Materials, data, and analysis scripts for "Not your average Clever Hans: Latent representation of psychological constructs in large language models"

Materials, data, and analysis scripts of the article "Not your average Clever Hans: Latent representation of psychological constructs in large language models" (submitted to Psychological Science). The archive contains the 66 psychological constructs and the 8,311 pairs of narratives that six writer LLMs composed for them, with their blind ratings and exclusion records; the 60 published experiments with their rewordings and the agent profiles; the read-out word lists of each construct's representation with the raters' choices; the materials for seven individual characteristics; the results that the GPU steps saved for each of the 16 examined LLMs (identification of the constructs, compressed exports of the internal states, responses in the silicon-sample experiments, effects of inducing and removing representations); the outputs of every laptop analysis; and code.zip, a snapshot of the code repository (https://github.com/schema-brand/latent-psychological-constructs). Unzip the archives into the folder of the code repository to reproduce every number, table, and figure of the article. README.md describes the contents and gives column dictionaries; MANIFEST.sha256 lists the SHA-256 checksum of every file; RAW_STATES_NOT_INCLUDED.md lists what is not included and why.