Marketing Letters

, Volume 25, Issue 1, pp 1–11

The same old song: The power of familiarity in music choice


    • Cox School of BusinessSouthern Methodist University
  • Joseph K. Goodman
    • Olin Business SchoolWashington University in St Louis
  • Julie R. Irwin
    • McCombs School of BusinessThe University of Texas at Austin

DOI: 10.1007/s11002-013-9238-1

Cite this article as:
Ward, M.K., Goodman, J.K. & Irwin, J.R. Mark Lett (2014) 25: 1. doi:10.1007/s11002-013-9238-1


Does "familiarity breed contempt" or is "to know you is to love you"? In this research, we explore the role of familiarity in music choice. We show that although consumers say they would prefer to listen to unfamiliar music, in actuality familiarity with music positively predicts preference for songs, play lists, and radio stations. Familiarity with music is at least as good, if not a better, predictor of choice as are liking, satiation (which actually positively predicts choice), and regret. We suggest that the need for familiarity is driven by consumers' low need for stimulation in the music domain, and show that when the need for stimulation decreases, the power of familiarity significantly increases. In addition to their theoretical contribution, these results are informative for music managers determining playlists, for the promotion of music events and products, and for advertisers selecting the most potentially lucrative music venues.


FamiliarityMere exposureOptimum stimulation levelMusicSong

1 Introduction

Imagine you are reaching for the radio dial or for your iPod. You have an almost limitless array of music at your fingertips and yet somehow you find yourself choosing that familiar old Motown song or comfortable Bach aria. You have just encountered the persistent tension in music choice between the opposing forces of the known and familiar versus the novel and new. People exhibit both tendencies in their consumption choices, and psychological theory has addressed the two forces extensively. However, despite the obvious managerial interest in the power of one force versus another, there is surprisingly little research examining which of these opposing forces will dominate in particular markets.

In this manuscript, we investigate the relative power of these two drives, the need for novelty and the power of familiarity, in the context of music choice. We show that familiarity exerts a very powerful force over music choice, one that would not be anticipated either by writing in the popular press on the subject, or by consumers’ own reporting of how much familiarity they desire.

We examine music choice for several reasons. Music is big business, an over $30 billion dollar industry, with many customers (e.g., Web radio now exceeds 57 million consumers each week). Instead of disappearing, traditional radio formats continue to endure even in the face of technological changes (Edison 2006), and the largest and most popular formats continue to be music-based ( In 2011, radio reached nearly 95 % of the U.S. population, and U.S. radio advertising netted $17.4 billion in revenues (Radio Advertising Bureau 2009). Newer formats such as Satellite Radio continue the tradition of providing music for customers; music stations comprise a large proportion of the Sirius XM radio stations ( Whether it is distributed via traditional radio, satellite radio, or Internet music sites, the music industry generates profits primarily from music sales and advertising revenue (eMarketer 2010). Offering and/or emphasizing songs consumers want is good marketing strategy, as is choosing to advertise in venues that play preferred music.

Despite the size of the current radio market and its continued growth, many commentators suggest that the future of radio is bleak because stations play too much of the same music (Dotinga 2005). Social media echoes this sentiment; for example, Facebook groups such as “I’m Tired of Hearing the same 10 songs on the Radio,” “Corporate Radio Sucketh,” “We Want the Airwaves Back,” and “Death to Corporate Radio” all espouse a similar message: “To bring back the radio as a means of hearing a variety of new… music.”

However, as we will show, the music consumers say they like may not predict what they choose when faced with the prospect of actually listening to the music. Across four studies, we show that familiarity is a stronger predictor of music choice than other prevalent measures such as liking and satiation. Furthermore, we show that consumers pick music that they are familiar with even when they believe they would prefer less familiar music. To our knowledge, our research is the first to quantify the effect of familiarity versus other forces (including liking) on consumer choice and to determine the power of these variables on actual market behavior.

2 Literature review

2.1 Seeking the less familiar

Maddi (1968) argued that consumers have an “internal drive” to seek out new experiences, and to select less familiar items as means of creating new feelings, experiences, and emotions (Mowen 1988). In experiential consumption decisions relevant to our inquiry, such as the domains of fashion (Campbell 1992) and art (Berlyne 1971), researchers find that people’s “desire for the new” often prompts them to choose items with novel aesthetic features.1

2.2 Familiarity-seeking

There is also ample evidence that consumers are drawn to the familiar and known. The mere exposure effect confirms that exposure to a stimulus can increase positive affect towards it (Zajonc 1968), across many types of stimuli (e.g., Berryman 1984). Perceptual fluency theory provides some explanation for the mere exposure effect. The theory asserts that the number of times a person has been exposed to a stimulus is positively related to the ease with which it comprehended, leading people to like it more simply because it is easy to process (Jacoby and Dallas 1981).

In sum, the psychology literature suggests both that novel items will be preferred and that familiar items will be preferred. Unfortunately, this extensive work does not provide much actionable managerial guidance for marketers as to which stimulus a consumer will actually choose in a particular product category. Based on research suggesting that consumers fear satiation but do not actually experience it to the extent they predict (Ratner et al. 1999) and that perceptual fluency may be especially powerful for brand and product choices (Ferraro, Bettman, and Chartrand 2009), we propose that, in the music domain, consumers will believe that they will prefer newer songs, but their actual choices will be driven by familiarity.

2.3 The power of familiarity: familiarity versus liking/satiation

Three components of preference are likely to play a role in song choice: the extent to which the consumer is familiar with the song(s) offered, the extent to which a consumer likes (or expects that they would like) a song, and the extent to which a consumer is satiated with (“sick of”) a song they have heard already. The latter components already play a role in corporate marketing researchers’ attempt to predict consumers’ preferences for music. “Auditorium music tests” (e.g., and “rate the music” surveys conducted by radio stations and online music retail outlets ask for both listeners’ expected liking and how “sick of” the songs they are. We propose that familiarity will be at least as powerful, if not more powerful, a predictor of music choice than either of these commonly used liking or satiation measures.

2.4 Optimum stimulation level

Although we are making overall predictions about which force will win in music choice, we also recognize that this main effect likely depends on differences in market context. Optimum Stimulation Level theory (OSL) (Berlyne 1960; Fiske and Maddi 1961) posits an inverted U-shaped function, intermediate levels of stimulation perceived as the most satisfying. Individuals’ vary on OSL, and individuals strategically adapt their consumption choices in order to maintain a desirable level of stimulation to remain engaged with (but not overwhelmed by) their environment. Thus, we predict that OSL will moderate preference for familiar music. When individuals are more (vs. less) stimulated by their context, they will prefer more (vs. less) familiar music.

3 Pilot study

In a pilot study, we assess radio listeners' opinions of the music they hear on the radio and predict that listeners will express a conscious desire for more novel music on the radio.

3.1 Procedure

We surveyed 386 U.S. individuals via Mechanical Turk (Goodman et al. forthcoming). Participants responded to four statements (1 = strongly disagree/7 = strongly agree): “Radio is too repetitive,” “Radio should play more new music,” “I’m sick of listening to the same music on the radio,” and “I look for new songs on the radio” and then answered an open-ended question about their impression of music played on the radio.

3.2 Results and discussion

Confirming our predictions, participants endorsed the idea that: radio stations should play more new music, M = 4.95; F(386) = 396.6, p < .001, they found themselves looking for new music on the radio M = 4.45; F(386) = 158.78, p < .001, the songs played on the radio are too repetitive, M = 6.03; F(386) = 479.61, p < .001, and they were sick of listening to the same music on the radio M = 5.08, F(386) = 458.38, p < .001. Likewise, many of their open-ended responses referenced their intent to abandon radio due to the lack of novel music. In our next studies, we test whether in fact people prefer novel music.

4 Study 1: choice study

In this study, participants rated their liking, familiarity, and satiation with 24 different songs. These questions are similar to how radio and music research is traditionally conducted (Little 2010). To simulate a radio listening choice experience, we asked participants to choose which songs they would prefer to listen to from pairs of songs. We predicted familiarity would predict song choice, above and beyond the effect of liking and satiation. The stimuli were songs that were currently played on the radio, and artists that would be familiar to the respondents so that they could predict their liking for the song.

4.1 Procedure

In a within-subject experiment, 190 undergraduate students viewed the names and artists of 48 songs that were currently being played on popular radio stations. The song list was purposely composed half of songs that were played more and half of songs played less on the radio according to published playlists. In the first of three sets of questions, participants rated how much they liked (expected to like) each song. In the second, they rated how familiar they were with each song, and in the third, they rated how satiated with (“sick of”) the song they were (all on seven-point scales). The 24 songs were randomized within the blocks of questions. After a 15-min filler task, participants viewed 24 pairs of songs containing one more familiar song and one less familiar song. To control for artist, 10 of the 24 choices contained two songs that were from the same artist; the remaining 14 pairs contained songs that were from different artists. The results were not affected by this difference. Participants chose which of the two songs they would prefer to listen to for 3 min.

4.2 Results

We obtained individual slopes by participant for familiarity, liking, and satiation. In the following sections, we report the mean slopes for each regression and computed the statistical significance of each slope (Table 1).
Table 1

Study 1: model slopes and R2


Independent variables
















Simple models













Multivariate models
































4.2.1 Simple models

Songs that were more liked β = .12, F(189) = 111.42, p < .001, more familiar, β = .06, F(189) = 277.55, p < .001, and, (surprisingly) that participants were sick of β = .01, F(189) = 4.79, p < .05 were more likely to be chosen.

4.2.2 Full multivariate regression models

Regressing choice onto liking and familiarity showed that familiarity continues to predict choice, β = .03, F(189) = 30.36, p < .001, above and beyond the significant effect of liking, β = .10, F(189) = 376.36, p < .001 (Fig. 1).
Fig. 1

Study 1: the effect of familiarity on choice by model

When we control for satiation, familiarity continues to significantly predict choice, β = .10, F(189) = 513.02, p < .001, as well as when we control for both satiation and liking, β = .05, F(189) = 67.07, p < .001. The surprisingly positive relationship between satiation and choice is driven by familiarity; satiation only becomes negatively related to choice when familiarity is controlled for in the model, β = −.06, F(189) = 111.09, p < .001.

Thus, familiarity is a strong driver of song choice, and does not drive song choice simply because it is related to liking. Notably, the typical satiation measure used extensively by marketers to negatively predict preference actually significantly positively predicts choice because it includes a component of the powerful familiarity variable.

4.2.3 Relative explanatory power of familiarity

When familiarity was added to the simple liking model, it continued to significantly predict preference M = .41 vs. M = .49, F(1,189) = 131.90, p < .001, showing that familiarity helps predict participants’ music choices even above and beyond the consumers’ own rated liking for the songs. This shows that our findings cannot be due to respondents being unable to rate their liking for less familiar songs; even holding liking constant, familiarity is a strong predictor of preference.

4.3 Discussion

In Study 1, consumers chose to listen to more familiar music over less familiar music. Furthermore, familiarity predicted choice above and beyond liking. Familiarity was such a strong predictor of choice that it even drove an unexpected positive relationship between satiation and preference.

Although the data are supportive of our predictions, there are limitations to Study 1. Participants did not have to actually listen to their chosen song. Also, one could argue that participants are choosing based on social endorsement, to be accepted or cool by choosing popular songs. In addition, respondents could be exhibiting ambiguity aversion, trying to minimize regret by avoiding risky and less familiar songs. In our next study, we address these alternative explanations.

5 Study 2: real choices

5.1 Procedure

In a within-participants experiment, 244 undergraduates made 16 choices between song dyads selected from a subset of the same songs used as stimuli in Study 1. The study was conducted via computer in the laboratory where participants were instructed that they would actually listen to the songs they chose at the end of the study. In each choice set, participants chose between a less familiar and more familiar song. As in the previous study, the songs were all currently played on the radio (at varying numbers of spins), and all of the artists were likely familiar to the respondents. Of the 16 dyads, 8 were between songs by the same recording artist (with order counterbalanced) and 8 were randomized pairs (with different artists and order counterbalanced) with a total of 128 possible combinations. There were no differences in the effects across these two types of stimuli. After making their choices, participants rated familiarity, liking, coolness, and how much they believed they would regret their choice if they picked the other song in the pair on 1–7 scales. At the end of the study, participants listened to one of their chosen songs.

5.2 Results

5.2.1 Simple models

As expected, songs that were more liked, β = .16, t(237) = 26.81, p < .001, or rated as more cool, β = .15, F(237) = 501.31, p < .001, were more likely to be chosen. Regret, which is a measure of uncertainty and ambiguity aversion, had a negative effect on choice, β = −.12, F(237) = 295.84, p < .001. As in the previous two studies, the more familiar a song was, the more likely it was to be chosen, β = .10, F(237) = 421.89, p < .001.

5.2.2 Full multivariate regression models

Again, we find support for our prediction that familiarity has a significant effect on choice above and beyond liking, β = .06, F(237) = 105.47, p < .001. We added the variables cool and regret to the model with both familiarity and liking, to test whether these variables explain our effects of familiarity on choice. Liking continued to predict choice in this model, β = .06, F(237) = 26.11, p < .001, and regret, β = −.03, F(237) = 6.86, p = .09, and cool, β = .02; F(237) = 3.34, p = .07, had a marginally significant effect on choice. Most importantly, familiarity continued to significantly predict choice, β = .06; F(237) = 61.78, p < .001 with all variables in the model (Fig. 2), showing that they are not the locus of the familiarity effect. Thus, social acceptance and ambiguity aversion do not explain our results.
Fig. 2

Study 2: the effect of familiarity on choice by model

5.2.3 Relative explanatory power of familiarity

The difference between the explanatory power of liking and familiarity was only marginally significant, F(1,233) = 2.80, p < .1, implying that familiarity is almost as strong a predictor as liking in predicting choice. Some of the participants in this experiment had only a nominal model fit for the full model including both familiarity and liking (i.e., R2 < .1), so we conducted a follow-up analysis excluding these participants. In this analysis, familiarity had marginally more explanatory power, M = .72, compared to liking, M = .68, F(1, 219) = 3.09, p = .08 in predicting choice.

More importantly, we again see an increase in the R2 when we add familiarity to the simple liking model, M = .65 vs. M = .86, F(1,233) = 166.97, p < .001, indicating that familiarity adds significant explanatory power when predicting participants’ song choice. In sum, the results provide additional evidence that familiarity is at least as good, if not a better, predictor of choice than liking in these song choices.

5.3 Discussion

Study 2 shows that people are likely to choose music based on familiarity, even when they will have to actually listen to the music, and that the effect of familiarity on choice is not due to avoidance of regret, the perceived coolness, or the social endorsement of a song. In addition, all of our analyses together show that familiarity’s dominance is not due to participants’ inability to rate their liking of less familiar songs. The results also confirm that liking does not drive the strong relationship between familiarity and preference. In fact, familiarity predicts choice above and beyond liking, has a stronger direct effect on choice than does liking, and in some tests has even more explanatory power than does liking.

Our psychological explanation for this effect is that people have a low OSL (Raju 1980) for music. In our next study, we manipulate stimulation level directly, using cognitive load (Furnham and Bradley 1997), to support our hypothesis. Cognitive load is an effective method to manipulate stimulation thresholds, and it is particularly relevant to music choice because musical usage occasions often involve other activities such as driving, reading, and exercising (North et al. 2004).

5.4 Study 3: optimal stimulation via cognitive load

We expect lowered OSL, operationalized using cognitive load, to moderate our familiarity results, resulting in even more desire for familiarity.

5.5 Procedure

Two hundred and seventy-six students participated in this study. We manipulated cognitive load by having participants memorize either 20 (high load) or 4 (low load) words. Participants then chose between five radio stations (of their preferred genre of music) that they could listen to while doing the task. The five stations were each described with common radio marketing phrases indicating whether the station played relatively familiar music or novel music (e.g., “The Hot Hits Station—we play the hottest top 10 songs all day long!,” “New songs all day long—we play the songs of tomorrow that you’ve never heard!”). Participants rated each station on 1–7 scales gauging their liking, familiarity, and potential for distraction (“how distracted they would be by the songs on the station”). Then, they recalled their memorized number.

5.6 Results

We ranked, within participant, the familiarity and likability of the songs for the playlists. Participants’ radio station choice was significantly predicted by their predicted song familiarity, χ2 (1, n = 276) = 6.55, p < .05, above and beyond the effect of their anticipated song liking. Controlling for liking, participants were still more likely to choose the radio station they judged to play the most familiar music when they were in the high, M = .52 versus low, M = .38, cognitive load condition, χ2 (1, n = 276) = 3.17, p = .075, affirming our predictions.

5.7 Discussion

Even under no load (Studies 1 and 2) and minimal load (the four word condition in study 3) respondents chose familiar music, suggesting that the OSL for music is quite low. Study 3 affirms that the OSL is low because when it is lowered further via cognitive load, the preference for familiar music increased.

6 General discussion

In three studies, we examined the power of familiarity on music choice and showed that familiarity is a more important driver of music choice than more obvious, and commonly tested, constructs such as liking and satiation (i.e., being “sick of”). We find this relationship even though consumers do not expect it, and it is especially strong when the usage situation already favors less stimulation, as it often does for many music usage situations.

It is natural to wonder if these effects are reflected in the music market. To explore this question, we collected and analyzed radio station data from Arbitron, which measures radio audiences across the USA. We collected data from radio stations playing the top ten formats (according to Arbitron’s Radio Today2007) located within the top 60 US markets for Fall 2008 (Arbitron 2008). We operationalized familiarity by collecting song “spins” (the number of times a station played its top songs in the current and prior week) for the top ten songs on each station. Although not a perfect measure, we expected that more spins means more consumer familiarity.

We found a significant positive effect of number of spins on market share, F(1,381) = 6.17, p < .01. We replicated this finding by looking at the top ten songs for the target week and the week prior and found the same positive relationship between number of spins and market share, F(1,381) = 4.78, p < .05. These market results are consistent with our experimental findings and suggest that familiarity is a major driver of actual music choice and market share.

6.1 Implications

Our results suggest that the emphasis on novelty in the music domain, by consumers and people protesting the current state of the music business, is misplaced. In the marketplace, and in our pilot study, consumers indicate that they want more novelty when in fact their choices suggest that they do not. These findings add to the list of unconscious preferences that keep consumers from being consciously aware of their actual desires (e.g., Wilson 2009). When testing whether consumers would choose particular music or a particular playlist, marketers would do well to bypass consumers’ notions of what they want, and to instead ask how familiar consumers are with the music. Familiarity is as powerful as, and sometimes more powerful than, any other measure of music preference in our studies. In addition, our results show that asking for satiation measures is counterproductive, at least for predicting reduced preference for music.

For music outlets with playlists, our results suggest the best strategy is to concentrate primarily on familiar songs, even if consumers say they want more novelty. When a new song is introduced, it should be played often (as the successful stations did in our actual market data) and offered to consumers through promotions, free samples, and incidental listening. For music outlets that allow the user to create a playlist, such as iTunes, our results suggest that marketers should heavily promote and make familiar music, easy to find for purchase, and should not emphasize unfamiliar music. Likewise, our results predict the success of Apps such as Spotify and Pandora, which offer newly released music that has many familiar elements, such as familiar artists, styles, and melodies.

6.2 Extensions and limitations

It would be useful to extend these results to other domains. We expect a powerful influence of familiarity in other artistic categories other than music, such as the entertainment, food, and the visual arts. Most popular movies include familiar actors and plots, and many popular restaurants seem to serve essentially the same food, suggesting a low optimal stimulation level in these domains as well.

Future research could further explore the relationship between OSL, familiarity, and context. Familiar stimuli have been found to be preferred to novel stimuli when people are focused on security (e.g., prevention focus or narrow categorization) as opposed to growth (e.g., promotion focus or broad categorization; Gillebaart et al. 2012). Perhaps a prevention focus reduces OSL because resources are needed to address whatever threat is being prevented. In the case of music, listeners may need their resources to prevent mistakes in other tasks they perform while listening. If listeners were to switch to a growth focus, perhaps by concentrating only on the music, they may open up resources to allow them to explore more novel music.

Future research might combine our findings with the work on assortment and variety, perhaps by addressing how consumers respond to the same song repeated several times in a short time period or how they assemble playlists (Ratner et al. 1999; Read and Loewenstein 1995). Our results show that consumers underestimate the power of their familiarity for a song on choice, a finding that correlates with the findings that consumers overestimate their desire for variety.


In this research, we concentrate on familiarity, and do not address variety or variety-seeking. Variety refers to the number of different items in an assortment (Broniarczyk et al. 1998; McAlister and Pessemier 1982; Ratner et al. 1999), and variety-seeking refers to the desire to consume a diverse set of items. A very diverse assortment could include all familiar or all unfamiliar goods, and a very homogenous assortment could likewise vary a great deal in familiarity. In other words, high variety does not imply low familiarity and vice versa.


Copyright information

© Springer Science+Business Media New York 2013