Estimating the number of jobs lost is very different than estimating the sign of a difference of a fixed "error-prone method for measuring jobs lost"

The former is asking "are my weighing scales accurate?" the latter is, "do I look like i'm loosing weight?"