next up previous
Next: Phenomenology Up: Observability Previous: Coping with inhomogeneity

Latent variables

In practice, however, homogeneity can be much more difficult to identify or refute, because other problems and processes can cause the same symptoms. Suppose for two otherwise homogeneous scripts B and C, there is yet another unknown rogue process interfering with B. In this case, the fact that B had to correct a problem twice does not indicate a sequencing problem with C, but is rather due to a hidden latent variable that has nothing to do with the performance of either script we are considering. Conversely, a successful test like the above can only indicate that homogeneity is preserved in one test case. Any such analysis only suggests that homogeneity may or may not be preserved, but does not indicate whether it is or is not preserved. Sadly, the latter are mathematical ideals that are meaningless in practice.

Alva L. Couch