I am female and an early-career PI (principal investigator) in a discipline that tends to attract women, and people with high levels of anxiety (this is documented). In my first years as a graduate supervisor, I have selected a balanced group of men and women from a variety of backgrounds based on their academic potential and research interests.

I feel strongly about the PI’s role in reducing the stress and pressure of graduate school. Each student is fully supported financially and has a funded research project they declared and interest in and agreed to, and are encouraged to develop their own line of research according to their interests for subsequent projects. Some have a co-supervisor and all have a committee of PIs to support the training plan and help make important decisions. I encourage students to develop workload and project management skills, and am always available if they need help or to discuss, but I do not pressure them. I make it clear that I highly prioritize mental health and encourage them to only work 35 hours/week, take vacations when they like, to exercise/socialize, and to avail themselves of university services (of which we have many).

The men in my lab are doing very well. They occasionally need reassurance, emotional support, or an adjustment in tasks/approach, and then they are able to get on with their research and are well on their way to developing into independent, capable, and kind researchers.

Almost all of the women have turned out to have pre-existing mental health issues (anxiety, depression, and in one case PTSD). These students become overwhelmed, emotionally volatile, keep changing their ideas about what they want to work on (even though their funding is connected to lab projects), sometimes become jealous of other students' progress, and in some cases become entirely non-functional at intervals. I scramble to adjust their workload for example hiring undergrads to help with their data collection or doing it myself, investing hours in counseling and reassuring them, problem-solving with their committee/co-supervisors, and losing sleep when I know they are feeling miserable. I don’t see what more I can do actually change about their graduate experience while having them progress and be successful in their degrees, nor do they have concrete, stable ideas about how I could help.

So far, I am surprised that this seems to be a highly gendered issue. I feel ill-equipped to provide the needed level of psychological and emotional support for this kind of high-anxiety student – but nor can I run a lab and attend to my other duties if half of my students are taking 90% of my time and are making limited progress.

As woman and a person with a strong interest in equity, diversity and inclusiveness I am horrified at the idea, but I find myself considering only accepting to supervise non-female students until the lab is established and there is less startup pressure.

How can I avoid this terrible solution and live up to my ideals, while not going nuts myself?

Clarifications: I do not attempt to provide counseling(!), but rather refer students to mental health services for specific personal issues, which they can access free of charge. Some students choose to make me aware of their struggles, usually when they are in distress, in which case I try to work with them to rearrange practical aspects of their situation, and consult with senior colleagues if necessary. Students in my lab already have a lighter load than others in similar situations as they are fully funded without being required to serve as teaching assistants - they only have to do light coursework and progress on their own research projects. I am not trying to justify this practice, nor to assert that it is a universal problem (which is why I think it is not useful to defend my observations with specific statistics, though I have referenced several papers), but rather to find better solutions in my specific situation after having tried to make various changes and accommodations. I am not that worried about discrimination complaints and legal issues - selecting one person over another with the amorphous selection criteria we use and respondents have suggested can always be justified, which is why sexism and discrimination is such a difficult thing to remove from our system (but I am going to keep this anonymized).

Comment on my own mental health: I have some excellent mentors myself and am well supported by my colleagues, but I am admittedly in a difficult career stage. There is maybe a personality factor in that I have difficulty not being upset myself when people I am responsible for are in distress and I can do nothing for them, and they continue to disregard guidance and make choices that make things worse for themselves and others. In my opinion, some have no business starting a graduate degree with the unresolved problems they have. I would have difficultly finding the time and energy to support the drain that comes with people with pre-existing anxiety and depression, though will need to accept students occasionally to keep up projects and funding commitments. So perhaps I have no business selecting people for whom there is a higher risk of this problem, and whom I would not be able to optimally support at this time.

Final note:

My values are to support EDI and promote women and underrepresented groups in science and to treat people as individuals. My practical considerations are that I am having trouble during a critical early career stage supporting an unexpectedly high proportion of students with pre-existing anxiety and depression, and don't want to take on any more of them until I can be sure I can provide them with the needed level of support and that they will not cause problems for themselves or other people. I should note that other groups may have a higher prevalence of other sorts of problematic issues; this is just the one that is using most of my time and energy.

I think the best combination of answers is:

  • recruit from amongst people I have had a chance to work with or a trusted colleague has (not just a letter of ref, which are often not very revealing)
  • screen as selectively as possible for evidence of resilience, stress management, and coping skills with thoughtful interview questions
  • make very sure students understand that it's going to be psychologically challenging
  • be kind but reasonably supportive, but clear and firm about my role as a supervisor and what are non-negotiable expectations regarding professionalism
  • accept that I can't fix people and I can only be responsible for their successful professional development if they want and let me
  • ride it out, as the lab collectively gains more maturity the senior students and our lab culture will be able to help stabilize and support new people

Thanks everyone for your opinions and for remaining civil. I like to think that academic training equips people to examine controversial points and consider alternative viewpoints. For the most part this seems true here.

  • 44
    Wow, this is going to get ugly. I am going to take the unusual step of preemptively marking this controversial and setting up a chat. Please see the post notice above before adding comments or answers.
    – cag51
    Commented May 27, 2020 at 23:57
  • 1
    I moved a handful of comments to chat again. If you want to discuss the premise of the question, please do it there. Please only use comments only if you expect them to lead to an improvement of the question. Other comments will be deleted without warning. Also see this FAQ before posting another comment. @Nemo: Please edit any clarification into your question.
    – Wrzlprmft
    Commented May 28, 2020 at 10:18
  • 6
    What's your sample size? A "balanced group" of 4 students (the minimum that would fit your use of plurals) would be statistically very different to a group of 20, and you say you're early-career suggesting low numbers
    – Chris H
    Commented May 28, 2020 at 13:48

14 Answers 14


Should I avoid admitting female students for now?

No. Assuming the information you have posted is correct, using a gender filter is not ethical if implemented for the reasons you have given, as you seem to already know.

Actions you can take:

  • Design a filter for prospective students that addresses your actual problem (bad behaviors).
  • Stop providing counseling to your students. It is not ethical or effective to try and have both the counseling role and the supervisory role, even if you are trained in both. Supervisors can provide some support to students, but it sounds like you're going too far.
  • Educate yourself further in managing students' problematic behaviors. This will include getting students outside support for their mental health.
  • Be willing to terminate the supervisory relationship when it's not working. Obviously this is not a desirable option, but it needs to be available to you.

Reasons I am skeptical of the premise:

  • It sounds like you've drawn a conclusion based on just the PhD students in the group of a "new prof", which is an unreasonably small sample.
  • You say you cannot attract top students, but you think you are able to discard the majority of your applicant pool. This is not logically consistent.
  • 9
    Thanks for your comments. To clarify, I absolutely do not try to provide psychological counseling and have referred students to mental health services, also noting to them where my role supporting their mental health and their progress as a graduate student may be in conflict. Small sample size, though striking separation. I am interested to know how to ethically screen for such things that are not visible in the application materials nor in an interview, such as resilience / ability to manage one's mental state and emotions despite any underlying personality traits or mental health issues.
    – Nemo
    Commented May 28, 2020 at 1:00
  • 15
    “Be willing to terminate the relationship” - this is good advice generally, but in Nemo’s situation with it being almost entirely women who are having problems, it would look just as bad (worse?) then a gender filter at hiring time - I doubt anyone can get away with terminating in a strongly gender skewed manner (even if for the right reasons).
    – Tim
    Commented May 28, 2020 at 8:27
  • 9
    "Design a filter for prospective students that addresses your actual problem (bad behaviors)" - have we established that 'bad behaviors' have any relevance here? I don't think you meant it this way but from a distance this looks like an excuse to discriminate, which I don't suggest.
    – Bryan Krause
    Commented May 28, 2020 at 8:32
  • 3
    @BryanKrause There is a vague list of bad behaviors in the question. Commented May 28, 2020 at 9:15
  • 4
    Unfortunately, this may come with its own set of problems in the case of actual mental illnesses. Rejecting a student because of something like depression or PTSD might be illegal, for example.
    – reirab
    Commented May 28, 2020 at 16:41

Should I avoid admitting female students for now?

Absolutely not. Excluding a entire group of student based on what is most likely legally protected characteristics such as gender, is unethical and maybe illegal. Students should not be discriminated against based on an inborn characteristic such as gender that they have no control over.

I feel ill-equipped to provide the needed level of psychological and emotional support for this kind of high-anxiety student

Nice job on recognizing your limitations. The demands of research and graduate studies can be stressful and some students may indeed require extra support to cope. It would be advisable to refer these students to a professional such as a therapist / counsellor / psychiatrist who are trained to help individuals with their mental health / coping skills. In addition, being a PI puts you in a position of power and requires being objective to properly assess your students achievements. Attempting to counsel your students at the same time can muddle the relationship and lead to being inappropriately close such that it would interfere with your professional role.

  • 7
    "Attempting to counsel your students at the same time can muddle the relationship and lead to being inappropriately close such that it would interfere with your professional role." -- A reason I (grad student) avoid discussing personal issues with colleagues and my advisers is ensuring any advice and evaluation coming from them is focused on my work and isn't biased because of knowledge of what is going on with my personal life. There are things that sometimes go through, but I try keeping it friendly (but not too close) and professional. I feel it's best this way at least until I'm finished.
    – Daniel
    Commented May 28, 2020 at 10:44
  • 3
    @Daniel that's a fine approach unless things are sufficiently bad as to impact your work or could progress to become such. There is generally no need to discuss the issues in depth, but it is often good for a supervisor to be aware of such issues as they develop.
    – Isaac
    Commented May 28, 2020 at 10:52

I think it is dangerous and prejudicial to suppose individuals of one category will perform a certain way. These are exactly the types of approaches that lead to all sorts of inequalities in the world, and even if there is an actual difference between men and women in your lab it is the surfacing of those inequalities (which are not simply escaped by entering a field where women are a majority); reacting to inequality by perpetuating it seems to be entirely in conflict with your stated goals.

If students under your supervision are having trouble with the stress and pressure of graduate school, then go ahead and take the PI's role you suggest and help them deal with that stress and pressure. You might find some of your approaches are counterproductive and revise them, and seek advice from others on how to better serve your students. Learn how to address some of the specific patterns you are seeing as problematic. But treat your students as individuals, and don't use labels to disqualify them.

  • 3
    As a corollary - how sure are you that the Men are not affected too? Since we appear to be allowing generalisations - Men tend to hide such things much better than Women, which is not a good thing.
    – MikeB
    Commented May 28, 2020 at 10:44
  • 3
    "even if there is an actual difference between men and women in your lab it is the surfacing of those inequalities" This is possible, but it's definitely not a certainty. There are differences between men and women which have nothing to do with "surfacing of inequalities." Pretending those differences don't exist is not helpful, but, of course, neither is excluding all female students from your graduate admissions. +1 for your conclusion at the end, though.
    – reirab
    Commented May 28, 2020 at 16:48
  • @reirab There may be measurable average differences between sexes, but besides some physical features like height there is an incredible overlap in the distributions. It's very problematic to apply those small differences to individuals.
    – Bryan Krause
    Commented May 28, 2020 at 16:54
  • 2
    @BryanKrause Completely agree that the differences are on average and most are not true for anywhere near every case. But that still doesn't make the assertion that any differences that show up are a "surfacing of inequalities" true. A difference that is true on average will still show up in a workplace. I 100% agree on your conclusion of treating each student as an individual and not stereotyping them, just not the part about any differences that show up definitely being a "surfacing of inequalities." Replacing "is" with "may be" would fine, though.
    – reirab
    Commented May 28, 2020 at 17:04

No you should definitely not avoid admitting female graduate students. As others have helpfully answered there are many reasons for not doing this.


How can I avoid this terrible solution and live up to my ideals, while not going nuts myself?

Of course you don't want to go nuts. Graduate school is a big commitment for both the student and the advisor. As you note in your comments, it's not always easy to evaluate a candidate based on their written materials (essays, transcripts, test scores, etc.) and an interview (often done remotely). Letters of recommendation or other types of references can help, but don't always.

So what to do? In my experience, one way to have a chance to see if a prospective talent is a good fit for your lab environment and the type of research you do is to take her/him on as an intern. Even 3-6 months of working with someone can be enough to gain insight into someone's personality and for them them to know you and your lab. Alternatively, you may use a masters program as a lower-commitment way of bringing a talented student on, with the prospect of converting to a PhD following an evaluation.

Everyone likes the opportunity to "walk away happy". If you take a prospective student on an internship, and you are clear in advance that there are no guarantees about what comes next, you can cut your losses if you feel things aren't working out. If they turn out to be awesome, you have a great opportunity to recruit someone for grad school who is a known quantity. I am also finding that many highly talented students like the opportunity to try out working in a research/academic environment without having to make an immediate commitment to graduate school.

Internships can also be relatively inexpensive, though not all organizations have ready funding to support them. Startup, summer salary, and departmental/university funds may help with this. Some universities even have special pools of funding available to support internships from underrepresented groups.

  • 2
    "(...) students like the opportunity to try out working in a research/academic environment without having to make an immediate commitment to graduate school" -- I did that and end up staying. I woulndn't have applied for grad school otherwise as at the time I wasn't sure I'd like it (right now it's a love-hate relationship).
    – Daniel
    Commented May 28, 2020 at 10:56
  • 4
    This might be the only answer (so far) that suggests a specific practical solution.
    – Isaac
    Commented May 28, 2020 at 11:21
  • A donwside of this approach is, that you might lose talented students, which don't want to waste time as an intern but start right away.
    – usr1234567
    Commented May 28, 2020 at 12:47
  • I like this idea, thank you. I got one of my students this way and they liked the lab, work, and how I do things, and have chosen to do a PhD with me. The person is actually someone I might not have picked otherwise as they do not have excellent grades (which are important for funding reasons in our situation), but their other qualities are very well-suited for graduate work.
    – Nemo
    Commented May 28, 2020 at 12:49
  • 1
    @usr1234567: Why would the internship be wasted time? There is nothing to prevent these results from going into the thesis. (Where I am, the largest difficulty would be to get funding for the internship - having a probationary period for the PhD job contract would be the more usual solution) Commented May 28, 2020 at 12:53

This is going to go into a bit of a different direction than most of the other answers, but ... have you considered that your expectations may be inconsistent?

You say that you

encourage them to only work 35 hours/week

Yet you also discuss "startup pressure" and "top students", indicating to me that there still is the (quite normal) expectation of excellence and going above and beyond in your research group. So is it possible that some of your students are freaked out because not only do you want them to compete with the best of the world, but do so by putting in substantially less hours than their peers elsewhere?

The gender difference you are seeing may be (and I am purely speculating here) that different people deal differently with such an inconsistency - some will bang their head trying to make inconsistent expectations work, while others will silently ignore one part of the medaillion (e.g., tell you that they are really relaxed and working only until 5PM, while secretly working on the weekend).

The reason why I bring this up is because I see a lot of the same behavior here in Sweden. Work-life balance is valued very highly (which is good), but at the same time people are not willing to accept that something has got to give if you expect staff to work less. If you compare your students with students in labs with less healthy work-life balance expectations, can you point your finger at some specific, time consuming things you do not expect from your students? Otherwise, this is where the problem may be and what you need to change.

And the sad news is that there simply may not be a good change that you could do. Many expectations on students are things you as a lone PI will not be able to fix (e.g., you can't waive coursework for them, or reduce the minimum requirements on a dissertation even if you wanted to), and even if you could it may still be to the detriment of your students overall (e.g., even if you are ok with your students working less and not competing for top publications, these students would end up in a dramatic disadvantage once on the job market).

Circling back to your titular question:

Should I avoid admitting female students for now (new PI question)?

No. You should revisit what the reason is for your students to get mental health issues (gender-independently), and see if this can be fixed. If it cannot, you are in the same boat as most of us - most of us have accepted that grad school is, and will for the forseeable future, remain a stressful and at times mentally taxing experience. You should still keep fighting the good fight whenever you can, but remain aware of what the realities today are. And then select for students that have the necessary mental state to deal with grad school as it is today, not what you think it should be (gender will again not really be a factor in this consideration).

  • 5
    FWIW, 35 hours/week is the official work week in France.... Commented May 28, 2020 at 10:56
  • 2
    The OP refers to pre-existing conditions, not to conditions developed during the PhD.
    – user117109
    Commented May 28, 2020 at 10:59
  • 1
    Yes, conditions were pre-existing. "If you compare your students with students in labs with less healthy work-life balance expectations, can you point your finger at some specific, time consuming things you do not expect from your students?" - my students are fully supported financially (quite a bit more than local set minimum) and are not expected to TA as part of their stipend, which most are in my location.
    – Nemo
    Commented May 28, 2020 at 12:39
  • 2
    "select for students that have the necessary mental state to deal with grad school as it is today" - how?
    – Nemo
    Commented May 28, 2020 at 12:42
  • 3
    @Nemo The TA thing is good, but I should warn you that "I pay better, so I can expect more" is a fallacy (unless students in other labs have to work side jobs to support themselves). And regarding how to select - if I knew how to do this consistently I would be a much happier prof. What I do now is tell people early and with force in the interview process that it will be stressful and challenging in many different ways, and that they will sometimes work beyond their expected hours. The reaction to this is often telling, I feel.
    – xLeitix
    Commented May 28, 2020 at 16:15

I'm in math, in the U.S., at an R1, and have for many years tried to be a good advocate for diversity, equity, and inclusivity.

Yes, in my several decades of experience observing (and managing) our grad program in math, I have certainly seen that onset of mental health issues is very significant for our grad students. No easy fix.

On the whole, our women have been more mature than the men, which is consistent with accepted appraisals of development.

Yes, in the U.S., women are not traditionally as discouraged from emotional expression as men. In my experience, the would entirely account (and more) for differences in observed expression of distress.

It may be statistically that the young people going in to math in the U.S. have a different psychological pattern than in other disciplines... but, in my several-decades observation of them, the women are generally more mature (if not necessarily more "happy").

So I do wonder about the interpretation of the events... considering my own experiences over some decades...

EDIT: Following @cag51's suggestion, I mean to suggest not changing admission policies in this regard until you have a bigger sample, as well as maybe seeing how things play out over longer periods of time. Adding yet-another platitude to my previous remarks, but which is very relevant here, the early-to-mid 20s is a very tumultuous time of life for many of us, and the person at 25 may be very different from the person who "went into that tunnel" at 22.

Also, in my observation, international students from many locations have, in effect, been "pre-filtered for toughness", by the system in which they grew up. There is somewhat less of that in the U.S. (although I am not competent to judge the situation for many traditionally-under-represented demographics...) Some of my colleagues have explicitly expressed preference for non-U.S. students, because "they don't complain". In my opinion "not complaining" is not a fundamental virtue, nor is "toughening people up"... by being a little mean or needlessly judgemental.

So, again, I'd not change admissions policies for now.

And, to be clear, some people who seem to have completely collapsed at one point do manage to get things together and do a good PhD and so on, while some people never seem to have issues but just don't finish. The real problem for me is that it is difficult to anticipate who is who... so some "gambles" seem to be necessary, to "give people a chance".

(And, again, this opinion is only based on anecdotal experience, rather than formal studies, ... as I've been a grad advisor and now-and-then Dir Grad Studies in Math, as well as paying attention to Diversity-Equity-Inclusivity for some decades...)

  • 11
    In my subjective observation, the women in graduate programs in math are overwhelmingly more mature as a population than the men - so much so that I've concluded that women with lower levels of maturity are discouraged from studying mathematics in a way that men with lower levels of maturity are not. Commented May 28, 2020 at 1:13
  • 3
    Thanks for your comment an for sharing your observations, there may indeed be factors that lead to different psychological profiles by discipline. (The males are from varied nationalities and seem comfortable sharing their stressors/doubts/imposter syndrome experiences, so I don't think it's an expression difference.)
    – Nemo
    Commented May 28, 2020 at 2:06
  • 5
    I don't think you can "explain away" all of Nemo's observed gender differences as differences in expression (implying that the prevalence and severity of the underlying psychological issues is roughly comparable between the genders). Becoming "entirely dysfunctional" goes beyond a mere expression and would be visible in men as well, no matter how well they have learned to not show emotions. Commented May 28, 2020 at 10:47
  • Re: Edit: True, perhaps I must just be patient and see how things evolve. I do note a difference in student attitudes / resilience when comparing notes with a colleague in an engineering discipline - I don't know if they are suffering just more quietly, but they seem to expect to work harder and longer, and they deal better with adversity.
    – Nemo
    Commented May 28, 2020 at 22:05

Extending on the aforementioned answers:

Between yes and no.

You should not implement policies that exclude a gender. But if you filter out certain quantifiable and problematic traits, and those simply happen to be more common in females, keep to those filters.

While it is easy to abuse my previous sentence to hide intentionally gender-based filters, the fear of such abuse should not result in abandoning those filters. To give an example: I once worked in a company that assessed certain mathematical problem-solving skills. In case of that company, it resulted in 80% of the workforce being male. As the assessment was relevant to the task at hand, it was ethical to conduct.

So your task is to find out what traits, that seem to correlate with gender, cause the problems, and how to assess them.

  • Yes, I think I need to design some kind of relatively objective selection process that can select out people with issues neutrally, even if the outcome is imbalanced.
    – Nemo
    Commented May 28, 2020 at 21:01

I can't comment, so I can't ask: do you have solid mental health support yourself?

This arrangement sounds incredibly stressful. In addition, you're taking on the burden of others' mental health.

In such a case, I would strongly recommend taking some time to talk to a mental health professional yourself. They can help you get and stay healthy and give you a safe place to release some of the pressure that you are under.

Without making judgments, you're probably putting some of this pressure on yourself. There's also a good chance that some of these behaviors are triggering a reaction in you that probably don't need to be triggered. Some childhood upbringings can make folks highly susceptible to other people's moods. The parenting styles that you were exposed to will translate into the management styles that you are using.

There's really no "good" or "bad" here... rather, look for a solution that is effective, fair and sustainable. The first place to look for that is within.

  • 2
    This is a good point. I have some excellent mentors myself and am well supported by my colleagues, but I am in a difficult career stage. I'm not sure it has anything to do with childhood, but maybe personality that I have difficulty not being upset myself when people I am responsible for are in distress and I can do nothing for them, and they continue to make choices that make things worse for myself and others. Some of these people have no business starting a graduate degree with the unresolved problems they have.
    – Nemo
    Commented May 28, 2020 at 16:36
  • 1
    I'm glad to hear you have good support. My experience with therapy isn't the classic "now... tell me about your childhood", but it proved enlightening. I'm an objective scientist by training; therapy helped me understand my blindspots. It also helped me process emotional triggers more effectively. Previously, I "processed" emotions by dismissing them as irrational distractions. Learning how to hold your own emotional state with self-respect and self-compassion is incredibly effective whenever you're working with people. Much of this stems from what you learned as a child observing adults.
    – Dancrumb
    Commented May 28, 2020 at 18:20
  • 2
    An anecdote, but I'm in the process of leaving grad school because my supervisor cared much more about my 'well being' and success then about the actual research I was doing. When I started protesting the I'm unhappy and we should talk more about the research he was really baffled because in his mind he was doing everything in his power to make me 'happy'. I think if he would cared less about me and my happiness things would have end up differently. Sometimes less is more.
    – tom
    Commented May 28, 2020 at 20:32
  • 1
    @tom Well, you don't hear that everyday! I had a mentor that did an excellent job adapting to each student's needs, which can be very different. I hope to learn how to do that, but people vary so much it's really not obvious.
    – Nemo
    Commented May 28, 2020 at 20:40
  • 3
    @tom, you might want to google "Ruinous Empathy" and see if you recognize that :)
    – Dancrumb
    Commented May 28, 2020 at 22:24


  • Gender is the wrong surrogate for "Do I want to hire this student?", not only for ethical and legal reasons, but also because it would prevent hiring women without psychological issues, i.e. the group in gender x mental health that is lacking in the group.
  • Not everyone does equally well in all aspects at spotting whom to hire or not. You may have a blind spot for recognizing female students with psychological issues - or a bias in favor of them.
    Also, having a hiring committee of several people and/or standardized/structured questionnaires may help.
  • There may be a several possibilities available within the work-health system at OP's university/in OP's country to spot, mitigate and deal with the observed issues.
  • Psychological issues are difficult, and you cannot expect to produce miracles.
  • Networking: I'm sure you are not the only fresh PI struggling, and a peer group may help.

Long version:


As someone already commented, if you have a reputation of being nice to students with problems (or to female students with problems), your applicant pool may already have more than your "fair share" of students with problems.

But different people differ in how prominently they recognize different traits in others. And this may very well lead to varying bias in hiring decisions. Bias in the neutral sense of the outcome being on average hiring people with certain traits. Depending on the trait, we want

  • a favoring bias (e.g. students who do well),
  • a disfavoring bias (students who are in danger of becoming ill from that particular type of work, physicall or mentally doesn't matter; or students who do not have the required field-specific knowledge for sucessfully doing a PhD on that project),
  • or no bias (gender, unrelated health issues/disabilities)

Actions that OP may consider taking:

  • find out whether she happens to be

    • good at spotting male students with psychological issues (does recognize as not suited to do a PhD ...)
    • biased against male students with psychological issues (... and correctly not hire)
    • blind for female students having psychological issues (does not recognize)
    • biased in favor of female students with psychological issues (... or hires nevertheless)

    and then work to improve their decision making process

  • have a hiring committee (2n eyes see more than 2)
  • Work to develop/get a structured questionnaire that helps to arrive at good hiring decisions.
  • Check for and work against unrealistic expectations on the side of the student already in the interview (see below).
  • Where I am, there is a probational period also for students hired on a PhD position. This should be used by both sides to test whether student, group, supervisor and project are a good fit.

Seeing the overall numbers of mental health issues in grad students, I'm afraid the problem will not go away any time soon, though.

Small sample size

Strange events permit themselves the luxury of occurring.

OP cites anxiety issues in 43 % of cis women and 34 % of cis men. For a group of 4 f + 4 m students (8 sounds a sizeable group to me for a early-career PI). That alone means a probability of 4 % to have at the same time 0 male students with anxiety and 3+ female students with anxiety. Also, of the two events 3+ out of 4 females with anxiety and 0 out of 4 males with anxiety, the 0 males are the more unlikely event under the cited probabilities.

At the same time it is of course good to get alert before you can show that there are statistically significant problems in the group.

Work health

It is perfectly OK to hire based on expected performance. If a disability or disease means the student cannot perform well (or at least on an average level) in the position it is OK to not hire them (no blind busdrivers). For mental health/psychological issues the line between objective inability to perform well and discrimination is very difficult, though.

In addition, there is the question of students developing health issues during their PhD. If a particular job would endanger this student's health, they should not be hired for that position.

Where I am, a prospective employer can (and often must) send the prospective employee to a medical examination to check that the employee is fit for the particular job. There are also regular checks later on. This may include pychological components of being fit for the job. The medical examination is confidential and the employer gets as answer only "OK", "cannot work this job" or "needs accomodations x, y and z".

While none of the examinations and check-ups I've had so far included psychological checks (at least I didn't recognize them if they had...), psychological issues among postgraduate students are anyways a well-documented problem. so maybe it's anyways time to include them:

As employer, OP can and should talk to work health services about work-related health issues she's concerned about and ask them what she can and should do.

Maybe sending everyone (no discrimination) to regular work health check-ups would be an option?

Some more thoughts:

These students become overwhelmed,

To some extent I think this experience is part of the normal experiences when growing up [professionally as a researcher].

  • Personally, I found it extremely helpful when colleagues/supervisors told me in actual clear words that the PhD is an extremely stressful time for most students - including their own PhD experience. That at least avoids the additional stress of thinking one is the only one affected like this.

  • I have met many students in what I call their "mid-PhD-crisis" when they realized their PhD thesis won't "rescue the world".
    IMHO, this problem may be greatly worsend by the sales pitch of the project creating totally unrealistic expectations about what is doable in the particular scientific project, what impact the work will (not) have, and the likelihood of failure/that research consists of far more finding out what does not work than what finally does work. Part of this may be that students will not recognize a sales pitch by a professor - after all, they've so far met professors only as teachers telling them reliable truths (almost always), but never in the position of someone selling an idea.

    I had a particular enlightning experience that showed me how optimistically biased I still was after 10+ years of professional experience so at a time when I had already told these "trade secrets" about realistic goals to many students: somenone from a funding agency for applied research told us that they fund only projects that they judge to have less than 20 % chance of success (!). (Higher probability of success -> no public money needed, industrial development project) These were the projects we were doing all the time, and the inner-academic point of view was that they are almost always successful!

emotionally volatile,

This sounds to me like unacceptable behaviour? You may have to tell some people that they are expected to behave professionally at work. You also have a responsibility to provide an acceptable working environment for the other members of your group.

Even a recognized disability or disease in the range of what still allows one to work excuses only so much misbehaviour (the same behaviour with a known and recognized reason is much easier to bear for the others. Think of a meltdown in the office by someone with a recent loss vs. a situation with a sneaking suspicion that there may be a tactical component).

Most students will cope with both points. But I think they do contribute to the stressful experience which proves too much for some - do you may want to avoid them.

keep changing their ideas about what they want to work on (even though their funding is connected to lab projects)

This may be a symptom of a fundamental conflict:

  • On the one hand, the thesis is exam-like in that it must be the student's very own work, and they will be judged on how well they do it and on what extent of supervision they did (not) need (PhD thesis prooves the ability to perfom longer-term research in a self-reliant way).

  • On the other hand, project funding and employment relationship put them into a subordinate position and legally speaking they must do what you tell them.

This conflict, in particular if the project does work out well (see unrealistic expectations) may create the symptom of a student trying to "break out" with their research topics when the supervisor tries to nail them down on the prescribed topic, without any mental health issues at all.

jealous of other students' progress

To me that's very much in the undesired behaviour category above, but I think it may help the whole group to make clear that a very few successes only once in a while is to be expected. (Wondering what happens if breakthroughs become celebrated group events? Obviously, either on cake that the successful one brings, or on group/project money)

become entirely non-functional at intervals.

over here, that should trigger your employer duties of care for the employee. In other words, you'd have to send them to a medical doctor or the work health services since there may be a medical problem.

I scramble to adjust their workload for example hiring undergrads to help with their data collection or doing it myself,

Please don't micromanage.

For one thing, for some this can create tension and stress them if you take away their proper work (see conflict between thesis and employee above). It can also rub in that they do not perform adequately.

Others don't have any issues with this, and will be happy that they can train you to do their work. Which is not what you want.

(There is nothing against doing a bit of your own research if that's what you like to do, but please don't pirate your students' projects)

investing hours in counseling and reassuring them, problem-solving with their committee/co-supervisors,

Management of your employees/students is your main job as PI. Treating their health issues is not.

losing sleep when I know they are feeling miserable.

not going nuts myself?

I feel ill-equipped to provide the needed level of psychological and emotional support for this kind of high-anxiety student

It is your job as employer to recognize that your students have problems with their job. If we are talking actual mental health issues, though, your job ends where you send the student to work health services. After that, you have to respect their privacy. And you anyways cannot have a professional employer - employee, supervisor - student and psychological counselor - patient relationship at the same time.

How can I [...] live up to my ideals

It may be that some ideals will have to go. I don't think gender equality is one of those, though. But I suspect you are overestimating what a supervisor and organizational measures can achieve.

The linked paper reports

  • high correlation between bad work-life balance and mental health issues. Of course, bad work-life balance may cause mental health issues. So it is good that you do not encourage unhealthy work practices. On the other hand, bad work-life balance may also be caused by pre-existing psychological issues. In the latter case, one cannot expect measures to restrict working hours to work miracles.

  • strong, supportive and positive mentoring relationships between graduate students and their PI/advisors correlate significantly with less anxiety and depression.

    Again, bad relationship may cause/worsen mental health issues - but such a correlation can also be caused by an underlying mental health issue that prevents the forming of a good professional relationship between student and PI. And in that case, there is again only so much that you can do.

Networking/PI Peer group

Except for the size of the problem you describe, they don't sound all that unusual to me.

I'm sure there are sufficiently many fresh PIs around that would be a suitable peer group to share problems, ideas, expertise and experience. Such a group may exist already. If not, maybe it's time you start one?
(Unrelated field, related technique: I'm in a "self-help" group for freelancers that is one of the activities of my professional society. The group is small (< 10 people from all over the country), and works under a strict confidentiality agreement. For a while, we further organized ourselves in subgroups of 2 - 3 partners that mentor each other.)

A similar idea would be to look around whether there is mentoring for fresh PIs. Or even ask some experienced professor whom you trust whether they'd agree to mentor you.

  • 1
    Thank you for your detailed and thoughtful response.
    – Nemo
    Commented May 28, 2020 at 20:54

The situation the OP describes goes way beyond different behavioural patterns in male and female students. I am not convinced that gender/ sex is the defining factor but I can only speculate for the reasons, let alone any suggestions.

I would focus much more on any pattern in the topics selected by male/ female students, and if there is a discernible connection to their CVs or background. It is plausible that the OP's area of interest attracts students of a certain background who pick certain topics, and that association revealed itself particularly strongly in those particular female students. In fields like psychology, for instance, a prime motivator to engage in research on a specific area are personal circumstances. I am not suggesting in any way a violation of someone's privacy or gossiping behind a student's back, but keeping an eye out for similar situations in the department or elsewhere, discussions in a common room, press articles and academic papers and most of all any interactions with the students themselves might reveal a lot.

It could also be accidental and the OP is simply observing a bad run - it is impossible to tell without knowing the number of cases. If there is a trustworthy colleague with experience in supervision, or an appropriate point of contact, a confidential discussion might help. I understand the OP's reservations and it is difficult for a young academic, but the school will notice if students start abandoning their PhDs, and the situation might be quite worse.

Stopping hires of female PhD students is out of the question. It is much more likely that the OP has some sort of bias (in the statistical sense) in the selection process that needs to be corrected, or a pattern has not been acknowledged.

  • 7
    I think the bias is more likely to be in the applicant pool rather than selection of people from the applicant pool. Perhaps the asker has a reputation for being a great mentor for students who have these challenges. Commented May 28, 2020 at 0:16
  • 3
    I'm inclined to suggest that OP go back and reread her students' application materials, looking for significant differences (other than gender) correlating with the observed differences in performance and mental health. If such differences can be found in the applications, they might be useful for assessing future applications. Commented May 28, 2020 at 0:24
  • 1
    @AndreasBlass Absolutely, that is my suggestion as well, and hopefully they are still available
    – user117109
    Commented May 28, 2020 at 0:25
  • 1
    Thanks for your comments. It might indeed be area-related, in that the males might in general come from a slightly more technical background, but it is not entirely consistent - the only consistency is a higher general level of anxiety, which seems also very unethical to try to screen for. I did mention my observation to a trusted senior mentor, who said they had observed something similar though to a lesser extent, but also said males tend to have a different set of issues (they did not elaborate).
    – Nemo
    Commented May 28, 2020 at 0:52
  • 2
    @Nemo In particular, I suggest you look carefully at the letters of recommendation: do these point towards (or omit pointing away from) signs of trouble?
    – avid
    Commented May 28, 2020 at 4:56

As you said it yourself in your own question, I'll just agree with you that discriminating based on gender is a terrible idea (not to mention that it is also illegal). But, if I take your problematic description (taking any gender out), I get this:

Almost all of the problematic students have turned out to have pre-existing mental health issues (anxiety, depression, and in one case PTSD). These students become overwhelmed, emotionally volatile, keep changing their ideas about what they want to work on (even though their funding is connected to lab projects), sometimes become jealous of other students' progress, and in some cases become entirely non-functional at intervals.

Now, the first part, the pre-existing mental health conditions should not be a problem all in itself (and discriminating based on that is probably also illegal, unless there are specific reasons to require somebody with no such conditions). The problem you are having is that they:

  • become overwhelmed
  • become emotionally volatile
  • keep changing their ideas
  • sometimes become jealous

These are bad characteristics for any PhD candidate, and your interview process apparently does not do a good enough job of screening for those characteristics. What you want instead is somebody who:

  • can perform well under pressure
  • can judge situations rationally
  • has independently developed a research idea or a larger project through to the end
  • works well in a team

All of these characteristics have been assessed at interviews at all the levels of academic positions I have every applied for (from PhD to Lecturer positions). There are fairly standard interview questions for that (i.e. "Describe a challenging situation you encountered in teamwork, and how you came to a resolution").

Finally, I'd like to say that I think it's a great thing you're asking this question. You have actually identified the problematic PhD student characteristics fairly well (the above bullet-points), and then your unconscious bias stepped in. The best way to correct such behaviours is by acknowledging your unconscious biases and actively working to overcome them. As this experience has (I think, understandably) left you with a bias towards female PhD candidate, may I also make the following suggestions:

  • make sure you re-focus your interview criteria to identify the above (un)desirable characteristics
  • but have a second or even a third opinion when evaluating the candidates, to ensure that you are evaluating the candidates based on their answers and evidence, and not your assumptions
  • Thank you, I like that there are practical points and suggested interviewing strategies. I fear it can be easier for people to give the right answer than actually behave optimally, but perhaps if you put them on the spot a bit there will be a difference in genuine vs. embellished responses.
    – Nemo
    Commented May 28, 2020 at 20:57
  • Yes, I guess assessing what you want through a limited time of an interview is always a challenge. The types of questions I am suggesting usually aim at recalling a past experience, and I that makes it a bit easier to judge if it's genuine. It can also give you a valuable hint about the person if they say they had no such experiences in the past (Q: "Describe a situation where you were in a disagreement, and how you handled it?" A: "Oh, I never get into any disagreements" -- might give you a hint about the interviewees perception of themselves)
    – penelope
    Commented Jun 4, 2020 at 11:24

The answer to the question in the title "Should I avoid admitting female students for now?" is simple and almost obvious: Yes!

The reason is the following:

1) You claim as an empirical fact that the probability that a female grad students is going to cause you trouble is very high.

2) You claim that you want to avoid this trouble.

Thus, assuming we accept 1), and you certainly do, then the answer is clear: yes. Avoid admitting women for now.

Clarifications: This is a question of values. You have your values, and others have their own values. If you ask this question whether to admit female students to your lab, it means I think that you are not entirely against implementing this strategy, hence you can live with such a decision, value-wise. Therefore, there is almost only one logical consequence in this story: do what helps you.

Comment: my answer is a formal answer to the simple question. I do not promote doing anything, or not doing anything, or any specific value system. I simply answer the question asked based on the most rudimentary logical reasoning. I am also not a law official, or an ideological cop. I am not in charge of telling people what to value, or what is the law. They can do whatever they want as far as I'm concernend. I can only warn them if they are going to get in trouble if they do something dangerous. But here, if the OP conceals the fact that she refuses to admit women to her lab, she will not be in trouble as far as I am aware of.

  • I think it is worth it to mention that following your advice can put the OP into deep legal issues.
    – user35129
    Commented May 29, 2020 at 19:42
  • 2
    As I have explained explicitly in the answer I don't expect this to be the case if she does not make her decision known to anyone. Of course, we are not law officials here (also noted in the answer).
    – Dilworth
    Commented May 29, 2020 at 21:20
  • 2
    Legal issues seem highly unlikely unless I put it in the job posting, which would be pretty stupid. Further comment in edit to original post as requested by moderator.
    – Nemo
    Commented May 29, 2020 at 23:45
  • 3
    @paulgarrett, I respectfully disagree of course :) I believe our goal here is not to serve as ideological or legal police. We need to apriori believe the premise, and even if we don't believe it, act as if we believe the premise given, and answer based on this premise. We should avoid preaching to the OP, or telling them what is right and wrong (using our own subjective values). I agree that legal issues should be discussed, but here it is obvious that there is no danger because the decision will not be made known (explicitly mentioned in my answer).
    – Dilworth
    Commented May 30, 2020 at 13:44
  • 1
    There is a statistical complication here that you are not dealing with. If we assume that the observed correlation is true, this gives a possible basis for "statistical discrimination" but only if the available information does not include individual information that renders that correlation moot. In a hiring process there are usually avenues for finding detailed information on candidates, so is it possible to get individual information on new candidates that would render the previously observed correlation moot? Anyway, +1 for giving an answer that goes against the grain.
    – Ben
    Commented Nov 19, 2020 at 1:36

You can have the desired result without doing what you proposed.

Before accepting a student(both m and f) explain to candidate that people who struggled had this list of characteristics (do not cause problems for your existing students by providing too much details!).

Explain to candidates that it is in their best interest to be honest with you regarding if they think this is a path for them. Nonproductive collaboration is not in anybody's interest.

Some candidates will lie, but I believe most people are honest if you discuss stuff like this with them in private meeting.

  • 3
    But what if someone reveals a condition such as generalized anxiety disorder and I choose not to hire them because of that? Is that not also problematic ethically (and professionally)?
    – Nemo
    Commented May 28, 2020 at 17:10
  • @Nemo: "not to hire them because of that" yes, that's unethical, unprofessional and in many legislations also illegal unless you can prove that a generalized anxiety disorder implies either that the student cannot possibly do their job or that the job poses unavoidable and particular health risks for someone with generalized anxiety disorder. Commented May 28, 2020 at 18:32
  • This is the fine line other responders are approaching in different ways. It's interesting to see the range of opinions. I guess a case could be made that people with high anxiety levels can be harmed by graduate school, but no one would want that decision made for them - at the same time, people have to choose to hire those who are going to be more likely to be successful.
    – Nemo
    Commented May 28, 2020 at 21:04
  • I found another interesting thread on the decision to take students with a mental health issue: academia.stackexchange.com/questions/42944/…
    – Nemo
    Commented May 28, 2020 at 21:16
  • 1
    @Nemo I personally would not consider that unethical, as for legality I am not qualified to judge. As to why I do not consider it unethical: you are probably saving yourself and potential candidate a lot of pain, actually unethical thing to do is pretend you do not have a way to avoid it and do the easy thing Commented May 29, 2020 at 5:52

...nor can I run a lab and attend to my other duties if half of my students are taking 90% of my time and are making limited progress.

I think you are looking at this the wrong way. Rather than ceasing to hire females as new students, it would be much better to deal with the existing female students that are giving you problems. There are requirements on all students to meet the standards required of their program, and if you think they are not doing this then there are mechanisms to put those students on notice for unsatisfactory performance, and ultimately remove them from their program if they cannot meet the required standard. The starting point for this is to put problem students on notice of their unsatisfactory performance and let them know what is required to improve to meet the requirements of their work. PhD candidates have regular formal reviews that act as checkpoints on their candidacy --- if they get an unsatisfactory review then this usually requires them to improve up to standard by the next review or they can be removed from their program.

If you deal properly with your existing disfunctional students then one of two things will happen: for each of those students, either the student will improve up to the required standard of work (in which case your problems will diminish), or they will have to leave their program (in which case your problems will diminish). Either way, you should be able to improve the functioning of your laboratory, and you will free up a lot of the time you are presesently using to deal with disfunction. It is likely that your present proposal (to avoid accepting any more females) comes out of stress and fatigue at dealing with difficult students, and ironically, you may be about to make a bad decision yourself due to your own stresses. If you remove or diminish that problem, you will be able to look at things with a clearer perspective.

As to the goal of hiring good candidates in the future, it would be worth making a greater effort to assess each candidate's past work history, since if they are difficult in your lab then they may also have had a difficult history. You might be able to successfully weed out bad candidates if you are more inquisitive with their referees, but obviously this is something where there is no infallable method of selection. For many reasons, I don't agree with your proposal to simply exclude female candidates. (One reason to bear in mind is that this is likely to be unlawful,* but even setting this aside, it is not a good proposal.) Instead, try to develop good review methods during hiring so that you take on candidates with a proven track-record of working collegially in lab settings even when they are under stress. You can also use the candidature review mechanisms available to you to ensure that new candidates meet program requirements.

Try to bear in mind that you owe a duty of fairness to new female candidates, just as with all other applicants. They need to be assessed on their individual merits, as best as these can be ascertained from the available information. What you are proposing here is a species of what is called "statistical discrimination", where you are using an observed statistical correlation (females in your lab have more mental-health problems) to make an inference about other females outside of the observed group. This kind of discrimination can certainly be rational in cases with highly limited information. However, in a hiring process you have plenty of avenues to obtain detailed information on candidates, and so this specific individual information usually over-rides outside statistical correlations. (In statistical parlance, the conditional statistical correlation, conditional on individual candidate information, becomes low, and so the inference is then weak.) It is likely that you can obtain information from candidate referees that will identify good and bad applicants based on their past education and work history.

  • Depending on your jurisdiction, it is important to note that anti-discrimination laws generally prevent sex discrimination in hiring and acceptance to educational programs, other than in a narrow set of exempted cases. There are also exemptions for "special measures" (i.e., affirmative action) in most legislation, but what you are describing does not sound like it would fall within an exemption. Consequently, if you were to cease hiring female candidates, it is likely that this would be unlawful, and the university could be sued for it.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .