That information can still be considered. For instance, if you have had a borderline performance in an exercise or competency, they may look back at earlier stages to see if there is evidence elsewhere that supports the decision (which ever way it goes). For instance, I had many instances where a written exercise wasn't ideal, and so we would look back at the pyschometric assessments (which then was a verbal reasoning assessment) and written style of their application to see if there were any concerns there. If there were, it tended to push a decision to a no, while if they were strong, it might push it to a yes.
However, I would say in most instances, this approach isn't needed. You can typically make a decision just based on the latest assessments you have undertaken. If a candidate has made it through the earlier stages, there is a minimum benchmark they have met anyway (and that tends to be quite high) and so trying to benchmark candidates based on their earlier performances is a bit futile as they are often quite similar.