The Danish Goldsmiths Musical Sophistication Index: Validation in samples with different degrees of representativeness

Abstract

Convenience sampling is often used in music psychology research, leading to an overrepresentation of young participants with high socio-economic status, and potentially compromising the generalizability of empirical findings to the broader population. Fortunately, analysis techniques enable matching biased samples to known population characteristics, e.g., age, gender, and employment status. This study’s aim is twofold. First, we validate the Danish version of the Goldsmiths Musical Sophistication Index (Gold-MSI-da), a self-report inventory assessing various aspects of musical sophistication on continuous scales in the general population. Then, we demonstrate the use of stratification techniques and quantify the influence of sample composition by comparing results from analyses performed on our biased sample to similar analyses performed using stratified subsampling and poststratification weights to increase representativeness. The bifactor structure of the Gold-MSI-da was consistently replicated showing similar subscale percentiles in three sample variants. Psychometric benchmarks of internal consistency aligned with benchmarks from other language versions, and although model fits were weaker in the most representative sample, they were still acceptable. Interestingly, the sizes of the effect of self-reported musical training and expertise on musical listening abilities diminished with increasing sample representativeness, indicating potential overestimation when relying on non-representative samples. We conclude that the Gold-MSI-da is a valid tool for assessing musical sophistication in the Danish general population.