{{ extends "global/Page.html" }} {{ block title }}Instructions - Part 2{{ endblock }} {{ block content }}

How Scores Relate to True Quality

For each score, higher numbers indicate better evaluations. You can think of each score as a noisy signal of the candidate's true underlying quality. A candidate may perform better or worse than their true ability at a particular stage, so no single score is perfectly accurate on its own.

Historically, these evaluation errors are normally distributed with mean 0, and the amount of noise across stages is shown below:

The figure below illustrates the distribution of evaluation errors for each stage. All three error terms are centered at zero.

A negative error means the stage score underestimates true quality, a zero error means the stage score equals true quality, and a positive error means the stage score overestimates true quality.

The noise at each stage is independent. This means that a candidate may perform better than their true quality in one stage, and worse than their true quality in another stage—each stage’s error is drawn fresh.

{{ if noise_distribution_uri }}
Distributions of Resume, Zoom, and In-person evaluation errors
{{ endif }}

Comprehension Check

{{ form.Q_score_interpretation.label }}

{{ form.Q_score_interpretation }}
{{ next_button }}
{{ endblock }}