How do you know a score from this quarter compares to one from last quarter?

Because the second attempt is not the same assessment, and it is not a random substitute either. It is a variant built from the same blueprint and checked before anyone sees it: just as hard, same skills, same question types, close to the same length.

We judge that across the whole assessment rather than question by question, since small differences even out over a full form.

After launch we watch how people actually score on each variant, and flag any that drifts.

Skip that step and a second score is an unknown mix of learning, memory, and luck of the draw.