Soundtracks of Our Lives: How Age Influences Musical Preferences

Alan Said; Arsen Matej Golubovikj; Bruce Ferwerda; Marko Tal\v{c}i\v{c}

REVIEW 3 major objections 4 minor 20 references

Soundtracks of Our Lives: How Age Influences Musical Preferences

T0 review · 3 major / 4 minor · reviewed 2026-08-04 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Using 15 years of Last.fm logs, this paper argues that individual musical taste broadens with age even as listeners become more alike, and that after about age 40 nostalgia for teenage-era tracks resurges.

desk verdict Useful descriptive longitudinal study, but the diversity metric is volume-confounded and the Gini-index based inter-user diversity claim is not supported by the presented evidence. read the letter →

arxiv 2509.08337 v1 pith:A75PTMOJ submitted 2025-09-10 cs.IR

Arsen Matej Golubovikj , Bruce Ferwerda , Alan Said , Marko Talčič This is my paper

classification cs.IR

keywords musicrecommendersystemslongitudinalusermodelingageandtastelisteningdiversitysong-specificnostalgiaLast.fmlogsLFM-2b

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that musical preferences change with age in systematic, measurable ways that recommender systems should model. It works with the LFM-2b dataset, covering about 542 million listening events from 42,883 Last.fm users across 15 years, and adjusts each user's one-time reported age to the moment of each listen. The analysis finds that as a person ages, their own listening diversity increases, while differences between listeners of the same age shrink. A song-specific age analysis shows that younger listeners strongly favor contemporary releases, while after about age 40 a second, nostalgia-driven peak appears for music from the listener's adolescence. If the findings hold, they give recommender systems a concrete age-based dial for tuning how much diversity or personalization to serve.

What carries the argument

The argument is carried by four instruments. (1) An age-adjusted longitudinal sample: each user's static age from LFM-2b is shifted per listening event by the year difference from Oct 31, 2013, so every event gets an adjusted age M. (2) Two [0,1] diversity ratios—track diversity U_O = unique tracks ÷ playcounts and artist diversity U_A = unique artists ÷ unique tracks—following the diversity definition of Schedl and Hauger [15] and inverted per Spear et al. [18]. (3) The Gini index applied to within-age-group diversity, measuring the 'diversity of diversity' and quantifying inter-user convergence. (4) Song-specific age (SSA), the listener's age at a track's release, with playcounts log-norma

What would settle it

A decisive check would be to rerun Figures 2, 3, and 5 on the subset of users whose reported age does not end in 0 or 5 and who have listening events in both 2005–2008 and 2016–2020; if the post-40 nostalgia peak and the falling Gini index shrink to noise, the findings would be artifacts of age adjustment and sample attrition rather than true aging effects.

Watch

Extended reading notes

Core claim

The paper's central claim: once users are tracked for years rather than weeks, age visibly reshapes musical taste. Track diversity (unique tracks per playcount) grows from about 0.32 at age 15 to 0.60 at age 40 and then plateaus, so each person's listening widens. Over the same span, the Gini index of diversity within age groups falls from about 0.33 to 0.17, so same-age listeners become more alike. Song-specific age—the listener's age when a track was released—peaks for current releases at every age, but after 40 a second, nostalgia peak appears at ages 15–20. The authors interpret this as a shift from broad, contemporary, socially shared listening to narrower, personalized, nostalgia-infus

Load-bearing premise

The clearest vulnerable premise is the age data: Last.fm users report their age once (2013–2014), and the study shifts that one number forward and backward by calendar year for every listening event, so if users misreport, round off, or share accounts, all the age curves would be distorted.

Editorial extensions

If this is right

Recommender systems can use age as a diversity/personalization dial: younger users benefit from broad, contemporary-heavy suggestions, while older users benefit from tighter, nostalgia-weighted ones.
Around age 40 is a preference tipping point; beyond it, consumption of current releases declines and adolescent-era music dominates, so age-aware models should weight historical taste more heavily.
Because inter-user diversity shrinks with age, older users' tastes are easier to model from their peers, while younger users need more exploration and less conformity.
Algorithmic feedback loops matter: if recommenders feed young users varied content and older users narrow content, the systems may amplify the very age divergence the paper describes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Inference: The paper stops at descriptive evidence; a direct algorithmic extension would be to feed adjusted age (or SSA band) into a recommender baseline and measure whether the diversity/personalization trade-off actually shifts, turning the observation into a deployable design rule.
Inference: Because the age variable is self-reported and snapshotted in 2013–2014, the longitudinal curves implicitly assume one user per account and stable truthful reporting; a robustness check on users with non-heaped ages and continuous fifteen-year histories would separate genuine aging effects from cohort and attrition artifacts.
Inference: The same song-specific-age machinery could be applied to films or books, where release-year-relative-to-age may produce an analogous nostalgia bump; finding one would suggest a general reminiscence effect rather than a music-specific quirk.
Inference: The diversity metric is a playcount-concentration ratio, so the exact magnitudes of age differences likely depend on metric choice; entropy- or coverage-based measures could shrink or grow the reported gaps even if the direction holds.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Worth a look for the longitudinal LFM-2b analysis and the SSA (song-specific age) curves, but the central diversity claims have a load-bearing measurement problem.

What's new: this is the first analysis of LFM-2b that assigns an event-level age by shifting the static age against a reference date, which lets them track users across 15 years. That's a reasonable and clever use of the dataset's timestamp structure. The SSA finding – a second listening peak around age 15-20 for users over 40 – is a genuine, interpretable pattern that aligns with the music psychology literature and is the most convincing part of the paper. The descriptive plots are clear, and the cleaning procedure is reproducible in principle.

The soft spots. The stress-test note is right and it's a serious one. U_O = unique_tracks / playcounts. The paper shows playcounts drop steeply with age (Fig. 1a), while unique tracks peak around 22 and then decline gently. So U_O increases with age for most of the range even if actual variety is flat or falling. The text literally says diversity grows 'significantly' between 15 and 40, but with no confidence intervals or significance tests anywhere, 'significantly' is doing no statistical work. Worse, the Gini index in Fig. 3 is computed on U_O values across users within an age group, not on the overlap of their actual music choices. That is a measure of dispersion of a volume-confounded ratio, not inter-user preference overlap. The conclusion that 'inter-user diversity decreases' (Section 4) is therefore not supported by the evidence as presented. The paper's own limitation section admits the metric is basic, but doesn't flag the playcount-volume confound, which is more serious than a nuance.

The age-accuracy limitation is real but secondary, and the paper already admits it. The survivor bias (users active in 2013-2014) is also acknowledged.

Net: as a descriptive, exploratory study, it's a decent contribution and a useful pointer for recommender-system people who want age-aware diversity or nostalgia handling. The SSA results alone justify a serious referee. The intra-user diversity increase claim needs either a better metric (e.g., entropy, or unique tracks controlled for playcount) or at least a clear caveat that it's partly a volume artifact. The inter-user diversity claim should be retracted or re-measured with actual taste overlap.

I'd send it to peer review, but with a strong steer that the diversity analysis needs rethinking before it can be published as-is.

Referee Report

3 major / 4 minor

Summary. The paper analyzes the LFM-2b listening-event dataset to study how music preferences and consumption behavior change with user age. The authors adjust a static age variable to event time, restrict the sample to users aged 10–64, and compute annual playcounts, unique tracks, unique artists, a track-based diversity ratio, an artist-based ratio, the Gini index of diversity, and a song-specific age (SSA) metric. They report that intra-user diversity increases with age, inter-user diversity decreases, younger users favor contemporary releases, and after age 40 a secondary 'nostalgic' peak appears. The paper draws recommender-system implications from these trends.

Significance. If the reported trends are valid, the paper addresses an under-studied longitudinal dimension in recommender-systems research using a rare large-scale dataset. Its descriptive contribution is potentially useful: the cleaning procedure is clearly described, the age-adjustment idea is sensible, and the SSA analysis connects directly to prior music-psychology findings. However, the central metric used to support the main claims is confounded by listening volume, and no statistical uncertainty is reported. The paper's value therefore depends on whether the main conclusions survive additional analysis.

major comments (3)

[§3.2, §3.3, Fig. 1, Fig. 2, Fig. 3] The central diversity measure U_O = |S|/R conflates taste breadth with listening volume. Fig. 1 shows median annual playcounts falling from 2,318 at age 19 to 560 at age 58, while unique tracks peak at 805 (age 22) and then decline. A user who halves playcounts while keeping the same unique-track set doubles U_O. The rise in U_O between ages 15 and 40 is therefore not by itself evidence of 'broader listening preferences' (Section 3.3). The same confound propagates to the Gini index in Fig. 3, which measures dispersion of this ratio across users, not overlap or distance between users' musical tastes. The statement in Section 4 that 'inter-user diversity decreases (Fig. 3)' is not supported by the presented evidence. Please report at least one volume-independent diversity measure (e.g., entropy, Simpson diversity, or unique tracks conditioned on playcount) and, for inter-user diversity, a
[§3.3, Figs. 1–5] The paper uses 'significantly' in describing the diversity increase between ages 15 and 40, but no confidence intervals, significance tests, or effect sizes are reported for any of the median trends. The regression lines in Fig. 1 are fitted without stating the procedure or uncertainty. Because the main claims are about monotone age trends in a noisy, long-tailed dataset, bootstrap confidence intervals on the medians by age would be a minimal and necessary addition.
[§3.1, §4.1] The entire analysis depends on the adjusted age variable M = M + year difference from a fixed reference date. The paper itself lists 'improbable user ages' as a limitation. The static age is retrieved in 2013–2014, and age at each event is computed by adding the year difference; if the static age is misreported or systematically biased, all age-trend conclusions are affected. A sensitivity analysis is needed: report the distribution of static ages before cleaning, show the effect of excluding users with suspicious or inconsistent ages, and check whether the results hold when restricting to users whose activity is consistent with a single reported age.

minor comments (4)

[§3.1] The equation for the adjusted age is garbled: 'ω𝑁 (𝑁𝑀 ,𝐿 𝐿 )' and related symbols appear corrupted. The quartile-based outlier formula also has a typo ('𝑂1 ↑ 1.5 · 𝑃𝑂𝑄' should presumably be Q1 − 1.5·IQR).
[Table 1 and §3.1] The text says 'Table 1 shows the number of users in each age group post-cleaning,' but Table 1 only gives global counts. Please include per-age-group user counts or remove the cross-reference.
[Fig. 1 caption] The caption lists '(b) Number of Unique Artists' and '(c) Number of Unique Tracks,' while the main text describes the subfigures in the opposite order. Please make the caption consistent with the panels.
[§4] The 'algorithmic confounding' paragraph is speculative and not tied to any data or prior result in the paper. It may be better framed as future work.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity; empirical study with independent metrics and contextual self-citation only.

full rationale

This is an empirical longitudinal measurement study, not a derivation. The central claims (intra-user diversity increases, inter-user dispersion decreases, SSA peaks shift) are summaries of observed medians computed from the LFM-2b dataset. The diversity metrics U_O and U_S are defined in Section 3.2 independently of the age variable and of the conclusions; the age adjustment M = M + ω(...) is a straightforward arithmetic shift. No parameter is fitted to a subset of data and then 'predicted' on a closely related quantity; the reported trends are direct aggregates. The only self-citation (Ferwerda et al., 2017, by two of the present authors) is used in Sections 1 and 4 to motivate the gap and to contextualize genre-based findings, not as a load-bearing proof or uniqueness theorem. The potential confounding of U_O by playcount volume (the skeptic's point) is a substantive validity concern about what the metric measures, but it is not circularity: the paper does not define the conclusion in terms of the metric's inputs, nor does any equation reduce to itself. No circular step can be exhibited.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The paper's central claim rests on the reliability of the LFM-2b age field, the completeness of Spotify release-year data, and the validity of the two diversity ratios as measures of musical diversity. The free parameters are methodological choices in the cleaning pipeline; none is reported as being tuned to the conclusions, but no sensitivity analysis is provided.

free parameters (6)

Reference date for age adjustment = October 31, 2013
Used to compute adjusted age for every listening event; chosen as midpoint of the 2013-2014 age retrieval window. A different reference date would shift all event-level ages, though by at most about one year.
Activity percentile threshold for user inclusion = 20th percentile of listening events per age group
Users below this threshold are excluded as inactive. The 20th percentile is hand-chosen; no robustness check is reported.
Outlier threshold multiplier = 1.5 IQR
Standard IQR rule used to drop users with unusually high or low listening event counts. Arbitrary and not tested for sensitivity.
Age range restriction = 10-64 years
Users outside this range are excluded, which may truncate patterns in childhood and old age.
Age bin width = 5 years
Users are grouped into 5-year intervals; the choice balances sample size and granularity but is arbitrary.
Playcount log-normalization scaling = log10 scaled to [0, 3.903]
Used for SSA visual comparison; the upper bound is derived from the maximum observed playcount and is not a scientific quantity.

assumptions (4)

domain assumption The static age field in LFM-2b is accurate enough for computing adjusted ages
Section 3.1 uses this field for the age adjustment formula; Section 4.1 acknowledges unreliable age data as a limitation.
domain assumption Spotify API provides complete release-year metadata for all tracks
Section 3.1 states release years were obtained via the Spotify API but does not report missing data rates.
domain assumption Unique tracks / playcounts and unique artists / unique tracks are valid diversity proxies
Section 3.2 defines these measures, adapted from Schedl and Hauger (2015) and Spear et al. (2021); the paper does not validate them against other diversity notions.
domain assumption Observed listening behavior reflects underlying preference rather than platform artifacts
The analysis interprets age trends as preference evolution; algorithmic confounding is only discussed qualitatively in Section 4.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Soundtracks of Our Lives: How Age Influences Musical Preferences." pith.science (2026). https://pith.science/paper/A75PTMOJ

@misc{pith2026250908337,
  author       = {Pith},
  title        = {Pith review of: Soundtracks of Our Lives: How Age Influences Musical Preferences},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A75PTMOJ}},
  note         = {Machine review of arXiv:2509.08337}
}

read the original abstract

The majority of research in recommender systems, be it algorithmic improvements, context-awareness, explainability, or other areas, evaluates these systems on datasets that capture user interaction over a relatively limited time span. However, recommender systems can very well be used continuously for extended time. Similarly so, user behavior may evolve over that extended time. Although media studies and psychology offer a wealth of research on the evolution of user preferences and behavior as individuals age, there has been scant research in this regard within the realm of user modeling and recommender systems. In this study, we investigate the evolution of user preferences and behavior using the LFM-2b dataset, which, to our knowledge, is the only dataset that encompasses a sufficiently extensive time frame to permit real longitudinal studies and includes age information about its users. We identify specific usage and taste preferences directly related to the age of the user, i.e., while younger users tend to listen broadly to contemporary popular music, older users have more elaborate and personalized listening habits. The findings yield important insights that open new directions for research in recommender systems, providing guidance for future efforts.

Figures

Figures reproduced from arXiv: 2509.08337 by the authors.

**Figure 2.** shows median track-based diversity (𝑈𝑂 ) and artist-based diversity (𝑈𝑆 ) by age. Both metrics start low at younger ages and increase as users mature, with 𝑈𝑂 rising faster than 𝑈𝑆 . Diversity grows signi$cantly between ages 15 (𝑈𝑂 = 0.316, 𝑈𝑆 = 0.189) and 40 (𝑈𝑂 = 0.599, 𝑈𝑆 = 0.328), re%ecting broader listening preferences with age. While both metrics show similar trends, di"erences are more pronounced in 𝑈𝑂 , whic… view at source ↗

**Figure 1.** Median (a) playcounts, (b) number of unique artists, [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

**Figure 3.** The Gini index of track-based diversity, based on [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: The median track-based diversity 𝑈𝑂,𝑃 in each age group per release year. Higher values indicate more diversity [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]

**Figure 5.** Figure 5: Median Log normalized playcounts in each age [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

20 extracted references · 16 canonical work pages

[1]

Joeran Beel, Stefan Langer, Andreas Nürnberger, and Marcel Genzmehr. 2013. The Impact of Demographics (Age and Gender) and Other User-Characteristics on Evaluating Recommender Systems. In Research and Advanced Technology for Digital Libraries (Lecture Notes in Computer Science) , Trond Aalberg, Christos Papatheodorou, Milena Dobreva, Giannis Tsakonas, and...

work page doi:10.1007/978-3-642-40501- 2013
[2]

Andy Bennett and Jodie Taylor. 2012. Popular music and the aesthetics of ageing. Popular Music 31, 2 (May 2012), 231–243. https://doi.org/10.1017/ S0261143012000013 Publisher: Cambridge University Press

work page 2012
[3]

Arielle Bonneville-Roussy, Peter Rentfrow, Kate Xu, and Je" Potter. 2013. Music Through the Ages: Trends in Musical Engagement and Preferences From Adoles- cence Through Middle Adulthood. Journal of personality and social psychology 105 (July 2013). https://doi.org/10.1037/a0033770

work page doi:10.1037/a0033770 2013
[4]

Arielle Bonneville-Roussy, David Stillwell, Michal Kosinski, and John Rust

work page
[5]

Bruce Ferwerda, Marko Tkalcic, and Markus Schedl. 2017. Personality Traits and Music Genre Preferences: How Music Taste Varies Over Age Groups. In RecTemp 2017 Temporal Reasoning in Recommender Systems . CEUR Workshop Proceedings

work page 2017
[6]

Jill Harrison and John Ryan. 2010. Musical taste and ageing. Ageing & Society 30, 4 (May 2010), 649–669. https://doi.org/10.1017/S0144686X09990778 Publisher: Cambridge University Press

work page doi:10.1017/s0144686x09990778 2010
[7]

Jan Hemming. 2013. Is there a peak in popular music preference at a certain song-speci$c age? A replication of Holbrook & Schindler’s 1989 study. https: //journals.sagepub.com/doi/10.1177/1029864913493800

work page doi:10.1177/1029864913493800 2013
[8]

Holbrook and Robert M

Morris B. Holbrook and Robert M. Schindler. 1989. Some Exploratory Findings on the Development of Musical Tastes. Journal of Consumer Research 16, 1 (1989), 119–124. https://www.jstor.org/stable/2489308 Publisher: Oxford University Press

arXiv 1989

Show all 20 references

[9]

Noam Koenigstein, Gideon Dror, and Yehuda Koren. 2011. Yahoo! music recom- mendations: modeling music ratings with temporal dynamics and item taxonomy. In Proceedings of the !fth ACM conference on Recommender systems. ACM, Chicago Illinois USA, 165–172. https://doi.org/10.1145...

2011
[10]

Song-Speci$c Age

Reinhard Kopiez, Jan Weigang, Friedrich Platz, and Nina Düvel. 2021. Farewell to Holbrook & Schindler’s (1989) "Song-Speci$c Age"? -Little Evidence for Lifelong In%uence of Age-Speci$c Musical Preferences. Music & Science 4 (April 2021), 1–20. https://doi.org/10.1177/20592043211001794

2021 doi
[11]

Xiangyu Ma. 2021. What are the temporal dynamics of taste? Poetics 84 (Feb. 2021), 101514. https://doi.org/10.1016/j.poetic.2020.101514

2021
[12]

Lorenzo Porcaro, Carlos Castillo, and Emilia Gómez. 2021. Diversity by Design in Music Recommender Systems. Transactions of the International Society for Music Information Retrieval (Nov 2021). https://doi.org/10.5334/tismir.106

2021 doi
[13]

Markus Schedl and Christine Bauer. 2019. Online Music Listening Culture of Kids and Adolescents: Listening Analysis and Music Recommendation Tailored to the Young. http://arxiv.org/abs/1912.11564 arXiv:1912.11564 [cs]

2019 arXiv
[14]

Markus Schedl, Stefan Brandl, Oleg Lesota, Emilia Parada-Cabaleiro, David Penz, and Navid Rekabsaz. 2022. LFM-2b: A Dataset of Enriched Music Listening Events for Recommender Systems Research and Fairness Analysis. In Proceedings of the 2022 Conference on Human Information Int...

2022
[15]

Markus Schedl and David Hauger. 2015. Tailoring Music Recommendations to Users by Considering Diversity, Mainstreaminess, and Novelty. In Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval. ACM, Santiago Chile, 947–9...

2015
[16]

Markus Schedl, Peter Knees, Brian McFee, Dmitry Bogdanov, and Marius Kamin- skas. 2015. Music Recommender Systems. In Recommender Systems Handbook , Francesco Ricci, Lior Rokach, and Bracha Shapira (Eds.). Springer US, Boston, MA, 453–492. https://doi.org/10.1007/978-1-4899-7637-6_13

2015 doi
[17]

Markus Schedl, Hamed Zamani, Ching-Wei Chen, Yashar Deldjoo, and Mehdi Elahi. 2018. Current challenges and visions in music recommender systems research. International Journal of Multimedia Information Retrieval 7, 2 (June 2018), 95–116. https://doi.org/10.1007/s13735-018-0154-2

2018 doi
[18]

Lawrence Spear, Ashlee Milton, Garrett Allen, Amifa Raj, Michael Green, Michael D Ekstrand, and Maria Soledad Pera. 2021. Baby Shark to Barracuda: Analyzing Children’s Music Listening Behavior. In Fifteenth ACM Conference on Recommender Systems . ACM, Amsterdam Netherlands, 63...

2021
[19]

João Vinagre, Alípio Mário Jorge, Marie Al-Ghossein, Albert Bifet, and Paolo Cremonesi. 2022. Preface to the special issue on dynamic recommender systems and user models. User Modeling and User-Adapted Interaction 32, 4 (Sept. 2022), 503–507. https://doi.org/10.1007/s11257-022-09341-y

2022 doi
[2017]

Conceptualization and empirical investigation

Age trends in musical preferences in adulthood: 1. Conceptualization and empirical investigation. Musicae Scientiae 21, 4 (Dec. 2017), 369–389. https://doi.org/10.1177/1029864917691571 Publisher: SAGE Publications Ltd

2017 doi

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.

[1] [1]

Joeran Beel, Stefan Langer, Andreas Nürnberger, and Marcel Genzmehr. 2013. The Impact of Demographics (Age and Gender) and Other User-Characteristics on Evaluating Recommender Systems. In Research and Advanced Technology for Digital Libraries (Lecture Notes in Computer Science) , Trond Aalberg, Christos Papatheodorou, Milena Dobreva, Giannis Tsakonas, and...

work page doi:10.1007/978-3-642-40501- 2013

[2] [2]

Andy Bennett and Jodie Taylor. 2012. Popular music and the aesthetics of ageing. Popular Music 31, 2 (May 2012), 231–243. https://doi.org/10.1017/ S0261143012000013 Publisher: Cambridge University Press

work page 2012

[3] [3]

Arielle Bonneville-Roussy, Peter Rentfrow, Kate Xu, and Je" Potter. 2013. Music Through the Ages: Trends in Musical Engagement and Preferences From Adoles- cence Through Middle Adulthood. Journal of personality and social psychology 105 (July 2013). https://doi.org/10.1037/a0033770

work page doi:10.1037/a0033770 2013

[4] [4]

Arielle Bonneville-Roussy, David Stillwell, Michal Kosinski, and John Rust

work page

[5] [5]

Bruce Ferwerda, Marko Tkalcic, and Markus Schedl. 2017. Personality Traits and Music Genre Preferences: How Music Taste Varies Over Age Groups. In RecTemp 2017 Temporal Reasoning in Recommender Systems . CEUR Workshop Proceedings

work page 2017

[6] [6]

Jill Harrison and John Ryan. 2010. Musical taste and ageing. Ageing & Society 30, 4 (May 2010), 649–669. https://doi.org/10.1017/S0144686X09990778 Publisher: Cambridge University Press

work page doi:10.1017/s0144686x09990778 2010

[7] [7]

Jan Hemming. 2013. Is there a peak in popular music preference at a certain song-speci$c age? A replication of Holbrook & Schindler’s 1989 study. https: //journals.sagepub.com/doi/10.1177/1029864913493800

work page doi:10.1177/1029864913493800 2013

[8] [8]

Holbrook and Robert M

Morris B. Holbrook and Robert M. Schindler. 1989. Some Exploratory Findings on the Development of Musical Tastes. Journal of Consumer Research 16, 1 (1989), 119–124. https://www.jstor.org/stable/2489308 Publisher: Oxford University Press

arXiv 1989