Psychometric challenges and proposed solutions when scoring facial emotion expression codes

Area of Science:

• Psychology
• Computer Science
• Data Science

Background:

• Automated software is increasingly used for facial emotion expression coding.
• There is a lack of standardized methods for scoring these automated codes.
• Understanding facial emotion expression is crucial in various psychological and communication studies.

Purpose of the Study:

• To discuss facial emotion expression theories and current coding methodologies.
• To identify challenges in scoring software-coded facial emotion expression data.
• To propose and evaluate statistical procedures for scoring and treating this data.

Main Methods:

• Review of emotion expression theories and coding methodologies.
• Analysis of a time series dataset on individual differences in facial emotion expression ability.
• Application and evaluation of four scoring methods and four data treatments for software-coded data.

Main Results:

• Loess smoothing and controlling for baseline expression and plasticity are recommended data treatments.
• The maximum score method is preferred for scoring facial emotion expression ability.
• The study provides insights into analysis decisions for emotion expression research.

Conclusions:

• Standardized scoring and data treatment are essential for reliable facial emotion expression research.
• The proposed methods can inform future analyses of emotion expression data.
• Further research is needed to refine scoring procedures in different experimental contexts.