Sentiment analysis and PANAS: Differential sensitivity to mood effects and personality traits

ersonality traits are known to relate to affective states, which are typically assessed using self-report questionnaires such as the Positive and Negative Affect Schedule (PANAS). The present study examined whether affective responses to mood induction are captured differently by PANAS and by sentiment analysis of brief free-text responses, and whether these measures show similar associations with personality traits. In an online randomized experiment, 95 participants were assigned to either a happy or sad mood induction procedure. Positive affect, negative affect, and linguistic valence were assessed before and after mood induction, and personality traits were measured using the NovoPsych Five Factor Personality Scale (NFFPS-30). Mood induction effects were detected by both PANAS and sentiment valence. Comparative analyses of standardized change scores indicated that sentiment valence showed greater sensitivity than PANAS Positive Affect, but not PANAS Negative Affect. By contrast, personality traits were more clearly associated with baseline self-reported affect than with linguistic valence, and only limited evidence was found for trait-based moderation of emotional reactivity. Overall, the findings suggest that free-text sentiment analysis may complement structured self-report scales by capturing mood-related change in partly distinct ways, whereas structured self-report measures may be more informative for personality–affect associations.