LPU
February 28, 2012
Reader jekka asks:
Are you advocating Least Publishable Unit papers here DM?
My answer?
Yes.
Naturally this comes with qualifiers. For now, however, I invite you to stretch yourself and
1) Define the Least Publishable Unit concept and manuscript type as cleanly as you can.
And…
2) explain to me what the cost/problem/drawback is in the PubMed era.
Finally, please assure me that you have never cited a paper you consider an LPU, never allowed* such a turd to shape, motivate or inform your research and for goodness sake never polluted a grant application with any such thing.
*naturally, I have done all these things. Repeatedly.
DrugMonkey 
February 28, 2012 at 1:45 pm
And as luck would have it….
http://blogs.nature.com/news/2012/02/university-of-sydney-sackings-trigger-academic-backlash.html
Grist!
LikeLike
February 28, 2012 at 3:57 pm
Its interesting that the first commenter is complaining that they are not crediting “chairing conference sessions” as productivity.
LikeLike
February 28, 2012 at 3:58 pm
One of the worst aspects of the LPU is that there is a lot of time, effort, and resources that goes into writing a manuscript, submitting it, dealing with reviews, resubmitting, checking page proofs, paying submission and/or publication fees, etc. And none of these things scales with the scope of the publication. So you are doubling all of these costs when you publish two manuscripts of data that you could have conbined into a single one.
LikeLike
February 28, 2012 at 4:36 pm
We have a responsibility to the post-docs who put their faith in us to put them in the best possible position to compete for the job for which they aspire. If this is a tenure track assistant professorship, they need a major career development grant and a major paper plus. Its doing them a disservice to publish LPUs. Grad students or my own project, this is fine but not for post-docs.
LikeLike
February 28, 2012 at 5:16 pm
Doubling down, I see. In addition to CPP’s point, there IS something to be said for making a better story using multiple parts. I’m not talking about stretching for a glamor pub, but just making something more comprehensive. It cuts down on ref chasing* and can build a truly better outcome.
Or are you suggesting that groups should collaborate, then retreat to their labs to write up their halves of the story?
*I know you’ll claim that in the intertoobs age there is no such thing!
LikeLike
February 28, 2012 at 5:28 pm
Ideally every grad sudent will write a paper, which can lead to LPUs. Otherwise I’m with CPP — the hassle of writing more papers feels very significant. (I wonder how much of that is psychological.) Also, think about it from the reader’s perspective. If the two LPUs are complementary enough to fit naturally into one paper then isn’t it easier just to look at one paper and to have the full story in one place? I guess the counterargument is that some of the smaller results might get lost in the big paper, especially if they don’t make it into the unified abstract…
LikeLike
February 28, 2012 at 6:08 pm
I try not to publish LPUs unless they are orphaned data, which can’t be fit onto another project. Sometimes though I keep them around until it’s their moment to shine, or become relevant to another study.
LikeLike
February 28, 2012 at 6:33 pm
Huh, Pp actually came up with a good reason. I’ll buy that one. Of course, part of the trainees’ development is dealing with that hassle as the first author…..
P-lS- sure, retreat to your labs and write up as a normal level collaboration. You can still co-submit if you want….
Namnezia- that’s nice. But in the mean time the grad student or postdoc is waiting around on your fancy…And your field cannot make use of your (presumably worthy) observation.
PS- are you going to pick and choose which trainee to put in a good position and which 5 to hose? Because that is what you are advocating. Far better to dismantle the all-or-none edifice.
LikeLike
February 28, 2012 at 7:16 pm
Drug Monkey-
Of course I pick and choose….according to what’s in the advisee’s best interest. We have a responsibility to put our advisees interests first. If this means bowing to the reality of the fact that a post-doc needs a high impact publication over my own ideology, then this is what I do. It’s what any responsible mentor would do. You can tilt at windmills all you want – I need to make sure my post-docs are competitive. So yes, I have different standards and different goals for different trainees.
LikeLike
February 29, 2012 at 4:14 am
It’s not about PubMed. It’s about reviewing, reading, and conveying science efficiently. As others have noted, LPUs might be key for grad students who need to show they can produce pubs, but they otherwise weigh everybody down. I get >300 articles in my feed per week for one of my study topics, and it’s a pain in the arse to get through the potentially juicy ones before I realize they’re very minor additions to something else. Really–you don’t need to publish a new study for each ANOVA, especially when the results aren’t that interesting. I’d wager the more synthetic, complete papers get more citations too.
LikeLike
February 29, 2012 at 4:17 am
Also, as PP noted, it’s very costly to spend time on boring spin-off papers. I tried to shove all my null results in the supplement of my last paper, but the editor made me strip out a story to shorten the manuscript. I’ve been polishing and sculpting this totally inane result for way too long to try to get it into a mid-tier journal. Would’ve been much better to move on to next exciting topic.
LikeLike
February 29, 2012 at 4:41 am
DM, you’re not taking this far enough. We could fix both the Open Access issues and the glamor mag BS at the same time. You could pioneer Twitter Science. Data up in an image, 140 character discussion, followed by 140 character acknowledgement. Boom. With advances like Storify it should be no problem for anyone savvy enough to be in science to piece it all together. In fact, I’m making sure we get credit for this idea so I can slap it on my CV.
Substance, P-l.*, Becca, Dr.*, (2012) Storification of science as the future. Drugmonkey Blog (Scientopia). Feb 29, 2012.
*equal contributors
LikeLike
February 29, 2012 at 4:44 am
Naturally, when I cite our brilliant comment on *my* CV, the order of the names will be switched.
LikeLike
February 29, 2012 at 4:45 am
I promise not to be judgey about it.
LikeLike
February 29, 2012 at 7:04 am
We have a responsibility to the post-docs who put their faith in us to put them in the best possible position to compete for the job for which they aspire
*snerk*
Sorry, that just made me snarf my coffee..
LikeLike
February 29, 2012 at 7:26 am
I totally agree with CPP. The Mosers have always said that editors want a complete story or a mindblowing story. The LPU is not that interesting to me because it is only part of the story. For a graduate student, one could imagine combining all chapters from the dissertation into a really nice Neuron or J Neurosci paper. That’s kinda how I roll. I didn’t start out that way– I was a big LPU person as a youngster. The Mosers and Stephen Lisberger (the Editor at J Neurosci at the time) really changed my mind.
The complete story often requires hardcore collaboration, where the dual first-authors are doing things that the other simply cannot do (e.g., one is doing dendritic recordings and the other is doing x-ray crystallography). I think the duallies have equal merit, but I agree that the field still looks down on it (unfortunately). I suppose a thoughtful study section or tenure committee should be able to identify the unique and critical contributions that each author made, and would therefore hold such papers in similar regard. Humans can be hurried, though, and revert to biases that would relegate the duallies into a lower tier than onesies.
To quote Gordon Shepherd (the papa, not the youngun), though: “Get the paper published and let time be the editor.” You don’t want to be the crazy kook that sits on data for a year or two just so that you can include into a big-time paper (especially if you’re an Asst Prof).
LikeLike
February 29, 2012 at 8:26 am
Your “complete story” is STILL only part of the story.
That’s science.
If you think *any* paper has exhausted all there is to know you either lack imagination or are so focused on a GlamourMag pub (as a published document) being the entire goal that you have lost sigh of what we are supposed to be doing.
Advancing knowledge.
LikeLike
February 29, 2012 at 8:28 am
….in 140 character +fig increments. I’m telling you, breakthrough.
LikeLike
February 29, 2012 at 12:39 pm
You still haven’t explained *why* your idea of an acceptably dense/comprehansive paper is “better”, Prof-like.
LikeLike
February 29, 2012 at 1:15 pm
Because, DM, even in this age of on-line publication, an Interested Reader is quite likely to miss some units of your 10 LPUs that make up One Big Story. For instance, an electrophysiologist might ignore all the behavior and neurochemistry LPUs you published on a similar topic, even if she pays attention to all your e-phys papers. However, if you put them all together, they will get read together. If the data from different neurogenres all support a common hypothesis, then the story is much more convincing taken together than when some small bit of it in Archives of Cholinergic Chemistry has to refer to last year’s LPU in Eur J of Muscoluskeletal Electrophysiology of Synapses in the Left Pinkie.
Yes, yes, you could easily argue that people are going to focus on data from their technique/field of interest even in a mega-mondo pub, but so what? People are *still* going to be more convinced by a nicely put-together story that seems to cover all the bases than by a string of obscure LPUs.
And anyway, this applies even if your 10 LPUs all use the same technique, with slight variations. If those LPUs really all support one unified hypothesis, then putting them all in 1 or 2 Big Papers will be much more convincing than scattering them around the literature. People will say, “wow, they really nailed it!” instead of “ok, that was a really incremental advance, why again should I care?” — and that last is only if they even read it.
Of course, if your Approach to Science is to take one technique and provide us with infinite variations that *don’t* advance a unified hypothesis, then the LPU game is the way to go. Just don’t expect the average reader to get beyond “why should I care?”
LikeLike
February 29, 2012 at 1:38 pm
Sure, Grumble, but just as often a beautiful result is packaged with a bunch of dross so that a “complete” “story” can be submitted to a glamour journal, since glamour journals increasingly are not interested in “complete” “stories” and not in beautiful experiments.
LikeLike
February 29, 2012 at 1:39 pm
Corrected:
Sure, Grumble, but just as often a beautiful result is packaged with a bunch of dross so that a “complete” “story” can be submitted to a glamour journal, since glamour journals increasingly are interested in “complete” “stories” and not in beautiful experiments.
LikeLike
February 29, 2012 at 2:22 pm
I’m with whoever the fucke suggested just making papers on twitter. Allz I read is the fucken titles anyway, so fucke itte.
LikeLike
February 29, 2012 at 4:46 pm
Amen proflike substance (and Dr. Becca)! But the model should be Wikipedia (but with peer review of some sort), not Twitter. That way, you get your big story all in one place, and list each piece of evidence behind the story as an experiment (or small set of experiments) conducted and analyzed by ONE PERSON. “Editors” or whatever who actually write up the larger story get separate credit.
This would also allow relevant experiments to be listed in support of multiple stories to which they are relevant. We actually already do this in citing papers–what we are really citing is not the paper as a whole, but the one figure in it that shows the published information on which the new evidence builds.
I realize that this model has some severe logistical limitations (each “story” could get too big to read; peer review and editing would be painful)…but there is probably a way to tweak the model to address these issues.
LikeLike
February 29, 2012 at 4:49 pm
That’s true, SN, but that’s really the fault of the Glamour Mags. You don’t see this (as much) in, say, J. Neurosci., but in that journal you do see a lot of complete stories that thoroughly test an interesting hypothesis.
LikeLike
February 29, 2012 at 6:04 pm
People DO only focus on the part that intersects with their interests (unless they are copy cat rumpsniffers), glumping it into one alleged “complete” story necessitates loss of oft-critical detail and the process also slows down the speed of communication.
LikeLike
March 1, 2012 at 4:46 am
Can we not conflate “complete story” with Glamor mag science? Doing so brings too much baggage to a conversation that is not relevant to the LPU argument unless you want to make the case that slapping a few LPUs together = Nature.
Alright, say I have two parts to a project, 1) a novel analysis method, and 2) new data that I use to both demonstrate the effectiveness of this method and show a novel finding. Should these be published separate? One could argue that they could in an LPU model, but the “whole” story (which does not explain my entire field, but merely links method and novel finding) is more effective. It gives the reader a solid demonstration of a new method to go with some interesting science. The alternative two stories with less general interest, fewer direct links and it pushes the burden on the reader to put the pieces together. With so much data and information available, even a savvy scientist may not make all the links that form the most interesting connections.
Unless of course you want to publish the method first while sitting on the analysis for 6-12 months so you can cite it. I’m sure the 1st author trainee has that kind of time.
LikeLike
March 1, 2012 at 5:38 am
See? this is why “LPU” is in the eye of the beholder. Personally I’d prefer that “methods” be wrapped in with something to which the methods are applied. Whether a purely methodological paper stands alone depends on the specifics. In my view novelty, fixing something generally applicable to the field or because you need to give the finger to stupid grant reviewers. Occasionally because the methodological stuff is just huge in terms of the number of figures that are necessary.
Nevertheless there are also cases in which I’ve seen people squeak stuff out that is slicing the sausage too thinly even for my taste but you know what? Those are still a contribution that I will, eventually, cite. I am not mortally offended to open up the PDF and find out how thin the data were. I file it and move on….perhaps after an idle thought that the postdoc must be about to depart the main project and be keen to get something out of it.
I still reject your notion of “burden”. I suppose it is possible we are just imagining different scenarios and talking past one another. But I just don’t experience this as a problem.The most recent paper should discuss current data in context of prior papers. that’s scholarship. So what if it happens to be a paper from the same lab or collaboration?
while sitting on the analysis for 6-12 months so you can cite it.
This is ludicrous. You send them off when ready. If you are really close, you can try citing the other one as “submitted” or some such. Then as the review process wends along, you update the status.
even a savvy scientist may not make all the links that form the most interesting connections.
If you feel like your colleagues aren’t smart enough to connect up your brilliance, write a damn review article.
LikeLike
March 1, 2012 at 5:47 am
Can we not conflate “complete story” with Glamor mag science? Doing so brings too much baggage to a conversation that is not relevant to the LPU argument unless you want to make the case that slapping a few LPUs together = Nature.
They are inherently linked issues. Growing obsession with Glamour is what took a somewhat arguable theme of a “complete story” (meaning dotting i’s and crossing t’s within a single methodological framework) off the deep end into “a bazillion different methods”. This, plus the space limitations, eventually created exactly what you describe. The N/S paper that is published is in fact a bunch of single-figures*, not even LPUs. You put them together and they somehow are supposed to be “complete”. They aren’t . You have to go to the supplemental data to get even close (hello! “burden” amirite?) aaaaand, often enough even with the supplemental materials, the methodological “completeness” is crappy.
*go back to the Science articles of 30-40 years ago. One or two figures of real genuine novel stuff that created an expanding followup of real studies in real journals in the wake. That is what those glossy rags should return to.
LikeLike
March 1, 2012 at 7:21 am
I completely agree that the completeness of “complete” stories in NatureScience is an aggravating fiction. But there are plenty of examples of lengthy, detailed articles in J Neurosci and Neuron and Eur J of Neurosci and many other journals, which do exactly what you say: cross the t’s and dot the i’s. In general I’m far more impressed by these than a series of little LPUs. Maybe my impressions are wrong, but I get the feeling that if the authors have put in the effort to really thoroughly test a hypothesis, the hypothesis is probably important. If instead they do just one experiment and publish, the hypothesis probably wasn’t all that weighty to begin with.
Again, maybe by putting together all those single experiments, the results suggest a much larger and more important hypothesis — but why make the reader hop around hunting for it? If you want to sell a Big Idea, then sell it as strongly as you can — don’t make your audience *work* to figure it out! In general I think scientists know that about salesmanship, so they will tend to put their best stories together in 1 paper. That’s why I think LPUs are suspect: they tend *not* to support a cohesive idea that’s really important to move the field forward.
LikeLike
March 1, 2012 at 7:47 am
Addressing the pros and cons of LPUs (and I concur, LPU can mean anything anybody wants it too) is pruning dead leaves on a sick tree. Given the importance of IF (or whatever other magic metric the kids are using these days) and pub counting as indicators of productivity in tenure decisions, the only consideration as to whether to put out LPUs is whether a little extra work could increase the impact substantially as to be worth holding back. When institutions start taking paper size, figure number, and the narrative arc of a paper into consideration, then TT faculty can certainly rethink their strategy.
As it is, I find some LPUs to be quite refreshing. The sort of wee papers with a straight forward single hypothesis (sans heavily arborized cascade of subhypotheses), one or two straightforward experiments, and then boom, a conclusion. It’s rare to be able to conduct a study as linear as that, so if you find yourself in that position, you may as well make the most of it, imho.
LikeLike
March 1, 2012 at 8:26 am
much larger and more important hypothesis — but why make the reader hop around hunting for it?
because there are significant costs in time and authorship credit. We started this particular discussion as an offshoot of one that dealt with the need for trainees to have first author papers and the lying sham that is the “equal contribution” footnote.
the other cost which I keep raising and nobody seems to be concerned with is the fact that collaborative science broadly writ, that where other competing laboratories step upon your stair and move on to build the next one, is hindered if you wait 5 years to publish a finding. (because let’s be honest, the same forces driving this competition for bigger and bigger scope frankensteinian GlamourMag style articles also drive a failure to present intermediate data at meetings)
Your general lack of concern persuades me that you think science proceeds best via the independent silo model. blech. the scientific episodes that are most exciting and productive stem from direct exchanges in the published literature. rapid exchanges. with pubs coming out ever year or more frequently to question or answer the next lab over.
LikeLike
March 1, 2012 at 10:48 am
They are inherently linked issues. Growing obsession with Glamour is what took a somewhat arguable theme of a “complete story” (meaning dotting i’s and crossing t’s within a single methodological framework) off the deep end into “a bazillion different methods”.
Perhaps we are talking past each other, because this is not the type of thing I’m referencing at all. You almost CAN’T have a complete story in a glamor mag pub anymore, as you well recognize. I am referring to a scenario in which the target is a well respected sub-field journal and the LPU targets would be society-level journals at the sub-4 IF level (if we are talking numbers as a general indication of journal scope). Perhaps our very different backgrounds (Biomed vs. Bio) color the discussion.
LikeLike
March 1, 2012 at 11:33 am
Dunno what fucken planet you are on, but here on Earth, many cutting-edge technical approaches are not capable of being conducted and analyzed by a single person.
Dude, you ever hear of the journal Nature Methods? Last I checked its impact factor was > 20.
LikeLike
March 1, 2012 at 6:52 pm
I am totally late to this party, due to a sick baby, so I’m going to offer up summaries of two recent conversations I’ve had on this topic.
Colleague#1: GlamourPubz contain a story that changes the way the field thinks about a particular problem. Lesser stories are incremental science that is of interest to those in a sub-field, but will probably not be seen by anyone else. These experiments are the next obvious thing to do and are therefore boring.
Colleague#2: If you haven’t had a paper in a journal with at least an IF of 10 in your last grant cycle, you shouldn’t get renewed b/c you aren’t significantly advancing the field.
My opinion on the “complete story” angle is that a high-profile paper should kick off a flurry of new questions/papers, not “close the book” on a topic.
I cite LPU turds because those bitchez review my grants.
LikeLike
March 1, 2012 at 8:58 pm
And I think if your shit isn’t related to “Health” in some actual way you shouldn’t get reviewed at the NIH no matter how many Science papers you have…
LikeLike
March 1, 2012 at 9:02 pm
What constitutes “related to health”? Human subjects only? Drug trials? I know some physical chemists who permanently lost NIH funding when study sections got restructured. Maybe they never should have had it.
LikeLike
March 1, 2012 at 9:04 pm
Also: do the aforementioned colleagues understand that the vast majority of glamour pubs get cited way, way less than the IF might suggest, and that IF is driven predominantly by a handful of papers, most of which are review articles?
“GlamourPubz contain a story that changes the way the field thinks about a particular problem.” Maybe. A lot of the time, they’re simply over-sold so as to get published in the big three.
In one of my three or so sub-fields, the *majority* — over half — of the glamour papers over the last dozen years or so have proven to be wrong or misleading. Not an exaggeration, not a joke. People outside that sub-field continue to cite them. People within the sub-field don’t. The papers advancing this sub-field are being published almost exclusively in IF 6-14 journals — think JBC to Cell/Nature specialist rags. Call it Tier 1.5 to 2. Papers published in PNAS have proven to be especially important in this area over the last few years.
So, IMO, your colleagues are full of shit.
LikeLike
March 2, 2012 at 7:53 am
“If you haven’t had a paper in a journal with at least an IF of 10 in your last grant cycle, you shouldn’t get renewed b/c you aren’t significantly advancing the field. ”
Presumably that >IF10 paper should not have a single reference to a <IF10 paper, though, right? Otherwise that would kind of stick a big fucking pointy stick in the argument that <IF10 doesn't advance science.
You know, while we're at it, football really should just can the concept of an offensive line. I mean, most of those fat, sweaty bastards hardly ever touch the ball anyway, and they barely move more than a yard, if they move forward at all. Their incremental contributions are surely of little interest to the general sports fan.
LikeLike
March 2, 2012 at 9:36 am
I pretty much agree with DrugMonkey. I recently wrote a review article where I looked at every paper I could find on a topic. I was surprised to find that the glamour mag pubs had more experiments in them, but for the types of experiments I was examining, were usually more poorly done than in singly-focused pubs (LPU). The experiments in the glamour mags often lacked adequate controls, but were used to support other experiments. The LPU were more likely to have used controls and generally had more complete data for the experiments I investigated.
In addition, with PubMed, I also agree that all journals are becoming more equivalent. In my field the impact factor of the lower rated journals is going up and for the higher rated ones are dropping. They may never equalize, but the impact of PubMed is obvious. The glamour mags were “better” because you got them in the mail and read them. The number you actually get in the mail now is declining.
In my opinion, if you want to present a complete story, write a review article.
LikeLike
March 2, 2012 at 10:43 am
In my field the impact factor of the lower rated journals is going up and for the higher rated ones are dropping.
Huh. Now that you mention it maybe this IS the reason. I had just assumed more papers meant more available citations….but I can see some gap-closing between the lowly and the slightly-better-than-average journals in my areas.
LikeLike
March 20, 2012 at 9:58 am
[…] most recently took up the issue of the Least Publishable Unit of science in the wake of a discussion about first authorships. In that context, the benefit of […]
LikeLike
June 5, 2013 at 6:51 pm
After 2011 the flamboyant baptism, adidas Originals by Jeremy Scott in 2012 continued with their wings and the Section of Zoology, the launch of two new Jeremy Scott x adidas Originals JS Wings 2. new white leather-based uppers for the substrate, coupled with the eye-catching American flag pattern, the latter introduced a large blue and white to keep up with the stars and pink and white horizontal stripes, will be applied to the factors of heroism in the stop. The start of the Jeremy Scott x adidas Originals can be explained as vintage series is by considerably the most impressive factor of the new items.
LikeLike