The NIH removes requirement for standardized scores in pre-doc fellowship applications

Jul 10 2015 Published by under Postgraduate Training

Huh.

NOT-OD-15-120:

To align with recent changes in the fellowship biosketch format,this Notice eliminates the requirement for inclusion of scores from standardized exams (e.g., MCAT, GRE) in the fellowship biosketch from the following funding opportunity announcements, effective immediately:

For reference, from PA-14-147:

Note that scores for standardized exams (e.g., MCAT, GRE) as well as a listing of the applicant’s courses and grades must be included in the Fellowship Applicant Biographical Sketch, and NOT in this attachment.

Anybody seen a rationale for this one?

The overall thrust of the Investigator Biosketch revamp seems to be to brag even more highly upon personal accomplishments, rather than suitability for the specific proposal. Also to allow people with non-traditional (non-published, say) accomplishments to brag on those.

Doesn't it seem like eliminating standardized scores works against this?

Can anyone think of why this would be a good thing for NIH to do?

Next point: I see where it says it is eliminating the requirement, not telling applicants not to include their scores. Fascinating.

First: If you have excellent standardized scores, I suggest you continue to put those in the pre-doc NRSA biosketch somewhere people.

Second: If you don't put them in there, the reviewer who is fond of such measures of your aptitude is going to assume your scores are really bad. Right?

Third: I think this is more evidence of NIH changes that will throw chaos into the system rather than really improving much.

10 responses so far

  • profduder says:

    NSF recently got rid of the GRE scores for their graduate fellowships. I believe the reason is that many reviewers were relying on these scores to rank applicants. The GRE is notoriously bad at predicting success in graduate school and easily gamed.

  • Jessica Tollkuhn says:

    when are they going to stop requiring undergrad grades for F32 apps? so useless.

  • JustAGrad says:

    Of course this happens right after I applied for the last time before I graduate. I got a few critiques about my "subpar" GRE scores that didn't seem to hinder my admissions or predict my graduate GPA.

    I welcome any chance to reduce the impact of a test that exists mostly to make ETS more money.

  • datahound says:

    I don't understand why NIH is changing the requirement in this context, but numerous studies have revealed that GRE scores are not very predictive of success in graduate school. They are most predictive of first-year course grades (which are hardly key outcomes). See http://grad.berkeley.edu/wp-content/uploads/GRE-Studies-Annotated-BIBILOGRAPHY-2014-December-5.pdf for an annotated list of references.

  • qaz says:

    The reason GRE scores are not predictive of grad school outcomes is because we only take the top slice of GREs.

  • qaz says:

    And because there are first-order effects that drive GRE scores (such as socioeconomic class). Most of the studies in that annotated list DH references (and the other ones I know of) find a small but significant relationship between GRE and success not just in first-year course grades, but in overall graduation rates and research productivity. Particularly the subject scores. (Correlations in the 0.1 to 0.25 range.) On the other hand, none of the studies I have found take socioeconomic rating into account. I bet that once you account for socioeconomic class GRE will be even more predictive.

    We've done analyses on our incoming admissions over the last 20+ years. We found that for our program the only thing that was predictive was "has had research experience" (meaning did a project, not published a paper). So we now require that. (Actually, we found that publishing a paper was negatively correlated with success because it depended on how the advisor ran things much more than on the student. We had been letting in students with lower GRE scores if they had "good productivity" [meaning published a paper]. Once you factored out GRE scores, publishing a paper as an undergrad had no relationship to success as a grad student.)

  • E rook says:

    Qaz, this is enlightening, thank you. I had been bending over backwards to try to get my undergrads' data into a paper if they expressed future interest in research for their career. I am running into difficulty with a student who did research with me for quite a while and is having trouble getting into med school. I wholeheartedly believe this person would make an excellent physician and I want to reach through the admin docs & slap the committees in the face. My fear is the pub record makes them seem too "research focused" when it should just be interpreted as "highly focused (on long & short term goals)."

  • Established PI says:

    Dropping the GRE scores benefits underrepresented minorities and first-generation college students, who tend to score lower on these tests. As noted above, a recent study found the scores to be irrelevant in predicting success anyway, so removing them from the evaluation seems like a plus.

    Now I wish they would eliminate grades from the F32s - I have seen great postdoc applicants downgraded because of a handful of sub-par undergraduate course grades, which I find preposterous.

  • Juan Lopez says:

    Qaz, "we found that publishing a paper was negatively correlated with success because it depended on how the advisor ran things much more than on the student". How did you determine this? I can see how you could find the association. But, how could you determine the cause?

  • qaz says:

    Fair enough. What we found was 1. that it was negatively correlated with success on its own; 2. that once we accounted for other factors (GRE, GPA, etc.), that correlation went to 0; and then 3. we looked at the undergrads in our department who were publishing papers, and found it seemed to depend on faculty style, on how faculty assigned authorships and encouraged papers. Middle author contributions are very obviously dependent on how PIs assign authorship, but we found that first author papers also depended on how PIs encourage, co-write, and push to publication.

Leave a Reply