Then it's worse than not useful, it's potentially very misleading - real-world data (and salaries in particular) do not always follow that distribution. There's a reason medians are often the preferred benchmark.