Using Large Language Models for Qualitative Analysis can Introduce Serious Bias

Julian Ashwin, Aditya Chhabra and Vijayendra Rao

No 10597, Policy Research Working Paper Series from The World Bank

Abstract: Large Language Models (LLMs) are quickly becoming ubiquitous, but the implications for social science research are not yet well understood. This paper asks whether LLMs can help us analyse large-N qualitative data from open-ended interviews, with an application to transcripts of interviews with displaced Rohingya people in Cox’s Bazaar, Bangladesh. The analysis finds that a great deal of caution is needed in using LLMs to annotate text as there is a risk of introducing biases that can lead to misleading inferences. Here this refers to bias in the technical sense, that the errors that LLMs make in annotating interview transcripts are not random with respect to the characteristics of the interview subjects. Training simpler supervised models on high-quality human annotations with flexible coding leads to less measurement error and bias than LLM annotations. Therefore, given that some high quality annotations are necessary in order to asses whether an LLM introduces bias, this paper argues that it is probably preferable to train a bespoke model on these annotations than it is to use an LLM for annotation.

Date: 2023-11-07
New Economics Papers: this item is included in nep-cmp
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
http://documents.worldbank.org/curated/en/09943331 ... 85d0ab10e497bd1f.pdf (application/pdf)

Related works:
Working Paper: Using Large Language Models for Qualitative Analysis can Introduce Serious Bias (2023)
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:wbk:wbrwps:10597

Access Statistics for this paper

More papers in Policy Research Working Paper Series from The World Bank 1818 H Street, N.W., Washington, DC 20433. Contact information at EDIRC.
Bibliographic data for series maintained by Roula I. Yazigi (ryazigi@worldbank.org).