Quantitative research in musicology

Quantitative methods and the public in musicology

Quantitative methods are a great help in investigating what moves the members of the public: the helpers, the receivers, the recoders, those providing the feedback for a successful performance, the re-users, and the recontextualisers of the music. The lack of interest in ­musicology in applying these methods could signify a lack of interest in the public. Indeed, much attention is often given to the musical product and its producers. With the exception of popular music studies, the public often comes off badly. This is puzzling since conveying is an undisputed aspect of the musical process.

Except for the study of musical material itself, musicologists do not often use quantitative methods. Reducing an individual to a set of data and a group of people to a data matrix is, admittedly, disrespectful. But quantitative methods can also be used to support the qualitative work. If the collection and analysis of quantitative data form a modest part of the research, they may enhance its significance, especially when attention to larger groups involved in the music studied is necessary. Data gathered in qualitative research and the results of the analysis of that data should be used to interpret or engage in dialogue about the quantitative data. And, the results of the quantitative data analysis may be used to throw light upon the qualitative material.

Audience research, then, is not particularly at the centre of musicological attention. Yet ethnographical methods, like those used in anthropology and musicology, are at the centre of attention of (media) audience research after the latter, like some other branches of social and cultural studies in the 1990s, made an ‘ethnographic turn’ (Livingstone 2003: 344):

Thus, the crucial transition was made from text to context, from literary/semiotic analysis to social analysis¹Morley, D. 1992. Television, audiences and cultural studies. London: Routledge.). Of course, these should not be posed as either/or options, for the moment of reception is precisely at the interface between textual and social determinations and so requires a dual focus on media content and audience response.

By nature, audience definition is one of the concerns of audience research. Livingstone (2003) outlines that in the second half of the twentieth century, the description of western audiences has seen a shift in conception and attention from the mass audience that passively consumed centrally broadcast content to an audience of individual members, all being constructors of meaning. Secondly, the widespread metaphor of communication transmission, which assumes that communication merely requires the more or less efficient transport of fixed and already-meaningful messages in a linear manner from sender to receiver, has been contrasted with the idea of a cyclical process, a circuit of articulated practices – production, circulation, reception, reproduction – each of which represents a site and aspect of meaning-making. It is stimulating to see that the emergence of the new, individualised media has brought the audience researchers to using ethnographic methods. How embedded the idea of a passive, mass audience was, and possibly still is, in the west is exemplified by the fact that in many fields of cultural studies, the central focus of attention was great artists or thinkers and their products. This expresses itself in the fact that in audience research some findings are considered to be new that reveal more than they bring, like: “The life of signs within modern society is in large measure an accomplishment of the audience” (Bruhn Jensen as quoted in Livingstone 2003: 343). The strange thing is that though this assertion may sound obvious to most musicologists, so little research has been done into that audience.

Though the research presented here was, for an important part, done in the 1980s, it addresses some of the questions, if paraphrased, posed by present-day audience research. How were relations among people framed as one of ‘audience’? Did certain kinds of texts produce certain kinds of audience? How did the social relations among people shape the communicative possibilities of gatherings with music, enabling some and inhibiting others? Which were the intersections and disjunctions between processes of encoding (makers) and decoding (public), contextualising both within a complex cultural framework? How did people’s tacit knowledge ‘fill the gaps’ or reframe the meaning of texts, resulting in divergent interpretations of the very same texts?

As Robert Allen (1987)²Allen, Robert C. (ed.) 1987. Channels of discourse. Chapel Hill: University of North Carolina Press. argued, once textual and literary theorists had made the crucial transition to a reader-oriented approach, context flooded in for two reasons: first, the shift from asking about meaning of the text in and of itself, to asking about the meaning of the text as achieved by a particular, contextualised reader (i.e. the shift, in Eco’s terms, from the virtual to the realised text); second, the shift from asking about the meaning of the text to asking about the intelligibility of the text (i.e. about the diversity of sociocultural conditions which determine how a text can make sense).³Livingstone (2003: 12).

Usefulness of surveys

One should not expect much new data from a survey. For the greater part, the result will be many ordered lists of answers given and a run-down of adjustments and confirmations of things already thought of or guessed at. The possibilities to combine data for analysis make a survey worthwhile. These form a good starting-point for further exploration of the subject and may serve to generate new ideas. It is mainly this exploratory side of surveys that is useful; we do not seek to prove anything with statistics.

A survey should be based on an adequate knowledge of the field, not on a lack of knowledge. A survey as a major method of data collecting or a survey held at the beginning of the research period will more often than not yield superficial, incomplete, or unreliable data and data interpretation.

Issues with surveys are many. To mention a few: the unfamiliarity of the informants with the interview situation and style, the rapport between interviewer and informant, the formulation of questions, the order of questions, the representativeness of the samples, the categorisation of answers, the handling of missing data, the use of appropriate analysis techniques and the checking of interpretations made on the basis of the data.
The returns, however, can be good, if the quantifications and the analyses are interpreted against the background of an independent source of information, usually the researcher’s own comprehensive qualitative research and possibly that of others. However, the help of local specialists in interpreting the results of a survey, as done in this research, is indispensable.

Data analysis

It may be a comfort to many musicologists that data analysts are not uncompromising statisticians. They even tend to say understandable things. According to Tukey, the pioneer of such data analysis, exploratory data analysis is an attitude and a flexibility much more than a catalogue of techniques: “a willingness to look for what can be seen, whether anticipated or not wholly anticipated” while recognising “that the picture-examining eye is the best finder we have of the wholly anticipated”.⁴Tukey as quoted in Gifi (1981: 27,28). In the cyclic process of conjectures and refutations that forms development and discovery, the task of exploratory data analysis is to come up with conjectures. As a result, the interpretability of representations of data is of prime importance. Practicable quantitative techniques of (exploratory) data analysis in musical preference research should yield visual representations of data that appeal to the eye’s pattern recognising ability.

Plots

At various places in this study, plots are used to visually represent quantitative data. In the plots, the relative positions of the categories of all variables in the analysis are represented in one space that, for reasons of interpretability, has only two dimensions. The first two dimensions often give us a fairly clear idea of the most important effects in the data.⁵Gifi (1981: 58).

The plots use the intuitively appealing image of distance for dissimilarity. Consequent­ly, the categories belonging to the ‘average set of answers’ find a place close to the origin (zero-point), and the more peculiar an answer (category) is, the larger the distance to the origin. In addition, the angle made by the lines drawn from the origin to points corresponding to the categories of two different variables is informative: when two lines point in the same direction, this indicates a positive relationship between the two categories involved; when they point in opposite directions, a negative relationship is indicated; when they make an angle of 90º, there is no relationship. Read more about the two forms of correspondence analysis used.