Stereotype Content

Brief Description Stereotype content in text refers to stereotypical attributes ascribed to social targets, defined here as persons, social groups, or social categories, in textual communication. According to the Stereotype Content Model by Fiske et al. (2002), stereotypes are organized along two dimensions: warmth (i.e., perceived intent, friendliness, and trustworthiness) and competence (i.e., perceived capability and effectiveness). Field of Application/Theoretical Foundation Stereotype content in text is grounded in the Stereotype Content Model (SCM; Fiske et al., 2002), which conceptualizes stereotypes along two fundamental dimensions: warmth and competence. The SCM predicts that different social groups occupy distinct positions in this two-dimensional space: high-warmth / low-competence portrayals (e.g., elderly people, people with disabilities) elicit pity; low-warmth / high-competence portrayals (e.g., wealthy outgroups) elicit envy; low-warmth / low-competence portrayals elicit contempt; and high-warmth / high-competence portrayals elicit admiration. These predicted emotional responses have implications for how media portrayals may shape audience attitudes toward social groups. In text-based communication research, the variable can be used to analyze how social targets are portrayed along these dimensions across contexts including entertainment, health communication, user-generated content, and platform communication. The approach is applicable to any textual corpus in which language about or attributed to social targets can be identified, including news articles, fiction, social media posts, and institutional documents. It is particularly useful for comparative designs examining how warmth and competence portrayals vary across groups, media genres, or national/cultural contexts and across language barriers. Importantly, automated measurement using validated dictionaries (e.g., Nicolas et al., 2021) allows researchers to scale these analyses to large corpora while preserving theoretical grounding in the SCM. References/Combination with other Methods of Data Collection Stereotype content can be assessed through both manual and automated approaches. Manual approaches use qualitative or quantitative content analysis to code warmth- or competence-related evaluations in text (e.g., Kroon et al., 2018). Automated approaches apply computational text analysis to approximate such content through linguistic indicators across larger text corpora, for example through dictionary-based methods (e.g., Nicolas et al., 2021) or computational modeling (e.g., Fraser et al., 2022). The present entry focuses on automated approaches to measuring stereotype content in text. Dictionary-based approaches provide an indirect measure of stereotype content by approximating linguistic markers of warmth and competence rather than capturing stereotypes as relational, context-dependent constructs directly (Nicolas et al., 2021). Although transparent, replicable, and scalable, they rely on word-counting or bag-of-words procedures that may miss sentence-level structure, including negation, irony, sarcasm, or reported speech. This can lead to misclassification in complex textual contexts, especially when dictionary-based methods cannot determine how words relate to each other, who is speaking, or which social target is being evaluated (Nicolas et al., 2021; van Atteveldt & Peng, 2018). More context-sensitive automated computational approaches can therefore complement dictionary-based measures by capturing stereotype content at finer linguistic levels (Fraser et al., 2022).Automated analyses of stereotype content in text should be accompanied by systematic validation procedures. For such validation to be meaningful, researchers need to specify the social target to which warmth- or competence-related evaluations are attributed, define the unit of analysis, and document how context-dependent cases are handled. Context-dependent statements include cases in which the meaning of evaluative language depends on speaker attribution, the social target being described, or whether the statement is ironic, negated, sarcastic, quoted, or ambiguous. Manual coding can provide such a benchmark by checking whether automated outputs assign warmth and competence signals to the intended social target and whether the resulting coding reflects meaningful stereotype content (Nicolas et al., 2021). This is particularly relevant in texts with multiple actors, quoted speech, or ambiguous references, where dictionary-based measures may otherwise misattribute evaluative language to a social target that is not actually being described (Hase, 2021). Thus, systematic validation is necessary to assess whether dictionary-based measures capture meaningful stereotype content rather than isolated or misattributed word use (Nicolas et al., 2021). Example Studies for Automated Content Analysis Table 1 provides examples of automated operationalizations of stereotype content in text. Angermayr and Scherr (2026) examine how mental health disorders are portrayed in Netflix movies and series across 13 production countries. Their analysis is based on full movie and series transcripts from 2001 to 2023 (N = 130; n = 80 movies, n = 50 series). Stereotype content is operationalized at the transcript level through warmth and competence scores. Using LIWC and the Stereotype Content Dictionary by Nicolas et al. (2021), the study aggregates directional dictionary values for each dimension, with higher scores indicating more positive warmth or competence portrayals. Zhu et al. (2026) examine whether warmth portrayals of medical crowdfunding beneficiaries differ across racial groups. Based on medical crowdfunding campaign messages, the study operationalizes the construct at the message level through warmth language. Using the Stereotype Content Dictionary by Nicolas et al. (2021), the authors calculate a warmth index by subtracting low-warmth from high-warmth word frequencies and dividing the result by the total frequency of both word categories. Fraser et al. (2022) examine how social groups are represented in Twitter discourse, using tweets about women and older adults as empirical cases. The study operationalizes stereotype content at the sentence level by mapping text onto a two-dimensional warmth–competence space. The model identifies sentences expressing group perceptions and assigns them warmth and competence values. This approach is validated against manually coded expert annotations and crowd-sourced stereotype data. Table 1: Measurement of “Stereotype Content in Text” using automated content analysis Author(s) Sample Procedure Formal validity check with manual coding as benchmark* Code Angermayr & Scherr (2026) Netflix movie and series transcripts Dictionary approach; machine learning Not reported Stereotype Content Dictionary by Nicolas et al. (2021): https://osf.io/yx45f/  Zhu et al. (2026) Medical crowdfunding campaign messages Dictionary approach Not reported Stereotype Content Dictionary by Nicolas et al. (2021): https://osf.io/yx45f/ Fraser et al. (2022) Twitter data on women and older adults  Computational modeling approach Reported; validated against expert annotation and crowd-sourced stereotype data https://github.com/katiefraser/computational-SCM Note. * This column shows whether the respective study reports a validation of the automated measure against manual coding. References Angermayr, K., & Scherr, S. (2026). Mental health disorders on Netflix: Analyzing stereotypes across 13 countries using the stereotype content model and machine learning. Psychology of Popular Media. Advance online publication. https://doi.org/10.1037/ppm0000660 Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82(6), 878–902. https://psycnet.apa.org/doi/10.1037/0022-3514.82.6.878 Fraser, K. C., Kiritchenko, S., & Nejadgholi, I. (2022). Computational modeling of stereotype content in text. Frontiers in Artificial Intelligence, 5. https://doi.org/10.3389/frai.2022.826207 Hase, V. (2021). Actors (automated content analysis). DOCA - Database of Variables for Content Analysis. https://doi.org/10.34778/1b Kroon, A. C., van Selm, M., ter Hoeven, C. L., & Vliegenthart, R. (2018). Reliable and unproductive? Stereotypes of older employees in corporate and news media. Ageing & Society, 38(1), 166–191. https://doi.org/10.1017/S0144686X16000982 Nicolas, G., Bai, X., & Fiske, S.T. (2021). Comprehensive stereotype content dictionaries using a semi-automated method. European Journal of Social Psychology, 51(1), 178–196. https://doi.org/10.1002/ejsp.2724 Van Atteveldt, W., & Peng, T.-Q. (2018). When communication meets computation: Opportunities, challenges, and pitfalls in computational communication science. Communication Methods and Measures, 12(2–3), 81–92. https://doi.org/10.1080/19312458.2018.1458084 Zhu, X., Kim, Y., & Chen, H.-y. (2026). Differential effects of counter-stereotypical portrayals of warmth across racial groups: Behavioral and experimental evidence. Communication Research. Advance online publication. https://doi.org/10.1177/00936502261430969