What is your advice on how to select a graduate advisor? What are the
must have,
good to have,
must avoid, and
"run, disaster ahead"?
traits in a PhD advisor? How does one get this information? It would be good if you can cite specific real world examples, especially your own experience.
You need three broad pieces of information, and all of them can be hard to get for one reason or another.
First, you need someone compatible with your learning style. The problem is, you may not know your learning style, even into your twenties. And assuming you do, there's little more anyone here can say except, "Try to match it." (How do you find out? Ask members of this person's research group.)
Second, you need someone who isn't outright abusive. Abuse can take many forms, by the way, it's not always as clear cut as demanding sexual favors or demanding unpaid work outside the university. It can be ongoing neglect. It can be habitually throwing you under the bus at group meetings. It can be constant disrespect in front of peers. It can be transparently treating you as a game piece on the way to their next goal. It can be abusing your trust or confidence. It can be credit stealing or playing games with authorship. It can be fobbing lecture duty off on their grad students. It can be a lot of different things.
You would think that would be easy to detect, but it isn't always.
I was hypervigilant for this, and still got trapped. (I theorize now, because everyone else was keeping their heads down, trying to get out. But I'll never know.)
Third, you need someone who has demonstrated that they can regularly move students through the process to either a master's degree or a PhD. This, you at least have a chance at. If you see someone who has a regular stream of graduates, that's a good sign. If you don't, that's a bad sign. A new-ish faculty member with two graduated PhDs and three in the pipeline is very different from someone who's been there 15-20 years with only two graduated PhDs. If possible, you want to look at the ratio of successful candidates to candidates. In some departments that will not be easy to get.
I like most of what Anonymous said in their answer, but I think there is even more.
In addition to Anonymous' points:
All that is to say, there are many, many things to consider, and much of that is hard to find out in advance. The best strategy is to (a) look very carefully at what your advisor has been publishing recently (assume that you will have to write papers similar to these), (b) try to talk to current or previous students of this advisor, and (c) remain flexible. At the end of the day your advisor will have characteristics that you don't love (just as you will not be the "perfect" student for your advisor). Ideally, both sides make small compromises on the way to a successful graduation.
P.S.: because of the need to make small compromises I would not recommend being "hypervigilant" (as Anonymous says in their answer) - be critical when selecting your advisor, but once there I would give them the benefit of doubt. Constantly being on the lookout for issues and escalating each disagreement is a difficult way to actually finish - in a 5-year close collaboration, you will get into arguments and you will see things differently at some point. Your advisor is also human, they will make mistakes, choose their words poorly, or have bad days. Resolving these conflicts constructively is a superpower, not only in academia but also in any other professional endeavour.
Try to talk to the grad students and post-docs who are currently in the department. Find out what they think of the other profs besides their own. Do ask them about their own but find out what the cross views are. See if you can get some views from people not directly affected by the prof you are thinking about.
Also, get them to "dish" on the other grad students and post-docs and such. You will be working nearby most of them for much of your degree. You are going to want to know which ones are helpful and which ones you need to avoid if possible.
For example, if you are looking at a prof doing XYZ, and the department also does PDQ that is quite far from XYZ, wander over to the PDQ students and ask them "What do you think of the other profs?" For me it was physics. And the department had a big particle physics group and a big laser physics group. The laser physics group was more than happy to "dish" on the profs in the particle physics group. They got a little more quiet when it came to their own profs.
Find out where students of this prof have wound up after graduating. I went through physics. There is a magazine called Physics Today that lists recent grads and where they got jobs. Check out the equiv for your subject area. Find where various profs got their students jobs. Or post docs or whatever. (Your uni library will most likely have some guidance on this question.)
So one prof when I was a PhD candidate graduated three people with PhDs during the last few years of my grad work. Two got post docs at major labs, CERN for one and FermiLab for the other. The third student got a post doc at a major US university. If you were doing particle physics experimental, that would probably be a big bright glowing indicator of promise.
A prof in another area (laser physics) also graduated three students. One became a high school teacher, one got a job at Ontario Science Centre. And the third was not listed as having a job. So you would have a different set of evaluations for that prof.
Choose someone who has the reputation for being a good mentor, providing well for their group, and for creating a healthy working environment. Graduate research is highly uncertain and can be extremely stressful. So, a constructive environment and stable support are very important. Your graduate advisor can also serve as a mentor and champion throughout your career.
First, do not judge by first impressions as these can be very wrong.
Second, you should feel comfortable talking to the person. This person is helping you develop into a researcher, you should not feel nervous knocking on their door and going to speak to them about some problem.
Third, you might want to check if the supervisor has a good track record with training students who go on to get permanent jobs in academia, rather than people who perhaps make it as far as postdoc and then no further than that. Some other people are suggesting that if the supervisor gets a lot of people to postdoc level than they are a good supervisor, but I think you should look longer term than that, as getting a postdoc is not as difficult as it used to be.
This answer should be considered together with the other answers on this page. Based on my own experience as a PhD student in an engineering field, here are some of the most important traits that a PhD supervisor should have:
1. Collaborates at a higher rate with their PhD students than with colleagues at their own level.
This, in my opinion, is by far the most important metric. What this trait means is that the PhD supervisor is able to consistently publish work with their own PhD students rather than with other professors or postdocs at their level. I've found that PhD supervisors who mostly publish with colleagues and not PhD students are often difficult to work with (e.g. does not provide reasonable guidance and/or micromanages you).
A possible reason for this is that, when working with colleagues, PhD supervisors tend to suppress their unproductive qualities (such as micromanagement), since they do not wield the same power over their colleagues as they do over their PhD students. Therefore, they are "forced" to compromise with their colleagues to work more towards their and their colleagues' common objective of doing good work.
However, when working with PhD students, their unproductive qualities often show up more, which could lead to a difficult time for the PhD student. That is, with PhD students, they let their unproductive qualities stand in the way of doing good work with their PhD students. The truth is, even with PhD students, some compromise is required in order to do good work, and micromanagement will work against this.
To determine whether a potential PhD supervisor collaborates at a higher rate with PhD students, check their Google Scholar profile to make sure that they publish mostly with PhD students rather than colleagues at their same level. That is, the proportion of papers co-authored with students should be high compared to the proportion of papers co-authored with colleagues at the same level. This supports their competency at mentoring PhD students. You can use an Excel spreadsheet to keep track of this, and, ideally, the proportion of papers published with PhD students should be at least 50-60%.
2. Once it has been established that the potential PhD supervisor publishes consistently with PhD students rather than colleagues, make sure that the difference in quality between papers co-authored with PhD students and papers co-authored with colleagues is not substantial.
If the difference in quality is significant, then it could mean that the potential PhD supervisor cannot properly mentor a PhD student, since they are not able to train them to be able to publish papers of the same quality as the ones that they publish with their colleagues.
Of course, some difference in quality is expected when a PhD student is just starting their degree, but by the end of their degree, the papers that they publish with their PhD supervisor should not be that different, in terms of quality, compared to the papers published with the PhD supervisor's colleagues.
Again, this can be checked from the potential PhD supervisor's Google scholar.
3. Number of PhD students graduated every year.
A good PhD supervisor consistently graduates 1-2 PhD students every year. Of course, this depends on whether the PhD supervisor is in the earlier or later stages of their career.
4. Most number of PhD students concurrently supervised.
A good PhD supervisor should be able to handle a large number (at least 3) of PhD students concurrently. If not, then it could mean that the PhD supervisor has poor people management or time management skills. Of course, this is not a general rule, and exceptions exist.
5. Where their PhD graduates end up.
A good PhD supervisor graduates students who end up working in jobs that you eventually want to pursue. However, in my experience, I've also found that, regardless of where you, the PhD student, wants to end up, where the PhD supervisor's graduates end up can tell a lot about the PhD supervisor.
For example, if most of their graduates end up in industry jobs, then there is not much to say about the PhD supervisor, since it could be the case that they did good research with their PhD student, but the PhD student chose to pursue an industry job.
On the other hand, if their graduates end up in academic jobs (e.g. postdoc or professorship at a university), then one could argue that a necessary condition for this to happen was that the PhD supervisor had to have done good work with their PhD student for the PhD student to get an academic job.
6. Participates in co-supervision as little as possible compared to sole supervision of PhD students.
A good PhD supervisor should be able to supervise PhD students themselves, and not rely on the support of another PhD supervisor to co-supervise a PhD student. Of course, exceptions exist. However, if the PhD supervisor graduates a majority of their PhD students via co-supervision with another PhD supervisor, then it may suggest that the PhD supervisor is not able to properly mentor a PhD student alone, and they may be difficult to work with.
Also, in my experience, co-supervision is difficult to get right, since each co-supervisor can decide to provide as little guidance as possible since they assume that the other co-supervisor will pick up the rest of the work. However, if both co-supervisors do this, then you, the PhD student, will not receive enough guidance and will suffer in the end. Therefore, avoid co-supervision if possible.
7. Diversity of papers.
Although this is not absolutely necessary, it is good to see that a PhD supervisor is able to publish papers of different kinds (e.g. papers with empirical data vs. papers with theoretical derivations). This suggests that they are flexible and are able to contribute to their area using different kinds of skills that you, the PhD student, should be able to learn from them.
8. Last, but certainly not least: when making a decision on who to choose as your PhD supervisor, don't let your emotions fool you.
A potential PhD supervisor who is a good teacher or has a friendly personality may be a terrible PhD supervisor. Therefore, don't rely on the strength of your social connection to a person as a good metric to evaluate their suitability as a PhD supervisor. Instead, use the objective metrics I mentioned above.
In addition to what I wrote above, here are some pieces of general advice:
Tenure-track vs. tenured professor
If a potential supervisor has not been tenured yet (e.g. assistant professors), then they often have more motivation to perform high-quality research than supervisors who are already tenured (e.g. associate professors and above). Moreover, supervisors who have not been tenured yet are often younger, and so are more likely to be more familiar with the latest technologies/research than older tenured supervisors, who have no incentive, other than personal interest, to stay up-to-date.
