Abstract
The Social Security Administration's Continuous Work History Sample is used to show a strong trend towards more frequent early retirement among men. Moreover, the effects of including retirements due to incapacitation with other retirements and of including partial retirements with full retirements are quantified. The extent of retirement has previously been measured by the application of residual methods to aggregated census bureau data on population and labor force. The assumptions underlying that technique are examined critically, and issue is taken with a published study that purported to demonstrate with residual methods that the age at retirement of men is not changing.