Cross sectional employment data is not random. Workers who survive to a longer level of tenure tend to have a higher level of productivity than those who exit earlier. Wage equations that use cross sectional data could be biased from the over sampling of high productive workers at long levels of tenure. The survival bias that arises in cross sectional data could possibly bias the coefficients in wage equations. This could lead to false positive conclusions concerning the presence of pay discrimination. Using 1989-2008 NBA data we explore the extent of survival bias in wage regressions in a setting in which worker productivity is extremely well documented through a variety of statistical measures. We then examined whether the survival bias affects the conclusions concerning racial pay discrimination. Key Words: NBA, survival bias, pay discrimination
