Consumer Esthetics Outside the Lab: Preliminary Report on a Musical Field Study

James J. Kellaris, University of Cincinnati
ABSTRACT - This paper reports preliminary analyses on a field study designed to explore responses to music as a function of its objective properties. Musical stimulus properties were manipulated in a quasi-experimental design using live music. Behavioral responses of audiences were unobtrusively observed. Hedonic response, operationalized as applause duration, was found to be a nonmonotonic function of musical tempo; however, the specific nature of this relationship varied across modes.
[ to cite ]:
James J. Kellaris (1992) ,"Consumer Esthetics Outside the Lab: Preliminary Report on a Musical Field Study", in NA - Advances in Consumer Research Volume 19, eds. John F. Sherry, Jr. and Brian Sternthal, Provo, UT : Association for Consumer Research, Pages: 730-734.

Advances in Consumer Research Volume 19, 1992      Pages 730-734

CONSUMER ESTHETICS OUTSIDE THE LAB: PRELIMINARY REPORT ON A MUSICAL FIELD STUDY

James J. Kellaris, University of Cincinnati

ABSTRACT -

This paper reports preliminary analyses on a field study designed to explore responses to music as a function of its objective properties. Musical stimulus properties were manipulated in a quasi-experimental design using live music. Behavioral responses of audiences were unobtrusively observed. Hedonic response, operationalized as applause duration, was found to be a nonmonotonic function of musical tempo; however, the specific nature of this relationship varied across modes.

INTRODUCTION

Aesthetic objects can evoke numerous affective, cognitive, and behavioral responses in consumers of the "aesthetic experience." While some forms of aesthetic communication stimulate changes primarily in internal states (e.g. visual arts), other forms are strongly associated with overt behavioral responses. For example, musical performances often encourage audience participation in the forms of rythmic hand-clapping, body swaying, dancing, singing along, shouting, and applauding. These common forms of participation in the music-consumptive experience are shared across many cultures.

Most previous studies in consumer esthetics have focused on affective and cognitive responses to aesthetic stimuli (e.g. Anand and Holbrook 1986; Kellaris and Kent 1991). The majority of these studies were conducted as controlled experiments involving forced exposure to aesthetic stimuli in laboratory settings. In many cases, artificial stimuli such as "tone sequences" (Crozier 1974) or digitally synthesized music was used.

While lab studies can offer valuable insight into the influence of music on consumers, it would seem desirable to study music where it actually touches the hearts and lives of listeners. As Pavelchak et al. (1988) stated, a real world setting is important because "context is an integral part of emotional experience and because results obtained in laboratory settings need to be externally validated" (p. 360).

The present study examines behavioral responses to an aesthetic object -- Greek music in this case -- in natural settings. This preliminary report focuses on the influence of two structural properties of music (tempo and modality) on hedonic response as measured by applause duration. Tempo and mode were selected for study because of their theoretic and practical importance as "design features" of musical "products." This study combines elements of previous consumer research to produce a unique contribution. Whereas previous field studies have examined influences of music as an environmental background feature (e.g. Milliman 1982, 1986; Yalch and Spangenberg 1990), and controlled lab experiments have examined self-reported responses to music as a focal stimulus object (e.g. Anand and Holbrook 1986; Kellaris and Kent 1991), the present study examines a behavioral response to music in natural settings where music is the focal stimulus object.

BACKGROUND & HYPOTHESES

As Bruner (1990) states, "music is not simply a generic sonic mass, but rather a complex chemistry of ... elements" (p. 94). Music is composed of multiple time-, pitch-, and texture-related variables (Dowling and Harwood 1986). Bruner proposes that musical variables are capable of producing main and interactive effects on consumers' thoughts, feelings, and behaviors. There is scattered evidence that supports this proposition, but Bruner concludes that "at this time the relevant body of literature is still meager" (p. 102). More research relating musical variables to listeners' responses is needed.

Among musical variables, tempo (the speed or pace of music) is believed to make a particularly important contribution to music's aesthetic character (Hevner 1937; Rigg 1940). While some studies report a positive linear relationship between tempo and affect (e.g. LeBlanc and McCrary 1983), there is ample reason to expect a nonmonotonic relationship (Anand and Holbrook 1986). Theory suggests that positive affect should be greatest at a moderate level of arousal (Wundt 1874; Berlyne 1974). Since arousal is closely linked with stimulus intensity, variation in affect can be anticipated across stimulus patterns (Vitz 1966).

This principle can be applied to musical stimuli. Specifically, very slow or very fast musical tempi should produce less positive affect than a moderate pace. Thus, across a wide range of musical examples, one might anticipate a nonmonotonic relationship between tempo and hedonic response. Although this study is exploratory in nature, there is sufficient basis to hypothesize the following general relationship:

H1: There will be a significant nonmonotonic relationship between musical tempo and hedonic response of listeners.

Furthermore, tempo may interact with other musical variables (Bruner 1990). A recent lab study by Kellaris and Kent (1991) found that the effects of tempo depend on the "modality" (key) of the music. Modality refers to the configuration of intervals between pitches that comprise a musical scale (Apel 1973). The influence of tempo on music's perceived appealingness differed across major, minor, and atonal modes in the Kellaris-Kent experiments. Given the expectation of a positive correlation between perceived appealingness and behavioral manifestations of hedonic pleasure, the moderating effect of modality should generalize to the tempo-applause relationship in a natural setting. Thus:

H2: Musical modality will moderate the influence of tempo on hedonic response.

The present study examines the influence of tempo and modality in the real world beyond the behavioral lab, using unobtrusively observed behavioral measures of hedonic response to live Greek music. Greek music was chosen because it employs a wide range of tempos and modes, and because it customarily involves active audience participation representing a broad range of overt behaviors (Holst 1975). The focus of this preliminary report, however, is solely on applause. Applause is a simple indicant of hedonic response which is nearly universal across musical audiences.

METHOD

Overview

The purpose of this study is to examine the influence of objective stimulus properties of music on audience responses in natural settings. Songs were selected from the standard Greek repertoire to manipulate musical variables in a quasi-experimental design. These songs were performed live at six different social events in hellenic communities in the United States. Behavioral responses of audiences were unobtrusively observed and recorded by the author, who participated as a musician in an orchestra. Musical properties and outcome variables were coded by experts and data were analyzed using the individual song performance as the unit of analysis.

Subjects

Audience members at six events served as subjects. The events included an AHEPA conference ball, two Greek Orthodox wedding receptions, and outdoor Greek Festivals in three southeastern cities. Audience sizes ranged from 200 guests at one wedding reception to over 1000 at the outdoor festivals. Subjects were not told that their behavior was being observed.

Stimulus Materials

Forty songs from the standard Greek dance music repertoire were used to manipulate musical properties. The songs, performed at each of six events, included traditional and modern pieces, both laiki (urban) and demotiki (rural mainland and island) genres, representing multiple dance rhythms (e.g. calamatiano, syrto, tsamiko, syrtaki, etc.). About a third of the songs were sung and the remainder performed as instrumentals. Most of the songs were familiar to Greek audience members, as supported by the observation that many "sang along" with the tunes, including those performed instrumentally.

At each performance the songs were presented in one hour "sets," with a brief pause between selections and a break of ten to fifteen minutes between sets. Each set consisted of about ten songs. Song durations ranged from just over one minute to almost seven minutes, with a median time of about 3.5 minutes. The order in which the songs were played varied across performances.

The individual song performance was the unit of analysis. Some cases were lost due to technical problems with recording equipment, and some could not be used due to missing data. The usable sample included 192 complete observations.

Procedure

The stimulus songs were performed by Nick Demos and the Greek Islanders, a Greek-American musical group that performs predominantly in the southeastern United States. The instrumentation of the orchestra included an eight-stringed electric bouzouki (played by the author), clarinet, electronic keyboard, bass, and kit of drums, including middle-eastern percussion instruments (e.g. doumbek).

The music was preselected and the sequence of performance determined in cooperation with the orchestra leader. Each performance was unobtrusively tape-recorded using a Toshiba RT-7016 to facilitate coding at a later time.

Measures and Coding

Independent variables included the following objective measures of musical stimulus properties: (X1) tempo, coded in beats per minute (BPM), (X2) duration of song, coded in seconds, (X3) meter, coded categorically as "even" (duple meter), "triple," or "odd/mixed," (X4) modality, coded categorically as "major," "minor," or "other" keys, (X5) presence of lyrics, coded as "1" for vocals and "0" for instruments, (X6) instrumentation, coded categorically according to which instrument was the principal melodic or solo instrument for a given song, and (X7) dance type (i.e., calamatiano, syrto, tsamiko, etc.), also coded categorically. These variables are a representative subset of the cantometrics taxonomy (Lomax 1976) used in empirical ethnomusicological research, and are similar to the set used by Stout and Leckenby (1988) in their ad music study, with obvious ethnic adaptations.

Both the researcher and a doctoral candidate in Musical Arts coded the independent variables for each song by listening to audio recordings of the performances. A Seiko digital metronome (SQM-342) was used to measure tempo in BPM. Disparities in coding judgments were reconciled by discussion.

To allow assessment of possible confounding effects, the following variables were also included in the data set: (X8) "gig," which specified the performance at which a given song was performed, (X9) "set," which specified the grouping in which a given song was performed, and (X10) order of presentation.

The dependent variable was hedonic response, measured as duration of applause in seconds. There is ample precedent in the social sciences for using applause as an indicant of audience pleasure or approval (e.g. Heritage and Greatbatch 1986). Applause duration was coded by the author by listening to tapes of each performance. As a check on the coding, a Musical Arts student coded a subset of the stimulus songs. There were no coding disagreements. The coding procedure was pretested using a video tape recording of a live performance. Pretest data were not included in the data set analyzed in this study.

ANALYSES

Confounding Checks

Lack of experimental control is the bane of quasi-experimental field studies such as the present one (Campbell and Stanley 1963). Because of this inherent lack of control, data analysis began by performing statistical confounding checks. A variety of metric and nonparametric tests were used since the independent set includes categorical, ordinal, and metric variables.

Based on conventional wisdom, one might anticipate a natural rise ("warm-up" effect) and fall ("fatigue" effect) in hedonic response over the course of a performance. This effect should be independent of the influence of musical variables; however, serial order of presentation may be correlated with musical variables in the data set. Thus, the first confounding check explored the relationship of serial order to tempo and mode. No statistical associations were found.

Other confounding checks explored relationships between tempo, mode, and other musical characteristic. Not surprisingly, tempo was confounded with two other time-related variables: meter (the division of musical pulse into strong and weak beats) and song duration; however, neither meter nor duration was statistically related to applause duration after controlling for tempo. Since these variables have no effect on applause independent of their association with tempo, this confounding is not problematic. Tempo was statistically unrelated to the other musical variables.

A final confounding check explored performance context effects. Since audience sizes and composition varied across performances, performance context may have influenced responses to the music. Although the distribution of tempi and modalities varied slightly across performances, this variation was not statistically significant. This suggests that data can be meaningfully aggregated across performances.

H1: Tempo Effects

Based on earlier work in empirical esthetics (e.g. Fetchner 1876) and more recent work in consumer esthetics (e.g. Anand and Holbrook 1986), a nonmonotonic tempo-response relationship was anticipated. But, whereas previous studies have reported several different types of nonmonotonic relationships, the present analysis explored applause duration as a logarithmic, power, and quadratic function of tempo using OLS regression.

A significant nonmonotonic (quadratic) tempo-applause relationship was found. This finding supports Hypothesis 1. Applause duration peaked around 147 BPM, then diminished with further increases in tempo. The fit of this curve was modest but statistically significant (R2 = .06; F = 6.15, d.f. = 192, p < .003), with significant contributions from both tempo (t = 2.41, p < .02) and tempo-squared (t = -2.21, p < .03). Although the R-squared is low relative to those found in lab studies, this is what one might expect to find outside the controlled environment of the behavioral lab due to extraneous "noise" (McGrath 1981).

H2: Moderating Effect of Mode

Given Kellaris and Kent's (1991) finding that modality moderates the influence of tempo, separate analyses were conducted for major, minor, and "other" (neither major nor minor) modes. Findings support Hypothesis 2.

A significant quadratic tempo-applause relationship was found for the 51 songs pitched in minor keys (R2 = .22; F = 6.85, p < .002), with significant contributions from tempo (t = 3.48, p < .001), tempo-squared (t = -3.38, p < .002), and a constant term (t = -2.55, p < .02).

For the 106 songs pitched in major keys, a different type of nonmonotonic tempo-applause relationship emerged. A logarithmic function produced the best fit (R2 = .07; F = 7.64, p < .01). Only the log of tempo contributed significantly (t = 2.76, p < .01) to the prediction of applause duration.

Among the 33 songs in non-major or minor modes, no statistically significant linear, logarithmic, exponential, or quadratic relationship was found between tempo and applause duration.

To summarize, the data provide support for both hypotheses. There is a nonmonotonic relationship between musical tempo and hedonic response. The specific nature of this relationship varies by mode.

DISCUSSION & CONCLUSION

This study examined a behavioral response to an aesthetic object as a function of its objective properties in natural settings. It found significant nonmonotonic relationships between the tempo of live Greek music and hedonic response of audiences as measured by applause duration. The strength and nature of the relationship varied across different modes. Consistent with Kellaris and Kent's (1991) recent lab findings, tempo and modality were found to exert an interactive influence on hedonic response. Specifically, applause duration was a quadratic function of tempo for songs pitched in minor keys, and a linear function of the logarithm of tempo for songs pitched in major keys. There was no statistical relationship between tempo and applause for songs pitched in non-major or minor keys.

Over a fairly wide range of tempi (68 to 178 BPM), and across all modes, applause duration followed an inverted U-shaped curve as tempo increased. This is the "Wundt-curve" pattern predicted by psychobiological theory (Berlyne 1974; Anand and Holbrook 1986). The finding indicates that listeners prefer a moderate level of stimulation. This quadratic response pattern held for songs pitched in minor keys, but changed for major and other modes.

The loglinear response pattern for songs in major keys may be explained in terms of the relative informational complexity of major vs. minor modes. Minor modes contain altered pitches -- notes that depart from the natural overtone series on which major scales are based (Apel 1973). In information theoretic terms, minor modes have higher "uncertainty levels" than major keys (Crozier 1974). Major keys, in contrast, may be characterized as less complex, or lower in uncertainty. For any mode, increases in speed create greater "informational density," i.e., more notes per minute. Affective arousal is, in part, a function of the amount of information contained in a musical stimulus (Berlyne 1974). Since major keys contain less varied information than minor keys, one might expect an eventual decrease in hedonic response to occur at a higher tempo. In other words, the observations in the present study may simply be on the "uphill" side of an inverted U-shaped "Wundt curve" for major keys. This explanation could be tested using a more extreme range of tempi; however, the songs in the present study represent a realistic range. Listeners are unlikely to encounter dance music performed at speeds greater than the max of 178 BPM in this study.

There was no statistical tempo-applause relationship among the non-major or minor songs. The relative novelty of these modes which used atypical combinations of pitches may have called more attention to the pitch dimension of the music, thereby mitigating effects stemming from the temporal dimension (i.e., from tempo). Alternatively, there may have been too few observations in this category to allow the presence of tempo effects to be detected.

The generality of these findings are limited by the design and setting. Since the data were collected in natural settings, rigorous experimental control was not possible. As a result, the design was nonorthogonal, cell sizes were uneven, and there was some confounding of independent variables with exogenous influences. The audiences were self-selected. But, whereas the present findings are consistent with theory and previous findings in consumer esthetics, the tempo-hedonic response relationship is apparently robust to the methodological perils of field settings.

Although a number of statistical checks were performed to assess potential confounding, these tests did not examine the relationship of objective musical attributes to qualitative variables such as subjective meaning. For example, faster and slower songs may have differed not only in tempo, but in meanings, associations, and imagery evoked.

Generality of findings is also limited by the type of music used in this study. Greek music tends to be highly rhythmic, and perhaps emphasizes the temporal dimension of music to a greater extent than do some other types of music. This may make temporal effects more likely to emerge. The effects reported here may generalize to other highly rhythmic forms of music, such as rock, hip-hop, or reggae, but may not generalize to all styles of music.

Having recognized these limitations, some preliminary implications can be drawn. The first implication concerns external validity. This study demonstrates the external validity of laboratory experiments that found objective properties of sound to influence consumers' responses to music. It is important to verify that such relationships exist outside the artificial world of the behavioral lab.

Second, the substantive findings on tempo's influence hold practical implications for parties who use music to influence consumers. There may be an optimal tempo at which hedonic response is maximized for different types of music. Perhaps the design features of music in ads or in retail environments can be "aesthetically engineered" to optimize music's positive influences on listeners.

Third, the finding on the moderating role of modality suggests that time- and pitch-related musical properties should be considered jointly. Several studies in marketing (e.g. Milliman 1982, 1986) have varied a single property of music without considering possible interactive influences. As Anand and Holbrook (1986) suggested, "The term 'music' denotes a wide range of stimuli ... that hang together in an extremely complex gestalt. One pulls apart this gestalt only at one's peril" (p. 655).

This study suggests several possible directions for future research. Certainly other types of music and other audiences must be studied to test the generality of the psychobiological principles underlying the music-aesthetic experience. While applause is probably an important indicant of an audience's pleasure, many other behavioral responses could be studied. It would be interesting to observe audience behaviors such as facial expressions, finger tapping, or even yawning. Other dimensions of applause could also be investigated, including loudness, reaction time or response latency, and applause "density," i.e., portion of audience applauding.

As consumer research efforts often do, the present study seems to raise more issues than it can address conclusively. It offers only one small tile to the Byzantine mosaic of consumer esthetics. Future research must contribute more "tiles" before a clearer picture of the influences shaping the music-hedonic consumptive experience can emerge.

REFERENCES

Anand, Punam, and Morris B. Holbrook (1986), "Chasing the Wundt Curve: An Adventure in Consumer Esthetics," in R. J. Lutz, ed., Advances in Consumer Research, vol. 13, Provo, UT, Association for Consumer Research, 655-657.

Apel, Willi (1973), Harvard Dictionary of Music, 2nd edition, Cambridge, MA, Belknap Press.

Berlyne, D. E. (1974), Studies in the New Experimental Aesthetics: Steps Toward an Objective Psychology of Aesthetic Appreciation, Washington, D.C., Hemisphere Publishing.

Bruner, Gordon C., II (1990), "Music, Mood, and Marketing," Journal of Marketing, 54, 4 (October), 94-104.

Campbell, Donald T., and Julian C. Stanley (1963), Experimental and Quasi-Experimental Designs for Research, Boston, Houghton Mifflin.

Crozier, J. B. (1974), "Verbal and Exploratory Responses to Sound Sequences Varying in Uncertainty Level," in D. E. Berlyne, ed., Studies in the New Experimental Aesthetics, Washington, D.C., Hemisphere Publishing, 27-90.

Dowling, W. Jay, and Dane L. Harwood (1986), Music Cognition, Academic Press.

Fetchner, G. T. (1876), Vorschule der Asthetik. Leipizig, Breithopf und Hartal.

Heritage, John, and David Greatbatch (1986), "Generating Applause: A Study of Rhetoric and Response at Party Political Conferences," American Journal of Sociology, 92, 1 (July), 110-157.

Hevner, Kate (1937), "The Affective Value of Pitch and Tempo in Music," American Journal of Psychology, 49, 621-630.

Holst, Gail (1975), Road to Rembetika: Music from a Greek Subculture, Anglo-Hellenic Publishing, Athens, Greece.

Kellaris, James J., and Robert J. Kent (1991), "Exploring Tempo and Modality Effects on Consumer Responses to Music," in R. H. Holman and M. R. Solomon, eds., Advances in Consumer Research, vol. 18, Provo, UT, Association for Consumer Research, 243-248.

LeBlanc, Albert and Jan McCrary (1983), "Effect of Tempo on Children's Music Preference," Journal of Research in Music Education, 51 (4), 283-294.

Lomax, A. (1976), Cantometrics: A Method of Musical Anthropology, Berkeley, CA, University of California Extension Media Center.

McGrath, Joseph E. (1981), "Dilemmatics," American Behavioral Scientist, 25, 2 (November/December), 179-210.

Milliman, Ronald E. (1982), "Using Background Music to Affect the Behavior of Supermarket Shoppers," Journal of Marketing, 46 (Summer), 86-91.

Milliman, Ronald E. (1986), "The Influence of Background Music on the Behavior of Restaurant Patrons," Journal of Consumer Research, 13 (September), 286-289.

Pavelchak, Mark A., John H. Antil, and James M. Munch (1988), "The Super Bowl: An Investigation into the Relationships Among Program Context, Emotional Experience, and Ad Recall," Journal of Consumer Research, 15 (December), 360-367.

Rigg, Melvin G. (1940), "Speed as a Determiner of Musical Mood," Journal of Experimental Psychology, 27 (November), 566-571.

Stout, Patricia A. and John D. Leckenby (1988), "Let the Music Play: Music as a Nonverbal Element in Television Commercials," in S. Hecker and D. W. Stewart, eds., Nonverbal Communication in Advertising, Lexington, MA, D. C. Heath, 207-223.

Vitz, Paul C. (1966), "Affect as a Function of Stimulus Variation," Journal of Experimental Psychology, 71 (1), 74-79.

Wundt, W. M. (1874), Grundzuge der Physiologischen Psychologie, Leipzig, Engleman.

Yalch, Richard F., and Eric Spangenberg (1990), "Effects of Store Music on Shopping Behavior," Journal of Consumer Marketing, 7, 2 (Spring), 55-63.

----------------------------------------