Indirect measurement of mobility trends: The case of male intragenerational mobility

Much of what has been written about trend in mobility is based on the use of indirect standardization to infer changes in unobserved rates of upward mobility. This method, which consists of interpreting the differences between actual and hypothetical destination percentages (d values), has not been critically examined. Here we investigate two procedures for applying indirect standardization. The first procedure is modeled after previous use in the literature. Applied to data on one‐year male intragenerational mobility, this procedure erroneously concludes that there was an increase in the overall probability of upward mobility. The second procedure, which is based on the decomposition of d values into weighted sums of changes in aggregate mobility rates, produces results that are generally valid though diffuse. For two reasons, previous use of indirect standardization has been biased toward overstating increases in upward mobility: d values are not sensitive to upward shift in origin distribution, and d values do not distinguish between increase in upward mobility and reductions in downward mobility.