adding to the independency of the factors is the presence of at least some hard core fans, or "adolescent music marginals" (Hakanen and Wells, 1990) or "univores" (Bryson, 1997), who don't like any other cluster of genres but their own. Statistically these groups may comprise only a small percentage of the total population, but their existence will have an effect on the factor solution.

The most confusing element of all, probably, is the independency of traditional highbrow forms of music — Classical Music, Cabaret, Jazz and Blues — and traditional lowbrow forms of music — the Dutch Schlager as well as Country and Western. For some researchers the orthogonality of factor solutions like these even provided sufficient empirical evidence to declare the old class-based opposition between high culture and low culture to be completely outdated (Van Wel and Van der Gouwe, 1990). However — because of the reasons mentioned above — the dimensions emerging from factor analyses like these, tend to represent clusters of genres rather than the coordinates of the cultural arena in which musical tastes are formed. Moreover, for technical reasons relating to the scalability of musical preferences, an ordinal principal component analysis may be more fit to order and visualize the patterns between clusters of genres (Tillekens, 1993; Stevens, 2001). The diagram of Figure 2 just does that by presenting the preferences for pop and rock genres along two dimensions that closely conform to the reconstructed model of Murdock and Phelps.

Figure 2: Two out of four dimensions from the Groningen data set, 1986 (n= 712; mean age: 16.2); Princals; dimension 2: tradition vs. modernity (x-axis; variance explained: 10.0%) and dimension 3: lowbrow vs. highbrow (y-axis; variance explained: 8.7%)

Figure 2 shows the matrix of pop music preferences, delineated by two dimensions resulting from an ordinal principal component analysis (Princals) on the Groningen data set, explaining 53.5% of variance. Comparing these outcomes with the model of Murdock and Phelps, again the crystallisation of genres within the domain of pop music and the new additions of Punk and Disco seem to be the most striking deviations. The field itself, showing the same criss-cross pattern, seems to unfold along the same coordinates. The labels, though, do not always apply. Indeed, at the end of the 1970's, "Heavy Rock" as Murdock and Phelps called it, was potent and Hard Rock and Heavy Metal were to become even more so. Taking the place of Underground, Punk however was equalling or even surpassing Heavy Metal in loudness and potency. Lowbrow versus highbrow thus seem to offer a better definition for this dimension, presented here as the vertical dimension. The horizontal dimension can be interpreted as constituting Murdock and Phelps' dimension of conventional and traditional versus unconventional, more modern genres. In combination, the x-axis and the y-axis offer a comprehensible picture of pair-wise oppositions, marking the genres as well as their adherents.

The youngsters at the left-hand, traditional side of the matrix, so the data show, are more conforming to the tastes of their parents and less taken in by the flow of youth culture. At the opposite end we find marginal groups, at the time said to constitute categories of "extreme youth-centrists," fully, some of them even exclusively engulfed in their own pop-oriented way of life, notably Punk and Hard Rock. As can be expected this dimension shows a gender-effect as boys are scoring higher on this dimension than girls, which fits in with other descriptions of the fan-base of Punk and Hard Rock. The favourite clothing style, accordingly, ranges from more to less conventional. Correlations, moreover, show these youngsters to experience more problems at school and to report more conflicts with their parents. These findings coincide with those reported by Roe (1992). The identification of the vertical dimension with the lowbrow-highbrow divide is affirmed by its correlation with back ground variables indicating social class — the work level of the father in particular — and the respondent's school level.

As for the genres themselves, the matrix indicates a shift over time. The x-axis, horizontally, shows the effects of a traditionalization and contraction of genres. Traditionalization occurs when genres fall from the grace of the younger generations and are mainly kept alive by the remaining attention of older age-groups and therefore cease to be "youth music." Consequently, they will shift from left to right on the dimension of tradition versus modernity, contracting in the end with the genres that are already located there to become part of the parent culture. This interpretation is supported by an age effect: how younger the respondents, the higher they are rating on this dimension. At the same time, their adoption of new genres makes this dimension into a time-axis for the genres themselves. Older genres become conventional and, losing their edges, they finally merge with the existing corpus of traditional genres. This contraction effect, that is made visible by Figure 2, may be an important indication, or even explanation for the merging of formerly opposite styles in cross-overs.

Introducing two additional dimensions. Most research projects on music preferences using factor analysis arrive at a solution of five dimensions. Such a number may seem too much to fathom. It doesn't mean, though, that we have to trade comprehensible Euclidean space for inconceivable Einsteinian relativity. Principal component analysis is not quite as exotic as string theory and four to five of these dimensions still can be represented in two-dimensional space. The model of Murdock and Phelps, reconstructed in Figure 1, for instance shows a pair-wise grouping of four clusters — A and B versus C and D; A and D versus B and C. These oppositions are covered by the above discussed dimensions. To this respondents with an exclusive preference for only one cluster at the cost of the remaining three — for instance A versus B, D and C — will define dimensions that constitute diagonal lines in the same matrix. If we add a central cluster of genres (E), yet another dimension arises: the middle against both sides. Both these dimensions also emerge from our ordinal principal component analysis. Figure 3 presents both these remaining dimensions.

Figure 3: Two out of four dimensions from the Groningen data set, 1986 (n= 712; mean age: 16.2); Princals; dimension 1: univorism vs. omnivorism (x-axis; variance explained: 27.0%), and dimension 4: receptiveness vs. activity (y-axis; variance explained: 7.8%)

Closer inspection of the first of these dimensions, presented on the x-axis of the diagram of Figure 3, reveals it to be a combination of the diagonal lines of which one is dominant. The individual scores on this dimension follow an erratic path stressing the diagonal lines, in this case accentuating a dominant opposition between softer and more intellectual genres like Blues and Jazz at one hand and the louder counter image of Hard Rock on the other. A supplementary analysis (Mudfold; see Tillekens, 1993) showed that these diagonal lines can be treated as unfolded scales, confirming that the preferences for genres can be ranked in order according to their relative distances from the position of each individual within the matrix of the genres. This dimension correlates with the first factor of the factor analysis to a point of being identical and is also responsible for the majority of the explained variance. Moreover, this particular dimension seems to coincide with the dividing line Peterson and Kern (1996) found and labelled as a new elitist form of "omnivorism," though one can question their interpretation (Van Eijck, 1999; 2001). Indeed, the category of the omnivores is only slightly more open to other genres than the majority of the respondents. This "new" omnivorism, like Van Eijck a.o. (2002), argue therefore should be interpreted not as much as "snobbery" as well as an effect of the growing cultural heterogeneity of the more conventional part of the upward mobile new middle classes. Besides, as one should keep in mind, this dividing line is also constituted by the disliking of their favourite genres by their antipodes, mainly the fan-base of Hard Rock and Heavy Metal.

Comparing the matrices of Figure 2 and Figure 3, one can easily identify the final dimension, presented on the y-axis of the diagram of Figure 3, with the "activity" factor of Murdock and Phelps. The genres rating high on this dimension, indeed, constitute the middle field of Figure 2. This fourth dimension incorporates "black dance," but added to genres like Reggae, Soul, Ska and Funk we now also find Disco and even New Wave. It proves to be the cluster to which the overwhelming majority of the sample by its ratings confesses to belong. Where those residing at the extremes of the dimensions of lowbrow versus highbrow culture and tradition versus modernity, taken together, only cover about twenty percent of the population, the remainder can be found in the middle field — which may be a special feature of the mid-1980's. Projected in Figure 2, this dimension can be shown to be centripetal, as it seemingly acts as a force joining these genres together and gravitating them towards a common centre of danceable music. The scores on this dimension correlate with spending time on window shopping and, if available, money to buy the trendy attributes of youth culture. Again, these activities point towards an attitude rather than excessive behaviour. It is indeed the middle ground of the pop music arena, organized around the peer group and the attitudes of youth-centrism, though the latter is not to be confused with the more intense forms of youth-centrism we find in the marginal clusters of Hard Rock and Punk.

Figure 4: A causal model for the Groningen data set (SEM); chi-square = 41.39; df = 34; p-value = 0.179; RMSEA = 0.018

The most relevant, i.e. significant, relations of the four dimensions with exogenous and intermediate variables are summarized in Figure 4 by means of a simple structural equation model (SEM). The model displays some important background variables indicating social class, together with some relevant characteristics of the respondents like gender, age and type of school. Some of those, mainly gender, do interact with each other. At that time, at least till the fourth grade of secondary education, girls were doing better at school than boys by reaching higher levels and — on top of that — on a younger age as the path coefficients in question in the model do confirm. The relations, though significant, between the background variables and the dimensions are rather weak. Homology may exist for the groups of hardcore fans, but can not convincingly be shown to exist for the audience at large. Rock music, however, clearly, is not immune to the experiences of social class. The position taken by respondents on the lowbrow-highbrow dimension is directly and indirectly influenced by father's social position as well as the youngsters' own educational achievements. The impact of family background even may be highly underestimated and education overestimated, as it is on attitudes on and participation in highbrow culture (Sieben, 2001; Van Eijck, 1997). Most important, though, is the finding that both social class and school-based selection contribute in explaining the diversity of taste in respect to the adoption of lowbrow and highbrow variants of pop and rock music.

The dimension of tradition versus modernity is affected by gender and age. The latter may be the reason why Christenson and Roberts (1998) typify age as the "elemental predictor" of musical preferences. Age, though, is not affecting the dimension that, there and then, explained most of the variance in music preferences. In effect, the dimension we baptized "omnivorism," mainly coincides with school level and gender. Differences between school types manifest themselves most strongly in the degree of "omnivorism" and — girls scoring far higher than boys — at least for 16-year-olds this dimension proved also to be the main gender divider. Being attracted to the centre, as indicated by a high score on the fourth dimension is only related to school level. The finding that the occupants of the lower reaches of secondary education in this respect surpass their peers at the upper levels deviates from Murdock and Phelps and Van Bork and Jacobs, which may be due to the upsurge of Disco music in the 1980's. Another remarkable and possibly related difference is the diagonal shift from an opposition between lowbrow traditional tastes and highbrow modern tastes to an antagonism between highbrow traditional and lowbrow modern preferences. This finding may be ascribed to the peculiarities of the questionnaire or reflect a real turn towards highbrow conservatism.

The rise of new genres. Music preferences usually are being researched by asking respondents to rate their likings on a Likert scale on a preformatted list filled with genre names. That is why this kind of research tends to miss new developing and not yet fully labelled genres and that is exactly what was happening by the end of the 1980's. Publishing the results of their studies at that time, many researchers of this first wave of Dutch popular music studies saw their outcomes being overtaken by the flow of time. At the end of the decade, and more so in the first half of the 1990's, Dance culture closely entwined with House music, together with Hip-Hop culture relating to Rap and Urban, did change the landscape of pop and rock music considerably, even up to a point of breaking the basic conventions of rock music itself and turning Rock into a genre of its own. Added to this, and as if in response, some conventional genres like Cabaret were updating their relevance, while more traditional audiences were adapting religious music to the idiom of rock and labelling the result as Contemporary Christian Music (CCM). Many authors also signalled a new attitude among the young, being less exclusive to one genre and thereby creating an inextricable mishmash of fads and fashions (Ter Bogt and Hibbel, 2001; De Leeuw, 2001). The confusion brought about by these changes, made Frith (2000) confide that he did not think the popular music world still worked in the ways which rock discourse assumed, and seems also to have paralysed research on youth music preferences as there are no data available yet for this decade — at least in respect to the Netherlands.

The situation at the middle of the 1990's, though, can be derived from the study of Frank Stevens (2001), who analyzed the preferences of Flemish youngsters (n = 4,722; age: 17-18) for eighteen different genres during the school year of 1996/'97. Though the outcomes show some notable deviations from the situation in the 1980's they still, as Stevens argued, could be interpreted along the lines of the Groningen model. For one thing the number of factors or clusters of genres, arising from the exploratory factor analysis, seemed to have been reduced from five to four. The first one even appeared to comprise "univore" and formerly opposite genres like Hard Rock, Heavy Metal, Punk, Grunge and even Rock. The second one ranged Classical Music, Folk, Reggae and by now also Sixties, World Music and Sound Tracks side by side. The third factor comprised next to Disco the new coming genres of Techno, House, Rap and New Age. Finally the fourth factor merged Top-Ten music with Dutch Schlagers. Sound Tracks, House, Rock Techno, Disco and Rap proved to be the most popular genres, while Dutch Schlagers, Heavy Metal and Punk — in that order — were counted as the least popular. Nurturing other research questions, Stevens refrained from analysing the mutual relations between these dimensions. To that end we will have to skip a few years by stepping to a Dutch data set, collected shortly after the Millennium change.

Figure 5: Two out of four dimensions from the HBSC data set, 2001/2 (n = 5,730; mean age: 13.9); Princals; dimension 1: tradition vs. modernity (x-axis; variance explained: 30.4%) and dimension 3: lowbrow vs. highbrow (y-axis; variance explained: 14.2%)

Figure 5 shows two dimensions, derived by means of an ordinal principal component analysis over fifteen genres from the Dutch HBSC data set (data collected in 2001/2; n = 5,730; mean age: 13.9) — the Dutch part of the Health Behaviour in School-Aged Children project. Over four dimensions, the variance explained totals to an impressive 71.0%, affirming their discriminatory power. The data set covers the whole range of secondary education. The dimensions of the diagram, shown in Figure 5, can be identified as the lowbrow-highbrow dimension (x-axis) and the tradition-modernity dimension (y-axis) we already encountered in Figure 2. Comparing both figures, we find some strong confirmation for Stevens' outcomes. Many things appear to have changed over the interval of fifteen years, the most obvious being the emergence of a whole new array of genres at the upper, respectively the lower end of these dimensions — forcing previous opposite genres like Dutch Pop and Jazz to contract at the other ends, almost to the point of a convergence of their audiences. On the narrowed dimension of tradition and modernity, now explaining most of the variance, Classical Music seems to have slightly recovered. Indeed, according to Ineke Nagel (2002), the attraction of traditional highbrow culture among Dutch youngsters had been slowly rising again after nearing an all-time-low by the end of the 1980's. This, of course, may also be attributed to the changes taking place within the institutions of high culture themselves. At the time of the Millennium change, symphony orchestras were happily performing jointly with pop artists, while theatre groups were freely making use of the pop idiom in their performances.

Figure 6: Two out of four dimensions from the HBSC data set, 2001/2 (n = 5,730; mean age: 13.9); Princals; dimension 2: univorism vs. omnivorism (x-axis; variance explained: 16.9%) and dimension 4: receptiveness vs. activity (y-axis; variance explained: 9.5%)

More impressive deviations even can be found in the locations of the genres along the two remaining dimensions as shown in Figure 6. On the dimension of univorism versus omnivorism (x-axis) we now find a complete split between the univore genres of Punk and (Alternative) Rock and Heavy Metal and its more stylish offspring Gothic on one hand and all the other genres at the other hand. As the main body of the respondents do not show themselves to be very fond of Punk and Heavy Metal, this opposition offers an indication of a common, silent revolt against these former marginal genres and accordingly a more open attitude towards the remaining ones. "Black dance" genres still scoring high, at first sight the distribution of the genres along the vertical dimension appears to deviate less from the situation in the 1980's. However, the prominent position of competitive "white dance" genres like House and Club at the other end certainly questions it being labelled as "receptiveness" versus "activity." As could be expected, correlations with background variables show this dimension to be related to immigrant status.

Figure 7: A causal model for the HBSC data set (SEM); chi-square = 51.16; df = 39; p-value = 0.092; RMSEA = 0.007; error variance for "work father," "work mother," "urban area," "immigrant status," "gender" and "dimension 1" = 1.00, "age" = 0.99, "type of school" = 0.90, "dimension 2" = 0.92, "dimension 3" = 0.98, "dimension 4" = 0.93

Despite the changing landscape of pop and rock music, the overall relations between the individual dimensions and relevant background variables presented in Figure 7, by and large, show the same patterns. Omnivorism still proves to be more of a girl thing than a male inclination. Due to the now almost all-encompassing character of omnivorism, the formerly strong impact of school level on this dimension, however, has disappeared and even been inverted. The opposition between tradition and modernity, still only is influenced by age. The effect of age, however, is rather small certainly if we consider the fact that the HBSC dataset spans a rather large age group. The narrowing down of this dimension here may play its role. The lowbrow-highbrow dimension still draws a large impact of differences in school level, though the direct influence from family status indicators has disappeared which may be an effect of the ongoing meritocratization of Dutch society. In respect to differences of social class, Stevens found a strong and widened cultural split in music preferences and media use dividing the upper and lower middle reaches of secondary education. This finding is affirmed here for the Netherlands by the large overall percentage of explained variance, the clear markings of different genres on the lowbrow-highbrow dimension as well as the direct influence of school type on this dimension. The fourth dimension, here still labelled "activity," is positively related to school, with girls showing a slightly higher rating than boys.

The data set adds two other relevant and variables to the analysis: immigrant status and level of urbanization. Both these variables are related, because the number of immigrants in the inner cities has grown to become a majority. As Figure 7 shows, immigrant minority status has a relevant positive effect on omnivorism as well as on activity. The level of urbanization already proved to be important in the 1983 data set of Van Bork and Jacobs (1986: 90). At that time notably Country and Western and Rock and Roll were preferred better by young people living in small villages, while larger city dwellers seemed to have more of a knack on Funk. The role of the latter genre now seems to be taken by Rap and the aptly named genre of Urban.

Summarizing our findings so far, by and large, we find the four dimensions resulting from the analysis themselves to be comparable to those of the 1980's. However, the distinctive diagonal criss-cross pattern constituted by the lowbrow-highbrow dimensions and tradition-modernity in the 1970's and the 1980's has vanished completely by the rise of new genres. Add to this the isolation of marginalized genres on the lowbrow-highbrow dimension and the split between "black" and "white" dance music in respect to "activity" and we can safely conclude that, all in all, the field of pop music has changed considerably. Looking at Figure 5, it even seems as if, by the end the 1990's, a whole new genre distribution was in effect slowly replacing the older genres, bottom-upwards on the lowbrow-highbrow dimension.

The coming of the tweeners. The birth rates may be declining in Western countries, but the teenager, apparently, still is an expanding social category. In the 1930's and 1940's people used to define the young as those who between the age of 18 and 24 had entered and not yet left the marriage market. By the 1950's, in the wake of the expansion of schooling and the rising importance of the "adolescent society" (Coleman, 1961), the starting age of adolescence had gone down to 16 year to dwindle downwards to 12 year since. The expansion of youth at the lower end of the age barrier has not halted yet. Transgressing the social boundary between secondary and primary education, the audience of pop music now definitely seems to incorporate the higher reaches of primary education. Moreover, these "tweeners", as they have been branded by marketers, not only adapt to the tastes of their older brothers and sisters but also develop their own preferences. Over the last ten years, their impact clearly has risen, as can be derived from the chart successes of Teen Pop. In this respect, discussing the activity factor of pop music preferences Van Eijck (2001: 1180) concludes about the importance of age: "We expect this variable to have become even more important as of the 1990's, given the rapid emergence of bands gainfully tapping into the Teen Pop market (e.g., *NSYNC, Britney Spears, Boyzone, Spice Girls)."

In school year 2000/1, during the fourth study of the Dutch biennial longitudinal research project (Prima-4) into the achievements of primary school pupils, a questionnaire concerning preferences for 25 music genres was completed by a cohort of fourth grade primary school pupils (n = 4,263; mean age: 10.0) and a cohort of sixth grade primary school pupils (n = 14,594; mean age: 12.0). Application of principal component analysis on these data at first produced graphs that could not easily be interpreted, possibly because of the intricate effects of social class and gender on this young age group. After being rotated (Konig, 2002), the results however again revealed coordinates comparable to those we found earlier on in the Groningen and HBSC data sets, now explaining 57.2% of variance. The outcomes testify to the rise of new genres like Urban, Latin and Teen Pop and the adaptation to their success of some older genres as visible in the transformation of Country into Alternative Country — implying yet another shift in the locus of the genre (Peterson and Di Maggio, 1975) and of Rock into Alternative Rock.

Figure 8: Two out of four dimensions from the Prima data set, 2000/1 (n = 18,857; mean age: 11.6); Princals (rotated); dimension 1: tradition vs. modernity (x-axis; variance explained: 20.8%) and dimension 3: lowbrow vs. highbrow (y-axis; variance explained: 10.9%)

In graphical form, Figure 8 presents the dimensions that can be interpreted as tradition versus modernity (x-axis) and lowbrow versus highbrow genres (y-axis). Most conspicuously, the new genres shown to be developing in older age group of secondary education, now cover the whole range of lowbrow versus highbrow. Actually it is all chart music, as the pupils confess to get most of their song knowledge and pleasure from MTV or its Dutch counterpart TMF. On this dimension, moreover, there are strong differences between "black"-oriented genres," like Rap and Urban, at the lower and their "white"-oriented variants — like Teen Pop — at the upper end with Latin playing an intermediating role (Cepeda, 2000). Rock, here taking its place next to Teen Pop to fill the space that had been left open in the outcomes of the HBSC data set, must be seen Alternative Rock.

On the dimension of tradition versus modernity, there proves to be a massive shift from the fourth grade to the sixth grade, and as a consequence the age between 10 and 12 year proves to be rather critical in this respect. Whereas only about one third of fourth grade pupils indicated to enjoy the new genres of pop music next to older genres, the mass of sixth grade pupils restrict themselves almost exclusively to the new genres. Discussing the "open-earedness" of young children to all kinds of music, Hargreaves and North (1999) conclude that children in the final stage of primary education seem to become deaf at the one ear they use listen to their parents' music. Christenson (1994), by the way, already traced this effect back to the mid-1980's, concluding that the attraction to genres of their parent likings proves to be the highest in third and fourth grade to collapse afterwards. Using high culture participation indices at 10 year, Nagel (2002: 226) arrives at similar conclusions. In contrast with the variation of the new genres, we see a strong contraction of the older genres that are being traditionalized and thereby stripped from their class-based connotations. Sixties Music, of course, by itself is a category in which former divisions have been dissolved. And, despite the success of groups like Slipknot, Heavy Metal — by now perhaps the longest surviving genre of pop music — at last seems to be crumbling and suffering the fate of traditionalization and contraction.

Figure 9: Two out of four dimensions from the Prima data set, 2000/1 (n = 18,857; mean age: 11.6); Princals (rotated); dimension 2: univorism vs. omnivorism (x-axis; variance explained: 16.1%) and dimension 4: receptiveness vs. activity (y-axis; variance explained: 9.4%)

Figure 9 shows the two remaining dimensions: univorism versus omnivorism (x-axis) and receptiveness versus activity (y-axis). The range of genres on the omnivorism dimension shows a strong resemblance with the outcomes of the other data sets with low ratings for Heavy Metal and Punk and high ratings for Classical Music. Added to the high rating genres, however, are the new genres of (Alternative) Country, Latin, Teen Pop and Urban, indicating the open attitude of most pupils towards these genres. These genres also converge on top of the receptiveness versus activity dimension.

Figure 10: A causal model for the Prima data set (SEM; final class; n = 14,594; mean age = 12.0); chi-square = 20.06; df = 18; p-value = 0.329; RMSEA = 0.003; error variance for "education mother,""ses," "education father," "gender" and "dimension 1" = 1.00, "age"= 0.94, "final aptitude test" = 0.90, "dimension 2" = 0.95, "dimension 3" = 0.94, "dimension 4"= 0.84

Figure 10 again summarizes the relations between the four dimensions and some relevant background variables, restricted to grade six cohort because of the availability of the final aptitude test. Compared to the equivalent models of the Groningen data set and the HBSC data set, the picture is more complex. The overall relations, though, by and large show the pattern we already encountered in both the other data sets. The effects of background variables indicating social class are stronger and the effects of school, respectively the scores on the final aptitude test are weaker, which probably can be attributed to the stronger cultural attachment of young children to their parents as well as to the external differentiation of secondary education. Most remarkable and intriguing for the Prima-4 data set is the strong relationship between gender and activity, confirming the feminization and "Divafication" of the new genres (Lister, 2001).

Comparing the genres' positions on the dimensions, there is yet another interesting observation to make. Whereas the new genres of Alternative Rock, Teen Pop, Latin, Urban and Rap so clearly diverge on the lowbrow-highbrow dimension, they closely converge on the dimension of receptiveness versus activity. It seems as if these dimensions, in fact often representing latent rather than manifest opinions, feelings and attitudes of the audience, can be treated as counteracting forces. As such, their combined effect may be highly informative about the fate of genres. The diagram shown in Figure 11, resulting from an analysis compressing the variance in preference ratings in just two dimensions makes some of these combined effects visible.

Figure 11: The convergence of the new genres (Prima data set, 2000/1; n = 18,857; mean age: 11.6); Alscal: analysis forced to two dimensions

Will the primary school pupils keep to their preferences when entering secondary education or even later on? Actually, it is difficult to say if the differences between both data sets of primary and secondary education reflect an age effect, a period effect or a generation effect. A powerful argument for an age effect can be derived from the strong gender effect in the Prima-4 data set on the "activity" dimension. One may easily expect that, sooner or later, most boys will catch up with the girls at the time of going through the first years of secondary education. There may be other things that will disappear by growing-up — the one-ear-deafness for older genres, for instance. However, the new genres in question kept dominating the charts since 2001 which, of course, points to a period effect and even strengthens the idea that we may be witnessing a generational shift with the youngest part of the audience, the girls in front, regrouping around genres like Rap, Urban, Latin and Teen Pop and even finding a common interest in these genres. In this respect it is interesting to note that these genres, over the last years, not only raised their chart positions considerably but also have been merging on the production side by interchanging producers as well as artists.

The changing parameters of pop music preferences. The history of pop music is a multi-layered process. One side is taken by the artistic process of creating songs, the other by the process of their adoption by the audience with the charts validating the winning matches in-between. Though well-documented and preserved, the charts will disclose little information about the changing tastes of the audience at large. Delving deeper and looking at the outcomes of empirical research, we've walked by some of these changes among pop music's primary audience of adolescents. Neglecting the question of national and regional differences (Wells and Tokinoya, 1998), we have been taking long steps over the past decennia with our data sets restricted to the Dutch situation and their dates arbitrarily set by their availability. Apart from the immanent problem of the varying genre pools, the presented data sets, admittedly, are quite dissimilar. The technique of principal component analysis too seems to have its shortcomings. For instance, the dimensions accounting for the largest part of explained variance in each analysis, clearly, attract some superfluous air that can be attributed to differences in the attitude towards music in general, be it distaste or critical distance. We left it in for technical reasons and to keep close to the original data. More important, genre labels always lag behind the development of the songs that, of course, constitute the real currency of the field (Ennis, 1992).

Despite all these shortcomings, but helped by the amazing stability of the factors of pop music preferences over time and place, we think to have located some important and comprehensible coordinates of the field of pop music along which artists, songs, genres and their audiences keeps moving. The songs, artists and genres of pop music may change, but over time their movements seems to remain within the framework of the same oppositions, delineated by the same four dimensions. This may be seen as the main conclusion of our argument. Despite all that is changing, the struggle on taste still takes place within the same arena. Our search to map the continental drift of pop music preferences along these coordinates may raise more questions than answers. Still, the outcomes largely confirm what is already known about pop music's course over generations of youth within the last fifty years: the fragmentation of the field in the 1970's constituting the typical criss-cross pattern of Murdock and Phelps; the assumed shift between the weight of the diagonals by the mid-1980's and the simultaneous filling up and expansion of the centre field; the budding of new competitive genres in the 1990's, ultimately destroying the criss-cross pattern and even threatening to throw out some older genres completely while the youngest part of the pop audience — girls in front — at the same time seems to restore some balance by adopting Teen Pop and its likes.

Only a small percentage of variance in music preferences actually is explained by the dimensions - though more so by the end of the 1990's than in the mid-1980's and at the Millennium change more within the older age groups in secondary education than within the younger age groups in primary education. Social class, schooling and gender, in turn, only explain the dimensions to a small degree. This clearly testifies to pop music's open and democratic character. For the main body of young people the genres serve as marking points in an accessible musical space they freely walk through rather than being exclusive sanctuaries of total identification. This openness, though, is not to be equated with the withering away of the effects of social class in the cultural domain. In respect to omnivorism Van Eijck a.o. (2002) rightfully point to the expansion of the professional middle class. High culture forms of old, like Classical Music, may have lost much of their distinctive power by the process of traditionalization and contraction. Their place and function, however, seem to be taken by highbrow variants of pop music, mediated by level of schooling and constituting the cultural split Stevens (2001) observed within the middle class itself. Gender, ethnicity and the level of urbanization also situate specific positions on the map of popular music preferences.

The four dimensions we have been discussing, not only make up coordinates but also act like forces joining genres together or dividing them. In this respect the dimension of "activity," despite being badly labelled, clearly deserves renewed theoretical attention as it seems capable of joining genres that are being kept apart by the social dividers of social class, gender and ethnicity. Undoubtedly, this dimension points towards the shared conditions, problems and practices that tend to unite youth as a social category and to which they apply most of their leisure and music. As some have argued, the social forms in which these common interests of youth manifest themselves seem to have changed over time — from subcultures into styles and then again into the new voluntary and temporary forms of sociality of neo-tribes (Hebdige, 1979; Maffesoli; 1988; Bennett, 1999). If so, this may offer some explanation, not only for the rise and fall, but also for the divergence and convergence of pop music's genres.

Author's Note

The Groningen data set was collected during a small-scale research project on school and leisure in the North of the Netherlands and reported by Tillekens (1993). The second data set is based on a questionnaire on music preferences collected during the Dutch part of the Health Behaviour in School-Aged Children project (HBSC) as reported by Ter Bogt a.o. (2003) and Mulder a.o. (2005). The third project (Prima-4) regards a survey of music preferences, inserted as an additional module in the fourth round of a regular biennial longitudinal research project — the Longitudinal Data Collection Primary Education — in the Netherlands. The basic outcomes of this project were reported by Driessen a.o. (2002) and Van der Veen a.o. (2002).

References

Anderson, Bruce, Peter Hesbacher, K. Peter Etzkorn, and R. Serge Denisoff (1980), "Hit record trends, 1940-1977." In: Journal of Communication, 30, 2, 31-43.
Arnett, Jeffrey Jensen (1996), Metalheads. Heavy metal music and adolescent alienation. Boulder, Colorado: Westview Press.
Bennett, Andy (1999), "Subcultures or neo-tribes? Rethinking the relationship between youth, style and musical taste." In: Sociology, 33, 3, 599-617.
Bryson, Bethany (1996), "Anything but Heavy Metal. Symbolic exclusion and musical dislikes." In: American Sociological Review, 61, 5, 884-899.
Bryson, Bethany (1997), "What about the univores? Musical dislikes and group based identity construction among Americans with low levels of education." In: Poetics, 25, 2/3, 141-156.
Cepeda, María Elena (2000), "Mucho loco for Ricky Martin, or the politics of chronology. Crossover and language within the Latin(o) music 'boom.'" In: Popular Music and Society, 24, 3, 55-71.
Christenson, Peter G. (1994), "Childhood patterns of music uses and preferences." In: Communication Reports, 7, 2, 136-144.
Christenson, Peter G., and Donald F. Roberts (1998), It's not only Rock'n'Roll. Popular music in the lives of adolescents. New Jersey: Hampton Press.
Coleman, James (1961), The adolescent society. The social life of the teenager and its impact on education. New York: Free Press of Glencoe.
De Leeuw, Kitty, Sjouk Hoitsma, Ingrid de Jager and Peter Schonewille (2001), Jong! Jongerencultuur in Nederland, 1950-2000. [Youth! Youth culture in the Netherlands, 1950-2000.] Zwolle: Waanders.
De Meyer, Gust (1980), Jeugd en popmuziek. [Youth and pop music.] Leuven: Centrum voor Communicatiewetenschappen.
Dheil, E. Roderick, Michael J. Schneider and Kenneth C. Petress (1983), "Dimensions of musical preference." In: Popular Music and Society, 4, 3, 41-50.
Driessen, Geert, Annemarie van Langen and Hermann Vierke (2002), Basisonderwijs: veldwerkverslag, leerlinggegevens en oudervragenlijsten. Basisrapportage Prima-cohortonderzoek. [Primary education: elementary data on pupils and parents from the Prima-cohort, 2000-2001.] Nijmegen: ITS.
Ennis, Philip H. (1992), The seventh stream. The emergence of rock'n'roll in American popular music. Hanover; London: Wesleyan University Press, 1992.
Finnas, Leif (1987), "Do young people misjudge each other's musical taste?" In: Psychology of Music, 15, 2, 152-166.
Fink, Edward L., John P. Robinson and Sue Dowden (1985), "The structure of music preference and attendance." In: Communication Research, 12, 3, 301-318.
Frith, Simon (1978), The sociology of rock. London: Constable.
Frith, Simon (2000), "The centre writes back. A response to Young." In: Young, 8, 2.
Grossberg, Lawrence (1992), We gotta get out of this place. Popular conservatism and postmodern culture. New York: Routledge.
Hakanen, Ernest A. (1998), "Counting down to number one. The evolution of the meaning of popular music charts." In: Popular Music, 17, 1, 95-111.
Hakanen, Ernest A., and Alan Wells (1990), "Adolescent music marginals. Who likes Metal, Jazz, Country, and Classical?" In: Popular Music and Society, 14, 4, 57-66.
Hakanen, Ernest A., and Alan Wells (1993), "Music preference and taste cultures among adolescents." In: Popular Music and Society, 17, 1, 55-69.
Hargreaves, David J., and Adrian C. North (1999), "Developing concepts of musical style." In: Musicae Scientiae, 3, 2, 193-216.
Hebdige, Dick (1979), Subculture. The meaning of style. Londen: Methuen.
Janssen, Jacques, and Maerten Prins (1991), "Jeugdsubculturen binnenste buiten. Een onderzoek naar homologie van jeugdsubculturen." [Youth cultures inside out. Homologies in youth subcultures.] In: Jeugd en Samenleving, 21, 2/3, 194-212.
Konig, Ruben (2002), "On the rotation of non-linear principal components analysis (Princals) solutions. Description of a procedure." In: ZUMA-Nachrichten, 26 (50), 114-120.
Lister, Linda (2001), "Divafication. The deification of modern female pop stars." In: Popular Music and Society, 25, 3/4, 1-10.
Lopes, Paul D. (1992), "Innovation and diversity in the popular music industry, 1969-1990." In: American Sociological Review, 57, 1, 56-71.
Maffesoli, Michel (1988), Le temps des tribus. Le déclin de l'individualisme dans les sociétés de masse. Paris: Méridiens Klincksieck, 1988.
Matza, David (1961), "Subterranean traditions of youth." In: The Annals of the American Academy of Political and Social Science, 338, 102-118.
Mizell, Lee, Brett Crawford and Caryn Anderson (2003), Music preferences in the US, 1982-2002. Report for the National Endowment for the Arts. Santa Monica, California: Lee Mizell Consulting.
Mulder, Juul, Tom ter Bogt, Quinten Raaijmakers and Wilma Vollebergh (2005), Pop music preferences and psychosocial well-being of adolescents. Amsterdam: Ascor (article submitted for publication).
Murdock, Graham, and Guy Phelps (1972), "Responding to popular music. Criteria of classification and choice among English teenagers." In: Popular Music and Society, 1, 3, 144-151.
Murdock, Graham, and Guy Phelps (1973), Mass media and the secondary school. London: Macmillan.
Mutsaers, Lutgard (1990), "Indorock. An early Eurorock style." In: Popular Music, 9, 3, 307-320.
Nagel, Ineke (2002), "Op welke leeftijd lijken kinderen het meest op hun ouders? Cultuurparticipatie tussen zes en achttien jaar." [From six to eighteen — at which age do children take most after their parents?] In: Mens en Maatschappij, 77, 3, 207-230.
Parker, Martin (1991), "Reading the charts. Making sense with the hit parade." In: Popular Music, 10, 2, 205-217.
Peterson, Richard A., and Paul Di Maggio (1975), "From region to class, the changing locus of country music. A test of the massification hypothesis." In: Social Forces, 53, 3, 497-506.
Peterson, Richard A., and Roger Kern (1996), "Changing highbrow taste. From snob to omnivore." In: American Sociological Review, 61, 5, 900-907.
Rentfrow, Peter J., and Samuel D. Gosling (2003), "The do-re-mi's of everyday life. The structure and personality correlates of music preferences." In: Journal of Personality and Social Psychology, 84, 6, 1236-1256.
Roe, Keith (1992), "Different destinies — different melodies. School achievement, anticipated status, and adolescents' music use." In: European Journal of Communication, 7, 3, 335-358.
Schwartz, Kelly D., and Gregory T. Fouts (2003), "Music preferences, personality style, and developmental issues of adolescents." In: Journal of Youth and Adolescence, 32, 3, 205-213.
Sieben, Inge (2001), Sibling similarities and social stratification. The impact of family background characteristics across countries and cohorts. Nijmegen: Katholieke Universiteit Nijmegen.
Sikkema, Paul (1988), Jeugd nu. [Today's youngsters.] Den Haag: SDU.
Stevens, Frank (2001), "Gemaakte keuzes? Een analyse van de muziek- en mediapreferenties van Vlaamse jongeren." [Choices of their own? Music preferences and media involvement of Flemish youth.] In: Sociologische Gids, 48, 2, 138-155.
Ter Bogt, Tom, and Belinda Hibbel (2001), Wilde jaren. Een eeuw jeugdcultuur. [Wild times. One hundred years of youth culture.] Utrecht: Lemma.
Ter Bogt, Tom, Saskia van Dorsselaer and Wilma Vollebergh (2003), Psychische gezondheid, risicogedrag en welbevinden van Nederlandse scholieren, HBSC Nederland 2002. [Health Behaviour in Dutch School-Aged Children.] Utrecht: Trimbos Instituut.
Tillekens, Ger (1993), "Het patroon van de popmuziek. De vier dimensies van jeugdstijlen." [The patterns of pop. The four dimensions of youth styles.] In: Sociologische Gids, 40, 2, 177-194.
Tillekens, Ger (1998), Het geluid van de Beatles. [The sound of the Beatles.] Amsterdam: Het Spinhuis.
Van Bork, Remko, and Jan Jacobs (1986), Popmuziek. Het geluid van jongeren. [Pop music. The sound of youth.] Muiderberg: Coutinho.
Van Eijck, Koen (1997), "The impact of family background and educational attainment on cultural consumption. A sibling analysis." In: Poetics, 25, 4, 195-224.
Van Eijck, Koen (1999), "Socialization, education, and life style. How social mobility increases heterogeneity of status groups." In: Poetics, 26, 5/6, 309-328.
Van Eijck, Koen (2001), "Social differentiation in musical taste patterns." In: Social Forces, 79, 3, 1163-1184.
Van Eijck, Koen, Jos de Haan and Wim Knulst (2002), "Snobisme hoeft niet meer. De interesse voor hoge cultuur in een smaakdemocratie." [No more need for snobbery. The interest in highbrow culture in a taste society.] In: Mens en Maatschappij, 77, 2, 153-177.
Van der Veen, Hendrina, Arjan van der Meijden and Guuske Ledoux (2002), School- en klaskenmerken basisonderwijs. Basisrapportage Prima-cohortonderzoek, vierde meting 2000-2001. [Primary education: elementary data on schools and classes from the Prima-cohort, 2000-2001.] Amsterdam: SCO-Kohnstamm Instituut.
Van Wel, Frits, and Daan van der Gouwe (1990), "Smaken verschillen. Jongeren over muziek, boeken en films." [Tastes do differ. Adolescents and their favourite music, books and movies.] In: Comenius, 10, 40, 543-554.
Wells, Alan, and Hiroshi Tokinoya (1998), "The genre preferences of Western popular music by Japanese adolescents." In: Popular Music and Society, 22, 1, 41-53.
Willis, Paul (1978), Profane culture. London: Routledge and Kegan Paul.