Limiting data loss in infant EEG: putting hunches to the test

Abstract

EEG is a widely used tool to study the infant brain and its relationship with behavior. As infants usually have small attention spans, move at free will, and do not respond to task instructions, attrition rates are usually high. Increasing our understanding of what influences data loss is therefore vital. The current paper examines external factors to data loss in a large-scale on-going longitudinal study (the YOUth project; 1279 five-month-olds, 1024 ten-months-olds, and 109 three-year-olds). Data loss is measured for both continuous EEG and ERP tasks as the percentage data loss after artifact removal. Our results point to a wide array of external factors that contribute to data loss, some related to the child (e.g., gender; age; head shape) and some related to experimental settings (e.g., choice of research assistant; time of day; season; and course of the experiment). Data loss was also more pronounced in the ERP experiment than in the EEG experiment. Finally, evidence was found for within-subject stability in data loss characteristics over multiple sessions. We end with recommendations to limit data loss in future studies.