Science2 publishers3 min readPublished
Survey of 289 geoscientists ranks gatekeeping and sabotage as fieldwork's most damaging harms
Eighty percent of respondents reported at least one harmful interaction during fieldwork, and three quarters of the worst incidents went unreported. Severe violence and misconduct were excluded by design, which makes the headline figure a floor rather than a prevalence estimate.
The Scientist · Science desk
What happened
- Anne Gold and colleagues anonymously surveyed 289 AGU members about how often harmful field interactions occur, which types hurt most, and why incidents go unreported, publishing the results in AGU Advances.
- Eighty percent of participants reported experiencing at least one harmful interaction during fieldwork, with the average respondent reporting five, and severe violence and misconduct were outside the study's scope.
- Three quarters of the instances respondents identified as most impactful were never reported to anyone.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- constraint Without a response rate or any measurement of the same behaviours away from field sites, the 80% cannot be used to argue that fieldwork is worse than campus work, which limits what a funder can justify spending on field-specific rules.
- decision Anyone writing a field safety plan now has evidence that the harms participants rate as worst are about credit and career access, so the default investment in a harassment briefing addresses the reported behaviour but not the reported damage.
- exposure The conditions that tracked early career stage are set by whoever books lodging and sets evening norms, which puts the burden on individual trip leaders rather than on the institution that wrote the code of conduct.
- contradiction The reporting frames federal guidance as the backdrop to persistent harm, yet nothing in the survey evaluates that guidance, so the tidy read that NSF's rules are the failure point is unearned by this evidence.
The arithmetic the summary leaves implicit: 289 respondents at a mean of five harmful interactions each comes to roughly 1,445 incidents [15]. Eighty percent of 289 is about 231 people who reported at least one [16], so if that mean is taken across everyone surveyed, the people who reported anything averaged closer to six apiece [17]. Neither writeup says which denominator the mean uses, and the two readings describe different problems: a thin background hum across a discipline, or repeat exposure concentrated in particular field seasons and particular crews.
This is a self-report instrument with an exclusion stated up front. Severe cases of violence and misconduct were left out of scope [2], which pushes the 80% down rather than up. What the sources do not carry is a response rate, a description of how respondents were recruited, or any measurement of comparable behaviour outside the field [18]. So the 80% is a share of the people who chose to answer, drawn from a professional society's membership rather than a census of everyone who does fieldwork [22], and with nothing measured in labs or offices the design cannot isolate what fieldwork itself adds. The word "disproportionately" is doing real work in the finding about women and nonbinary researchers [7], and neither writeup gives the counts or effect sizes behind it [19].
The result worth keeping is the ranking. Identity-based behaviours were commonly reported, but respondents consistently put professional-conduct harms, gatekeeping, sabotage and devaluation of work, at the top for negative impact [6]. That is what a self-report survey is actually built to measure: not how often an event occurred, but how much it cost the person it happened to. Those three are career mechanics, and they outranked the behaviours that safety modules usually cover.
On federal guidance, the sources put two things side by side. The National Science Foundation, described as the largest agency funding field-based research, provides fieldwork safety guidelines, and geoscience researchers nonetheless report bullying, discrimination and sexual harassment [12]. The sources also name shared housing, uneven power between students and professionals, remote sites and long workdays as conditions that raise the odds of harm [13]. They do not examine what the guidelines say or test them against outcomes [20]. The idea that the guidance misses housing and hierarchy is a plausible hypothesis; this survey did not test it.
The survey's internal patterns are the closest available support. The two harms that tracked early career stage were uncomfortable sleeping situations and pressure to drink alcohol [8], both decided by whoever plans and provisions a trip, and pressure to join unsafe activities and bullying were more common at remote worksites [9]. Against that, the remedies the summary emphasises are improved training, enforced codes of conduct and independent reporting pathways [11]. Two of those three act after an incident exists, which is the right response to 75% of the most impactful instances going unreported [10] and a weak lever on who sleeps where.
What to watch
- Whether the full AGU Advances paper reports a response rate and recruitment method, which decides whether 80% is a rate or a tally of volunteers.
- Whether funders or field programs change trip logistics such as housing and lodging assignments, rather than adding another training module.
- Whether the paper's tables carry counts and effect sizes behind the disproportionate impact on women and nonbinary respondents.