Homophily in European Research Council Reviews: When Shared Citizenship and Workplace Influence Scores
Reviewers at Europe’s most prestigious research funder, the European Research Council (ERC), tend to give higher scores to applicants who hold their passport and work in their country, according to a groundbreaking study published on 13 August in Nature Human Behaviour. Unlike previous analyses focusing on success rates, this study reveals that the same research proposal receives different evaluations depending on whether the reviewer shares citizenship and country of work with the applicant. This subtle yet statistically significant effect raises important questions about fairness and impartiality in research funding.
Nicola Fuchs-Schündeln and Felix Holub, researchers at the WZB Berlin Social Science Center, analyzed a comprehensive dataset of 37,073 ERC applications submitted between 2019 and 2023. Their findings indicate that when an applicant and a panel reviewer share both citizenship and country of work, the reviewer’s first-stage project score increases by 0.164 points on the ERC’s five-point scale. To put this in perspective, this difference represents about 16 percent of a standard deviation. By contrast, sharing the same gender raised scores by only 0.024 points, a much smaller effect.
The Methodology Behind the Analysis
The ERC review process involves 27 panels, each comprising approximately 15 senior researchers who individually score assigned proposals during the first stage. Each reviewer scores both the project and the principal investigator on a scale from 1 to 5, in half-point increments. After individual scoring, the panel collectively assigns grades A, B, or C, with only those graded A advancing further.
The authors leveraged the fact that each proposal is evaluated by multiple reviewers and each reviewer scores multiple proposals. This enabled them to use a fixed-effect model controlling simultaneously for the proposal’s intrinsic quality (which is unobservable) and the reviewer’s general scoring strictness. As a result, the remaining variation isolates the effect of reviewer-applicant matches on scores, allowing the authors to make a causal claim—a significant methodological advancement over previous descriptive studies.
The Impact of Shared Citizenship and Country of Work
Breaking down the influence of shared attributes, the study found that sharing only citizenship (without the same country of work) increased first-stage project scores by 0.075 points, while sharing only the country of work (without citizenship) increased scores by 0.056 points. When both citizenship and country of work matched, the increase was more substantial at 0.164 points. Scores for the principal investigator followed a similar pattern but were generally smaller in magnitude.
The authors highlight that these single-attribute effects are two to three times larger than the gender effect, and the combined dual match effect is nearly seven times greater than the gender match effect. This suggests that nationality and workplace location are far more influential in scoring than previously recognized demographic factors like gender.
To rule out innocent explanations, the researchers controlled for intellectual proximity by comparing the semantic similarity between proposals and reviewers’ most cited papers, checked for coauthorship links, and accounted for keyword overlaps in research topics. None of these controls diminished the observed effect, suggesting that shared citizenship and country influence scores beyond shared academic interests or knowledge.
Further evidence against a purely informational advantage comes from the distribution of scores: matched reviewers were more likely to assign top scores but not bottom scores, indicating a skew rather than broader variability expected if familiarity led to better-informed, more accurate assessments.
Interestingly, the study found no evidence that reviewers favor applicants from institutions of similar prestige. Classifying the top ten institutions per panel as elite, the data showed no statistically significant bias based on shared institutional status.
Panel Decisions Reflect Weaker Associations
While individual reviewer scores are the primary input for funding decisions, the analysis of panel-level outcomes showed weaker and more correlational effects. Each proposal receives a single panel verdict per stage, making fixed-effect modeling impossible at this level.
Nonetheless, the presence of one additional panel member sharing both citizenship and country of work was associated with a 1.7 percentage point increase in the probability of receiving the top “A” grade at the first stage, against a baseline success rate of 32.6 percent. If the panel chair shared these attributes, the increase was 3.1 percentage points, with a p-value marginally above the conventional threshold (0.055). However, at the second stage, no statistically significant effects emerged for either the panel grade or the final funding decision.
When considering both stages together, the authors estimate that having one extra doubly-matched panel member corresponds to a 1.0 percentage point higher chance of funding, against a baseline funding rate of 14.2 percent.
Focusing on the five countries submitting over half of all proposals—Germany, the UK, France, Italy, and Spain—the study notes applicants from these countries face panels with higher shares of members sharing both citizenship and country of work (4.6% versus 1.5% for other countries). This difference translates to an estimated funding-probability advantage of 1.7 percentage points compared to 0.8 points for applicants from elsewhere, a meaningful gap equating to roughly 6% of the baseline success rate.
Gender Effects Are Statistically Significant but Small
The study also examined gender homophily in scoring. While same-gender matches did yield statistically significant increases in individual scores, these effects were modest—ranging from 2.4 to 4.7 percent of the relevant standard deviations. The authors explicitly caution that such small effects are unlikely to influence final funding decisions.
Moreover, gender effects varied by discipline: they appeared in life and physical sciences but were not significant in social sciences and humanities. Surprisingly, at the panel decision level, the coefficient for the proportion of same-gender panel members was negative (p = 0.081), indicating a marginal tendency in the opposite direction, albeit not strong enough to draw firm conclusions.
Expert Analysis Enabled by Exclusive ERC Data Access
Crucial to this research was the unique access to confidential ERC application data, which only the ERC Executive Agency can provide. Both authors served as contracted experts for the ERC, with full disclosure that their research was a direct outcome of this engagement. Nicola Fuchs-Schündeln, also a professor at Goethe University Frankfurt, is a former ERC grant recipient herself, holding a €1.6 million Consolidator Grant awarded in 2018.
While the authors have shared their code along with pseudo-data to enable workflow replication, the actual application-level data remain confidential and are accessible only through ERC’s discretionary approval. This exclusivity limits external replication, a common challenge in high-stakes, sensitive data research.
The authors acknowledge several limitations: panel-level results are correlational and may be influenced by unobserved factors; the individual-score differences are small on an absolute scale and no single grant’s fate is shown to hinge on these effects; and no correction for multiple hypothesis testing was applied, a caveat disclosed transparently.
Peer review for the article was conducted by experts Matthias Egger and Alexander Petersen, whose reports have been published alongside the article itself. As of 21 August 2026, no formal corrections or comments had been posted.
Recommendations to Mitigate Homophily in Grant Review
The authors propose procedural changes rather than moral judgments. One suggestion is to limit reviewer assignments that pair applicants and reviewers sharing citizenship or country of work—similar to how the World Trade Organization excludes disputants from arbitration panels. Increasing panel diversity could also reduce the frequency of such matches.
Other proposals focus on refining the scoring process: implementing a second assessment for proposals near the funding cut-off, anonymizing proposal materials to blind reviewers to investigator identity, separating evaluation of the project and the investigator, or adopting a partial lottery system for marginally rated proposals, as some have advocated.
Given the ERC’s mission—established in 2007 and managing a €16 billion budget from 2021 to 2027—to fund projects based solely on scientific excellence, the study’s findings highlight a structural bias that may advantage applicants from larger, well-represented research systems while disadvantaging those from smaller countries. The first stage of review is particularly critical, acting as a gatekeeper that determines which proposals proceed to further scrutiny.
For the full study and additional details, see Here.
