Elamur Research⌕
☰
home / papers / 2024 / Psychometric Alignment: Capturing Human Knowledge Distributions via Language Models

Psychometric Alignment: Capturing Human Knowledge Distributions via Language Models

Paper · 2024 · Joy He-Yueya, Wanjing Anya Ma, Kanishk Gandhi et al.

Summary

Language models are increasingly used to stand in for human populations, for example when designing educational materials. In these uses, a model should reproduce how knowledge varies across people, not just give correct answers. The authors propose "psychometric alignment" to measure this. Humans and LM instances answer the same test items. A 1PL (Rasch) Item Response Theory model is then fitted separately to each group, and the metric is the Pearson correlation between the two sets of item difficulty estimates. Simulations show the metric stays reliable when groups differ in ability, falls as more items function differently between groups, and becomes more stable with larger groups. It also reveals differences that comparing accuracy alone misses.

The authors tested it on three domains: WordBank (first-language acquisition), Duolingo (second-language learning) and a new mathematics dataset, Eedi. Current LMs were clearly misaligned with humans. Persona-based prompting helped, and smaller models aligned better than larger ones. Fine-tuning on student data gave mixed results across domains. The metric offers a way to detect representation failures when LMs simulate people.

Key points

  • Eedi dataset: derived from the NeurIPS 2020 Education Challenge data. It contains responses from 2287 students aged 11–12 to 573 multiple-choice math questions. Code and data are on GitHub (joyheyueya/psychometric-alignment).
  • Simulation results: with 200 respondents per group and no DIF items, alignment is about 0.95. It drops to roughly 0.37 when 50 items are made to function differently between groups.
  • Prompting: persona prompting improves alignment but stays well below the ceiling set by comparing human subgroups from the same population. Smaller LMs tend to outperform larger ones.
  • Fine-tuning setup: Mistral-7b, Llama-8b and Deepseek-7b were fine-tuned with LoRA (r = 32, lora_alpha = 64) on student personas and response histories.
  • Fine-tuning results: the fine-tuned models beat the best prompting baseline only on WordBank, where they reach the alignment that 10–20 real students' data on the test items would provide. They do not beat it on Eedi or Duolingo.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Stanford UniversityStanford · United States434 works · 314,419 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksDecision Sciences · 209 worksManagement Science and Operations Research · 108 works

🧭 Topics 1Research subjects and languages the work is about.Language models & pretraining · 3,320 works

Cite

Cite as Joy He-Yueya, Wanjing Anya Ma, Kanishk Gandhi, Benjamin W. Domingue, Emma Brunskill and Noah D. Goodman. 2024. Psychometric Alignment: Capturing Human Knowledge Distributions via Language Models. arXiv:2407.15645.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2407.15645 · arXiv 2407.15645

@misc{heyueya2024psychometric,
  title = {{Psychometric Alignment: Capturing Human Knowledge Distributions via Language Models}},
  author = {Joy He-Yueya, Wanjing Anya Ma, Kanishk Gandhi, Benjamin W. Domingue, Emma Brunskill and Noah D. Goodman},
  year = {2024},
  doi = {10.48550/arxiv.2407.15645},
  eprint = {2407.15645},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2407.15645}
}

References

40 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Gati V Aher, Rosa I Arriaga, and Adam Tauman Kalai. Using large language models to simulate multiple humans and replicate human subject studies. In International Conference on Machine Learning, pages 337–371. PMLR, 2023. find
  2. Usman Anwar, Abulhair Saparov, Javier Rando, Daniel Paleka, Miles Turpin, Peter Hase, Ekdeep Singh Lubana, Erik Jenner, Stephen Casper, Oliver Sourbut, et al. Foundational challenges in assuring alignment and safety of large language models. arXiv preprint arXiv:2404.09932, 2024. find
  3. Lisa P Argyle, Ethan C Busby, Nancy Fulda, Joshua R Gubler, Christopher Rytting, and David Wingate. Out of one, many: Using language models to simulate human samples. Political Analysis, 31(3):337–351, 2023.
  4. James Brand, Ayelet Israeli, and Donald Ngwe. Using gpt for market research. Available at SSRN 4395751, 2023. find
  5. Gregory Camilli. Test fairness. Educational measurement, 4:221–256, 2006. find
  6. R Philip Chalmers. mirt: A multidimensional item response theory package for the r environment. Journal of statistical Software, 48:1–29, 2012. find
  7. Chaoran Chen, Bingsheng Yao, Yanfang Ye, Dakuo Wang, and Toby Jia-Jun Li. Evaluating the llm agents for simulating humanoid behavior. 1st HEAL Workshop at CHI Conference on Human Factors in Computing Systems, 2024. find
  8. Yun-Shiuan Chuang, Agam Goyal, Nikunj Harlalka, Siddharth Suresh, Robert Hawkins, Sijia Yang, Dhavan Shah, Junjie Hu, and Timothy T Rogers. Simulating opinion dynamics with networks of llm-based agents. arXiv preprint arXiv:2311.09618, 2023. find
  9. Michael C Frank, Mika Braginsky, Daniel Yurovsky, and Virginia A Marchman. Wordbank: An open repository for developmental vocabulary data. Journal of child language, 44(3):677–694, 2017. find
  10. Joy He-Yueya, Noah D Goodman, and Emma Brunskill. Evaluating and optimizing educational content with large language model judgments. arXiv preprint arXiv:2403.02795, 2024. find
  11. John J Horton. Large language models as simulated economic agents: What can we learn from homo silicus? Technical report, National Bureau of Economic Research, 2023. find
  12. Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685, 2021.
  13. Hyoungwook Jin, Seonghee Lee, Hyungyu Shin, and Juho Kim. Teach ai how to code: Using large language models as teachable agents for programming education. In Proceedings of the CHI Conference on Human Factors in Computing Systems, pages 1–28, 2024. find
  14. John P Lalor, Hao Wu, and Hong Yu. Learning latent parameters without human response patterns: Item response theory with artificial crowds. In Proceedings of the Conference on Empirical Methods in Natural Language Processing. Conference on Empirical Methods in Natural Language Processing, volume 201… find
  15. NIH Public Access, 2019. find
  16. Antonio Laverghetta Jr, Animesh Nighojkar, Jamshidbek Mirzakhalov, and John Licato. Can transformer language models predict psychometric properties? arXiv preprint arXiv:2106.06849, 2021. find
  17. Nian Li, Chen Gao, Yong Li, and Qingmin Liao. Large language model-empowered agents for simulating macroeconomic activities. arXiv preprint arXiv:2310.10436, 2023. — [arXiv:2310.10436](https://arxiv.org/abs/2310.10436) find
  18. Ryan Liu, Howard Yen, Raja Marjieh, Thomas L Griffiths, and Ranjay Krishna. Improving interpersonal communication by simulating audiences with language models. arXiv preprint arXiv:2311.00687, 2023. find
  19. Frederic M Lord. Applications of item response theory to practical testing problems. Routledge, 2012. find
  20. Xinyi Lu and Xu Wang. Generative students: Using llm-simulated student profiles to support question item evaluation. In Proceedings of the Eleventh ACM Conference on Learning@ Scale, pages 16–27, 2024. find
  21. Wanjing Anya Ma, Adam Richie-Halford, Amy Burkhardt, Klint Kanopka, Clementine Chou, Ben Domingue, and Jason D Yeatman. Roar-cat: Rapid online assessment of reading ability with computerized adaptive testing. 2023. find
  22. David Magis, Sébastien Béland, Francis Tuerlinckx, and Paul De Boeck. A general framework and an r package for the detection of dichotomous differential item functioning. Behavior research methods, 42(3):847–862, 2010. find
  23. Julia M Markel, Steven G Opferman, James A Landay, and Chris Piech. Gpteach: Interactive ta training with gpt-based students. In Proceedings of the tenth acm conference on learning@ scale, pages 226–236, 2023. find
  24. R Thomas McCoy, Shunyu Yao, Dan Friedman, Matthew Hardy, and Thomas L Griffiths. Embers of autoregression: Understanding large language models through the problem they are trained to solve. arXiv preprint arXiv:2309.13638, 2023. find
  25. Joon Sung Park, Lindsay Popowski, Carrie Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. Social simulacra: Creating populated prototypes for social computing systems. In Proceedings of the 35th Annual ACM Symposium on User Interface Software and Technology, pages 1–18, 2022. find
  26. Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. Generative agents: Interactive simulacra of human behavior. In Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology, pages 1–22, 2023. find
  27. Max Pellert, Clemens M Lechner, Claudia Wagner, Beatrice Rammstedt, and Markus Strohmaier. Ai psychometrics: Assessing the psychological profiles of large language models through psychometric inventories. Perspectives on Psychological Science, page 17456916231214460, 2023. find
  28. Georg Rasch. Studies in mathematical psychology: I. probabilistic models for some intelligence and attainment tests. 1960. find
  29. André A Rupp and Bruno D Zumbo. Understanding parameter invariance in unidimensional irt models. Educational and psychological measurement, 66(1):63–84, 2006. find
  30. Mustafa Safdari, Greg Serapio-García, Clément Crepy, Stephen Fitz, Peter Romero, Luning Sun, Marwa Abdulhai, Aleksandra Faust, and Maja Matarić. Personality traits in large language models. arXiv preprint arXiv:2307.00184, 2023.
  31. Shibani Santurkar, Esin Durmus, Faisal Ladhak, Cinoo Lee, Percy Liang, and Tatsunori Hashimoto. Whose opinions do language models reflect? In International Conference on Machine Learning, pages 29971–30004. PMLR, 2023.
  32. Burr Settles, Chris Brust, Erin Gustafson, Masato Hagiwara, and Nitin Madnani. Second language acquisition modeling. In Proceedings of the thirteenth workshop on innovative use of NLP for building educational applications, pages 56–65, 2018. find
  33. Omar Shaikh, Valentino Chai, Michele J Gelfand, Diyi Yang, and Michael S Bernstein. Rehearsal: Simulating conflict to teach conflict resolution. arXiv preprint arXiv:2309.12309, 2023. find
  34. Megha Srivastava and Noah Goodman. Question generation for adaptive education. arXiv preprint arXiv:2106.04262, 2021. find
  35. Ilia Sucholutsky, Lukas Muttenthaler, Adrian Weller, Andi Peng, Andreea Bobu, Been Kim, Bradley C Love, Erin Grant, Jascha Achterberg, Joshua B Tenenbaum, et al. Getting aligned on representational alignment. arXiv preprint arXiv:2310.13018, 2023. — [arXiv:2310.13018](https://arxiv.org/abs/2310.13018) find
  36. Zichao Wang, Angus Lamb, Evgeny Saveliev, Pashmina Cameron, Yordan Zaykov, José Miguel Hernández-Lobato, Richard E Turner, Richard G Baraniuk, Craig Barton, Simon Peyton Jones, et al. Instructions and guide for diagnostic questions: The neurips 2020 education challenge. arXiv preprint arXiv:2007.120… — [arXiv:2007.12061](https://arxiv.org/abs/2007.12061) find
  37. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35:24824–24837, 2022.
  38. Mike Wu, Richard L Davis, Benjamin W Domingue, Chris Piech, and Noah Goodman. Variational item response theory: Fast, accurate, and expressive. arXiv preprint arXiv:2002.00276, 2020. — [arXiv:2002.00276](https://arxiv.org/abs/2002.00276) find
  39. Songlin Xu and Xinyu Zhang. Leveraging generative artificial intelligence to simulate student learning behavior. arXiv preprint arXiv:2310.19206, 2023. — [arXiv:2310.19206](https://arxiv.org/abs/2310.19206) find
  40. Eric Zelikman, Wanjing Anya Ma, Jasmine E Tran, Diyi Yang, Jason D Yeatman, and Nick Haber. Generating and evaluating tests for k-12 students with language model simulations: A case study on sentence reading efficiency. arXiv preprint arXiv:2310.06837, 2023. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/he-yueya-2024-psychometric-alignment-capturing-human-knowledge/