MétaCan
Menu
Back to cohort
Record W4415779721 · doi:10.33423/jabe.v27i5.7857

Approaching Human-Level Data Coding? A Systematic Comparison of OpenAI, DeepSeek, and Human Coders in Qualitative Analysis of Customer Reviews

2025· article· W4415779721 on OpenAlexvenueno aff
Yu Liu

Bibliographic record

VenueJournal of Applied Business and Economics · 2025
Typearticle
Language
FieldSocial Sciences
TopicComputational and Text Analysis Methods
Canadian institutionsnot available
Fundersnot available
KeywordsCoding (social sciences)Qualitative analysisQualitative researchScalabilityQualitative property

Abstract

fetched live from OpenAlex

The rapid growth of online reviews presents both opportunities and challenges for qualitative research. Human coding ensures contextual accuracy but is difficult to scale. This study compares human coding with AI-assisted coding using OpenAI GPT-4o and DeepSeek r1 on customer reviews of a complex DIY product. Results show both platforms capture underlying relationships, with OpenAI aligning more closely with human coding and DeepSeek demonstrating stronger internal consistency. Systematic AI errors mainly take the form of conservative Type I errors. Findings suggest AI can complement, rather than replace, human coders to enhance scalability and efficiency.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.550
metaresearch head score (Gemma)0.771
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesMetaresearch
Consensus categoriesMetaresearch
DomainCandidate signal: Methods · Consensus signal: Methods
Study designCandidate signal: Observational · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: none
Teacher disagreement score0.450
Threshold uncertainty score0.555

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.5500.771
Meta-epidemiology (narrow)0.0020.002
Meta-epidemiology (broad)0.0030.002
Bibliometrics0.0110.010
Science and technology studies0.0050.013
Scholarly communication0.0070.009
Open science0.0030.011
Research integrity0.0020.003
Insufficient payload (model declined to judge)0.0020.001

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.244
GPT teacher head0.471
Teacher spread0.227 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; the direct Gemma label and the distilled Codex classifier agree on what is shown here.

Study designObservational
DomainMethods
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations0
Published2025
Admission routes1
Has abstractyes

Explore more

Same venueJournal of Applied Business and EconomicsSame topicComputational and Text Analysis MethodsFrench-language works237,207