# What is a good scheme to determine percentage contributions to a paper

Sometimes researchers are asked to define percentage contributions to a paper with more than one author. E.g., usually the supervisor in charge of the project will determine in which percentage each of the researchers who coauthored the study contributed to it (in my institution the supervisor is left out in order to prevent abuse). These percentages can then be used to distribute per-paper publication bonuses or for other (usually administrative) purposes.

Which is a good scheme to fairly determine these percentages, considering the inherent difficulties in numerically quantifying intellectual work? Are there "standard" procedures to do it?

• This sounds awful. Don't let the bean counters win. Commented May 6, 2015 at 10:06
• Use the money to buy flaming torches and pitchforks, storm the administration building, and toss the policy into a paper shredder. Commented May 6, 2015 at 10:10
• @StrongBad Whether it sounds awful or not this is what they do, and I cannot change it. Given that fact, I am looking for the most fair way to quantify how much each author contributed, so that, for instance, publication bonuses are distributed as fairly as possible. Commented May 6, 2015 at 10:24
• I think percentages are bad practice, confidence intervals are much better. Or p-values with null hypothesis that the piece each author worked on could've been produced by a random monkey. Commented May 6, 2015 at 11:18
• @MarcClaesen 93% of statistics are made up on the spot... Commented May 6, 2015 at 12:06

This does sound awful. However, I understand trying to make the best out of a bad situation.

As you write, it is extremely hard to numerically quantify intellectual work. As such, I would strongly recommend going with an extremely simple weighting scheme, so as to minimize the acrimony of discussions that will certainly come. You don't want people to discuss why somebody got 15% of a paper, while somebody else got only 10%, because people are extremely fast in figuring out what this 5% difference means in \$\$\$.

For each paper, get someone impartial to administer this. Then have all authors quantify everyone's contribution on a scale of 1-3. Don't go any finer, because anything finer is impossible. Let the impartial person pull all these together, average everyone's estimate of everyone's contribution and assign percentages based on the final number of points.

• This is simple and understandable.
• It requires no special software, apart from a few spreadsheets.
• And it gets the job done.

• It can easily be gamed. If someone consistently gives himself top marks, the impartial facilitator may want to have a discussion with this person.
• How can a junior author, perhaps an undergraduate student, assess the contribution of the PI and how much time he spent writing grants to get the money?

So it may well be that this is impractical, given the disadvantages. In which case you may need to fall back on someone senior unilaterally assigning contributions.

In such a case, I'd again work only with three-point scales. If your scale only includes three possible grades, you can meaningfully grade contributions into "minor", "medium" and "major". Anything more fine-grained, like a four-level scale, and people will start arguing why they only got a 2 when they obviously deserved a 3.

Everyone, please feel free to point out further shortcomings in the comments.

(And: try to get this policy revoked. It is horrible. Either the bonuses are too small to spend much time on such a scheme, and to fight over it - and someone will always pick a fight over this, no matter how trivial the sums are. Or the bonuses are high enough to matter, in which case the fighting will be worse. Better not to give bonuses based on paper contributions at all. Scientists should be motivated to write papers for other reasons, like tenure or recognition, which still seems to motivate enough people to write but should have drastically lower chances of poisoning your work environment.)

• I like this. A possible refinement would be that every author gives each author a 1-3 rating; it will reduce the reliance on a single person's opinions. And if I want to keep working with John I should probably be generous in acknowledging his contribution or next time I want help Is get a "no". Thus over time this self-adjusts. Incidentally a similar problem occurs when patents are written, especially in countries where inventors are due the royalties (even if they work for an employer). In those cases a lot of money can be at stake - not just a publication bonus. Commented May 6, 2015 at 12:05
• When you say "get this policy revoked", what are you suggesting? Giving equal bonus portions to every author? Not giving bonuses at all (one of the departments I'm affiliated with doesn't)? Maybe I should point out that the bonuses are not big enough that people have actually been fighting over percentages. Commented May 6, 2015 at 12:05
• @Miguel: I edited the answer to clarify what I mean by "get this policy revoked". Commented May 6, 2015 at 12:14