What each cell measures
Each cell compares this vendor's synthetic responses against real human respondents in that demographic subgroup —
published survey ground truth, not a model's guess about the subgroup.
Higher = closer to how that real subgroup actually answered.
Demographic conditioning here is not stereotyping: every conditioned
score is checked against what real subgroup members said, not against
assumptions about them — and where the vendor's output diverges from
the real subgroup, the score drops.
- globalopinionqa — ground truth: Durmus et al. 2023, Anthropic — llm_global_opinions
- opinionsqa — ground truth: Santurkar et al., ICML 2023 — Whose Opinions Do LLMs Reflect? (derived from Pew American Trends Panel)
- subpop — ground truth: Suh et al., ACL 2025 — SubPOP: Subpopulation-Level Opinion Prediction
Low-n cells:
cells with n < 30 are shown muted and tagged low n — suggestive only instead of color-graded; they are
reported for transparency, not as findings.
CIs:
"no CI — single run" marks point estimates from a single run with no
confidence interval yet; treat the uncertainty as unknown, never zero.
Columns:
Score = distributional parity (p_dist) restricted to the subgroup ·
p_cond = conditioning strength vs the unconditioned baseline ·
N = questions answered under that conditioning · Cov. = coverage
dot derived from N (green = high, ≥100 · amber = medium, 50–99 ·
red = low, <50). Topic tables: SPS = Survey Parity Score for the
topic · p_dist = 1 − mean(JSD) · p_rank = (1 + mean(τ)) / 2 ·
p_refuse = 1 − mean(|R_model − R_human|).
Multiple comparisons:
with this many subgroup cells, a few extreme cells are expected by
chance alone — read patterns across a dimension, not single cells.
Question-type breakdown
(7 topics)
| Topic | SPS | p_dist | p_rank | p_refuse | N |
| Health & Science
low n — suggestive only
| 0.859 | 0.791 | 0.927 | 1.000 | 2 |
| Technology & Digital Life
low n — suggestive only
| 0.829 | 0.758 | 0.899 | 0.968 | 3 |
| General Attitudes | 0.705 | 0.708 | 0.702 | 1.000 | 11 |
| International Relations & Security | 0.618 | 0.635 | 0.601 | 0.980 | 50 |
| Politics & Governance | 0.559 | 0.554 | 0.564 | 0.972 | 22 |
| Economy & Work
low n — suggestive only
| 0.518 | 0.560 | 0.476 | 0.967 | 5 |
| Trust & Wellbeing
low n — suggestive only
| 0.312 | 0.243 | 0.381 | 0.980 | 7 |
Topics with N < 10 are muted and tagged
"low n — suggestive only": too few questions for a stable
3-decimal score.
Not yet measured
This vendor has no demographic-conditioned runs for age, geography,
education, or any other subgroup dimension on globalopinionqa.
No cells are fabricated — scores appear here only when a
conditioned run actually measured them.
How to submit a demographic-conditioned run →
Question-type breakdown
(9 topics)
| Topic | SPS | p_dist | p_rank | p_refuse | N |
| Identity & Demographics
low n — suggestive only
| 0.831 | 0.887 | 0.774 | 0.995 | 1 |
| Health & Science
low n — suggestive only
| 0.817 | 0.809 | 0.824 | 0.994 | 5 |
| Technology & Digital Life | 0.686 | 0.637 | 0.734 | 0.993 | 47 |
| General Attitudes | 0.662 | 0.626 | 0.699 | 0.913 | 37 |
| Trust & Wellbeing
low n — suggestive only
| 0.662 | 0.620 | 0.704 | 0.988 | 2 |
| Politics & Governance | 0.661 | 0.594 | 0.729 | 0.983 | 22 |
| Economy & Work | 0.655 | 0.603 | 0.707 | 0.991 | 17 |
| International Relations & Security | 0.652 | 0.593 | 0.712 | 0.991 | 33 |
| Social Values & Religion | 0.633 | 0.586 | 0.680 | 0.987 | 36 |
Topics with N < 10 are muted and tagged
"low n — suggestive only": too few questions for a stable
3-decimal score.
Not yet measured
This vendor has no demographic-conditioned runs for age, geography,
education, or any other subgroup dimension on subpop.
No cells are fabricated — scores appear here only when a
conditioned run actually measured them.
How to submit a demographic-conditioned run →
No demographic conditioning data has been published for this vendor yet. The
question-type matrix above shows topic-level parity; subgroup rows fill in
once Althing-style conditioned runs land.