Social polarization in global cities: measuring changes in earnings and occupational inequality. Regional Studies. There is a paradox in the scholarly contributions to the social polarization debate. Studies that rely on occupational descriptions to measure employment change produce results that take the form of a professionalizing pattern of growth. By contrast, studies that rank occupations by income to measure employment change produce evidence in support of social polarization. This study applies both methods to measure employment change in greater Johannesburg (South Africa) from 1996 to 2012 in order to demonstrate why these different methods produce such different results. The results have important implications for how statistical evidence is used to test the social polarization hypothesis.