基于大语言模型的跨文化个性测量表（CPAI-2）人际取向结构探索：一项初步研究

李文冰; 朱天娇; 甘怡群

doi:10.12100/j.issn.2096-5494.226010

心理学通讯 >

2026 , Vol. 9 >Issue 1: 62 - 70

DOI: https://doi.org/10.12100/j.issn.2096-5494.226010

研究方法

基于大语言模型的跨文化个性测量表（CPAI-2）人际取向结构探索：一项初步研究

李文冰 ,
朱天娇 ,
甘怡群

展开

北京大学心理与认知科学学院,北京 100080

李文冰,2022年毕业于西南大学心理学部,获学士学位。目前就读于北京大学心理与认知科学学院,专业方向为临床心理学。研究领域主要聚焦于应激应对和数字干预。

^*甘怡群ygan@pku.edu.cn

网络出版日期: 2026-10-23

基金资助

重点研发计划项目（2024YFC3606800）; 国家自然科学基金（32471135）

收起

Exploring the Structure of the CPAI-2 Interpersonal Relatedness Factor Using Large Language Models: A Pilot Study

LI Wenbing ,
ZHU Tianjiao ,
GAN Yiqun

Expand

School of Psychological and Cognitive Sciences, Peking University, Beijing 100080, China

^*GAN Yiqun, ygan@pku.edu.cn

Online published: 2026-10-23

Fold

摘要

目的：探究跨文化个性测量表（Cross-Cultural Personality Assessment Inventory,CPAI-2）中人际取向（Interpersonal Relatedness）维度的模型结构。
方法：以100名中国成年个体为研究对象,邀请被试线上作答跨文化个性测量表人际取向量表和开放式语言任务问卷,以考察被试的自然语言表达中是否能反应出CPAI-2中人际取向维度的模型结构。采用开放式语言任务结合大语言模型（Large Language Models,LLM）ChatGPT-5进行数据编码、分析和模型建构,并使用余弦相似度进行分析。
结果：（1）两轮试验结果具有较高的相似性。（2）ChatGPT所建立的模型与原模型具有相似性和差异性,其中“和谐性”和“人际触觉”未在试验结果得到验证,这可能与文本含义和编码等级有关,可在未来的研究中进一步探索。
结论：该研究通过开放式语言任务的方法,结合大语言模型编码,拓展了CPAI-2的验证路径,为人格测评方法的多元化提供实践依据,推动了CPAI-2的应用和发展。

关键词： 跨文化个性测量表（CPAI-2）; 大语言模型（LLM）; 主题分析

本文引用格式

李文冰 , 朱天娇 , 甘怡群 . 基于大语言模型的跨文化个性测量表（CPAI-2）人际取向结构探索：一项初步研究[J]. 心理学通讯, 2026 , 9(1) : 62 -70 . DOI: 10.12100/j.issn.2096-5494.226010

Abstract

Objective: This study aimed to explore the structural model of the Interpersonal Relatedness factor in the Cross-Cultural Personality Assessment Inventory (CPAI-2).
Methods: A sample of 100 Chinese adults was recruited through an online survey. Participants completed the Interpersonal Relatedness scale of the CPAI-2 and an open-ended language task. The study examined whether the structural features of the Interpersonal Relatedness factor could be reflected in participants' natural language expressions. Responses from the open-ended task were coded and analyzed using the large language model ChatGPT-5 for model construction. Cosine similarity analysis was conducted to evaluate the consistency and correspondence of the derived structures.
Results: (1) The results of two independent coding rounds demonstrated a high degree of similarity. (2) The model constructed by ChatGPT showed both convergence and divergence relative to the original CPAI-2 model. Specifically, the dimensions of Harmony and Social Sensitivity were not supported in the present findings. This may be related to differences in textual meaning and levels of coding and warrants further investigation.
Conclusion: By integrating an open-ended language task with large language model coding, this study extends the validation approach of the CPAI-2, provides empirical support for methodological diversification in personality assessment, and contributes to the continued application and development of the CPAI-2.

Key words： Cross-Cultural Personality Assessment Inventory (CPAI-2); Large Language Models (LLM); Thematic Analysis

参考文献

[1] Cheung, F. M., & Cheung, S. F. (2003). Measuring personality and values across cultures:Imported versus indigenous measures.Online Read Psychol Cult, 4(4), 5.
[2] Cheung F. M., Cheung S. F., Zhang J. X., Leung K., Leong F., & Kuang H. (2008). Relevance of openness as a personality dimension in Chinese culture:Aspects of its cultural relevance.J Cross-Cult Psychol, 39(1), 81-108. doi:10.1177/0022022107311968
[3] Cheung F. M., Leung K., Fan R. M., Song W. Z., Zhang J. X., & Zhang J. P. (1996). Development of the Chinese personality assessment inventory.J Cross-Cult Psychol, 27(2), 181-199.
[4] Cheung F. M., Leung K., Zhang J. X., Sun H. F., Gan Y. Q., Song W. Z., & Xie D. (2001). Indigenous Chinese personality constructs:Is the five-factor model complete? J Cross-Cult Psychol, 32(4), 407-433. doi:10.1177/0022022101032004003
[5] Cheung S. F., Cheung F. M., Howard R., & Lim Y.H. (2006). Personality across the ethnic divide in Singapore:Are “Chinese traits” uniquely Chinese? Pers Individ Dif, 41(3), 467-477. doi:10.1016/j.paid.2005.12.023
[6] Costa, P. T., & McCrae, R. R. (1985). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI):Professional manual. Odessa, FL:Psychological Assessment Resources.
[7] Cui T., Kam C. C., Cheng E. H., & Ho M. Y. (2022). Distinguishing between trait desirability and item desirability in predicting item scores:Is informant evaluation of personality free from social desirability? Pers Individ Dif, 196, 111708. doi:10.1016/j.paid.2022.111708
[8] De Paoli, S., & Mathis, W. S. (2025). Reflections on inductive thematic saturation as a potential metric for measuring the validity of an inductive thematic analysis with LLMs.Qual Quant, 59(1), 683-709. doi:10.1007/s11135-024-01950-6
[9] Devlin J., Chang M. W., Lee K., & Toutanova K. (2019). BERT:Pre-training of deep bidirectional transformers for language understanding. In J. Burstein, C. Doran, & T. Solorio (Eds.), Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics:Human Language Technologies(pp. 4171-4186). Minneapolis, MN:Association for Computational Linguistics. doi:10.18653/v1/N19-1423
[10] Fan W., Li M., Leong F., & Zhou M. (2021). West meets East in a new two-polarities model of personality:Combining self-relatedness structure with independent-interdependent functions.Front Psychol, 12, 739900. doi:10.3389/fpsyg.2021.739900
[11] Guo Q., Cao J., Xie X., Liu S., Li X., Chen B., & Peng X. (2024). Exploring the potential of ChatGPT in automated code refinement:An empirical study. In Proceedings of the IEEE/ACM 46th International Conference on Software Engineering(pp. 1-13). New York, NY, USA:ACM. doi:10.1145/3597503.3623306
[12] Habib F., Ali Z., Azam A., Kamran K., & Pasha F. M. (2024). Navigating pathways to automated personality prediction:A comparative study of small and medium language models.Front Big Data, 7, 1387325. doi:10.3389/fdata.2024.1387325
[13] Jain N., Adeyinka S., Roseman L., & Allsop A. (2025). Multi-LLM thematic analysis with dual reliability metrics:Combining Cohen's kappa and semantic similarity for qualitative research validation[Preprint].arXiv. doi:10.48550/ARXIV.2512.20352
[14] Kao P. Y., Chen M. H., Chang W. A., Pan M. L., Shu W. D., Jong Y. J.,.. Lin Y. S. (2023). A genome-wide association study (GWAS) of the personality constructs in CPAI-2 in Taiwanese Hakka populations.PLoS One, 18(2), e0281903. doi:10.1371/journal.pone.0281903
[15] Kharchenko J., Roosta T., Chadha A., & Shah C. (2024). How well do LLMs represent values across cultures? Empirical analysis of LLM responses based on Hofstede cultural dimensions[Preprint].arXiv. doi:10.48550/ARXIV.2406.14805
[16] Lawrence H. R., Schneider R. A., Rubin S. B., Matarić M. J., McDuff D. J., & Jones Bell M. (2024). The opportunities and risks of large language models in mental health.JMIR Ment Health, 11, e59479. doi:10.2196/59479
[17] Lin C. E., Tan E., & Wang K. T. (2021). Perfect believer concepts and wellbeing among Buddhists, Christians, and Muslims in China.Ment Health Relig Cult,26(9), 873-889 doi:10.1080/13674676.2023.2282075
[18] Liu Y., He H., Han T., Zhang X., Liu M., Tian J.,.. Ge B. (2025). Understanding LLMs:A comprehensive overview from training to inference.Neurocomputing, 620, 129190. doi:10.1016/j.neucom.2024.129190
[19] Naeem M., Smith T., & Thomas L. (2025). Thematic analysis and artificial intelligence:A step-by-step process for using ChatGPT in thematic analysis.Int J Qual Methods, 24, 16094069251333886. doi:10.1177/16094069251333886
[20] Ornelas T., Araújo A. A., Araújo J., Araújo M., Trinkenreich B., & Kalinowski M. (2025). LLM-assisted thematic analysis:Opportunities, limitations, and recommendations[Preprint].arXiv. doi:10.48550/ARXIV.2511.14528
[21] Ortloff A. M., Fassl M., Ponticello A., Martius F., Mertens A., Krombholz K., & Smith M. (2023). Different researchers, different results? Analyzing the influence of researcher experience and data type during qualitative analysis of an interview and survey study on security advice. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems. New York, NY, USA:ACM. doi:10.1145/3544548.3580766
[22] Parkington K., Teferra B. G., Rouleau-Tang M., Perivolaris A., Rueda A., Dubrowski A.,.. Bhat V. (2025). Human vs. LLM-based thematic analysis for digital mental health research:Proof-of-concept comparative study[Preprint].arXiv. doi:10.48550/ARXIV.2507.08002
[23] Prescott M. R., Yeager S., Ham L., Rivera Saldana C. D., Serrano V., Narez J.,.. Montoya J. (2024). Comparing the efficacy and efficiency of human and generative AI:Qualitative thematic analyses.JMIR AI, 3, e54482. doi:10.2196/54482
[24] Rathje S., Mirea D. M., Sucholutsky I., Marjieh R., Robertson C. E., & Van Bavel, J. J. (2024). GPT is an effective tool for multilingual psychological text analysis.Proc Natl Acad Sci, 121(34), e2308950121. doi:10.1073/pnas.2308950121
[25] Raza M. Z., Xu J., Lim T., Boddy L., Mery C. M., Well A., & Ding Y. (2025). LLM-TA:An LLM-enhanced thematic analysis pipeline for transcripts from parents of children with congenital heart disease[Preprint].arXiv. doi:10.48550/ARXIV.2502.01620
[26] Wang Q., Erqsous M., Barner K. E., & Mauriello M. L. (2025). LATA:A pilot study on LLM-assisted thematic analysis of online social network data generation experiences.Proc ACM Hum-Comput Interact, 9(2), 1-28. doi:10.1145/3711022
[27] Zhang Y., Huang F., Mo L., Liu X., & Zhu T. (2025). Suicidal ideation data augmentation and recognition technology based on large language models.Acta Psychol Sin, 57(6), 987. doi:10.3724/SP.J.1041.2025.0987
[28] Zhou M., Huang D., Ren F., Fan W., Mu W., Li F.,.. Cheung F. M. (2021). Short forms of the Cross-Cultural (Chinese) Personality Assessment Inventory:Reliability, validity, and measurement invariance across gender.Front Psychol, 12, 709032. doi:10.3389/fpsyg.2021.709032
[29] Zhou M., Li F., Mu W., Fan W., Zhang J., & Zhang M. (2023). Round outside and square inside:The latent profile structure and adaptability of Chinese interpersonal relatedness.Acta Psychol Sin, 55(3), 390. doi:10.3724/SP.J.1041.2023.00390

Options

文章导航

模态框（Modal）标题

摘要

本文引用格式

Abstract

参考文献