An often neglected aspect in discussions of reproducible research: software and data need to be licensed. If you want your software and data to be reused, you need to provide an explicit license that explains exactly how the software and data may be reused.
I’m going to try to explain the issues and give suggestions about licenses to consider. But I’m no expert, and I’m definitely {\nhilit not a lawyer. I don’t guarantee that this is entirely correct.
If you will be sharing data on human subjects, or, for that matter, just working with data on human subjects, you need to be extra careful. I’ll try to sketch the basic concerns.
I’m talking just about the United States here. I understand US law a bit, but I don’t know the law in other countries much at all.
Use licenses to make reusability clear
Karl -- this is very interesting,
however you used an old version of
the data (n=143 rather than n=226).
I'm really sorry you did all that
work on the incomplete dataset.
Bruce
This is an edited version of an email I got from a collaborator, in response to an analysis report that I had sent him.
I try to always include some brief data summaries at the start of such reports. By doing so, he immediately saw that I had an old version of the data.
Because I’d set things up carefully, I could just substitute in the
newer dataset, type “make”, and get the revised report.
This is a reproducibility success story. But it took me a long time to get to this point.
\addtocounter{framenumber}{-1}
Intellectual property is property (ie, someone can own it) that is not an actual thing but more the idea of the thing. For example, it’s not the actual physical book, but the text in the book. It’s not an actual physical art work, but any depiction of the content of that artwork. This can get pretty complicated; it’s best to move on.
Most of what academics produce is intellectual property.
Different kinds of intellectual property are protected in different ways. I’m going to focus on copyright.
An important point to mention here is that an {\nhilit idea, {\nhilit fact} or {\nhilit algorithm} can’t be copyrighted. Ideas and algorithms can be protected with a patent, but {\nhilit facts} (including individual data points) can be neither copyrighted nor patented.
So, for example, copyright protection says that you may not be able to copy and use someone’s exact code, but you can grab all of the ideas and re-implement them in your own way. Unless the ideas or algorithms are patented, and then you have to get their permission to use them, even if it’s your own implementation. }
In academics, it is customary that researchers control copyright \onslide<2->{* At UW-Madison:
\begin{quote} \footnotesize \lolit “Except as required by funding agreements or other university policies, the university does not claim ownership rights in the intellectual property generated during research by its faculty, staff, or students.” \end{quote} }
Since 1978, works you create are automatically copyrighted. That includes data, software, papers, books, talks, posters, course syllabi, and lecture notes. Since 1989, you don’t need to include a copyright notice.
In many job situations, your employer will own the copyright on works that you create as part of your employment. But in academic settings, it is traditional that researchers retain copyright on the works they create. But there are exceptions, and universities are increasingly interested in generating income from the intellectual property of their faculty.
Also, students may be treated somewhat differently from other employees. Academic staff may be treated differently from faculty, for that matter. Ideally, the exact rules are written down somewhere. Find the rules and read them, and ask questions about them.
At UW-Madison, faculty, staff, and students own the intellectual property of the works they produce except in some special cases, such as instructional materials produced with significant university resources. And if you’re going to patent something, it must be through the Wisconsin Alumni Research Foundation (WARF).
Copyright protection gives the author exclusive rights to the work and to derivatives of the work.
Thus, the default is that no one can copy, modify, or redistribute your code.
Reproduction for criticism/commentary, teaching, and research
There are important limitations to copyright protection.
We are allowed to reproduce portions of a work as part of a criticism or commentary, in teaching, or for research.
The rules aren’t precise.
Quoting from a work is okay. Posting the full thing on the web is not.
\begin{frame}[c]{}
\centerline{Breaking copyright \quad $\longleftrightarrow$ \quad plagiarism}
\onslide<2->{\centerline{These are totally different things.}}
Just in case it’s not clear: the tradition of citing one’s sources is really totally different from following copyright law.
Works in the public domain should still be appropriately cited.
If you don’t indicate a license for your software, others { can’t reuse it. You need to be explicit about whether and how your software may be reused, by providing a license.
I choose between the MIT license and the GNU General Public License (GPL). The MIT license is as open as possible: do whatever you want, just don’t sue me. The GPL is “viral” (they say “copyleft”) in that extends to derivative works: software that incorporates code under the GPL must also be under the GPL.
The Creative Commons licenses may work for data, but they’re more for text, music, video, and such things. They {\nhilit should not} be used for code, as they are not compatible with other standard software licenses, such as the GPL. That means that you wouldn’t be able to mix in code that was licensed under the GPL. }
\begin{frame}[c]{}
\centerline{\large Pick a license, any license}
{\textendash} Jeff Atwood
I can’t emphasize this enough. If you release your software without a license, no one can modify it or incorporate it into their own software, as it’s under copyright protection.
If you want your software to be reused, pick a license, and make the licensing absolutely clear.
Copyright (C) <year> <copyright holders>
Permission is hereby granted, free of charge, to any person
obtaining a copy of this software and associated documentation
files (the "Software"), to deal in the Software without
restriction, including without limitation the rights to use,
copy, modify, merge, publish, distribute, sublicense, and/or
sell copies of the Software, and to permit persons to whom the
Software is furnished to do so, subject to the following
conditions:
The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY
KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE
WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE
AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
OTHER DEALINGS IN THE SOFTWARE.
The MIT license basically says: do whatever you want with the software, but be sure to include this notice, and don’t sue me.
Those are the key things you want: protect yourself from liability, and make plain that people can be free to reuse the software.
There is an older GPL-2. Use the GPL-3. It was updated to close some loopholes.
Key additions vs MIT license: distributions of the work or derivatives must include source code, and derivatives must also be licensed under GPL-3.
<line with the program's name and a brief idea of what it does.>
Copyright (C) <year> <name of author>
This program is free software: you can redistribute it and/or
modify it under the terms of the GNU General Public License as
published by the Free Software Foundation, either version 3 of
the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public
License along with this program. If not, see
<http://www.gnu.org/licenses/>.
To license your software under the GPL, include a notice like this.
The Creative Commons licenses are really useful for things like manuscripts, data files, videos, web sites, and such.
You shouldn’t use them for software, as they can’t be mixed with the GPL, and they make no explicit mention of source or object code. In the FAQ at Creative Commons, they explicitly recommend {\nhilit against the use of CC licenses for software.
BY means people must cite you as the originator.
SA means that derivative works must be distributed under the same license.
ND means the work must be distributed in its entirety, without any changes.
NC means the work can’t be used in a commercial setting. }
There are a lot of issues to consider.
I’d recommend avoiding ND and probably also NC.
Personally, I’m going with CC0 (by academic tradition, people should still cite you) or CC-BY. It means that a company could grab my stuff and make money off of it. But I’m fine with that. I’d rather see the results of my efforts put to further use.
Data are viewed as facts and so they {\nhilit can’t be copyrighted.
Your data file or database, though, {\nhilit can} be copyrighted, if its compilation involves some creativity, and that would generally be true for scientific data files.
So people can’t redistribute your data files unless you say it’s okay. But they may be able to extract and reformat the data and then distribute that.
If you want to prevent extraction and redistribution, the data files need a license, which would say the end user is prohibited from extracting data for uses other than intended. }
Statisticians, in particular, should want data to be openly available. And so you should cite the source of data and any relevant papers, not just because that’s the academic tradition, but also because we want to reward, as much as possible, people who make data accessible.
Even if it’s perfectly legal for you to re-distribute data, you should talk to the originator of the data before doing so. You don’t want them to get annoyed and then stop distributing data in the future.
If you want data to be reused, just put it in the public domain. Don’t add any extra complexities, like CC-BY.
If you want to control reuse or redistribution, talk to a lawyer. It seems really complicated.
\only<1-2|handout 0>{* Avoid human subjects research * [] \only<1|handout 0>{\color{background} (just kidding!)} } \only<3->{ * If there are humans involved, they’re human subjects * e.g., surveys * Human subjects research must be reviewed by an Institutional Review Board (IRB) * Not everything is research * e.g., data used solely in a course * Most things are research * If you publish a paper about it, it’s research * Anonymized data may be exempt * But the IRB wants to make that determination }
Any research on human subjects must be reviewed by an IRB.
If you’re considering publishing a paper about it, and if humans are involved, it’s human subjects research. That includes things like surveys. So informed consent, and review by IRB, with a clearly defined protocol and protection of data.
You can do a survey {\nhilit within a class and it may not be research, but then you can’t publish the results.
The NIH considers the analysis of anonymized human data to be “not human subjects research,” and it may be exempt from full IRB review, but IRBs generally want to make such determinations themselves: you need to fill out some amount of paperwork. }
HIPAA is really important, but it’s also a real pain.
The key thing is that medical data with {\nhilit any identifying information needs a whole bunch of paperwork if transferred/disclosed, and there need to be special security measures.
And the definition of “identifying information” is surprisingly broad. }
If you don’t {\nhilit license your software, it can’t be modified or reused.
Make data open, and be sure to reward those who make data and software accessible.
Be careful with human data, particularly if there’s anything remotely identifiable. }