You might call them academia's silent workforce - toiling for days on end in labs, libraries and the field, gathering much of the information that makes progress in research possible.
Quite how much PhD students contribute to general research has always been something of a mystery. But now a recent Canadian study has put a figure on it.
The research, which assessed papers published over an eight-year period by authors in Quebec, shows that doctoral students contribute as much as one-third of the province's total research output. It also suggests that doctoral students who become authors of published research are more likely to complete their degrees and go on to a career in academia.
The study's author, Vincent Larivière, assistant professor at the University of Montreal in Quebec, performed his analysis on Quebec students only, but believes the results might be reflected elsewhere. "I do think that we would obtain similar numbers in other countries - at least in the rest of Canada, the US and the UK," he says. "The research systems are comparable."
In the past, most studies attempting to quantify the amount of research performed by doctoral students have suffered from small samples or samples only within specialist fields.
One unusually large survey carried out in 1996 showed that roughly half of US graduate students had some published research to their name, with engineering and humanities students most likely to be published authors.
But even this study was limited in that its figures could have been swayed by the inclusion of conference papers and posters, which are not held in the same esteem as journal articles. Perhaps more importantly, it could not reveal the proportion of total research authored by doctoral students.
Not slave labour
Dr Larivière's study, which is due to be published in the peer-reviewed journal Scientometrics, breaks new ground. He took the names of all ,000 doctorate students in Quebec between 2000 and 2007, and compared them with authors of papers listed in Thomson Reuters' Web of Science - an online citation index - over the same period.
He found that roughly one-third of health, natural sciences and engineering papers had contributions from PhD students. The proportion was somewhat lower in other disciplines - about one-fifth in the social sciences and about one-eighth in the arts and humanities.
That result is no surprise. Publishing a paper is much more likely to be par for the course for PhD students in science and engineering subjects than in social sciences, arts and humanities.
Higher education experts say the results highlight problems with doctoral programmes. Paula Stephan, professor of economics at Georgia State University, who studies the careers of scientists and engineers, says that there has traditionally been an "implicit contract" by which students become trained and carry out work for an institution, in return, ultimately, for a permanent research job.
But while the number of graduate students has risen, she says, the probability of getting a research job has declined. "Many students feel that they have worked hard and will not be able to get a meaningful research position," she says. "And they have a real point."
Dr Larivière himself thinks his results should encourage people to recognise the importance of doctoral programmes. He believes PhD students are generally well rewarded, and that their work is not slave labour for permanent faculty members, an often-heard complaint.
"One must not forget that PhD students and faculty members do not do the same job," he says. "PhD students are still in the process of learning and are, generally, not yet ready to pursue independent research...Of course, one can always have a cynical standpoint and say that the student did most of the job, but this cannot be confirmed nor informed by my data."
What Dr Larivière's data do show, however, is a positive link between PhD students who are co-authors of published work and those who complete their degree and, ultimately, get a career in research. This implies - although it doesn't prove - that students can bolster their chances of success by getting involved with publications and socialising with other academics.
Jens-Christian Smeby, a sociologist at Oslo University College, argues that selection mechanisms can be at work. "It is often the most promising students that are invited to become co-authors, even though it may also be a way to help weak students," he says.
Dr Smeby's point raises the question of whether institutions should encourage all students to become co-authors to help them become more successful. Dr Lariviere thinks so: he says policies on doctoral programmes should, where possible, try to integrate students into the "collective dynamics" of research.
In the social sciences, arts and humanities, students are often left on their own to pursue their own projects and, possibly as a result, they take a long time to finish, Dr Larivière argues. He contrasts this with the natural sciences, where he says students tend to be involved with groups and finish their programmes much faster. "It is undisputedly more difficult to keep an original research project going if one is alone rather than part of a team," he says.
Problems with the dataset?
Catherine Millett, a researcher at the Education Testing Service, a non-profit US institution that develops exams for education, authored the 1996 study that examined the fraction of US doctoral students who became published authors.
She agrees that policy should take account of the benefits of publishing, as long as students are not pushed so hard that their chances of finishing their PhDs decline. "You want to get students into a doctoral programme and have them being a part of the research culture, understanding what it's all about," she says.
Not everyone agrees. Professor Stephan at Georgia State believes that there is a problem with Dr Lariviere's dataset, in that it mixes students who are well into their doctoral programmes with those just starting.
These newer students would have had less to publish, and might not have been set to finish by 2007 (the end of the analysis period) anyway. "Perhaps all it tells us is that more advanced graduate students (closer to completing their programme) write more and are more likely to graduate," Professor Stephan says. "I would certainly not want to base policy on this."
Nonetheless, Dr Larivière is now extending his study to assess the contributions of postdocs, who he believes account for an even higher proportion of total research than PhD students.