The raw dataset mistakenly recorded the label of frequency of five discriminatory situations