
Full text loading...
Abstract
Psychometrics involves the indirect measurement of latent constructs, including aspects of cognition and emotion, and Likert-type scales are a common tool to operationalize and quantify these constructs. One threat to the psychometric properties of such scales is the administration of surveys across multiple languages, which presupposes the translation of the survey instruments. While multiple recommendations exist on best practices in translation, implementation does not always satisfy such guidelines. This article employs Monte Carlo simulation to explore the potential effects of translation on survey measurement and psychometric properties. Three possible challenges are explored, namely ambiguity, shift in valence, and issues with negation. These translation effects are statistically modeled as increased variance, change in skewness, and reverse coding, respectively. Additionally, the simulation examines the value of multi-item scales over single-item measurement. Overall, the results illustrate how survey translation can impact exploratory factor analysis and reliability of measurement.
Article metrics loading...
Full text loading...
References
Data & Media loading...