Comment on ‘Local model uncertainty and incomplete-data bias’