Junior scientists whose NIH grant proposals landed just below the funding line later produced more highly cited work than similarly rated applicants who landed just above it. In a 2019 study published in Nature Communications, the authors’ statistical model estimated that a near miss increased the probability of publishing a top-5%-cited paper during the following decade by 6.1% among researchers who stayed active.

The same setback also pushed people out. Near misses were substantially more likely to disappear from the NIH system, meaning the apparent benefit applied only after conditioning on who remained.

This is one study, not settled consensus.

The researchers compared scientists on either side of a payline

Yang Wang, Benjamin F. Jones and Dashun Wang linked NIH grant records with Web of Science publication data. Jones and Dashun Wang were affiliated with Northwestern University’s Kellogg School of Management, and Yang Wang was a Kellogg postdoctoral researcher when the work was described by the school.

The team examined junior principal investigators who applied for R01 grants from 1990 through 2005. An R01 supports a defined independent research project and remains NIH’s most commonly used grant program.

NIH reviewers scored the applications, while paylines determined which proposals were likely to receive money. The researchers narrowed their main sample to applicants within five normalized points of a funding cutoff: 623 “near misses” and 561 “narrow wins.” Before the decision, the groups were statistically indistinguishable across 11 measured characteristics, including career age, previous applications, publication output, prior high-impact papers, team size and institutional reputation.

That threshold made a regression-discontinuity design possible. In plain English, two applicants very close to an externally imposed line should be more comparable than a funded star and a proposal that scored poorly. Crossing the line sharply changed the chance of funding, while the applicants could not precisely manipulate reviewers’ scores.

The 6.1% estimate is not the raw difference between two percentages

The study defined a “hit paper” as one in the top 5% of citations for its field and publication year. Among active scientists, 16.1% of papers from the near-miss group became hits during the first five years, compared with 13.3% from the narrow-win group. That is a raw difference of 2.8 percentage points, or 21% in relative terms.

The number in the headline came from a different calculation. In the regression-discontinuity analysis covering 10 years, the paper reported that a near miss increased the modeled probability of publishing a hit paper by 6.1%, with a p-value of .041. It is the authors’ estimate around the funding cutoff, not a restatement of 16.1 minus 13.3.

Near misses and narrow wins published similar numbers of papers. Yet near-miss papers collected more citations on average in both five-year windows, and the researchers recovered the same broad pattern when they changed the definition of a hit, matched applicants on observed characteristics and used other citation measures.

That is a long-lasting difference in this dataset. It is not evidence that every rejected scientist became more productive.

The other number is 12.6%

The analysis first focused on “active” principal investigators, defined as people who later applied for or received NIH grants. The near-miss group had 11.2% fewer active investigators in the year after the decision. The regression-discontinuity model estimated that an early near miss produced a 12.6% chance of disappearing permanently from the NIH system over the following decade.

That condition matters.

Leaving the NIH system is not identical to leaving science. The records could not capture every move into industry, overseas institutions, other funding streams or research roles without another NIH application. But attrition still separates the people whose later papers were counted as evidence of improved performance from those no longer visible in the system.

A quick reading might therefore celebrate the survivors and erase the cost paid by everyone who did not survive within this career track.

The authors tested survivor selection, but could not identify a mechanism

Perhaps rejection simply filtered out some near-miss scientists, leaving behind an unusually capable or determined subset. The authors took that possibility seriously. In one deliberately conservative test, they removed the lowest-performing narrow winners until the two groups had equal attrition. The surviving near misses still had an advantage in hit-paper probability and average citations.

They also matched applicants on prior performance and demographic characteristics, restricted the analysis to lead-author papers, controlled for collaborator status, and tested whether near misses changed institutions, shifted research direction or moved toward hotter topics. No measured explanation accounted for the full performance gap.

The design supports a local causal interpretation if its assumptions hold, but it cannot tell us what changed inside a scientist after rejection. Greater effort, a revised strategy and stronger resolve are plausible explanations raised by the authors. They were not directly observed. In my reading, “the setback taught them” remains an interpretation rather than a measured psychological result.

Early success still creates powerful advantages

This result sits beside a large literature on cumulative advantage in science, often called the Matthew effect: early recognition and resources make later success easier. A 2018 Proceedings of the National Academy of Sciences study used another grant threshold and found that early winners accumulated more than twice as much research funding over the next eight years as nonwinners just below the line.

Funding pays for people, equipment and time. The near-miss study did not find that those resources were useless. Narrow winners began with a large financial advantage, and the near-miss group received less NIH and National Science Foundation funding during the decade. What stood out was that the active near misses produced roughly as many papers and more highly cited ones despite that disadvantage.

The finding should not be converted into a policy of creating artificial hardship. The paper itself says its results do not imply that institutions should place roadblocks before junior scientists. Too many people were lost from the system for that reading to be defensible.

A close rejection is a poor verdict on future value

The sample was narrow: already accomplished junior biomedical researchers, close to one NIH R01 payline, whose applications were filed between 1990 and 2005. The result cannot automatically be generalized to students, other professions, repeated failures, applicants far below a cutoff or setbacks involving health and financial security.

Citations are also an imperfect proxy for quality. The authors found related advantages on measures tied to clinical translation, but no bibliometric measure can tell us whether every highly cited paper was more rigorous or useful.

The clearest implication belongs to funders. Scientists separated by a few review points had comparable records before the decision, yet some near misses later produced unusually influential work. A narrow rejection was not evidence of low potential. It was a consequential funding decision made at the edge of an uncertain ranking system, one that some researchers overcame and others did not.