Does your calculation try to recreate the 0.83 $ statistic by assuming the proportion of the workforce of a given (gender, race) is the same as the population as a whole?

Because that's not necessarily a valid assumption.