In my experience of mentoring over 100 students I have found that the most effective way of attempting to assess the suitability of a student is an interview and, if possible, a series of meetings (for internal candidates, e.g., involving an undergraduate on a project). With the ubiquity of Skype (or Google+, etc.) it is pretty easy to set up a series of interviews with short-listed candidates. Then you can ask them pointed questions about their knowledge base, give them a thought exercise to see how they think on their feet. In other words and use the interview as a small trial run to see if they have the skills and personality you are looking for.

For the internal candidates where I can have multiple meetings I always give them a paper to read and then come back a week later and we discuss it. Their performance in critiquing and analyzing the paper, in my experience, is often a great indicator not only of future performance but also of their staying-power when times might get tough on a project.

I've found that some of my best research students have not necessarily had the highest scores (on various standardized tests) but they have been the ones that have clearly shined in these interviews.

For those who are reading this from the interviewee perspective I would recommend asking for an interview, and practicing beforehand so you can clearly articulate your previous research experiences and your current interests.

I'm also facing a similar problem but from the other side. I am continuously applying for doctoral positions in India and abroad but haven't made it yet. Don't exactly know what's happening wrong. I've good grades, continuously performing, thesis is fine enough , have published papers (though not in journals with impact factors, but I agree these are not fraud and that I can prove too), doing projects continuously (I agree not of very high standard, but yes these are enough to show that I'm capable to implement what I've learned and guide students), know programming and so many other things which make me a good doctoral candidate.....Traditional method of selecting a candidate is good, I mean it helps you in selecting a student with previous knowledge in the field and his/her report can show his/her caliber to perform too....No doubt.....but hey, a student doesn't actually work on the project or report as he only learns with the work he perform in any project but the project may or may not be his/her own idea (after all he was being guided throughout) ....and the previous knowledge in the field depends upon so many factors(like options available to him for a master thesis topic, who were the guides available and what were those guides working on)...yes, grades can show the true caliber of a student but hey research never demand for a student with good grades, as far I know it, research essentially needs a student who is capable of implementing the knowledge he has gained so far or what he is learning t solve some particular problem. It needs a innovative student not intelligent student...................Look for a innovative candidate not an intelligent candidate. Yes, just don't try to look what you want from a candidate indeed look what you can get from that candidate and/or what you can make him do for you.........Direct methods of selection have been fine since years but now with changing times indirect methods of selecting a candidate are important.....after all with these social networking websites you can judge a candidate better than any test.........also, you'll have to find a good candidate as all good candidates are not aware how to search a good guide........

I'm attaching a link which is not meant for any competition but is to know ....what I should do to get a good position ...

Ruediger Rentsch, I afraid I've made this a different problem...

https://sites.google.com/site/kapilnarwalnow/

This is a nice question.

I used to select candidates for projects years ago, and some people did disappoint me a lot not only for their skill levels but also for their personalities. However, it is not easy even to only test applicants' skills, not to mention their underlying personalities.

But form another viewpoint, behaviors are "controllable" in some sense, and proper encourage is important for the performance of a team member.

Considering these risks, I chose the "redundant strategy" that selecting more people at the beginning, and wash out people with low performance even though I had tried to encourage them.

I think many of us share your frustration. There are things you can do, and probably have, that will reduce the changes but not eliminate them. A relatively detailed task analysis with assessments, interview items, and work history/references aligned to them are likely o be most helpful. Some of the tools developed based on Holland's career fit theory may help in some way (or may not). Also note that someone can be a great candidate and develop issues that interfere with performance after you have them on a project. There is little that can be done to per-screen for those things as I would see it.

In my experience of mentoring over 100 students I have found that the most effective way of attempting to assess the suitability of a student is an interview and, if possible, a series of meetings (for internal candidates, e.g., involving an undergraduate on a project). With the ubiquity of Skype (or Google+, etc.) it is pretty easy to set up a series of interviews with short-listed candidates. Then you can ask them pointed questions about their knowledge base, give them a thought exercise to see how they think on their feet. In other words and use the interview as a small trial run to see if they have the skills and personality you are looking for.

For the internal candidates where I can have multiple meetings I always give them a paper to read and then come back a week later and we discuss it. Their performance in critiquing and analyzing the paper, in my experience, is often a great indicator not only of future performance but also of their staying-power when times might get tough on a project.

I've found that some of my best research students have not necessarily had the highest scores (on various standardized tests) but they have been the ones that have clearly shined in these interviews.

For those who are reading this from the interviewee perspective I would recommend asking for an interview, and practicing beforehand so you can clearly articulate your previous research experiences and your current interests.

John Swaddle, I agree with you, multiple meetings in person and/or on Skype or Google+, pointed questions, exercises, performance in critiquing and analyzing the paper or in other words a small trial is the best way to select an applicant for research. It is indeed a respectable selection process in which the individual capabilities of an applicant are more important than his/her background and/or surroundings.

Some selection criteria are:

1 - Academic History: will take into account previous Higher Education in related subjects.

2- Professional experience and research: research articles published in scientific journals, participation in research projects, knowledge of other languages ​​for scientific interest, experience related to the Master.

3- Other merits (motivation, interest in pursuing doctoral studies, academic ability, reputation of the school of origin, references, other languages​​, aspects of emotional intelligence and personal history, etc.).