Sampling biases in studies of gender and schizophrenia.