A Double Machine Learning Approach to Estimate the Effects of Musical Practice on Student’s Skills

Abstract

This study investigates the dose–response effects of making music on youth development. Identification is based on the conditional independence assumption and estimation is implemented using a recent double machine learning estimator. The study proposes solutions to two highly practically relevant questions that arise for these new methods: (i) How to investigate sensitivity of estimates to tuning parameter choices in the machine learning part? (ii) How to assess covariate balancing in high‐dimensional settings? The results show that improvements in objectively measured cognitive skills require at least medium intensity, while improvements in school grades are already observed for low intensity of practice.

Returns to Higher Education and Dropouts: A Double Machine Learning Approach

“…In doing so it may be possible to shed light on such compositional changes. (Knaus, 2018) One of the extensions to Farrell (2015) proposed by Knaus (2018) is a method for systematically addressing potential sensitivity of the model to tuning parameter choices. This extension is based on the One Standard Error (1SE) rule proposed by Breiman et al (1984).…”

“…The extension proposed by Knaus (2018) is based on observation of the following; in the region around the penalty value indicating the minimum cross-validated MSE, values of the cross-validated MSE are similar. Thus, there is clearly a degree of uncertainty about the penalty term which minimises the MSE given there may exist a number of plausible values.…”

“…While based on the 1SE-rule proposed by Breiman et al (1984), the extension to Farrell (2015) that Knaus (2018) suggests is a little more comprehensive; the idea is to consider a set of models within a range plus or minus one of the estimated SE. Thus, the researcher considers not only a less complex model, but also a more complex model.…”

“… See Heigle & Pfeiffer (2019), who use DML with LASSO to estimate the effects of dropout on student outcomes in Germany, orKnaus (2018) who uses post-LASSO to estimate the returns to musicality in childhood.…”

Kausalanalyse mit maschinellem Lernen

“…Zur praktischen Illustration von DML werden als Abschluss dieses Beitrags drei empirische Anwendungen aus unterschiedlichen Forschungsfeldern andiskutiert. Knaus (2018) untersucht den Einfluss von musikalischer Betätigung auf die kognitiven und nicht-kognitiven Fähigkeiten von Jugendlichen, gemessen anhand schulischer Leistungstests und Persönlichlkeitsmerkmalen. Die Analyse basiert auf knapp 6900 Beobachtungen und fast 380 potenziellen Kontrollvariablen aus dem Jahr 2010 des "Nationalen Bildungspanels" für Deutschland und die Ergebnisse deuten im Allgemeinen auf positive Effekte von musikalischer Betätigung hin.…”