Realistic Productivity - Expectations from PhD Students (Research Assistants)

[kerprof]

I am new to mentoring and working with PhD students (funded Research Assistants). My first PhD student started 12 months back. I am in Computer Science area.

Please advise what realistic expectation I can have in terms of  number of publications (including peer reviewed conference publications) from first year, second year, and third year PhD students from their research work.

[kerprof]

I am new to mentoring and working with PhD students (funded Research Assistants). My first PhD student started 12 months back. I am in Computer Science area.

Please advise what realistic expectation I can have in terms of  number of publications (including peer reviewed conference publications) from first year, second year, and third year PhD students from their research work.

In addition to the number of publications accepted ... any guidance on number of publications submitted will also be helpful...

[Liquidambar]

This is completely field dependent.  You'd do better to talk with people in your department.

In my field (modeling related), students aim to graduate in 5 years, and a PhD dissertation corresponds to about 3 papers.  Paper 1 is probably finished after 3 years and probably published before graduation.  Paper 2 is probably submitted after the 4th year but is probably still in revision.  Paper 3 is probably mostly done at the end of the 5th year, enough to include the results in the dissertation but maybe not quite ready to submit yet.

[mamselle]

Even within computer science, I suspect that's going to vary by subspecialty, school type, etc.

Maybe look at the websites on labs comparable to those you want to set up, and/or hope to emulate, and the CVs of both the grad students and the PI(s) from those labs?

You could tell over a couple of years, say, how many first-author, non-first-author, and last-author articles there were and get a sense from those.

You could also see how many conference papers go to accepted article status how quickly; that might help set timeliness parameters.

You'd probably want to compare 3-5 labs to get a good sense of things, but it would help build a matrix of understanding, helpful for knowing whose labs are producing and at what pace,, too...

M.

[Durchlässigkeitsbeiwert]

In addition to using other groups as a baseline, I would suggest to correct your expectations based on
1) nature of the project you do (how fast results are produced)
2) departmental requirements on student's time (number of courses to be taken etc)
3) how advanced the project design is before student arrives: there is big difference between "modify and apply this tool to that problem" and "we need to do something within this broad topic"

[research_prof]

As you know, Computer Science is probably among the most challenging fields out there, where people work and compete like crazy. So, I would say productivity pretty much depends on where you want your work to gets published. If you aim for the best conference in your field, where acceptance rates may be well below 20%, then you will likely have to re-submit the same paper (or revised versions of the same work) several times before getting it accepted. In my experience, journals are a bit more flexible in the sense that you can revise your work and respond to comments/concerns from the reviewers.

My expectation from my PhD students is that they produce (not publish because that's a different beast in CS as I explained above) 1 publication that can make it to a top-tier conference and 1 more high-quality publication that may or may not make it to a top-tier venue (but it can definitely make it to a good conference or journal) per year. Now, if I have a student that produces a paper that makes it to a top-tier conference, but does not produce another paper in a year, I will definitely not be unhappy.

So far, my first PhD student who works very hard and he is a very smart guy (he is in his second year) is able to follow this trajectory with quite a bit of paper writing help from me (but that's ok). I had a couple more PhD students, who could not keep up with that pace and they either left on their own when they realized that they needed to work hard or I had to let them go because they simply would not like putting in the necessary effort.

To a large extent, it has to do with you and your own career goals. You need to be a role model for your students if you expect them to work really hard--you cannot just disappear or not spend time with them, but expect your students to work harder than you. At the end of the day, it is your research group. You need to work harder than your students.

[fizzycist]

look up a few profs in your sub-field at comparable universities who you know to have similar publication strategies. Estimate their number of publications per year where they are corresponding/senior author (not sure how easy this is to do for CS but in experimental lab science usually this is the last author), divide by the size of the research group.

Unless there is something really different about CS, I predict you will find that # of papers/student/yr << 1. In my area it is more like 0.25 papers/student/yr.

It's up to you to decide what is a realistic expectation, but if it differs from the average then you are setting yourself up for some serious frustration.

[lightning]

If the grad students are making you look good, without to much fuss from you, then it's safe to say that they are being productive.

[research_prof]

look up a few profs in your sub-field at comparable universities who you know to have similar publication strategies. Estimate their number of publications per year where they are corresponding/senior author (not sure how easy this is to do for CS but in experimental lab science usually this is the last author), divide by the size of the research group.

Unless there is something really different about CS, I predict you will find that # of papers/student/yr << 1. In my area it is more like 0.25 papers/student/yr.

It's up to you to decide what is a realistic expectation, but if it differs from the average then you are setting yourself up for some serious frustration.

0.25 papers/student/yr in your area? Are you sure you are not in engineering? :-P

[kerprof]

As you know, Computer Science is probably among the most challenging fields out there, where people work and compete like crazy. So, I would say productivity pretty much depends on where you want your work to gets published. If you aim for the best conference in your field, where acceptance rates may be well below 20%, then you will likely have to re-submit the same paper (or revised versions of the same work) several times before getting it accepted. In my experience, journals are a bit more flexible in the sense that you can revise your work and respond to comments/concerns from the reviewers.

My expectation from my PhD students is that they produce (not publish because that's a different beast in CS as I explained above) 1 publication that can make it to a top-tier conference and 1 more high-quality publication that may or may not make it to a top-tier venue (but it can definitely make it to a good conference or journal) per year. Now, if I have a student that produces a paper that makes it to a top-tier conference, but does not produce another paper in a year, I will definitely not be unhappy.

So far, my first PhD student who works very hard and he is a very smart guy (he is in his second year) is able to follow this trajectory with quite a bit of paper writing help from me (but that's ok). I had a couple more PhD students, who could not keep up with that pace and they either left on their own when they realized that they needed to work hard or I had to let them go because they simply would not like putting in the necessary effort.

To a large extent, it has to do with you and your own career goals. You need to be a role model for your students if you expect them to work really hard--you cannot just disappear or not spend time with them, but expect your students to work harder than you. At the end of the day, it is your research group. You need to work harder than your students.

Given the acceptance rate of the top-tier conferences and resubmissions needed for good journals, I am wondering if three to four submissions instead of two accepted publications be OK... at least for the first year PhD student

[research_prof]

As you know, Computer Science is probably among the most challenging fields out there, where people work and compete like crazy. So, I would say productivity pretty much depends on where you want your work to gets published. If you aim for the best conference in your field, where acceptance rates may be well below 20%, then you will likely have to re-submit the same paper (or revised versions of the same work) several times before getting it accepted. In my experience, journals are a bit more flexible in the sense that you can revise your work and respond to comments/concerns from the reviewers.

My expectation from my PhD students is that they produce (not publish because that's a different beast in CS as I explained above) 1 publication that can make it to a top-tier conference and 1 more high-quality publication that may or may not make it to a top-tier venue (but it can definitely make it to a good conference or journal) per year. Now, if I have a student that produces a paper that makes it to a top-tier conference, but does not produce another paper in a year, I will definitely not be unhappy.

So far, my first PhD student who works very hard and he is a very smart guy (he is in his second year) is able to follow this trajectory with quite a bit of paper writing help from me (but that's ok). I had a couple more PhD students, who could not keep up with that pace and they either left on their own when they realized that they needed to work hard or I had to let them go because they simply would not like putting in the necessary effort.

To a large extent, it has to do with you and your own career goals. You need to be a role model for your students if you expect them to work really hard--you cannot just disappear or not spend time with them, but expect your students to work harder than you. At the end of the day, it is your research group. You need to work harder than your students.

Given the acceptance rate of the top-tier conferences and resubmissions needed for good journals, I am wondering if three to four submissions instead of two accepted publications be OK... at least for the first year PhD student

In my comment, I talked about "produced"/"submitted" papers. As I said, publishing a paper at a top-tier venue is a different story. You might need 2-3 years from the first submission to the actual publication (due to several rejections and resubmissions). By "3-4 submissions ", do you mean 3-4 different papers? Or revisions of the same paper?

Producing 3-4 different papers per student, at least in my subfield, is very hard/impossible in a year, unless you compromise quality.

[kerprof]

As you know, Computer Science is probably among the most challenging fields out there, where people work and compete like crazy. So, I would say productivity pretty much depends on where you want your work to gets published. If you aim for the best conference in your field, where acceptance rates may be well below 20%, then you will likely have to re-submit the same paper (or revised versions of the same work) several times before getting it accepted. In my experience, journals are a bit more flexible in the sense that you can revise your work and respond to comments/concerns from the reviewers.

My expectation from my PhD students is that they produce (not publish because that's a different beast in CS as I explained above) 1 publication that can make it to a top-tier conference and 1 more high-quality publication that may or may not make it to a top-tier venue (but it can definitely make it to a good conference or journal) per year. Now, if I have a student that produces a paper that makes it to a top-tier conference, but does not produce another paper in a year, I will definitely not be unhappy.

So far, my first PhD student who works very hard and he is a very smart guy (he is in his second year) is able to follow this trajectory with quite a bit of paper writing help from me (but that's ok). I had a couple more PhD students, who could not keep up with that pace and they either left on their own when they realized that they needed to work hard or I had to let them go because they simply would not like putting in the necessary effort.

To a large extent, it has to do with you and your own career goals. You need to be a role model for your students if you expect them to work really hard--you cannot just disappear or not spend time with them, but expect your students to work harder than you. At the end of the day, it is your research group. You need to work harder than your students.

Given the acceptance rate of the top-tier conferences and resubmissions needed for good journals, I am wondering if three to four submissions instead of two accepted publications be OK... at least for the first year PhD student

In my comment, I talked about "produced"/"submitted" papers. As I said, publishing a paper at a top-tier venue is a different story. You might need 2-3 years from the first submission to the actual publication (due to several rejections and resubmissions). By "3-4 submissions ", do you mean 3-4 different papers? Or revisions of the same paper?

Producing 3-4 different papers per student, at least in my subfield, is very hard/impossible in a year, unless you compromise quality.

Yes. By 3 to 4 submissions, I  meant resubmissions/revisions included....

So when you say two papers per year do you mean two high quality submissions (that one or both may end up getting accepted) per year per student....

Actually for the PhD student I am talking... we submitted two papers in "Top-Tier" conferences over last year and both got rejected and the student is working on revising it addressing the comments.

[research_prof]

Right. I meant 1-2 new papers every year. It depends on the comments that you get from reviewers as well. If the comments require major revisions/extensions/new experiments, then they take time. If you get comments that can be addressed relatively easily by running quick new experiments, then the student will get that done sooner and have more time to work on "new" stuff.

[kerprof]

Right. I meant 1-2 new papers every year. It depends on the comments that you get from reviewers as well. If the comments require major revisions/extensions/new experiments, then they take time. If you get comments that can be addressed relatively easily by running quick new experiments, then the student will get that done sooner and have more time to work on "new" stuff.

Makes sense. Thanks for the clarification!

[kerprof]

To a large extent, it has to do with you and your own career goals. You need to be a role model for your students if you expect them to work really hard--you cannot just disappear or not spend time with them, but expect your students to work harder than you. At the end of the day, it is your research group. You need to work harder than your students.

Please advise what do you mean by "You need to work harder than your students"....   What are the activities that you think an ideal advisor/supervisor should do that will inspire a PhD student and produce good research work...