Rating certainty when the target threshold is the null and the point estimate is close to the null
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
When one initially targets the null effect and the point estimate falls close to the null, two challenges exist in rating certainty of evidence. First, when the point estimate is near the null and the data, therefore, suggests little or no effect, rating certainty in a benefit or harm is misleading. Second, since in general the narrower the confidence interval (CI) the more precise the estimate, if the CI is narrow, rating down for imprecision due simply to crossing the null is inappropriate. This paper addresses these issues and provides a solution: to revise the target of certainty rating from a non-zero effect to a little or no effect. This solution requires estimating a range in which the minimal important difference (MID) for benefit and an MID for harm might lie, and thus establishing a range that represents little or no effect. If GRADE (Grading of Recommendations, Assessment, Development, and Evaluations) users are confident that the point estimate represents an effect less than the smallest plausible MID, they will revise their target and rate certainty to a little or no effect. If the entire CI falls within the range of little or no effect, they will not rate down for imprecision. Otherwise (if the CI includes an important effect), they will rate down. Using the solution provided in this paper GRADE users can make an optimal choice of the target of certainty rating.
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.
Direct model labels (unvalidated)
Per-model category and study-design labels from the labeling rounds. They are machine output, unvalidated, and the disagreement between models ships as data. No study design here is MEDLINE-validated yet.
| Model arm | Categories | Study design | Confidence |
|---|---|---|---|
| gemma | Metaresearch Domain: Methods · Genre: Methods About the Canadian research system: no · About a Canadian topic: no | Theoretical or conceptual | high |
| gpt | Metaresearch Domain: Methods · Genre: Methods About the Canadian research system: no · About a Canadian topic: no | Theoretical or conceptual | high |
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.223 | 0.137 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.003 | 0.001 |
| Bibliometrics | 0.000 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.004 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.007 | 0.001 |
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 it