How individuals with high or low power think, feel, and act is a core focus in organizational behavior and social psychology research. Many studies use power recall manipulations, asking participants to recall and write about a time they possessed or lacked power. While previous work has validated this method through manipulation checks and successful predictions, we argue that this is insufficient. We systematically assess both the recall experience and the produced text using memory characteristics scales and automated text analysis (compliance, sentiment, dictionary analyses, and topic modeling). Across multiple samples, we find: (1) 12.71% of participants write text that large language models flag as inconsistent with instructions, varying by platform and condition; (2) sentiment analysis reveals expected differences but with substantial variation and non-negative values for low-power conditions; (3) topic modeling shows both high- and low-power participants focus on professional topics, but low-power participants more frequently discuss personal relationships, health crises, and financial issues. These findings highlight distinct differences in recall experiences and event types between high- and low-power participants.
