Four routes work entirely remotely: self-directed work on public data, open-source and citizen-science projects that credit contributors, remote-by-design programmes like MIT PRIMES and CrowdMath, and journals that accept submissions from anywhere. Only the third has eligibility gates. The first has none at all and is the one most students skip.
Students look for remote research and find a market selling remote programmes. The best remote routes are not products, and three of the four below cost nothing.
Route 1: self-directed work on public data
No application, no eligibility, no deadline. This is the one most students skip and it is the strongest.
The fields where a laptop is the entire apparatus:
| Field | What the raw material is |
|---|---|
| Economics | Census microdata, labour and price series |
| Political science | Election returns, survey waves, legislative records |
| Public health | Mortality, natality and surveillance files |
| Psychology | Repeated national adolescent and attitude surveys |
| Linguistics | Corpora, archives, subtitle and social text |
| Computer science | Public repositories, APIs, data you log yourself |
| Sociology and urban studies | Local authority, housing and transport data |
| History | Digitised newspapers, parliamentary records, collections |
Our list of free datasets covers twenty-two specific sources, and research without a lab sets out which corners of each field are genuinely open.
If you are applying from outside the UK or the US, there is an advantage here worth naming: a question about your own country, answered with your own country's data. You have access, context and language that nobody else in the applicant pool has, and no programme sells it.
The thing you have to supply yourself is the deadline. That is the honest weakness of this route, and it is also the only weakness.
Route 2: projects that credit contributors
Real research, publicly recorded, with no selection step.
Citizen science. Platforms like Zooniverse host projects where volunteers classify images, transcribe records or annotate data, and several have produced published papers that credit contributors. You are not a co-author on a whim — you contribute enough that the project acknowledges it, and the record is public and checkable.
The step most students miss is going further than classifying. The interesting work is asking your own question of the project's open data once you understand it. That converts a volunteering line into a research project.
Open source. An accepted pull request to a scientific software library — a bug fix, a test, documentation, a new function — is a permanent public record with your name attached, reviewed by maintainers who did not have to accept it. For anyone heading toward computer science or computational research this is more substantial than most summer programmes and costs nothing but competence.
Transcription and annotation projects. Archives, herbaria and historical collections run open transcription projects. Tedious, genuinely useful, and they put you inside a real dataset, which is where questions come from.
Route 3: structured remote programmes
Fewer than the marketing implies, and mostly nationality-gated.
MIT PRIMES is free, runs a full year, and is remote. It requires US residency. Crucially, PRIMES directs international students to CrowdMath instead — an open online collaborative research project with no residency requirement, where participants work together on genuine open problems. That is one of very few structured, free, globally open routes in existence, and almost nothing written for students mentions it.
MITES Semester is free and runs online over six months, but requires US citizenship or a green card.
Note the pattern: the free structured tier is largely closed by nationality on both sides of the Atlantic. The UK equivalents — UNIQ, Nuffield, In2scienceUK, Sutton Trust — are restricted to UK students and mostly to under-represented UK backgrounds. Our guide for students applying to UK universities from abroad covers what remains.
And one correction worth repeating, because it appears on dozens of high-school opportunity lists: NASA's OSTEM internships require enrolment at an accredited college or university. They are not open to school students, remote or otherwise.
Route 4: publishing from anywhere
Every venue worth submitting to accepts submissions remotely, and several accept them from any country.
The International Journal of High School Research states that "High school students and their mentors from any country may submit advanced research," with rolling deadlines and two or more outside reviewers. Vanderbilt's Young Scientist Journal runs a 1 November to 1 December window. The Oxford Journal of Student Scholarship takes high school and early undergraduate work across all disciplines, including literature reviews.
Our comparison of every student research venue sets out fees, reviewer counts and deadlines side by side.
The thing remote work cannot give you
A reader, and this is worth being honest about.
Most students who attempt independent research do not fail on intellect. They fail because nothing external requires a draft by a particular Thursday and nobody with expertise reads it when it arrives. A remote programme sells you that reader; the free routes do not include one.
Three ways to get one without paying:
A teacher with a doctorate at your own school. The most underused resource anywhere. One conversation.
A graduate student, asked one bounded question. "Would you read my methods section and tell me whether the design is sound?" has a defined end, which is why it gets answered. Location is irrelevant to an email — our guide to emailing a professor has the version that works.
A cold-emailed academic. Long odds, stated honestly by one Stanford lab that receives "tens of such requests per year" and takes a high school intern "only once every 4 to 5 years." The University of Arizona suggests approaching three to five faculty rather than one.
Does remote work count for less?
By one institution, explicitly, and it is a routing decision rather than a judgement.
Caltech reports that 42% of admitted students submitted research, specifying it should come from "a laboratory in a university or research-affiliated organization" with "a letter of recommendation from site of research." Independent research is directed to a portfolio instead.
Cutting the other way, Yale asks applicants to state "your specific contributions if the work was done in collaboration with others" — a question a solo remote researcher answers in three words and a student who was the fourth pair of hands in a lab finds much harder.
Most universities publish nothing this precise. What is not guesswork is that a tutor can question you about your project for as long as they like, and where you sat while doing it does not come up.
A first week that works
Day 1. Pick a field from the table above. Pick one dataset. Download it.
Days 2–3. Read the codebook. Not skim — read, until you find a variable you did not expect to exist.
Day 4. Write three candidate questions, each naming two variables and a population. Score them against six tests and keep one.
Day 5. Email five people asking each for one bounded read. Expect four not to reply.
Day 6 onwards. Start writing. Not reading — writing. The students who finish write from week one.
Where we fit
The Oxford Centre for Advanced Research runs 1:1 mentorships with PhD candidates and early-career researchers at Oxford, Cambridge, other leading UK universities and Ivy League institutions, entirely remotely. The Oxford Scholar Programme is £2,000 — about $2,600 — for ten contact hours over ten to fourteen weeks.
What we sell is the reader and the deadline described above. Everything else on this page — the data, the citizen-science projects, CrowdMath, the journals — is free and needs nothing from us.
If you have ever finished a long piece of work nobody set you, do this alone and keep the money. Our comparison of mentored and self-directed research sets out the honest test for which you are.
Today
Open one dataset and read its documentation for an hour.
Thirty emails to labs is roughly a one-in-a-hundred-and-fifty proposition on the published numbers. One hour inside a well-documented dataset reliably produces a question. The second is a better use of the same evening.
Frequently asked questions
Can you do real research from home?
In several fields, routinely. Economics, political science, public health, psychology using survey data, linguistics, computer science and statistics all run on public datasets that anyone can download. The constraint is the data, not the location, and in those fields the data is already free.
Are there structured remote programmes?
Some, and most are nationality-gated. MIT PRIMES is free and year-long but requires US residency, directing international students to CrowdMath instead. CrowdMath is an open online collaborative research project with no residency requirement, which makes it one of the few genuinely open structured routes.
Can I get credited on a real project without a placement?
Yes, through citizen science and open source. Zooniverse projects have produced papers crediting volunteer contributors, and an accepted pull request to a scientific software library is a permanent, checkable public record with your name on it. Neither requires anyone to select you.
Do remote projects count for less?
Caltech routes independent work to a portfolio rather than treating it as supervised lab research, which is a routing decision rather than a judgement. Most universities publish nothing this precise. What holds up at interview is whether you can defend the work, not where you sat while doing it.
What about remote lab internships?
Mostly they do not exist, and that is policy rather than preference. Most universities bar under-eighteens from the lab work students want to do, and a lab cannot supervise bench work remotely anyway. The computational corner of your field is the realistic version.
How do I get someone to read my work remotely?
Ask for one bounded thing by email. 'Would you read my methods section and tell me whether the design is sound' gets answered far more often than a request for supervision, and the recipient's location is irrelevant. Graduate students answer more often than professors.