Unequal effects of the COVID-19 pandemic on researchers: evidence from Chile and Colombia

In early 2020, COVID-19 led to campus closures and severe disruptions to academic work globally. Prior evidence from the United States and Europe showed reduced working hours and, especially, research time. Latin American universities, which operate with comparatively lower R&D investment and more precarious research infrastructures, faced particularly long and restrictive lockdowns in 2020, alongside reduced public funding for higher education and research. Chile and Colombia experienced nationwide closures of universities and schools for much of 2020, constraining access to labs, field sites, archives, and childcare. Against this backdrop, the study investigates: (a) the overall impact of the pandemic on research time among Chilean and Colombian researchers; (b) whether effects differed by discipline and individual characteristics; and (c) whether personal gender-role beliefs relate to changes in research time. Understanding these questions is important given pre-existing gender inequalities in the region and expectations that lockdowns would exacerbate caregiving burdens and widen disparities in research productivity.

Prior studies documented heterogeneous impacts of COVID-19 on scientists’ workloads and research productivity, with bench sciences (lab-dependent) showing larger declines than desk disciplines. Some fields (e.g., economics) even saw productivity increases, often driven by senior male researchers publishing on pandemic-related topics. A consistent literature highlights gender disparities in academia across hiring, publications, funding, and promotion, with women often facing a productivity penalty, partly explained by unequal household labor and caregiving responsibilities. During the pandemic, women’s research time and outputs declined more than men’s, and researchers with young children—regardless of gender—were especially affected, though women bore a greater burden. There has also been concern about early-career researchers’ vulnerability due to contract instability. This study extends the literature by examining Latin America, focusing on Chile and Colombia, and innovates by including personal gender-role beliefs as a potential factor influencing time allocation under lockdown.

Design: Anonymous online survey of researchers based in Chile or Colombia during strict lockdowns, fielded mid-July to early October 2020. Targeted researchers working in universities or research centers with a graduate degree (master’s, Ph.D., or equivalent) who reported research hours before and during the pandemic. Sampling and recruitment: Institution-centered approach using national and international rankings to identify leading research universities (18 in Colombia, 31 in Chile). Collected 28,871 emails from institutional websites (13,876 Chile; 14,995 Colombia). Invitations sent via Qualtrics with two reminders; an anonymous link encouraged snowball sharing. No personal identifiers collected; Qualtrics prevented multiple responses from the same email. Instrument: 46-item questionnaire (Qualtrics). Items adapted with permission from Myers et al. (2020) and ISSP measures for gender-role beliefs. Pre-tested with Chilean and Colombian scholars. IRB-approved by P. Universidad Católica de Chile. Participation and sample: 4904 unique visitors (3489 via email; 1415 via other means). 4253 consented; 3839 completed. Inclusion required affiliation with a university/research center, graduate degree, and reported research hours before or during the pandemic; gender had to be reported as male/female for analysis. Final analytic sample N=3257. Completeness rate 83.36% (2715). No weighting or imputation. Coverage estimates: Using RICYT counts (Chile 15,438; Colombia 16,796), coverage approximated at 12.68% (Chile) and 7.73% (Colombia), with caution due to lack of comprehensive population lists. Measures: Dependent variable was percentage change in weekly research hours before vs during lockdown at survey time. Key covariates included discipline, gender, age, academic position, contract type, teaching load, country, household composition (children under 12; adults 65+; partner), and a gender egalitarianism index. Gender-role beliefs index: Constructed from two ISSP items and one study-specific item (5-point Likert responses) capturing agreement with traditional vs egalitarian statements. Responses recoded to conservative vs egalitarian and summed to four categories: very conservative, conservative, egalitarian, very egalitarian. Analysis: One-way ANOVA with Bonferroni tests assessed differences across groups. Lasso regression (with interactions by gender for select variables) identified predictors with the greatest explanatory power for changes in research time; selected variables were then tested via OLS for statistical significance.

Overall impact: 54.1% of researchers reported a decline in research time; 24.3% no change; 21.7% an increase. Average weekly research hours fell from 14.0 pre-lockdown to 11.0 during lockdown, a mean decline of 16.97%. Discipline differences: Bench and site-dependent fields showed the largest declines (e.g., Astronomy −52.24%; Chemistry/Chemical Engineering −34.82%; Biological Sciences −27.25%). Fields requiring in-person interaction (e.g., Sociology, Communication) also declined. Health sciences showed smaller declines (e.g., Medicine −2.46%; Nursing −7.51%; Other Health Sciences −8.02%), and arts/architecture had minimal declines (Arts −1.66%; Architecture/Design/Urbanism −0.63%). After controlling for individual factors (Lasso), some health and arts fields exhibited increases in research time, while bench sciences remained negatively affected. Gender and caregiving: Women experienced a larger decline than men (women −22.13% vs men −12.42%), equivalent to −3.56 vs −2.49 hours, respectively (significant). Parenthood intensified impacts: women without children −14.42%; women with ≥1 child under 12 −34.49%; women with 3+ children −43.97%. Men without children were the least affected (−4.84%). Other individual factors: Living with a partner was associated with larger declines for women; having one older adult (65+) in the household mitigated women’s declines, but not with two or more. Teaching load mattered: no teaching or only one class was advantageous, particularly for men; teaching two or more classes increased declines. Contract type differences were small; part-time adjuncts (fee-based) appeared less affected likely because they reported few research hours overall. Assistant professors showed greater declines, followed by full/tenured, potentially reflecting higher pre-pandemic research hours. Lasso regression highlights: Being female was associated with an additional −5.35% decline, controlling for other variables. Having three or more children under 12 was associated with −12.89% change; being childless associated with +16.53%. For women, cohabiting with a partner and having two or more other adults remained disadvantages; for men, no or one class taught was an advantage. Most affected profile: women aged 40–49 with young children and a partner. Least affected: men aged 50–59, childless, teaching no more than one class. Gender-role beliefs: Overall egalitarianism categories did not differ significantly in aggregate. However, by gender, very conservative women saw the steepest declines (−40.35%), while conservative men had smaller declines (−10.07%). Very egalitarian women also declined more than very egalitarian men; egalitarian men had a larger penalty than conservative men, consistent with a greater sharing of house/care work among egalitarian men. Several group differences were statistically significant in ANOVA with Bonferroni corrections.

The study answers its research questions by showing that research time in Chile and Colombia decreased substantially during 2020 lockdowns, with pronounced heterogeneity by discipline and, more importantly, individual characteristics. Bench sciences and site-dependent fields lost the most time, while some health and arts disciplines were less affected or even gained time when controlling for individual factors. The most consequential predictors of reduced research time were gender and caregiving responsibilities, particularly having young children at home. Even after adjusting for discipline and other factors, being a woman carried an additional time penalty. The inclusion of gender-role beliefs clarified that cultural norms within households may exacerbate or mitigate time losses, with conservative beliefs associated with larger declines for women and egalitarian beliefs linked to comparatively higher declines among men, likely due to more equal sharing of care responsibilities. Compared with evidence from higher-income countries, Chilean and Colombian researchers started with fewer research hours pre-pandemic, and despite a smaller percentage decline than in Europe/US, their absolute research time during the pandemic was very low. These results underscore the intersection of structural constraints (lower baseline investment and institutional support) with household-level dynamics in shaping research time during crises.

This study provides evidence from over 3,200 researchers in Chile and Colombia that the pandemic’s effects on research time were uneven, with discipline-level differences but stronger disparities driven by gender and caregiving responsibilities. Women, especially mothers of young children, experienced the largest declines; childless researchers often gained time. Gender-role beliefs further differentiated outcomes across men and women. Findings suggest universities and research centers should design targeted mitigation policies (e.g., adjusted teaching loads, extensions on promotion/tenure clocks, and caregiver-specific research support). Future research should examine longer-term consequences for outputs (publications, grants, promotions), extend analysis through 2021 when lockdowns persisted, leverage improved regional databases for probabilistic sampling, and consider real-time time-use tracking to reduce recall bias.

Non-probabilistic sampling limits representativeness and introduces potential self-selection bias. Reliance on self-reported, retrospective estimates of research hours during a stressful period may introduce recall error. Internet-based survey methods can bias samples, though less likely among university-affiliated researchers. The study excluded respondents identifying as “other” gender due to small numbers, limiting gender diversity analyses. Missing data were not imputed, and no weighting was applied. Coverage estimates are approximate due to lack of comprehensive population lists. The sampling frame focused on leading research universities, potentially underrepresenting researchers from other institutions. Cross-sectional timing (mid-2020) may not capture evolving impacts through 2021.