Mozart or Pelé? The Effects of Teenagers' Participation in Music and Sports

Publication Type:

Journal Article


SOEPpapers, Issue 749 (2015), p.80 (2015)



Abstract taken from the article:

Using data from the German Socio-Economic Panel, this paper analyses the effects of spending part of the adolescents’ leisure time on playing music or doing sports, or both. We find that while playing music fosters educational outcomes compared to doing sports, particularly so for girls and children from more highly educated families, doing sports improves subjective health. For educational outcomes, doing both activities appeared to be most successful. The results are subjected to an extensive robustness analysis including instrumental variable estimation and a formal sensitive analysis of the identifying assumptions, which does not reveal any serious problems.


Research Question

With their paper on differential effects of doing sports vs. doing music on adolescents’ cognitive and non-cognitive skills and subjective health, the authors strive to fill a research gap: They state that for music and sports, being two of the most important education-oriented extracurricular activities of children (Cabane et al. 2015: 3), a large body of research addresses the isolated but not the differential effects of the activities as competing uses of leisure time.

Additionally, research on the impact of musical (or athletic) activities often suffer from methodological weaknesses as they neglect the specification of the counterfactual: “No study explicitly compares the effects of alternatively using the available time for doing music or doing sports, although this question appears very relevant given the time and budget constraints of parents and their children” (ibid.: 5).

The research interest bears some implications both on the idea of arts (here: musical) education and related policies. As the authors state, extra-curricular music activities are highly institutionalized in Germany, and musical activities for children and adolescents are granted subsidies in many European countries. In line with the arguments of actors and interest groups in favour of musical education, which often refer to positive side effects of musical activities for participants as well as with parents’ and/or children’s expected related returns on investment, the authors aim to investigate the differential outcomes of doing music vs. sports.

Theory, Data and Methodology

As the authors find that results from previous studies on the effects on musical or sports activities are not comparable because of different outcome variables, possible selection bias and/or different compositions of the according control group (ibid. 8p.), they derive some hypotheses on causal effects of music and/or sports activities from theoretical considerations. As for the effects of playing music, they name, amongst others, improvements in cognitive skills, improved judgement in assessing one’s own abilities, a more positive attitude towards school and performance, the acquirement of social skills and cultural capital and better school grades due to signalling effects (ibid. 9p.).

Furthermore, the authors note that apart from general outcomes, which can be induced by either music and sports, there do exist some activity-specific capabilities such as musical self-efficacy and a musical self-concept, whereby the former tends to be mistaken for the latter:

“(T)he findings of (…) literature are consistent with both the activity specific and the general channels (…). Nonetheless, in most cases, these results are misleadingly presented as being activity-specific” (ibid. 9).

Using a data set from the German Socio-Economic Panel (SOEP) and the related youth questionnaire, the authors exploit individual and parental information on adolescents pursuing musical and/or athletic activities at the age of 17 on a substantial level regarding cognitive and non-cognitive skills, subjective health measures and measures of personality traits. As outcome variables they employ school types, grades and plans to attend university for educational success, standardized test results for cognitive skills, questionnaire results according to the Big Five personality traits and different opinion/attitude measures for non-cognitive skills and some health measures (ibid. 12p.). Additionally, they use information on family characteristics to control for socio-economic and educational factors determining the selection of music and/or sports activities.

The authors employ econometric approaches to compare the effects of playing music and/or doing sports on adolescents. In order to “disentangle the effect of participating in a specific activity from the influence of differences in socio-demographic and other background characteristics” (ibid. 19), they opt for a selection-on-observables approach towards identifying causal effects and taking potential confounders into account. This concerns especially the several driving factors for the decision to take up an extracurricular activity as well as possible constraints.

The authors employ a propensity score matching estimator to compute treatment effects with respect to the treatment groups according to the activities music vs. / and sports. Additionally, they estimate the outcome differences with respect to gender and socio-economic characteristics to take effect heterogeneity into account. For a robustness check, the authors use the instrumental variables approach via the instrument of parental artistic activities, as they “are a strong predictor for adolescents’ music participation (…)” (ibid: 25).


The results from descriptive statistics indicate that musically active adolescents differ from inactive individuals in many respects. They achieve higher results with respect to the outcome variables “At least one parent with a university degree”, “Average monthly labour market income of parents present in household”, “Recommendation for upper secondary school”, “Average grade” and “Average cognitive skills” compared to those not playing music (ibid. 18p.). The highest results however are achieved by those who are both musically and athletically active.

This is basically confirmed by the results from the estimation of the propensity score. However, they also suggest that “parental education plays a much smaller role in choosing between sports and music than in the general decision to become musically or athletically active” (ibid. 27) and that “parental income is not statistically significant” (ibid.). Furthermore, they find that being female and having received a recommendation for upper secondary school are strong predictors of playing music rather than sport, or in case of the latter, of doing both activities instead of just one.

When testing for effect heterogeneity, the authors find different effects of playing music according to gender – where playing music appears more beneficial to girls with respect to language-related skills and school grades – and the socio-economic background of the adolescents of interest. They find that for adolescents whose parents are richer or more highly educated, as well as for higher-abled adolescents, playing music yields more advantages with respect to many outcomes than doing sports. In contrast, music and sports are similarly beneficial for children from disadvantaged social backgrounds (ibid. 33). The authors refer to Bourdieu’s hypothesis of cultural reproduction as a possible explanation for that finding.