Predicting attitudes toward ambiguity using natural language processing on free descriptions for open-ended question...

Area of Science:

• Psychology
• Natural Language Processing
• Computational Linguistics

Background:

• Individual differences in reactions to ambiguity are conceptualized as ambiguity tolerance.
• Traditional psychological research relies heavily on numerical rating scales for measurement.
• Advancements in natural language processing (NLP) enable new methods for assessing psychological states.

Purpose of the Study:

• To investigate the feasibility of measuring attitudes toward ambiguity using open-ended responses.
• To compare NLP-based analysis of open-ended responses with traditional questionnaire methods.
• To explore a novel, potentially less burdensome methodology for psychological assessment.

Main Methods:

• Collected online responses from 591 participants to three ambiguity-related situations presented in an open-ended format.
• Utilized bidirectional encoder representations from transformers (BERT) for text analysis.
• Correlated NLP-derived scores with scores from a conventional ambiguity tolerance questionnaire.

Main Results:

• A significant moderate positive correlation was found between the NLP-derived scores and the conventional questionnaire scores.
• Demonstrated that attitudes toward ambiguity can be effectively measured through open-ended responses.
• Validated a novel methodology for assessing psychological traits.

Conclusions:

• Open-ended responses analyzed with NLP provide a valid measure of attitudes toward ambiguity.
• This novel methodology can be expanded to other psychological scales.
• The approach has potential applications in educational and clinical settings due to its minimal participant burden.