A randomized crossover trial in older adults found that hearing-loss severity — not cognition or prior hearing-aid experience — predicts preference for slow versus fast compression; speech recognition in quiet and noise was unchanged by compression speed.