Voice not met; carry-forward met2,872 people, 2024 held out
The test at a glance
- What it tests
- Accuracy against reality: welfare-spending answers by party and question wording.
- Data
- General Social Survey 1972-2024 (Release 3a): 2,872 respondents in 2024; methods fitted on 2016, 2018, 2021 and 2022.
- Held out
- 2024 answers by party and wording. The overall 2024 shares had been seen and are reported as seen.
- Compared against
- The 2022 shares carried forward; a regression on the same facts; a small language model answering as each person; the anchored voice.
- Bar
- Six-cell error at most 3.0 points AND each party's wording gap inside its 95% interval.
- Result
- Carry-forward 2.46: met. Regression 2.84, but one wording gap outside its interval: not met. Voice 58.9 off.
On welfare spending, across 2,872 people in the 2024 General Social Survey, the last measured shares carried forward came within 2.5 points across six party-and-wording groups and met the registered rule. A regression on the same facts came within 2.8, but one party's wording gap fell outside its interval, so it did not. A small language model answering as each person missed by 58.9 points, saying nearly everyone thinks too much is spent. Registered before any 2024 answer was read by group. This is why Living Worlds fits lives to records first. The 2024 overall shares were seen before registration and are reported as seen.
Protocol, result and scorer