Comment by lowercased

> However, it’s worrying that a system of age-based payouts has such uncertainty in its data.

The data itself may have to be interpreted, which I would classify as 'suboptimal', but seemingly 'normal' for most projects I work with. I often have to join together various tables, remembering to include or exclude specific data via conditional logic. The conditional logic may be context-dependent, and documenting those cases is really key. Why include/exclude specific subsets of data to answer questions XYZ? Have those criteria changed over the years (and if so, why?)

Looking at raw data tables it's often quite easy to come up with ways to show the data to support whatever case you're trying to make.