Evidence Teardown — DSIT Feasibility Study, May 2025
The Government's headline synthesis on smartphones, social media and children's mental health is cited as if it settled the question. This assessment routes one question at every weak point in its evidence base, records a verdict, and attaches the source's own words as proof. No claims, just receipts.
What this is
This does not argue that social media is safe, and it does not relitigate the harms question. It tests one thing: whether the sample underneath the report can bear the weight of the causal claims placed on it. The answer is recorded against each weak point, on the record.
It does not assert that smartphones or social media are harmless to children. It does not draw the threat model or run the science. It does not dispute that real individual harm occurs.
It interrogates the evidence base of the umbrella review, records a verdict at each failure point, and attaches the report's own wording as evidence. Proof, not claims.
The screening
The umbrella review in Appendix 1 is the empirical core of the report. This is its screening funnel. The story is not the volume at the top, it is the quality split inside what survived.
Source: Orben et al. (2025), DSIT Feasibility Study, App. 1, Results: Review Selection, p. 89; PRISMA flow diagram, Fig. 3, p. 90. Primary-study overlap moderate, CCA 0.056, p. 90.
The findings
Each weak point is a question put to the sample. The answer is recorded as a four-state verdict. Open any finding for the evidence and a line ready to post.
For the record · thecontractor.io · ref TEARDOWN-DSIT-2025-01