Scientific premise has become the latest headache of uncertainty in NIH grant crafting and review. You can tell because the NIH keeps having to issue clarifications about what it is, and is not. The latest is from Office of Extramural Research honcho Mike Lauer at his blog:

Clarifying what is meant by scientific premise
Scientific premise refers to the rigor of the prior research being cited as key support for the research question(s). For instance, a proposal might note prior studies had inadequate sample sizes. To help both applicants and reviewers describe and assess the rigor of the prior research cited as key support for the proposal, we plan to revise application instructions and review criteria to clarify the language.

Under Significance, the applicant will be asked to describe the strengths and weaknesses in the rigor of the prior research (both published and unpublished) that serves as the key support for the proposed project. Under Approach, the applicant will be asked to describe plans to address weaknesses in the rigor of the prior research that serves as the key support for the proposed project. These revisions are planned for research and mentored career development award applications that come in for the January 25, 2019 due date and beyond. Be on the lookout for guide notices.

My first thought was…great. Fan-friggin-tastic.

You are going to be asked to be more pointed about how the prior research all sucks. No more just saying things about too few studies, variances between different related findings or a pablum offer that it needs more research. Oh no. You are going to have to call papers out for inadequate sample size, poor design, bad interpretation, using the wrong parameters or reagents or, pertinent to a recent twitter discussion, running their behavioral studies in the inactive part of the rodent daily cycle.

Now I don’t know about all of y’all, but the study sections that review my grants have a tendency to be populated with authors of papers that I cite. Or by their academic progeny or mentors. Or perhaps their tight science homies that they organize symposia and conferences with. Or at the very least their subfield collective peeps that all use the same flawed methods/approaches.

The SABV requirement has, quite frankly, been bad ENOUGH on this score. I really don’t need this extra NIH requirement to be even more pointed about the limitations of prior literature that we propose to set about addressing with more studies.

The latest Journal Citation Reports has been released, updating us on the latest JIF for our favorite journals. New for this year is….


provision of the distribution of citations per cited item. At least for the 2017 year.

The data … represent citation activity in 2017 to items published in the journal in the prior two years.

This is awesome! Let’s drive right in (click to enlarge the graphs). The JIF, btw is 5.970.

Oh, now this IS a pretty distribution, is it not? No nasty review articles to muck it up and the “other” category (editorials?) is minimal. One glaring omission is that there doesn’t appear to be a bar for 0 citations, surely some articles are not cited. This makes interpretation of the article citation median (in this case 5) a bit tricky. (For one of the distributions that follows, I came up with the missing 0 citation articles constituting anywhere from 17 to 81 items. A big range.)

Still, the skew in the distribution is clear and familiar to anyone who has been around the JIF critic voices for any length of time. Rare highly-cited articles skew just about every JIF upward from what your mind things, i.e., that that is the median for the journal. Still, no biggie, right? 5 versus 5.970 is not all that meaningful. If your article in this journal from the past two years got 4-6 citations in 2017 you are doing great, right there in the middle.

Let’s check another Journal….

Ugly. Look at all those “Other” items. And the skew from the highly-cited items, including some reviews, is worse. JIF is 11.982 and the article citation median is 7. So among other things, many authors are going to feel like they impostered their way into this journal since a large part of the distribution is going to fall under the JIF. Don’t feel bad! Even if you got only 9-11 citations, you are above the median and with 6-8 you are right there in the hunt.

Final entry of the day:

Not too horrible looking although clearly the review articles contribute a big skew, possibly even more than the second journal where the reviews are seemingly more evenly distributed in terms of citations. Now, I will admit I am a little surprised that reviews don’t do even better compared with primary review articles. It seems like they would get cited more than this (for both of these journals) to me. The article citation mean is 4 and the JIF is 6.544, making for a slightly greater range than the first one, if you are trying to bench race your citations against the “typical” for the journal.

The first takeaway message from these new distributions, viewed along with the JIF, is that you can get a much better idea of how your articles are fairing (in your favorite journals, these are just three) compared to the expected value for that journal. Sure, sure we all knew at some level that the distribution contributing to JIF was skewed and that median would be a better number to reflect the colloquial sense of typical, average performance for a journal.

The other takeaway is a bit more negative and self-indulgent. I do it so I’ll give you cover for the same.

The fun game is to take a look at the articles that you’ve had rejected at a given journal (particularly when rejection was on impact grounds) but subsequently published elsewhere. You can take your citations in the “JCR” (aka second) year of the two years after it was published and match that up with the citation distribution of the journal that originally rejected your work. In the past, if you met the JIF number, you could be satisfied they blew it and that your article indeed had impact worthy of their journal. Now you can take it a step farther because you can get a better idea of when your article beat the median. Even if your actual citations are below the JIF of the journal that rejected you, your article may have been one that would have boosted their JIF by beating the median.

Still with me, fellow axe-grinders?

Every editorial staff I’ve ever seen talk about journal business in earnest is concerned about raising the JIF. I don’t care how humble or soaring the baseline, they all want to improve. And they all want to beat some nearby competitors. Which means that if they have any sense at all, they are concerned about decreasing the uncited dogs and increasing the articles that will be cited in the JCR year above their JIF. Hopefully these staffs also understand that they should be beating their median citation year over year to improve. I’m not holding my breath on that one. But this new publication of distributions (and the associated chit chat around the campfire) may help with that.

Final snark.

I once heard someone concerned with JIF of a journal insist that they were not “systematically overlooking good papers” meaning, in context, those that would boost their JIF. The rationale for this was that the manuscripts they had rejected were subsequently published in journals with lower JIFs. This is a fundamental misunderstanding. Of course most articles rejected at one JIF level eventually get published down-market. Of course they do. This has nothing to do with the citations they eventually accumulate. And if anything, the slight downgrade in journal cachet might mean that the actual citations slightly under-represent what would have occurred at the higher JIF journal, had the manuscript been accepted there. If Editorial Boards are worried that they might be letting bigger fish get away, they need to look at the actual citations of their rejects, once published elsewhere. And, back to the story of the day, those actual citations need to be compared with the median for article citations rather than the JIF.