GratisAIcareer · field guide · local-first · no account
rating caliper

Spot Interview Ratings That Need Panel Discussion — Free

Averages can hide disagreement about the same interview evidence. Use rating spread to identify criteria needing a panel conversation, without automatically rejecting or ranking candidates.

The proof surface

Interview practice tools coach an applicant’s answer. This helps a panel review disagreement in a shared rubric, not score a person with AI or infer suitability.

Inputcriterion | lowest panel rating | highest panel rating; spread that triggers discussion
Rare deviceInterview practice tools coach an applicant’s answer. This helps a panel review disagreement in a shared rubric, not score a person with AI or infer suitability.
Output artifactrating points of spread per row with transparent workings
Cost$0 · local-first · one free run · samples labelled
Sample, not your facts: Illustrative sample: Technical example | 2 | 5; Collaboration evidence | 3 | 4; Role knowledge | 4 | 4. With spread that triggers discussion set to 2, total is 4.00 rating points of spread.

Why the flat version breaks

Averaging away disagreement

The same average can conceal very different evidence judgments.

Treating spread as candidate quality

It describes panel disagreement, not competence.

Forcing consensus

A different rating may reflect valid evidence worth recording.

How to work the rating caliper

Use one anchored scale

Every panelist must have used the same criterion definitions and scale. This tool accepts non-negative rating values but cannot detect incompatible rubrics.

Enter extremes for one candidate

Use anonymous criterion labels, not candidate names or protected characteristics. Review candidates separately; this is not a cross-candidate ranking table.

Subtract the low from the high

A technical example rated between two and five has a three-point spread. Flag spreads at or above the chosen threshold, including exactly two in this sample setting.

Discuss evidence before revising scores

Ask what observation supports each rating and whether the rubric was interpreted consistently. Preserve independent initial ratings in the authorized hiring system; agreement alone does not remove bias.

What the rating caliper replaces

QuestionBeforeVisible working
Averaging away disagreementThe same average can conceal very different evidence judgments.Enter extremes for one candidate
Treating spread as candidate qualityIt describes panel disagreement, not competence.Subtract the low from the high
Forcing consensusA different rating may reflect valid evidence worth recording.Discuss evidence before revising scores

A local arithmetic aid replaces hand calculation, not expert review. No paid AI service is needed.

Run it on the samples, right here

FIRST-LOAD

Input is processed locally and a draft is saved automatically in this browser profile when storage is available. Reset to sample clears that draft; Pro history has its own clear button. A state link encodes your inputs in its URL: share only non-sensitive rows. Browser history, clipboard and anyone receiving the link may retain it. Optional AI use below leaves this device; it is not required.

Human review aid only; not an automated employment decision, bias audit or legal compliance check. Use an approved fair hiring process.

Data note: Everything runs in this browser tab on the rating caliper: your criterion | lowest panel rating | highest panel rating stays on this device, nothing is uploaded, and the reading is rebuilt only when you press run.

Go deeper: the companion app files the same reading as a meter band gauge

The article demo above runs without limits. The companion app keeps a local history, exports the rows as CSV, prints the rating caliper reading, and holds your drafts on this device — one free run, then $ 4 one-time for the layer that keeps filing.

The rating caliper reading is complete for free. The optional $4 layer adds print, row CSV and the last five local reading summaries; it does not add hidden answers. Checkout is not configured yet; the article demo remains unlimited.

Open the spot interview ratings that need panel discussion companion

Boundary

Human review aid only; not an automated employment decision, bias audit or legal compliance check. Use an approved fair hiring process.

What this is built on

Before: The same average can conceal very different evidence judgments. After: the rating caliper shows the working beside each named row so the reader can change the assumption and inspect the consequence.

Optional AI formatting, not calculation

For this rating caliper, use the free AI Studio interface only if available to you, with no paid API key. Manual entry completes the same workflow for free. Supply only fictional or non-sensitive notes. Review its output against the source; never paste unresolved questions into the numeric rows.

Format my non-sensitive notes for a rating caliper. Return plain rows only: criterion | lowest panel rating | highest panel rating. Preserve supplied quantities exactly. Do not guess missing values; list questions separately. Do not calculate or add advice. I will check every row before pasting into the local tool.

Accepted schema: criterion | lowest panel rating | highest panel rating. No AI response is executed as code.