Not quite normal: Consequences of violating the assumption of normality with regression mixture models Academic Article uri icon