Codebook | AI in Academia - Dataspheres AI

Codebook - AI in Academia A deductive backbone of eight themes, built from the corpus's own structure before coding began. Definitions carry inclusion and ...

Codebook - AI in Academia A deductive backbone of eight themes, built from the corpus's own structure before coding began. Definitions carry inclusion and exclusion criteria so a second coder can apply them independently - that is the test of a codebook. Inductive codes may be added as coding proceeds; log them here with a date. Code Definition Include when Exclude when integrity - Academic Integrity Cheating, plagiarism, detection, honor policy, trust between institution and student. Claims about whether/how AI use is honest, detectable, or permitted. General worry about learning quality (code pedagogy instead). assessment - Assessment Redesign Rethinking assignments, exams, grading, and evaluation because AI exists. Claims that current assessment breaks under AI, or proposals to redesign it. Integrity enforcement of existing assessments (code integrity). pedagogy - Pedagogy & Human Judgment Teaching practice, learning outcomes, the human role in instruction. Claims about how teaching or learning changes, tutor efficacy, student voice. Institutional strategy without classroom implications (code governance). acceleration - Research Acceleration AI speeding discovery, analysis, or scholarly productivity. Claims about faster science, tooling productivity, AlphaFold-class results. Speed claims about grading/admin (code the relevant theme). labor - Labor & Deskilling Work, jobs, deskilling, and whose labor AI displaces or degrades. Claims about teacher/researcher labor, methodology that erases workers. Student skill formation (code pedagogy). equity - Equity & Access Who gets access, algorithmic bias, resource gaps between institutions. Claims about unequal access, bias, or divergent institutional capacity. General policy absence (code governance unless inequity is explicit). governance - Governance & Policy Institutional policy, regulation, national strategy, environmental cost. Claims about policy presence/absence, adoption strategy, oversight. Classroom-level rules (code assessment or integrity). publishing - Publishing & Peer Review Journals, authorship, review integrity, scholarly communication. Claims about AI writing/reviewing papers, publisher policy, review erosion. Student writing (code pedagogy or integrity). Coding rules Code the claim, not the item: one recording usually carries several codable claims. One primary theme per claim; if two genuinely compete, split the claim into two rows. Stance is coded per claim, not per source - a balanced show can make a critical claim. Every row must carry an evidence basis. Nothing enters the matrix from memory. Amendment v1.1 (after intercoder review) Statistics rule: quoted statistics and factual observations are balanced unless the segment itself contains evaluative language - the surrounding argument’s stance belongs to its own segments. Honor-code rule: honor codes, pledges, and enforcement mechanisms are integrity even inside assessment-redesign arguments; assessment requires the segment to address evaluation design itself. Inevitability rule: adoption-inevitability claims are governance when aimed at institutional response, pedagogy only when prescribing classroom practice. Full analysis: Intercoder Reliability .