Need to take a closer look at whether this is a triage bug or an aequitas bug, but creating an issue here for the time being to make sure we do so. When a numeric field is specified as a bias attribute (even if the field is stored as a varchar in the database), pandas will cast this to a numeric type in the dataframe that gets sent to aequitas, which breaks the crosstabs. Instead, we should specify a string datatype for all the fields being read into the dataframe.
Need to take a closer look at whether this is a triage bug or an aequitas bug, but creating an issue here for the time being to make sure we do so. When a numeric field is specified as a bias attribute (even if the field is stored as a varchar in the database), pandas will cast this to a numeric type in the dataframe that gets sent to aequitas, which breaks the crosstabs. Instead, we should specify a string datatype for all the fields being read into the dataframe.