cvprofiles · construct-validity profile
Admissible measurement set under a stated nomological network, and the image of β on that set. Empty sets and wide ranges are scientific features.
[L, U] · method = min/max B*
[-0.317486, -0.317486]
Width = 0.000000 · point-ID · bootstrap percentile band = [-0.394007, -0.199811]
Target β = corr_y on outcome gps_trust_cell. Range uses only measures in M*; rejected β values are shown below but never enter [L,U].
| run_id | 9461b368443ff5ad950c9b466d2ccc178b14655997f1408e2b0f8f88f59a9a77 |
|---|---|
| package | 3.0.2 |
| scores_hash | ff11b172db7ec3ea13d700b6abad6724c7ad68ecb27585a1caf542397f92499b |
| network_hash | d0afa9002b03c34746a448be1fbb9affc2c0ca99f9b3f2dd5956160912e218d0 |
| beta_hash | 19fe88c41c4644897cee8c770b625bd57adbae206e1efc2774599c116eacd51e |
| δ | 0.0 |
| seed | 20260814 |
| network | trust_cells_llm_extension |
Units-only Efron resampling; percentile band over non-empty replicates. The headline [L, U] above remains min/max β(M*) on the full sample.
| replicates | 1000 |
|---|---|
| seed | 20260814 |
| quantiles | [0.025, 0.975] |
| non-empty | 1000 |
| empty | 0 (0.000) |
| degenerate | 0 (0.000) |
| band | [-0.394007, -0.199811] |
Per-side α/2 quantiles over non-empty bootstrap replicates (α = 0.1). Heuristic band — not a confidence interval, not a holdout-robustness band. The headline [L, U] above remains min/max β(M*) on the full sample.
| α | 0.1 |
|---|---|
| κ | 2.0 |
| quantiles | [0.05, 0.95] |
| band | [-0.382417, -0.239745] |
| empty rate | 0.000 |
| degenerate rate | 0.000 |
| measure | margin | SE | boundary |
|---|---|---|---|
| m_trust_general | -0.176854 | 0.049025 | no |
| m_trust_in_group | -0.268407 | 0.043041 | no |
| m_trust_out_group | -0.274989 | 0.051229 | no |
| m_trust_institution | -0.627369 | 0.039714 | no |
| m_trust_composite | -0.473703 | 0.045587 | no |
| m_llm_trust_llama | -0.077267 | 0.040003 | yes |
| m_llm_trust_phi | 0.132638 | 0.043311 | no |
| m_noise | -0.171912 | 0.048268 | no |
| measure | p̂m |
|---|---|
| m_trust_general | 0.000 |
| m_trust_in_group | 0.000 |
| m_trust_out_group | 0.000 |
| m_trust_institution | 0.000 |
| m_trust_composite | 0.000 |
| m_llm_trust_llama | 0.022 |
| m_llm_trust_phi | 1.000 |
| m_noise | 0.000 |
Heuristic uncertainty band over non-empty bootstrap replicates (selection uncertainty on the pooled sample; holdout verdict is a full-sample point finding outside the band). Not a confidence interval; not a holdout-robustness band. Boundary attribution: |margin_m| <= kappa * SE_m.
λ scales threshold magnitudes only. λ is never auto-selected; the headline is always λ=1.0. This grid is not part of the freeze preimage.
| λ | |M*| | empty | L | U |
|---|---|---|---|---|
| 0.5 | 2 | no | -0.317486 | 0.179299 |
| 1.0 | 1 | no | -0.317486 | -0.317486 |
| 1.5 | 1 | no | -0.317486 | -0.317486 |
| 2.0 | 0 | yes | — | — |
One pre-data anchor per restriction (hash 07533ac8ce87…). The engine enforces completeness + hash; "pre-data" is a process commitment, not a temporal claim.
| restriction | kind | pre-data | anchor |
|---|---|---|---|
| conv_edu_cell | literature | yes | falk2018_table_v — Same education implication applied within country (cell means). θ=0.20. |
| mono_edu_cell | literature | yes | falk2018_table_v — Leftover within-country rank-monotone education restriction. θ=0.15. |
M* size = 1 / 8. Slack rule: admit if sr(m) ≥ −δ for all r.
| measure | status | β(m) | failed restrictions |
|---|---|---|---|
| m_trust_general | rejected | 0.052428 | conv_edu_cell |
| m_trust_in_group | rejected | 0.223931 | conv_edu_cell |
| m_trust_out_group | rejected | 0.211423 | conv_edu_cell |
| m_trust_institution | rejected | 0.284723 | conv_edu_cell |
| m_trust_composite | rejected | 0.317611 | conv_edu_cell |
| m_llm_trust_llama | rejected | 0.179299 | conv_edu_cell |
| m_llm_trust_phi | admissible | -0.317486 | — |
| m_noise | rejected | 0.014662 | conv_edu_cell |
Positive / zero ≈ satisfied at δ=0.0. Negative = violation magnitude.
| measure | conv_edu_cell | mono_edu_cell |
|---|---|---|
| m_trust_general | -0.176854 | -0.131341 |
| m_trust_in_group | -0.268407 | -0.233353 |
| m_trust_out_group | -0.274989 | -0.224794 |
| m_trust_institution | -0.627369 | -0.552204 |
| m_trust_composite | -0.473703 | -0.414333 |
| m_llm_trust_llama | -0.077267 | -0.065537 |
| m_llm_trust_phi | 0.148808 | 0.132638 |
| m_noise | -0.171912 | -0.119007 |
Engine does not invent restrictions. Agent may author oracle R for synthetic DGPs only.