Cross-Cultural Analysis of Dream Reports and Mental Health

Abstract

In this project, we will analyze dream reports and mental health measures from a cross-cultural survey conducted in the United States and Argentina. Participants provided free-text dream reports, responses to nightmare and dream-related items (e.g., nightmare frequency, dream intensity, emotional tone), and completed standardized mental health questionnaires: the Patient Health Questionnaire (PHQ-9) and the Generalized Anxiety Disorder scale (GAD-7). We will apply natural language processing to the dream reports to compute structural features, including type–token ratio (lexical diversity), parse tree depth (syntactic complexity), graph connectedness metrics (edges, largest connected component, strongly connected component, diameter, average shortest path, recurrence), and coherence scores (sentence embedding similarity). Dream, badream, and nightmare frequency and distress will be derived from self-report items. Our analyses will combine traditional null-hypothesis significance testing (e.g., regression, linear mixed models with random intercepts for participants) and supervised machine learning approaches (e.g., logistic regression for clinical cutoffs) to examine how dream structure, nightmare frequency, and distress predict PHQ-9 and GAD-7 scores. Cross-cultural differences will be tested by conducting a holistic process above, including country as a fixed effect and as an interaction with dream features.