Insufficient Effort Responding and Adolescent Respondents: Measurement, Extent, and Prediction

Abstract

This article addresses insufficient effort responding (IER), an issue in survey research affecting data quality. We focus on predicting reactive-check-based IER classifications through nonreactive measures and estimating the proportion of respondents classified as IER in a pupil population, using a sizeable adolescent survey. The analysis highlights IER as a nonmarginal issue that varies considerably by gender, migration status, and school type. Utilizing Random Forest models, we evaluate nonreactive measures’ predictive power for reactive-check-based IER classifications, notably response time, intraindividual response variability, and Mahalanobis Distance. The findings show these measures’ future research value, emphasizing the strong influence of response time. We also explore the relationship between predictors and our target variable and find that shorter response times and less response variability correspond to a greater likelihood of IER. This study illustrates the potential of nonreactive measures and advanced machine learning techniques for identifying IER and highlights the necessity for further research.