Procedures

The procedure actually followed, in sequence.

1. Research setup

Research problem, PICO specification, and research questions defined. Target population fixed as manufacturing SMEs enrolled in the K-Hitech DT training programme.

3. Synthesis

Retrieved literature processed through ScholarRAG, a staged review-automation pipeline, in systematic-review mode. Output: the integrated conceptual framework v1.0.

4. Data preparation

Raw responses screened for missingness; the missingness mechanism tested before deletion. Constructs assembled and reliability assessed. See Data.

5. Sequential model estimation

Three nested specifications estimated in order — linear main effects, first-order interactions, then second-order curvature — so that each addition can be tested incrementally rather than asserted. See Method.

6. Response surface analysis

Surface statistics computed under the Edwards (2002) protocol; the stationary point located and bootstrapped for confidence intervals.

7. Configurational analysis

Fuzzy-set QCA run on the upper quartile of the outcome, with theoretically anchored calibration and a sensitivity re-run at a stricter consistency cutoff.

8. Triangulation

Five methods applied to the same substantive question so that agreement or disagreement between them is informative rather than assumed.

9. Drafting

Five-chapter manuscript assembled, with cover letters to the two highest-ranked target journals.

10. Internal peer review

Structured review conducted against the draft, producing a five-document package and a Major Revision recommendation.

11. Revision — pending

Not started. The revision sequence is fixed by dependency: the dependent-variable decision comes first because it determines every downstream estimate; the panel restriction and the common-method diagnostic follow; multiple-comparison correction is last because it is re-tabulation rather than re-analysis.