LLMs use fewer extreme grades than human examiners, with grade-dependent opposing biases compressing grades toward the middle¶
Claim · All claims
Evidence · 1 study · designr2·q3
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly
Subclaims¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q3 i? LLM grading showed central tendency: fewer A and F grades than humans, upward adjustment of lower grades and downward adjustment of higher grades. → Brevik A 2026
Evidence¶
Brevik A 2026¶
Brevik A, Jerpseth H and Lafontan SR (2026). Large language models as grading assistants in public health education: a method-comparison study of essay-style exam assessment. Front. Educ. 11:1904452. https://doi.org/10.3389/feduc.2026.1904452
q3 · i? · design · r2
Bland-Altman analysis (Figure 3) comparing the average of five ChatGPT sessions with human assessment. The plot showed "opposing biases across the grading scale"; the average numeric grade was 3.63 for humans versus 3.73 for the five ChatGPT parallels, a non-significant difference.
"Thus, the small overall mean difference concealed opposing biases across the grading scale: upward adjustmen t of lower grades and downward adjustmen t of higher grades. This pattern is consistent with a central tendency or grade-compression effect, whereby ChatGPT concentra ted assessments toward the middle of the grading scale."
Discussion¶
Related Claims¶
- LLM-generated BREQ responses capture overall item-mean patterns but are more extreme and less variable than human responses — related
- LLM agreement with human examiners is limited in fast mode and improves in reasoning or thinking modes — related
- ChatGPT in thinking mode shows the strongest agreement with human examiners on essay-style exam grading (weighted kappa 0.718) — related
- In one reviewed study, GPT-4 automated essay scoring showed score inflation, with AI–human rating discrepancies exceeding 1.5 standard deviations in 38% of graduate-level bibliographies — related
- Human-LLM agreement on a complex multi-label codebook falls well below human-human agreement, while LLM-LLM agreement is comparable to human-human agreement — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
llm-central-tendency-grade-compression |
title |
LLMs use fewer extreme grades than human examiners, with grade-dependent opposing biases compressing grades toward the middle |
description |
LLMs use fewer extreme grades than human examiners, with grade-dependent opposing biases compressing grades toward the middle |
status |
review |
evidence_strength |
moderate |
newest_evidence |
2026 |
recheck_by |
2029 |
generated |
by process:wiki-ingest on 2026-10-11 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — brevik-a-2026 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.