When
Where
Announcing:
August H. Nilsson
Oslo Metropolitan University, Norway
Language-Based Assessments in Psychology
Language-based AI (e.g., ChatGPT) is taking the normal and scientific world by storm. In psychology, language-based assessments, that is assessments that use modern AI techniques for measuring constructs such as well-being and personality, are beginning to revolutionize the field. In this talk, I will present research illustrating two of the most vital advantages of language-based assessments: accuracyand insights. Regarding accuracy, when predicting the corresponding rating scale or other ground truth, language assessments have yielded correlations approaching (and sometimes exceeding) typical scale reliabilities (r = .70+), for outcomes including well-being, depression, and implicit motives, and have excelled in predicting external criteria. However, accuracy is only half the story. By understanding the language features (e.g., specific words, groups of words or predicted valence of the language) that drive a relationship or differentiate between groups, we get deeper insights into why a psychological association or phenomenon exists. For example, language features of resilience, sociality and implicit need for affiliation in Reddit posts predict well-being. The talk will close with a discussion on how we are moving into an era of optimizing language-based assessments to help approaching true states, moving beyond simple convergence with rating scales.
Please Note:
This colloquium will be offered in person.
You can also join via Zoom.
Accessibility & More:
With questions about access or to request any disability-related accommodations, please contact Vanessa Noonkester at:
Please join the Psychology Colloquium Listserv to receive news and updates.
To join, please follow these steps:
- Send a blank message from your email account to LIST@LIST.ARIZONA.EDU
- In the Subject line, please enter: "Subscribe PSYCOLLOQ YourFirstName YourLastName" (Example: Subscribe PSYCOLLOQ Wilma Wildcat)
Updated: 11/13/25