Constructs set? Find and adapt validated scales
Settled constructs do not make a survey ready. Yila checks each construct’s definition and boundaries, retrieves the original scale literature — Davis, Jian, Bhattacherjee and more — organizes an evidence matrix separating classic sources from generative-AI studies, then adapts items into your questionnaire with source, rewrite, and rationale logged. Anything not verified against the original text is marked pending — never invented.
Boundary check complete: PU, PEOU, trust, and continuance intention are clearly defined with no overlap. One caution flagged: the object of trust should be scoped to students who have actually used generative AI.
Search complete: 18 sources, all 18 DOIs verified — classic originals include Davis (1989) for PU/PEOU, Jian et al. (2000) for trust, and Bhattacherjee (2001) for continuance intention, plus recent generative-AI studies. Nothing invented.

The evidence matrix is ready: 21 sources classified as classic originals, follow-up reviews, or generative-AI context studies, with the priority original paper and reason for each construct.

The concept map is done: the three paths are labeled with their TAM/ECM, TAM-extension, and trust–adoption theoretical links, all marked “to be tested”, with a note that arrows are theoretical paths, not proven causation.

The adaptation worksheet is ready: every item lists its source, construct meaning, contextual rewrite, and rationale, and anything not verified against the original is marked “pending verification” — never presented as a validated scale.
The checklist is ready: back-translation secures semantic equivalence, expert review covers content validity, cognitive interviews catch misreadings, the pilot plus item analysis filter weak items, and reliability and validity testing close the loop — each step maps to a concrete risk.
Yila task replay complete.
Start with this method