A BDS dataset includes all observed questionnaire scores, derived endpoint rows, and imputed rows, even though only a subset is used for the primary analysis. What is the main benefit of including the broader set?