1. Measuring Stereotypes with Computational Methods
This paper suggests an efficient procedure to feasibly extend the specification of stereotype content of millions of text strings without using the traditional costly approach of questionnaire surveys. Using Google News Word2Vec embeddings, this paper develops a new semantic differential model of social perception to construct a Stereotype Content Dictionary for 3 million English words and phrases by reducing the semantic space of 300 dimensions to the two dimensions of the influential Stereotype Content Model (SCM). The new procedure is based on the associations among 100 antonymous pairs we developed based on selected seed words from the list of words to illustrate the two theoretical dimensions of stereotype content. This study utilizes Nicholas et al.’s gold-standard classification of Rosenberg et al.’s 64 personality traits to assess and compare the validity and performance of our model and Fraser et al.’s word embedding model. The results reveal that the trait classification by the Stereotype Content Dictionary correctly predicts 75% of the gold standard and significantly improves accuracy by over 40% compared with the random chance. Our model achieves 1.6 times higher in predicated performance than Fraser et at.’s model. Our model generates not only the first-ever Stereotype Content Dictionary for public use but also an efficient and feasible tool for stereotype research based on big data across societies and contexts.
- Qin, X., & Tam, T. (2023, September). Stereotype Content Dictionary: A Semantic Space of 3 Million Words and Phrases Using Google News Word2Vec Embeddings. In International Conference on Social Computing, Behavioral-Cultural Modeling and Prediction and Behavior Representation in Modeling and Simulation (pp. 12-22). Cham: Springer Nature Switzerland.
References
Code:
Data:
2. Integrate Social Perception Models
To be continued..