Re: “Invited Commentary: How Big Is That Interaction (In My Community)—And in Which Direction?”
Bibliographic record
Abstract
In their invited commentary on an article by Turner et al. (1), Panagiotou and Wacholder (2) discuss the estimation of population attributable risks (PARs), or population attributable fractions as they call them, when multiple risk factors are involved. In particular, they consider the difference between the PAR associated with the combined effects of 2 factors and the sum of the effects for each factor separately. In referring to much earlier work of mine on this topic (3), they state that I showed that “for a rare outcome” (2, p. 1154) this difference would be zero under an additive model of disease risks. In fact, in my earlier paper (3), I established quite general conditions under which these 2 methods of calculating PARs would be equivalent, and, contrary to Panagiotou and Wacholder's claim, no rare outcome assumption is required. In particular, my equation results show that the combined versus summative formulations of 2 PARs will be equal if the disease risks are additive or if no one is exposed to both factors (see equation 9 in my paper (3)). The only sense in which rarity of outcomes might be invoked in practice is if the additive model becomes empirically untenable—for instance, if it were to predict risks greater than 100% for doubly exposed persons, based on the risks associated with single exposures. Note that my paper also considered an important related issue—specifically, identification of the conditions under which a marginal estimate of PAR for a risk factor of interest would be valid in the presence of other risk factors. (If the conditions are satisfied, then a marginal approach to PAR estimation might be adopted, with only more limited data then being required.) These conditions were that either the exposures are independently distributed in the population or the other factors do not increase risk for persons unexposed to the factor of interest. When there are only 2 risk factors, one of these conditions must necessarily be true; however, in the situation of more than 2 risk factors, analogous conditions yield validity of a marginal PAR—but then the conditions need not necessarily be true. Again, no rare outcome assumption is required for any of these results. Conflict of interest: none declared.
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
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 | no category Domain: not available · Genre: Commentary About the Canadian research system: no · About a Canadian topic: no | Not applicable | low |
| gpt | no category Domain: not available · Genre: Commentary About the Canadian research system: no · About a Canadian topic: no | Not applicable | high |
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 | 0.013 | 0.098 |
| Meta-epidemiology (narrow) | 0.002 | 0.003 |
| Meta-epidemiology (broad) | 0.004 | 0.004 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.013 | 0.007 |
| Scholarly communication | 0.012 | 0.010 |
| Open science | 0.006 | 0.006 |
| Research integrity | 0.157 | 0.148 |
| Insufficient payload (model declined to judge) | 0.015 | 0.016 |
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, unvalidatedLabeled directly by 2 models reading the full record.
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".