Virtual experiments in megastudies: A case study of language and emotion

Area of Science:

• Cognitive Psychology
• Psycholinguistics
• Computational Linguistics

Background:

• Lexical megastudies provide large datasets for virtual experiments.
• Conflicting theories exist regarding word valence and recognition speed.

Purpose of the Study:

• Introduce a bootstrapping approach for hypothesis testing in virtual experiments.
• Evaluate the relationship between word valence and recognition speed.
• Assess the impact of methodological choices on experimental outcomes.

Main Methods:

• Utilized a bootstrapping approach on data from two lexical decision megastudies.
• Conducted multiple virtual experiments with factorial and regression designs.
• Analyzed item selection criteria, statistical power, collinearity, and dataset choice.

Main Results:

• The bootstrapping approach allows principled sampling and hypothesis testing.
• Demonstrated the influence of stimuli selection criteria on the valence-recognition speed relationship.
• Highlighted the impact of statistical power and dataset choice on outcomes.

Conclusions:

• The proposed bootstrapping method offers a robust framework for virtual psycholinguistic research.
• Methodological decisions significantly affect the interpretation of word recognition data.
• Further research should consider these factors for more reliable findings.