I was so proud of my little sister when she began getting admission offers from PhD programmes. But, as I talked to her, I found myself warning her away unless she was certain. The truth is that, as a fourth-year PhD student, I can recommend this path only reluctantly to someone I love because I have become increasingly concerned about its effects on students¡¯ mental health.

I am in many ways extremely lucky. I have a supportive adviser, reliable funding and research projects that I¡¯m devoted to. But I have at times been unhappy, and I have talked to many other unhappy PhD students; studies consistently show a high rate of mental health problems such as and among doctoral candidates ¨C roughly as in other highly educated populations.

To better understand the problem, I put out a survey asking PhD students to describe their specific experiences with mental health issues related to their doctorates and to propose ways to improve the situation. I received hundreds of responses from all over the world, primarily in the sciences. Here are some particularly striking experiences from students who agreed to share them:

• ¡°I have had trouble sleeping, and woke at 6am in December to work on a paper. I tripped and fell into my concrete apartment and broke my nose. That epitomises my experience.¡±
• ¡°I can no longer enjoy ANY moment of sleep, family time, or relaxation without feeling overwhelmingly anxious and guilty that I should be working.¡±
• ¡°During my PhD I had a panic attack that sent me to the emergency room. It was caused by being forced to work with a postdoc who constantly bullied me, publicly humiliated me, and generally made my life hell.¡±
• ¡°I remember myself taking long showers with cold water, crying and thinking that I was a fraud and I didn¡¯t deserve the opportunity to do science because of my lack of talent.¡±
• ¡°To invest everything that you have to an uncertain end tears one apart.¡±
• ¡°There is a certain sense of glorification of pain in academia. It is socially accepted that people go through depression.¡±
• ¡°My mom sent me a package for my birthday¡­Inside were a series of cute, brightly coloured decorative boxes with cookies inside¡­I cried alone in the parking lot for about 20 minutes, and felt about as low as I ever have.¡±

?

It is true that doing independent research is inherently hard and stressful, but we should not accept serious, long-term mental health issues as the price of doing a PhD. I believe there are several measures that we can adopt to address this. To come up with my suggestions, I first drafted a short list based on a review of the literature on PhD mental health. Next, I read each of the hundreds of responses to my survey, developed a standard set of categories to describe each one and then expanded and revised the original list of ideas on that basis. Finally, I circulated the expanded list to PhD students, professors and a mental health expert to get feedback.

ÖйúAƬ

Because most of my survey responses came from academics in the sciences, these recommendations are targeted at academics in the sciences as well. However, some apply more broadly.

1. Increase awareness that many PhD students suffer from mental health problems and improve treatment

ÖйúAƬ

This was one of the most commonly cited survey themes.

Currently, students are often for being honest about their mental health issues. ¡°I told one of my profs that I was getting an anti-anxiety med to help with the stress of [PhD exams]. She told me that anyone who needs a pill to survive grad school doesn¡¯t belong in grad school,¡± one survey respondent said. When another respondent ¡°met my supervisor and told him I had just seen my doctor [who had told me that] all my issues were stress-related, my supervisor answered that stress is good (he also constantly jokes that no one needs to sleep)¡±.

Students, advisers and administrators should be educated about the prevalence of mental health issues among PhD students, in order to reduce stigma and encourage people to speak up and seek help. They should also be made aware of the resources that students should seek out if they are suffering, and greater care should be taken that those resources are actually helpful. A of PhD students carried out in 2017 by the journal Nature found that only 35 per cent of those who sought help found helpful resources at their own institution.

2. Recognise and reduce discrimination

¡°Workplace sexism started the downward spiral,¡± one survey respondent wrote. A second described ¡°the time when I had a baby and my supervisor emailed me to inform me that women tend to lose interest in everything else once they become mothers¡±.

??these anecdotes, making it clear that discrimination persists in academia. Discrimination to depression and anxiety, which women report at???than men. Similarly, racial minorities in academia unique discrimination-related stress.

Combating discrimination requires realising that it is not always overt. For example, with respect to gender, a recent National Academy of Sciences finds that subtle discrimination is far more common than overtly sexual coercion. Combating discrimination also means making academics aware that they personally ¨C not anonymous others ¨C are prone to implicit bias, even if they are well-meaning and generally rational.

Resources directed specifically to underrepresented groups also help mitigate the effects of discrimination. These include gatherings and conferences for minority students, and regular check-ins with minority students to see how they¡¯re doing and to make sure that they feel like they belong. Efforts to recruit and retain more diverse faculty and students also help fight discrimination.

3. Facilitate supportive social connections

PhD students are . Many survey respondents emphasised that collaborating with peers, as opposed to competing with them, reduced loneliness and stress. Supervisors should help create collaborative environments by encouraging students to work together, rewarding students for being team players, avoiding favouritism and scheduling lab social outings. It is important to work hard from the outset to develop positive lab cultures because they become self-perpetuating. Developing simple social rules can also make the lab culture more positive and inclusive.

ÖйúAƬ

Survey respondents also emphasised connections with loved ones, who supported them irrespective of their professional achievements. ¡°Dogs and partners¡­help lots,¡± one remarked. The lack of such support can be devastating, as noted by an overseas student doing a PhD in a foreign country: ¡°One day, after having a [tough] day at office, I came back home and found no one waiting for me¡­I really had serious suicidal thoughts and I¡­cried for hours.¡±

Source:?

Getty montage

?

4. Improve work-life balance

It¡¯s true that PhD life requires periods of intense focus, when work-life balance becomes hard to achieve. But prioritising care for students is worthwhile even in these circumstances.

The Nature survey of PhD students found that 55 per cent viewed maintaining work-life balance as one of their top concerns, and my own respondents also attest to its importance. ¡°My mental health starts to deteriorate whenever my PhD starts to take over my life and becomes all-encompassing,¡± one wrote.

Studies show that depression and anxiety are more common among students with an unhealthy work-life balance. Many respondents to my survey recommended having hobbies, especially exercise, to provide some escape from work worries. Supervisors can also help to keep their students from working too hard. Model healthy behaviour: take real vacations, and encourage students to take them, too. Don¡¯t glorify lack of sleep; don¡¯t email students asking them to do things when they¡¯re on vacation; don¡¯t deliberately schedule meetings on Mondays, for which students will have to work through the weekend. Set clear expectations about how hard you expect students to work, and check in to ask whether you¡¯re doing things that unnecessarily stress them.

5. Pay students a living wage

¡°I went to the chair of my [programme] because our pay is based on 1950s rates, and I was not eating for days because I couldn¡¯t afford to,¡± one student wrote. ¡°She looked at me and said: ¡®I don¡¯t understand what your issue is. Lots of hours for little pay is just part of academia. If you don¡¯t like it, leave.¡¯ I¡¯m a 30-year-old man, and I had to go to the bathroom to burst into tears. I told this woman I couldn¡¯t afford to eat. She couldn¡¯t have cared less.¡±

Of respondents to the Nature survey, 50 per cent listed financial issues as one of their top concerns. In addition to paying students enough to live, universities should remove unnecessary uncertainty about their income streams. ¡°I¡¯m an international student in the US. Sometimes my advisor [threatens to stop] paying my tuition¡­This is very stressful!¡± one of my survey respondents said.

6. Provide students with more frequent and positive feedback

Many survey respondents complained that the feedback they received was too negative, vague or infrequent. They asked for concrete milestones, more structure and guidance, and more focus on successes, not just on critique. Advisers, of course, can be one source of this feedback. But some survey respondents said that they had benefited from having other mentors, including some who were not directly linked to their research but could provide broader life advice and emotional support. Additional mentors also protect students who have negative relationships with their advisers by reducing their reliance on a single person.

ÖйúAƬ

7. Conduct regular, anonymous, institution-wide surveys of PhD student mental health and well?being

This would help each institution track important issues, such as how rates of anxiety and depression change over the course of grad school and whether their interventions are effective. The survey could be extended to faculty, who, research suggests, also suffer from high levels of stress. As academics, we collect data on everything else; we should be rigorous about our own well-being. Such surveys have already been implemented at the University of California, and University, for instance.

Source:?

Getty montage

?

8. Help and incentivise academics to be better managers

This was the most commonly mentioned theme among survey respondents. Students who had positive relationships with their advisers testified to how much it mattered. ¡°I spent 6-12 months coming into the office every day and just not being able to do a thing,¡± one remarked. ¡°Thank the Lord that I had an understanding [adviser].¡±

But many respondents said that their relationship with their adviser harmed their mental health. One respondent ¡°would just walk into my building with dread¡­It was the fear of facing my advisor.¡± Another faced ¡°extreme anxiety leading up to meetings with my advisor, and panic attacks afterward. One time I just found a corner of a hallway I didn¡¯t think anyone would walk down and lay on the floor and cried after a meeting.¡± A third student said that because their adviser ¡°has no tolerance for seeing wrong results, [all the time], even during the night, I wake up and check my codes, derivations etc¡±.

The Nature survey found that guidance from their adviser was the top determinant of PhD students¡¯ satisfaction, and that 23 per cent would swap advisers if they could.

So why do students frequently struggle with their advisers? Some responsibility lies with the student, and useful for postgraduates have been written. But some responsibility also lies with advisers. All industries harbour some senior figures who are not skilled managers, but several factors make it uniquely difficult for professors to learn management skills. They are not generally trained or selected as managers, and there is often no formal feedback system for professors, so even well-meaning ones may be unaware that their students are unhappy. Moreover, professors just aren¡¯t incentivised to treat students well (although, of course, many of them do so anyway). They are put under enormous pressure to publish, and not penalised for passing that pressure on to their students.

Several survey respondents advocated management training for principal investigators. A second possibility is to have a formal, institution-wide system requiring that professors get annual feedback from the people they supervise. That feedback could be made part of their promotion process, the way it often is in the tech world. To alleviate students¡¯ fears of retaliation if they provided honest feedback (because anonymity is hard to preserve in small research groups), it may be best if the feedback were reported to the PI by a third party, rather than allowing them to read it directly. This would help to equalise the student-professor power imbalance (which survey respondents cited as one cause of abuse).

Students should also be encouraged to seek out information about whether a PI is a good manager (perhaps from their current students), and it should be made easier to switch advisers. These measures would incentivise all professors to become good managers ¨C and reward the many who already are.

9. Be honest about the negative aspects of doing a PhD

Present would-be students with statistics about job prospects and quality of life at the outset. The number of PhD students is increasing, and the number of tenure-track faculty positions is . The Professor Is In, an honest examination of the realities of academia by former tenured professor Karen Kelsey ¨C which has become a resource for many students ¨C notes that knowledge of this would ¡°empower you¡­to make informed choices about your career, and protect your financial security and your mental health.¡± Being honest about the difficulty of obtaining a tenure-track job would make nasty surprises less likely, and would also encourage students to explore careers outside academia.

It is also important to be honest about the ubiquity of failure. A year ago, when I had two papers rejected in one week, I took comfort in speaking to a friend who¡¯d had three papers rejected that week. We¡¯d all benefit from more honesty about the fact that research is really hard and that we all spend a lot of time feeling incompetent.

Such honesty would help to mitigate impostor syndrome, which is common among PhD candidates, and keep students from feeling, as one survey respondent put it, ¡°like I am the worst student in the group¡±. Give PhD students older students to confide in about their failures; have panels and online forums where successful students and professors . Celebrate honesty about mistakes rather than making students afraid to admit them. After all, as one survey respondent put it: ¡°We are all just advanced primates trying to do our best.¡±

It is important to bear in mind that, according to the Nature survey, more than three-quarters of PhD students remained at least somewhat satisfied with their decision to pursue a PhD despite the high rates of stress. ¡°The PhD gives me life,¡± one of my respondents wrote, describing their experience with depression. ¡°Every supervision meeting, every seminar, every conference, every debate about societal issues with colleagues over coffee, every rant with colleagues about the PhD, every day I wake up and do something to get me to the PhD finish line ¨C I resurrect over and over again.¡±

If the measures I have outlined were adopted, I am hopeful that many more students would feel the same.?

Emma Pierson is a Rhodes Scholar and computer science PhD student at Stanford University. She would like to thank Tim Althoff, Timnit Gebru, Dan Jurafsky, Scott Kamino, Clarke Knight, Pang Wei Koh, Hima Lakkaraju, Jure Leskovec, Alejandro Martinez, John Mikhael, Chris Olah, Leah Pierson, Paul Pierson, Nat Roth, Jacob Steinhardt, Pratiksha Thaker, Stephanie Vatz and Christopher Yau for helpful conversations and comments, and to all surveyed PhD students.

Reaching out: how the survey was done

The survey was disseminated on Twitter and Facebook. Current and former PhD students were asked to rate whether their mental health had improved since beginning their PhD, to describe their experiences with depression and anxiety, and to propose ideas for improving PhD student mental health.

Because unhappy PhD students may be more likely to respond to a mental health survey ¨C as have noted ¨C it is possible that my survey over-sampled them. There is some evidence, for example, that my survey had an unhappier respondent population than the that I mention in the main article, although questions in the two surveys were not identical. For example, in the Nature survey, 31 per cent of respondents reported that their level of satisfaction had worsened since beginning their PhD; in my survey, 50 per cent of respondents reported that their overall happiness had decreased since beginning their PhD.

However, the goal of my survey was not to estimate the rate of unhappiness, but rather to understand why students are unhappy and what they think might improve the situation. Thus, over-sampling unhappy PhD students increases sample size for this population of interest.

ÖйúAƬ

Emma Pierson