Minimum important difference is minimally important in sample size calculations
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
Performing a sample size calculation for a randomized controlled trial requires specifying an assumed benefit (that is, the mean improvement in outcomes due to the intervention) and a target power. There is a widespread belief that judgments about the minimum important difference should be used when setting the assumed benefit and thus the sample size. This belief is misguided - when the purpose of the trial is to test the null hypothesis of no treatment benefit, the only role that the minimum important difference should be given is in determining whether the sample size should be zero, that is, whether the trial should be conducted at all.The true power of the trial depends on the true benefit, so the calculated sample size will result in a true power close to the target power used in the calculation only if the assumed benefit is close to the true benefit. Hence, the assumed benefit should be set to a value that is considered a realistic estimate of the true benefit. If a trial designed using a realistic value for the assumed benefit is unlikely to demonstrate that a meaningful benefit exists, the trial should not be conducted. Any attempt to reconcile discrepancies between the realistic estimate of benefit and the minimum important difference when setting the assumed benefit merely conflates a valid sample size calculation with one based on faulty inputs and leads to a true power that fails to match the target power.When calculating sample size, trial designers should focus efforts on determining reasonable estimates of the true benefit, not on what magnitude of benefit is judged important.
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.
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 | 0.024 | 0.772 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.006 | 0.001 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.000 |
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