Simulation parameter ranges for the simulation experiments reported in Fig 6.

For each simulated data-set in each of the three experiments, we generated labels for both the paired and random cases, using a regular sequence of “before” and “after” labels in the paired case, and a random sequence in the random case (where, for each one of the n positions in the sequence, we randomly sampled a “before” or “after” label). Each data-set was generated with a distinct autocorrelation coefficient (ρ) value randomly sampled from the ranges described in the third column (autocorrelation) of the table, and with a distinct sample size (n) value randomly sampled within the range {30, 32, 34, …, 800}. (Note we always sampled even values of n to make sure we have complete “before” and “after” pairs in the paired case).