That's the point of the study: the real increase in wages to workers isn't the nominal statutory increase. For example, many people in the low-wage labor market in Seattle were already making more than the minimum wage.
Sure, but when the nominal increase is 18% and the selected group only shows 3% there's a big question about group selection. If you can cherry pick groups and variables then you can use phenomena like Simpson's paradox to prove pretty much anything with statistics.