Baseline correction

Often change score is an ineffective way of normalizing/scaling your data. Instead, when using a pre-post design, you might want to normalize your treatment effect by the baseline (pre-). This applies to both psychological scales as well as time-series.

Generally:

1. Compute the mean of the pre-treatment time series ${\mu }_{pre}$
2. For each post-treatment measurement at index $i$ subtract the baseline value: $X_{corrected}=x_{post,i}-{\mu }_{pre}$

---

Related

References