Evaluating LLM Autograders Without Ground Truth

Huijun Mao; Christopher J Piech

Evaluating LLM Autograders Without Ground Truth

Correct Metadata for

Use this form to create a GitHub issue with structured data describing the correction. You will need a GitHub account. Once you create that issue, the correction will be reviewed by a staff member.

⚠️ Mobile Users: Submitting this form to create a new issue will only work with github.com, not the GitHub Mobile app.

Important: The Anthology treat PDFs as authoritative. Please use this form only to correct data that is out of line with the PDF. See our corrections guidelines if you need to change the PDF.

Title Adjust the title. Retain tags such as <fixed-case>.

Authors Adjust author names and order to match the PDF.

Abstract Correct abstract if needed. Retain XML formatting tags such as <tex-math>. You may use ... for bold, ... for italic, ... for underline, <sc>...</sc> for small-caps, <tt>...<tt> for typewriter text, <url>...</url> for URLs, <a href=...> for hyperlinks, and <par/> for paragraph breaks.

Verification against PDF Ensure that the new title/authors match the snapshot below. (If there is no snapshot or it is too small, consult the PDF.)

Authors concatenated from the text boxes above:

ALL author names match the snapshot above—including middle initials, hyphens, and accents.

Abstract

LLM autograders are usually evaluated against human “ground truth,” a single reference label provided by one annotator or constructed by aggregating multiple annotations. However, human grading is often variable and subjective. We instead propose to evaluate LLM graders by treating humans and LLMs as two grading systems and characterize their agreement and disagreement. We operationalize this approach using distance correlation and generalizability theory. Across two programming-grading datasets, we find higher Human–LLM agreement on more objectively specified rubric items. On items involving greater subjective judgment, LLMs exhibit strong internal consensus, diverging more from human graders.

Anthology ID:: 2026.emnlp-main.2465
Volume:: Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
Month:: October
Year:: 2026
Address:: Budapest, Hungary
Editors:: Hung-Yi Lee, Sunipa Dev, Jessy Li, Desmond Elliott
Venue:: EMNLP
SIG:: SIGDAT
Publisher:: Association for Computational Linguistics
Note:
Pages:: 55976–55986
Language:
URL:: https://aclanthology.org/2026.emnlp-main.2465/
DOI:
Bibkey:
Cite (ACL):: Huijun Mao and Christopher J Piech. 2026. Evaluating LLM Autograders Without Ground Truth. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 55976–55986, Budapest, Hungary. Association for Computational Linguistics.
Cite (Informal):: Evaluating LLM Autograders Without Ground Truth (Mao & Piech, EMNLP 2026)
Copy Citation:
PDF:: https://aclanthology.org/2026.emnlp-main.2465.pdf

PDF Cite Search Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{mao-piech-2026-evaluating,
    title = "Evaluating {LLM} Autograders Without Ground Truth",
    author = "Mao, Huijun  and
      Piech, Christopher J",
    editor = "Lee, Hung-Yi  and
      Dev, Sunipa  and
      Li, Jessy  and
      Elliott, Desmond",
    booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.emnlp-main.2465/",
    pages = "55976--55986",
    ISBN = "979-8-89176-459-0",
    abstract = "LLM autograders are usually evaluated against human ``ground truth,'' a single reference label provided by one annotator or constructed by aggregating multiple annotations. However, human grading is often variable and subjective. We instead propose to evaluate LLM graders by treating humans and LLMs as two grading systems and characterize their agreement and disagreement. We operationalize this approach using distance correlation and generalizability theory. Across two programming-grading datasets, we find higher Human{--}LLM agreement on more objectively specified rubric items. On items involving greater subjective judgment, LLMs exhibit strong internal consensus, diverging more from human graders."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="mao-piech-2026-evaluating">
    <titleInfo>
        <title>Evaluating LLM Autograders Without Ground Truth</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Huijun</namePart>
        <namePart type="family">Mao</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Christopher</namePart>
        <namePart type="given">J</namePart>
        <namePart type="family">Piech</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-10</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Hung-Yi</namePart>
            <namePart type="family">Lee</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Sunipa</namePart>
            <namePart type="family">Dev</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jessy</namePart>
            <namePart type="family">Li</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Desmond</namePart>
            <namePart type="family">Elliott</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Budapest, Hungary</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-459-0</identifier>
    </relatedItem>
    <abstract>LLM autograders are usually evaluated against human “ground truth,” a single reference label provided by one annotator or constructed by aggregating multiple annotations. However, human grading is often variable and subjective. We instead propose to evaluate LLM graders by treating humans and LLMs as two grading systems and characterize their agreement and disagreement. We operationalize this approach using distance correlation and generalizability theory. Across two programming-grading datasets, we find higher Human–LLM agreement on more objectively specified rubric items. On items involving greater subjective judgment, LLMs exhibit strong internal consensus, diverging more from human graders.</abstract>
    <identifier type="citekey">mao-piech-2026-evaluating</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.2465/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>55976</start>
            <end>55986</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T Evaluating LLM Autograders Without Ground Truth
%A Mao, Huijun
%A Piech, Christopher J.
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F mao-piech-2026-evaluating
%X LLM autograders are usually evaluated against human “ground truth,” a single reference label provided by one annotator or constructed by aggregating multiple annotations. However, human grading is often variable and subjective. We instead propose to evaluate LLM graders by treating humans and LLMs as two grading systems and characterize their agreement and disagreement. We operationalize this approach using distance correlation and generalizability theory. Across two programming-grading datasets, we find higher Human–LLM agreement on more objectively specified rubric items. On items involving greater subjective judgment, LLMs exhibit strong internal consensus, diverging more from human graders.
%U https://aclanthology.org/2026.emnlp-main.2465/
%P 55976-55986

Download as File

Markdown (Informal)

[Evaluating LLM Autograders Without Ground Truth](https://aclanthology.org/2026.emnlp-main.2465/) (Mao & Piech, EMNLP 2026)

Evaluating LLM Autograders Without Ground Truth (Mao & Piech, EMNLP 2026)

ACL

Huijun Mao and Christopher J Piech. 2026. Evaluating LLM Autograders Without Ground Truth. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 55976–55986, Budapest, Hungary. Association for Computational Linguistics.