The study section convenes on a Tuesday morning. Your proposal is on page 47 of the reviewer’s stack. Study Section Member B has read your Specific Aims page twice. She has questions.
Not about the science. About the design.
The science is sound — you know that. You have been working on this problem for four years. The preliminary data are compelling. The team is right. The lab is right. The question is right.
But Member B is looking at Aim 2, and she is not reading the study you want to run. She is reading the study the proposal describes. And those are not the same study.
“The PI reads the proposal as the study they want to run. The reviewer reads the study the proposal actually describes.”
That gap — between intention and demonstration, between what the PI knows and what the proposal shows — is structural. It does not disappear with better writing. It does not close with a longer methodology section. It is invisible to everyone who has been close to the work for long enough to stop seeing it clearly.
Why Proximity Creates Blind Spots
The PI’s deepest familiarity with the research is the same thing that makes structural gaps impossible to see from the inside.
The lab has been working on this project for two years before the proposal was written. The co-investigators know the assays. The statistician knows the power calculation ran at 85% — they were in the room when it happened. The postdoc who ran the preliminary study knows every limitation of that n=12 dataset and why the effect sizes are still directionally reliable.
None of that is in the proposal. Or rather: it is in the proposal the way all assumed knowledge is in writing — incompletely, through implication, readable only to someone who already knows it.
The reviewer does not know it. The reviewer has a 2-day window to read 14 proposals and generate scored critiques for each. She is reading the proposal as a document that must demonstrate, not imply. She cannot score what the proposal intends. She scores what it shows.
The structural weakness the study section sees is almost always something the PI and team have been reading past for months — not because they are careless, but because familiarity has made the gap invisible.
This is not a writing problem. It is a distance problem. The PI cannot generate the distance required to read their own proposal as a skeptical reviewer who has never met them and has no investment in the study’s success.
The Four Most Common Structural Weaknesses
These four appear in the majority of proposals that fail to clear the fundable range on first or second submission.
1. Power Analysis Gaps
The power analysis is present. It appears in the Statistics section of the Approach. The problem is not absence — it is foundation. A power calculation based on effect sizes from a small preliminary study (n=8 to n=15) will almost always produce inflated effect size estimates due to small-sample bias. Reviewers in quantitative fields know this.
The proposal describes a power analysis. The reviewer scores whether the power analysis will actually detect the true effect at the proposed sample size. When the preliminary data are small-n, those are different questions.
“The applicants cite 85% power based on effect sizes from their preliminary data. The preliminary sample (n=12) is underpowered to produce stable effect size estimates. The proposed sample of 40 participants would have substantially less than 80% power to detect an effect of the magnitude observed in the published literature. This concern is not addressed.”
The fix is not explaining the power calculation in more detail. The fix is using conservative effect size estimates from published literature, not pilot data, and either increasing sample size or narrowing the aim to match the detectable effect.
2. Methodology-Design Mismatch
The study aims to establish causal evidence. The proposed design is observational. Or: the study aims to generalize across a clinical population, but the proposed recruitment excludes the majority of that population. Or: the study proposes to detect within-subject change, but the measurement instrument has a test-retest reliability coefficient that makes within-subject change detection unreliable at the proposed interval.
The researcher knows why the design is still adequate. The proposal does not demonstrate it. The reviewer reads the mismatch between the stated aim and the proposed method and scores accordingly.
A methodology-design mismatch is the most common single reason a proposal with excellent preliminary data fails to reach the fundable range — because the design cannot answer the question the Specific Aims page poses.
3. Overclaiming Innovation
The Specific Aims page describes the work as a “novel paradigm,” a “first-of-its-kind approach,” or a “transformative advance.” The Approach section describes an incremental application of established methods to a new population or context.
Reviewers score Innovation on whether the proposed methods would actually constitute a conceptual or methodological advance — not on the language used to describe them. The mismatch between claim and method produces a low Innovation score regardless of how strong the science is.
The fix is not moderating the language of the innovation claim. It is either ensuring the methods actually constitute the claimed advance, or accurately scoping the innovation claim to match what the methods deliver.
4. Missing Preliminary Data
Aim 3 depends on a key assay that has not been validated in the proposed model system. The proposal cites published literature showing the assay works in a related context. The reviewer wants to see preliminary data showing it works in this context.
This is not about volume of preliminary data. It is about the specific gap: does the PI have evidence that the core technical approach is feasible for this study? A reviewer who does not see that evidence will score Approach accordingly, regardless of how strong the preliminary data are for Aims 1 and 2.
The Self-Assessment Framework
The proposal must answer these questions without the PI in the room to explain.
Before the proposal reaches study section review, the PI should run this test: read each aim as if you have never seen the data behind it. Ask only what the proposal demonstrates — not what the PI knows.
For each primary aim:
- Does the design actually answer the stated question, or does it answer a related but different question?
- Is the power calculation based on effect sizes from the published literature, or from the PI’s own preliminary data? If the latter, what is the sample size of that preliminary study?
- Does the proposed innovation claim match what the proposed methods would actually produce?
- Is there preliminary data for every key technical step, including the ones that feel obvious to the lab?
For the proposal overall:
- If a competent skeptic read only the Approach section and had no other information, would they believe the study is feasible and appropriately powered?
- Does the proposal contain any sentence that requires the reader to already believe the study will work in order to accept the claim?
That last question is the most important one. Any sentence that requires prior belief in the study’s viability is a structural gap — because the reviewer does not have that prior belief, and the score reflects only what the proposal demonstrates on the page.
What Adversarial Pre-Review Surfaces
The PI needs someone who reads the proposal as a reviewer, not as a collaborator who wants it to succeed.
The PI’s own team cannot generate this reading. They have been in the lab. They know the assays work. They know the power calculation is conservative for reasons that are obvious to anyone who has read the same literature they have spent two years reading. The blind spot is shared — which is why the team cannot catch it.
Adversarial pre-review applies The Gate: a structured scoring process that evaluates the proposal across all five NIH review criteria — Significance, Investigator(s), Innovation, Approach, Environment — and returns a scored assessment that mirrors the study section’s output.
The Gate does not produce general feedback. It produces a scored review:
OVERALL IMPACT SCORE: 6
APPROACH: 7
CRITICAL CONCERN: The power analysis for Aim 2 uses effect sizes
from the PI's preliminary data (n=12). Small-sample bias routinely
inflates effect size estimates by 30-50% at this sample size. The
proposed n=45 would have approximately 55% power to detect the true
effect. This gap was not flagged in the original submission.
Proposal requires either expanded recruitment or a narrowed Aim 2
hypothesis with a corresponding sample size revision.
INNOVATION: 5
The methods represent an incremental application of an established
approach. The Specific Aims page describes a "novel paradigm" that
the proposed experimental design does not deliver. The innovation
claim should be scoped to what the methods actually produce.This is what the PI cannot see. The Approach score of 7, driven by the power analysis gap, is pulling the overall impact score up. The Innovation score of 5, driven by the mismatch between claim and method, adds a second drag. Two structural weaknesses — both invisible to the team — are enough to keep the proposal unfunded.
Adversarial pre-review surfaces the structural weaknesses the study section will score, before the study section scores them.
The difference between a proposal that clears The Gate and one that does not is not writing quality. It is whether the proposal demonstrates what it claims, independently of what the PI knows the study will show.
The Case Study: From 6 to 4
A PI in clinical outcomes research submitted an R01 that scored 6. The reviewer objection: the primary outcome relied on a single self-report measure with no behavioral or objective corroboration. The PI resubmitted, adding a paragraph expanding the psychometric validation of the self-report measure. The score remained 6. The objection returned in the same form.
Adversarial pre-review on the resubmission identified the structural gap: the single-measure design carried well-documented validity concerns for this clinical domain, and the preliminary data used effect sizes from that same measure alone. The reviewer was not objecting to the measure’s psychometrics — they were objecting to the design’s reliance on a single measure type in a domain where multi-measure convergence is expected.
The structural fix was adding a secondary behavioral outcome measure that provided convergent validity and revising the power analysis to account for multi-measure correction. The PI ran the revised proposal through The Gate. The Approach score moved from 7 to 4. The overall impact score moved from 6 to 4 — from the 50th percentile to the 20th, from unfunded to fundable.
The PI had addressed the reviewer’s language twice. The structural weakness remained both times. Adversarial pre-review found it in one pass.
Simulate the Study Section Before the Study Section Simulates You
The study section will convene. Member B will read Aim 2. She will score what the proposal demonstrates.
The PI cannot be in the room to explain that the power calculation is actually conservative, that the assay has been validated informally in the model system, that the team knows the design is adequate. None of that counts. The score reflects the proposal.
The question is whether to find out what the score will be before the proposal reaches the study section — or after.
The Gate applies The Gate’s logic now, to a draft, when structural corrections still change the outcome. The study section applies it later, when the only option is a resubmission cycle that starts at month seven.
A proposal that does not clear The Gate has the same profile as a proposal that will not clear the study section: competent science, one or two structural design gaps, a score that reflects what the proposal shows rather than what the study will do.
The PI knows the study will work. The proposal has to demonstrate it to someone who doesn’t.
If your proposal is going into study section review with structural weaknesses your team cannot see, the score will reflect that. Request a grants intake at axion.activewizards.com/grants-pilot or reach us at [email protected].