What is a Good Calibration Question?
Bibliographic record
Abstract
Weighted aggregation of expert judgments based on their performance on calibration questions may improve mathematically aggregated judgments relative to equal weights. However, obtaining validated, relevant calibration questions can be difficult. If so, should analysts settle for equal weights? Or should they use calibration questions that are easier to obtain but less relevant? In this article, we examine what happens to the out-of-sample performance of weighted aggregations of the classical model (CM) compared to equal weighted aggregations when the set of calibration questions includes many so-called "irrelevant" questions, those that might ordinarily be considered to be outside the domain of the questions of interest. We find that performance weighted aggregations outperform equal weights on the combined CM score, but not on statistical accuracy (i.e., calibration). Importantly, there was no appreciable difference in performance when weights were developed on relevant versus irrelevant questions. Experts were unable to adapt their knowledge across vastly different domains, and in-sample validation did not accurately predict out-of-sample performance on irrelevant questions. We suggest that if relevant calibration questions cannot be found, then analysts should use equal weights, and draw on alternative techniques to improve judgments. Our study also indicates limits to the predictive accuracy of performance weighted aggregation, and the degree to which expertise can be adapted across domains. We note limitations in our study and urge further research into the effect of question type on the reliability of performance weighted aggregations.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.102 | 0.390 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.004 | 0.003 |
| Science and technology studies | 0.002 | 0.003 |
| Scholarly communication | 0.007 | 0.012 |
| Open science | 0.002 | 0.004 |
| Research integrity | 0.004 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".