AltRA-Test: An Alternative Annotator Test for LLM Judges Without Shared Annotators
Abstract
A statistical test, AltRA-Test, determines when large language models can replace human annotators by comparing inter-annotator agreement in resampled mixed human-LLM panels, and evaluations show it often requires about 100 items and makes better replacement decisions than baselines.
Large language models (LLMs) are increasingly used as judges and graders to replace human annotators, raising the question of when such replacement is statistically justified. To support such decisions, the Alternative Annotator Test (Alt-Test) was introduced for dense settings where all items are annotated by the same annotators. However, in many large-scale annotation settings, items are annotated by different subsets of annotators. We therefore propose the Alternative Annotator Test with Resampled Annotations (AltRA-Test), a statistical test that does not require that all items are annotated by the same annotators. AltRA-Test uses a bootstrap that resamples items, draws k human annotations for each item, and substitutes one of them with an LLM annotation. This produces paired all-human and mixed human-LLM panels. A replacement is supported when the inter-annotator agreement (IAA) of the mixed panel is statistically non-inferior to the IAA of the all-human panel, i.e., when the LLM preserves the signal that the human annotators share. We establish that AltRA-Test is a consistent test of asymptotic level alpha. We then evaluate it through simulations and real-world experiments in which we test 18 LLMs and 5 prompting techniques on 20 datasets across text, image, audio, and video. Across these evaluations, about 100 items often suffice for the test. In simulations, where we know whether the LLM is a better or a worse annotator than the humans, AltRA-Test makes more correct replacement decisions than existing baselines. In real-world data experiments, LLMs pass the test in about half of the setups, depending mainly on the task. Since replacement is not always justified, we encourage researchers to test it for their task.
Community
AltRA-Test is a statistical test of whether an LLM judge can replace human annotators, and it needs no shared annotators, only three human annotations per item.
I like that this works when different people annotated different items, which the abstract describes as common in large-scale labeling. Asking whether the mixed panel agrees as well as the all-human panel feels like a fair bar for an LLM judge. If the test does not support replacement, is it usually a few ambiguous items or a broad drop across the dataset?
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- LLM Judge Validation Under Sparse Overlap: From Inference to Design (2026)
- Human-Residual Steering: Aligning Frozen LLM Judges with Human Judgment Distributions (2026)
- LLM Validation Makes Minimal-Pair Benchmarks Easier for Language Models (2026)
- Can This LLM Annotate My Data? Estimating Zero-Shot LLM Performance (2026)
- A Plausible Explanation Is Enough to Mislead Multimodal Judges (2026)
- Calibrate, Don't Retrain: Decomposing the Value of Annotator Disagreement in Cyberbullying Severity (2026)
- Large Language Models for Machine Translation Quality Annotation: Humans and Models Are Both Challenged (2026)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: @librarian-bot recommend
Get this paper in your agent:
hf papers read 2611.26073 Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash Models citing this paper 0
No model linking this paper
Datasets citing this paper 1
Spaces citing this paper 0
No Space linking this paper
Collections including this paper 0
No Collection including this paper