The effects of war on Ukrainian research

Russia’s full-scale invasion of Ukraine on 24 February 2022 caused large-scale civilian casualties, displacement, and infrastructure destruction. The war has disrupted Ukraine’s science sector via university closures or destruction, forced migration, and career changes among scientists. The paper’s purpose is to quantify how the war affected the Ukrainian scientific community, focusing on emigration (brain drain), time devoted to research, occupational mobility, and access to research infrastructure among those who stayed and those who emigrated. The study reports results from an online survey of Ukrainian scientists employed in higher education institutions (HEIs) or public research organizations (PROs) at the onset of the war, conducted between 21 September and 8 December 2022.

The literature on scientific human capital emphasizes mobility’s role in shaping national competitiveness and growth, with wars and shocks accelerating high-skill emigration (“brain drain”). Prior work shows emigrant scientists often gain productivity through exposure to new knowledge, networks, and resources, though outcomes depend on host environments and integration. Emigration can harm home-country productivity via loss of top scientists, weakened local collaborations, and reduced mentoring capacity, but can also create benefits through international knowledge bridges, collaboration, and potential returns that diffuse practices and know-how. While emigration has been studied extensively, the experiences of scientists who remain in war zones are comparatively underdocumented. War can impede research through psychological trauma, physical danger, destruction of infrastructure, power outages, and displacement, as well as shifts of effort toward caregiving, internal migration, income supplementation, or service in military/civil defense. The Ukrainian context involves historically separated research and higher education systems, evolving incentives for university research, limited funding, and war-related destruction of universities. Early surveys and bibliometric analyses suggested nontrivial shares of scientists abroad, continued but reduced affiliation with Ukrainian institutions, temporary host arrangements, and declines in publication output.

Design and target population: Online survey administered in four waves between 21 September and 8 December 2022. Target population: research-active employees of Ukrainian HEIs and PROs who worked in Ukraine when the war began (24 February 2022). Inclusion criteria required at least 3 hours/week of research on average over the three pre-war years.

Sampling and recruitment: Total of 3231 respondents. Exclusions: 413 were not employed by a Ukrainian HEI/PRO at the start of the war; 259 reported less than 3 hours/week of research pre-war. The analytical sample used across analyses is approximately 2,559–2,592 scientists, depending on item nonresponse.

• Wave 1: Contacted publishing scientists identified from Web of Science/Scopus (2011–2021) via publicly available emails; obtained 1183 responses before filtering. Adjusted response rate ~19.3% after accounting for 829 undeliverable emails.
• Wave 2: Emails sent to corporate addresses of all universities (list from the Ministry of Education and Science of Ukraine); additional manual harvesting of faculty-level contacts for top institutions.
• Wave 3: Outreach via social media (Facebook Messenger, LinkedIn, Telegram, Viber, Twitter) and via the National Research Foundation of Ukraine community.
• Wave 4: Dissemination through mailing lists of funding agencies, humanitarian programs, and Science4Ukraine. Waves 2–4 allowed referrals (snowballing).

Representativeness and weighting: The combined approach broadened coverage but induced imbalances (e.g., oversampling women, ages 30–59, professors/senior researchers; over-representation of social sciences; uneven regional coverage with concentration in Kyiv, Kharkiv, Lviv). Post-stratification weights were computed using iterative raking (DeBell & Krosnick, 2008; R package anesrake) to match sample margins to population benchmarks derived from Scopus author records (2019–2021) across multiple characteristics (gender, age, rank, field, region). Weights were recomputed for each analysis; not applied to sub-samples (stayers/emigrants), where the relevant population differs from the national frame. Assumptions acknowledge potential bias from omitted variables and nonresponse; estimates likely conservative if the most affected scientists were less likely to respond.

Measures and analyses: Outcomes include migration status (left Ukraine), exit from academia/research, and changes in weekly research time (categorical bands). Access to institution and research inputs were captured for stayers. Analyses used linear probability models (for migration and exit) with demographic and research productivity covariates (e.g., research hours >20 pre-war; top decile publications; doctoral degree), and ordered logit models for changes in research time among stayers. Reported statistics include marginal effects and F-tests for joint significance. Descriptive distributions summarize host arrangements and contract types for emigrants.

• Migration rate: By Fall 2022, an estimated 18.5% of Ukrainian scientists had left the country. Migration rates for scientists were broadly similar to those of the general population during wartime displacement.
• Selection into migration: More research-active and productive scientists emigrated at higher rates. Compared to others, those with >20 hours/week of research were about 6 percentage points more likely to leave; scientists in the top 10% by publications were about 12 points more likely; those with Ph.D. or Sc.D. were about 11 points more likely (joint F-test = 7.417). Female scientists had a roughly 5-point higher probability of leaving.
• Exit from academia/research: Approximately 17.5–17.6% of scientists no longer worked in academia or research. Emigration was associated with a 17-point higher likelihood of leaving academia/research. Scientists from South and East regions were significantly more likely to leave academia/research than other regions, but not more likely to emigrate.
• Research time contraction: The representative weekly research time fell from about 13 hours pre-war to about 10 hours during the war, implying roughly a 20% loss of research capacity. About 40% of stayers reported a decline in research time. Among stayers, 10.1% stopped research entirely, and 18.6% reduced research below 3 hours/week (threshold for non–research-active).
• Internal displacement and access constraints (stayers): 19.9% of stayers relocated within Ukraine. 23.5% reported losing access to important research inputs. Around 20.7–20.8% could not access their institution in its original location; reported access patterns included relocation of institutions (~14.5%) and online-only access for some.
• Emigrant scientists’ host arrangements: 57.5% hosted by HEIs/PROs; 15.7% by non-profits; 14.6% by for-profit companies; 5.4% self-employed; 4.6% job searching (n=280). Among those at HEIs/PROs (n=161): 18.1% formal contracts, 29.2% paid visiting positions, 44.7% scholarships/grants, 14.9% unpaid/informal. Only about 12% had long-term contracts (>1 year), indicating precariousness.
• Scientific engagement abroad: Over 75% engaged with host scientists; many reported substantial exposure to new ideas, tools, methods, and data; 87% believed the stay would improve their scientific abilities. Some moved partially into new fields (25–40% spent >10% of time in entirely new domains).
• Return intentions and ties: Among migrants, 40.5% said they will return, 24.8% are likely to return, 18.5% undecided, 7.2% unlikely, 0.5% will not (8.6% missing). Some evidence of reduced engagement with scientists in Ukraine compared to before.
• Long-run loss: Assuming half of those who paused academic activities do not return, Ukraine may have already lost roughly 7% of its scientists. Loss of highly active scientists risks a “lost generation” of Ph.D. trainees due to diminished mentoring capacity.

The study addresses how the war reshaped Ukraine’s scientific workforce via emigration, occupational exits, and reduced research time. Findings indicate that the war induced significant brain drain, disproportionately among the most active and productive scientists, while stayers faced severe constraints: internal displacement, loss of access to institutions and inputs, and psychological and infrastructure shocks that reduced research time. Although many emigrants report skill gains and exposure to new ideas, their precarious contracts hinder full integration and long-term contributions abroad. The combination of brain drain and accessibility constraints at home threatens Ukraine’s near- and long-term research capacity, Ph.D. training, and international collaboration networks. Policy responses should therefore be dual-tracked: stabilize migrant scientists through more and longer scholarships and pathways to secure contracts; and bolster stayers through remote visiting programs, access to digital libraries and computing resources, collaborative grants, mentoring, and institutional support to mitigate infrastructure and access barriers. Strengthening cross-border ties can convert migration into knowledge bridges and help preserve Ukraine’s scientific community for post-war recovery.

A nationally oriented, multi-wave survey of Ukrainian scientists documents substantial wartime impacts: approximately 18.5% emigrated by Fall 2022; research capacity fell by about 20%; and a notable share exited academia/research. Emigration has been selective of the most research-active, amplifying risks to Ukraine’s scientific system and doctoral training. While migrant scientists often gain exposure and skills, many face precarious, short-term arrangements. Stayers encounter severe research disruptions due to safety, infrastructure damage, power outages, and loss of access to institutions and inputs. Immediate policy priorities include stabilizing migrant scientists (longer scholarships, secure positions) and supporting stayers (remote access to resources, collaborative grants, visiting programs). Future research should track longitudinal outcomes—return rates, integration, productivity effects, and evolving access constraints—and assess psychological and organizational impacts not fully captured here.

• Cross-sectional snapshot reflecting conditions during September–December 2022; intentions (e.g., to return) may change over time.
• Potential nonresponse and coverage biases: severely affected scientists may be less likely to respond, making estimates conservative; uneven regional response distribution; over-/under-representation by gender, age, rank, and field.
• Weights correct observable imbalances but cannot address unobserved factors (e.g., localized conflict severity), leaving residual bias possible.
• Item nonresponse and categorical measurement of research time may underestimate declines within bands.
• Ambiguity in respondents’ interpretation of “leaving academia,” especially among public-sector employees.
• Some sub-sample analyses (stayers vs. emigrants) not weighted, as the reference population differs from national benchmarks.