For each score, the higher numbers indicate better evaluations. You can think of each score as a noisy signal of the candidate's true underlying quality. Sometimes a candidate will perform a bit better or worse than their true ability in a particular stage (for example, they might have a great or bad day in one interview), so no single score is perfectly accurate on its own. Historically, it is known that each evaluation score has an error term that is normally distributed with mean 0 and standard deviation {{ sigma_resume }}.
