Surveying mentally retarded persons: responsiveness and response validity in three samples.