Trends in American scientists’ political donations and implications for trust in science

The paper investigates whether and how American scientists’ political orientations, as reflected in campaign donations, have shifted over time and what implications this has for public trust in science. Motivated by longstanding claims of liberal bias in academia and concerns that such bias undermines trust in science, the authors analyze the political donations of scientists across sectors using FEC data. Prior studies on faculty ideology often suffer from sampling and methodological issues; more robust work finds the professoriate to be largely moderate to slightly liberal. Given evidence that many scientists identify as Democrats, the study examines longitudinal donation behavior (1979–present) among scientists in academia and industry to clarify trends, contrast them with the broader academic workforce, and relate them to partisan trust in science.

The authors review debates over alleged liberal bias in academia, noting methodological flaws in many critiques (e.g., selective sampling, small samples, reliance on voter registration). Robust surveys (e.g., Gross and Simmons, 2006) find moderates slightly outnumber liberals among faculty. Pew (2009) reported scientists lean Democratic. Prior work has used donation data to assess professions’ ideologies (e.g., lawyers). The literature on trust in science shows increasing partisan polarization since the 1970s, with a reversal whereby Democrats now report higher trust than Republicans (Gauchat; Pew; Gallup). Explanations include education’s liberalizing effect, Republican skepticism toward science (e.g., climate change, COVID-19), right-wing populism’s portrayal of scientists as elites, and potential tensions between scientific norms (communism, universalism) and conservative worldviews.

- Data sources: Federal Election Commission (FEC) individual donation records from 1979 onward; employer and occupation fields available from 2002. General Social Survey (GSS) waves through 2021 for trust in science. Scopus API (Elsevier) to classify academic donors’ scientific fields. The analysis code pipeline is available at https://github.com/lue/political-donations-by-scientists. - Party classification: Donations were weighted by amount and categorized as to Republicans, Democrats, or third parties combined. Donations with party affiliation marked as unknown were excluded (low proportion). - Academic sector identification (post-2002): Donors with employers containing “college” or “university” and occupations containing “professor,” “faculty,” “scientist,” or “lecturer.” Administrators identified by the occupation keyword “administrator.” Analyses include all college/university employees, the professoriate subset, and a subset of scientists within the professoriate via Scopus matching. Institutions were further split into Ivy League and Council for Christian Colleges and Universities (CCCU) members. - Industry sector identification: Focus on 10 large energy companies (Exxon Mobil, Chevron, Marathon Petroleum, Phillips 66, Valero Energy, Energy Transfer, World Fuel Services, ConocoPhillips, Exelon Corporation, Plains GP Holdings). Executives/management identified by occupation keywords (“supervisor,” “chairman,” “CEO,” “COO,” “VP,” “executive,” “president”). Scientists/engineers identified by keywords (“engineer,” “geologist,” “chemist,” “geophysicist,” “scientist,” “professor,” “researcher”). Donations from top 10 technology-sector companies were used for comparison in figures. - GSS trust measures: Party identification based on the standard GSS question; independents/other/no answer omitted. Trust in science measured as the fraction reporting “a great deal” of confidence in the scientific community over time. - Field classification: Scopus API used to retrieve donors’ subject areas (physical, health, social, life sciences). For scholars with multiple fields, donations were fractionally allocated. Geographic information from Scopus was not used due to mobility and residence-workplace mismatch. For common names with multiple entries, the top search result was taken. - Sample sizes: Approximately 100,000 professors making 1,000,000 unique donations since 2002 identified; 80,000 matched via Scopus; restricting to unique, non-ambiguous matches yields ~28,000, without qualitative changes to results. - Temporal plotting: Lines begin only when sufficient data are available (not immediately at 1979 or 2002). - Stated limitations: Self-reported names/occupations may be inaccurate; recent donation practices (e.g., online platforms, small-dollar giving) complicate comparisons over time; donors are more politically active than the broader scientist population. Authors focus on donation ratios (R vs. D) and carefully selected groups to mitigate biases.

- Academia overall: From 1984–2000, college/university employees’ donations to Republicans were roughly stable at ~40%, indicating only a slight Democratic lean. From 2000–2021, the Republican share dropped sharply to <10%. - Professoriate: Since 2016, professors gave about 5% of donations to Republicans, lower than other university employees; Ivy League professors gave about 2% to Republicans. Aggregate donation volumes rose sharply in 2019 while Republican share was at a historic low. - Administrators vs. professors: Administrators donated more to Republicans than professors, suggesting administrators are less liberal or less estranged from Republicans than the professoriate. - Institution types: Declines in Republican donations were observed for both secular and sectarian institutions, including CCCU-affiliated colleges; Ivy League showed consistently lower Republican shares. - Disciplines: Across physical, health, social, and life sciences, donations to Republicans declined; physical sciences showed a slightly higher Republican fraction than other fields, though precise cross-disciplinary comparisons are limited by matching methods. - Energy sector (industry): Overall Republican share among energy-sector employees dropped from ~60% (1990–2000) to ~30% in 2020. Among energy-sector scientists/engineers, Republican donations fell from ~50% (2008–2012) to ~10% (2018–2020). A post-2020 uptick occurred during a low-volume period. - Executives vs. scientists (energy): In 2020, executives/management donated to Republicans about five times more frequently than scientists/engineers, indicating strong internal divergence akin to (but larger than) the administrator-professor gap in academia. - Third parties: Third-party donation fractions remained small and stable (~1%) across groups and time, comparable to recent Republican shares among professors, and did not display large trend shifts. - Trust trends (GSS): Partisan reversal since the 1970s–2000s: Republicans once reported higher trust in science than Democrats; by 2021, 65% of Democrats vs. 32% of Republicans reported “a great deal” of confidence in the scientific community.

The pronounced shift of scientists’ donations toward Democrats is observed in both academia and industry, suggesting it is not solely an artifact of academic culture. Education’s liberalizing effect helps explain why scientists (highly educated) lean left relative to the general population and their organizational leaders (administrators/executives), but it does not account for the recent, sharp decline in Republican support. The authors argue that Republican Party antagonism toward science (notably on climate change and COVID-19) and a broader populist framing of scientists as out-of-touch elites have likely alienated even moderate or conservative scientists. Concurrently, Democrats have increased their trust in science and integrated pro-science messaging. GSS data show a partisan reversal in trust in science, aligning temporally with scientists’ declining donations to Republicans. The evidence suggests scientists have turned away from the Republican Party due to its oppositional stance toward science, rather than Republicans distrusting science because scientists are anti-Republican. Concerns that scientists’ political leanings bias scientific validity lack empirical support; replication analyses in psychology show no partisan replicability differences. However, heightened perception among conservative Republicans that scientists are liberal, combined with ideological polarization and opposition to certain policy implications of science, likely contributes to reduced Republican trust. Strategically, the authors propose elevating centrist and conservative scientists’ public engagement to leverage trusted-messenger effects and potentially improve Republican trust in science.

The study demonstrates a substantial, post-2000 realignment of U.S. scientists’ political donations away from the Republican Party and toward Democrats, evident across academia and the energy industry. The divergence between scientists and organizational leaders (administrators/executives) underscores differences linked to education and possibly hierarchical roles. The timing of these donation shifts coincides with a partisan reversal in trust in science, supporting the interpretation that Republican Party antagonism toward science has driven scientists’ estrangement. Policy and communication implications include fostering public engagement by centrist and conservative scientists to enhance trust among Republicans and counter anti-science messaging. Future research should refine data linkage between FEC and Scopus to assess disciplinary variation more precisely, examine sectoral differences beyond energy and technology, and explore causal pathways between party elites’ rhetoric, public trust, and scientists’ political behavior.

- Self-reported names, employers, and occupations in FEC data may be inaccurate. - Donation practices have changed (e.g., rise of online platforms and small-dollar donations), complicating cross-period comparisons. - Donors are a politically active subset of scientists and may not represent all scientists’ views. - Post-2020 analyses note very low donation volumes, reducing interpretability of short-term fluctuations. - Scopus matching challenges: common names, multiple entries; using top search result and excluding ambiguous matches may introduce bias, though authors report no qualitative change in results. Geographic data were not used due to mobility/residence issues. - Pre-2002 FEC records lack employer/occupation fields, limiting earlier period disaggregation by role or discipline. - Analyses rely on donation ratios (R vs. D) and selected occupation/employer keyword filters, which may misclassify some donors.