>>12306052>heteroskedasticityMemeword used almost exclusively by those who lack practical experience. The central limit theorem is robust against most forms of heteroskedasticity (as a limit theorem should be), and sometimes you can even estimate heteroskedastic parameters using methods such as GLS.
The real concerns are omitted variables, missing data, measurement error, and determination of convergence rates. In contrast, you rarely hear of studies getting challenged on grounds of heteroskedasticity.
t. economist
