遇见数据集

Agency (process vs outcome agency) — de-identified raw data archive, 2026-08-06 project freeze

收藏
Zenodo2026-08-07 更新2026-08-20 收录
官方服务:

资源简介:

RESTRICTED ACCESS. De-identified raw data archive for the Agency project ("The Agency Trade-Off: Generative AI Reshapes Two Dimensions of Human Agency in Opposite Directions"), frozen 2026-08-06. 44 files, 49,725 participant-level rows across three studies, eight pilot rounds, and the full MongoDB backend of the caption-contest application. Covers: Study 1, a pre-registered factor analysis of process vs outcome agency on Prolific (AsPredicted #301,596, N = 505 raw / 495 analytic), plus its 2026-02 pilot (N = 399); Study 2, the Alex/library-petition 2x2 vignette (AsPredicted #288,196, N = 3,663); Study 3, the cartoon caption-contest experiment (AsPredicted #288,197, N = 1,533 submitters); and eight earlier Prolific / CloudResearch Connect pilot rounds. Studies 2 and 3 are not reproducible. Both were fielded on the Walton Family Foundation-Gallup Voices of Gen Z Wave 4 nationally representative panel (May 2026). Respondents span ages 12-29, so the sample includes minors, who participated under parental consent plus child assent. No further Gallup wave is planned, so neither collection can be re-run or re-fielded. De-identification. IP addresses, latitude/longitude, Qualtrics recipient name/email fields, Gallup panel household roster keys, state, and population-density measures were removed. All Prolific, CloudResearch Connect and Gallup/Qualtrics participant identifiers were replaced with stable study-local pseudonyms; the re-identification crosswalk is held locally and is not part of this record. Free text was scanned for emails, phone numbers, IP addresses, personal names and pasted platform IDs; the single hit found (a participant who pasted their own Prolific ID into a feedback box) was redacted. The Study 2 file is an Agency-scoped 39-column extract of a 232-column Gallup client delivery; the full Gallup client files are not included in this record. The exact de-identification code is included as deidentify.py. See README.md for the full file manifest, provenance, variable dictionary and exclusion notes.

提供机构:
Zenodo
创建时间:
2026-08-07
二维码
社区交流群
二维码
科研交流群
商业服务