Accounting for motion in resting-state fMRI: What part of the spectrum are we characterizing in autism spectrum disorder?

•Exclusion of high-motion subjects reduces rs-fMRI artifacts but may bias the study sample.•Autistic children with usable data had less severe symptoms and were older than the original sample.•Among children with usable data, symptom severity and age were related to functional connectivity.•Using doubly robust targeted minimum loss based estimation (DRTMLE) to address these biases, we observe more extensive group differences.•If quality control changes the distribution of a variable related to the outcome, we recommend using DRTMLE.