If you are in a class I am teaching or working with me, here is a description of the standards by which I try to work (over and above those required by my employer, host institutions and professional bodies to which I belong). While I understand that different cultures have different standards on issues such as plagiarism and authorship, I have outlined a non-exhaustive acceptable set of practices below.
If you want to work with me, please email this page back to me with a statement that you have read it and agree to try to stick to the guidelines.
Table of Contents
- Plagiarism and collusion
- Acknowledgements and funding bodies
- References, citations and ‘personal communications’
- IRB/Ethics approval
- Double publishing
- Conflicts of interest
- Source code and open source
- Presentations, conferences and other ‘publications’
- Sharing results before publication
- Working expectations
An interesting and readable article on these issues, can be found in the New Yorker. You can also find a guide to Oxford’s Plagiarism Policy and Disciplinary Procedures here. Note the distinction between unintentional and deliberate plagiarism. While the former is less egregious, it is still unacceptable, and should be remedied immediately (in less than a week) on discovery. This doesn’t avoid disciplinary action, but it may lessen the magnitude of the repercussions.
Durham University has a good definition of plagiarism and collusion:
“Plagiarism: unacknowledged quotations or close paraphrasing of other people’s writing, amounting to the presentation of [another] person’s thoughts or writings as one’s own. This includes material which is available on the world-wide web and in any other electronic form”.
This also includes algorithms and computer code, even if you are translating them into another language. Of course, some algorithms are so standard (such as the FFT) that they do not require citation, although the specific implementation does require credit. From my point of view, this is just good scientific practice, since not all FFTs are the same. It’s important that the algorithm has been benchmarked and the specific version is noted in case bugs are found later which invalidate or weaken your conclusions.
Durham University’s definition of collusion is as follows:
"Collusion: working with one or more other students to produce work which is then presented as one’s own in a situation in which this is inappropriate or not permitted and/or without acknowledging the collaboration".
Neither is acceptable. Even if you are working in teams, you must describe who did what - see ’authorship’ below.
Many journals do not specify what constitutes authorship on a scientific article. I think the "Authors’ contributions" section of Biomedical Engineering Online provides an excellent description:
"An ’author’ is generally considered to be someone who has made substantive intellectual contributions to a published study. To qualify as an author one should 1) have made substantial contributions to conception and design, or acquisition of data, or analysis and interpretation of data; 2) have been involved in drafting the manuscript or revising it critically for important intellectual content; and 3) have given final approval of the version to be published. Each author should have participated sufficiently in the work to take public responsibility for appropriate portions of the content. Acquisition of funding, collection of data, or general supervision of the research group, alone, does not justify authorship.
All contributors who do not meet the criteria for authorship should be listed in an acknowledgements section. Examples of those who might be acknowledged include a person who provided purely technical help, writing assistance, or a department chair who provided only general support.
List the source(s) of funding for the study, for each author, and for the manuscript preparation in the acknowledgements section. Authors must describe the role of the funding body, if any, in study design; in the collection, analysis, and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication."
Different fields consider the subject of authorship in different manners though. Clinical publications often carry authors who ’just’ collected the data to provide sufficient credit for the enormous effort this involves.
If you work in my research group you will be expected to co-author with other researchers and academic or clinical staff. Unless the article is in a completely unrelated discipline, you will be expected to pass all potential publications to me before presenting in public, placing on any website, submitting to a journal/conference/workshop etc, or presenting/publishing in any public way. This is to protect the professional reputations of all involved. If you wish to claim sole authorship of a work (such as an article or patent), this will be considered by myself in the first instance, and then if I disagree, by a departmental committee.
Authorship order is often contentious. Generally the first author has to contribute the majority of the intellectual input, work on the project, and work in the manuscript and the senior (last) author is the person who ran the project, designed a lot of the data collection and analysis. The rest of the authors tend to get listed 2nd to N-1th in order of contributions, although it’s not worth haggling about these positions. This formula is not always true though. Although it’s a good idea to set out authorship near the start of a project, this is not always done or possible. The senior author/PI has the final word in assigning authorship, although my general policy is to appoint a first author, and ask them to draft author order based on their own perceptions of contributions. It should be noted, however, the primary author must continue to take the lead on research and publication throughout, even throughout submission, review and final reporting. Faliure to do so will mean first authorship is assigned to another author, who is willing to step up, except in rare cases (such as serious illness during the later stages - e.g. submission and review).
One caveat. If you do not answer emails in a reasonable period of time (or leave a forwarding email), I may choose to publish some of your work without your agreement. This is to ensure that our research progresses despite your omission.
Timing is everything
Once you’ve created your shortlist of authors, contact your supervisor before the other authors, or if you are the PI on the study, the other grant PI(s) and decide if you agree on the list. Resolve these differences before involving all the authors. Err on the generous side initially. Push disputes up the line, not sidewards to those uninvolved in the process. This all needs to happen at least three weeks in advance of any deadline. Any less is rude and doesn’t give time for those on holiday to respond or check for the inevitable errors. Check your contracts too - some research agreements require you to provide a draft copy of the intended publication to the sponsor (particularly if the research is partly or fully industry-sponsored) so that it can be reviewed for patentability, trade secrets etc. In these cases you need to start this process at least three weeks *plus* that review period in advance of any deadline. If you work for me and don’t do this, then I’ll refuse to let you submit. You do need your PI’s permission to submit anything - their name is on this too (or should be), so they have a right to a reasonable review time. If you have more than one supervisor, go to the one who you consider would be the senior author, but keep the other one in the loop, even if they are not an author.
The following is from Biomedical Engineering Online’s instructions for authors:
Anyone who contributed towards the study by making substantial contributions to conception, design, acquisition of data, or analysis and interpretation of data, or who was involved in drafting the manuscript or revising it critically for important intellectual content, but who does not meet the criteria for authorship should be named in an acknowledgements section. Please also include their source(s) of funding. Please also acknowledge anyone who contributed materials essential for the study. The role of a medical writer must be included in the acknowledgements section, including their source(s) of funding. Authors should obtain permission to acknowledge from all those mentioned in the Acknowledgements.
It is important that all (key) relevant materials are cited in your publication. This includes anyone who’s work you have used, and any work you have considered but rejected. When citing the work you should be careful to indicate why you have used or rejected the material. Note that if the source you cite is not primary, you need to cite the primary source as well (or instead). Note, you cannot cite a paper to justify or explain a technique which does little more than cite another paper to justify or explain a technique. It is important that the format of the references remains consistent. Although some journals require the Harvard style, I prefer Vancouver. The following is from Biomedical Engineering Online’s instructions for authors (and is the Vancouver style):
“All references must be numbered consecutively, in square brackets, in the order in which they are cited in the text, followed by any in tables or legends. Reference citations should not appear in titles or headings. Each reference must have an individual reference number. Avoid excessive referencing.
Only articles and abstracts that have been published or are in press, or are available through public e-print/preprint servers, may be cited; unpublished abstracts, unpublished data and personal communications should not be included in the reference list, but may be included in the text.
Obtaining permission to quote personal communications and unpublished data from the cited author(s) is the responsibility of the author. Journal abbreviations follow Index Medicus/MEDLINE. Citations in the reference list should contain all named authors, regardless of how many there are.
Web links and URLs should be included in the reference list.” They should be provided in full, including both the title of the site and the URL, an author (if known) and the date accessed (since the web is ephemeral). Note that a thesis can be considered a peer-reviewed publication to cite, but it is of little use if the thesis is not readily available. If a paper exists, it is best to cite that instead (or as well), or if the thesis is publicly available, then a URL can be provided.
If your work involved living subjects other than yourself, you (or whoever collected the data) must have obtained ethical approval for the study. Again, BMEO’s policy provides an excellent guide which I paraphrase here:
Any experimental research that is reported in the manuscript must have been performed with the approval of an appropriate ethics committee. Research carried out on humans must be in compliance with the Helsinki Declaration, and any experimental research on animals must follow internationally recognized guidelines. A statement to this effect is generally required in the Methods section of the manuscript (particularly in medical journals), including the name of the body which gave approval, with a reference number where appropriate. Informed consent must also be documented. Manuscripts may be rejected if the editorial office considers that the research has not been carried out within an ethical framework, e.g. if the severity of the experimental procedure is not justified by the value of the knowledge gained.
The NHS provides a quick lookup sheet here. There appear to be some circumstances in which ethics approval is not required, but you need to check the current / local rules and note that IRB/ethics approval requirements change from time to time.
Any manuscripts submitted to a journal must not already have been published in another journal or be under consideration by any other journal, although it may have been deposited on a preprint server. Manuscripts must not have already been published in any journal or other citable form, with the exception that the journal is willing to consider peer-reviewing manuscripts that are translations of articles originally published in another language. In this case, the consent of the journal in which the article was originally published must be obtained and the fact that the article has already been published must be made clear on submission and stated in the abstract. Manuscripts that are derived from papers presented at conferences can be submitted unless they have been published as part of the conference proceedings in a peer reviewed journal. Authors are required to ensure that no material submitted as part of a manuscript infringes existing copyrights, or the rights of a third party.
You can take concepts from previous articles and develop them, or write expanded versions of conference articles in some journals. It varies from journal to journal and field to field. Medical journals, although much more relaxed on authorship, generally don’t allow you to publish any concept (in an expanded form or otherwise) that has previously been published or presented at a conference.
See Cal Tech’s resource and this article for definitions and a discussion on how much you can borrow before you are infringing on copyright. For example, you can photocopy or print a certain percentage of many copyrighted journals or books for personal or teaching use.
Some journals require you sign over your copyright to them. I would discourage this practice. See this webcast of Hal Abelson from MIT for a good justification on why we shouldn’t trade our IP for relatively low up-front publishing costs. Other journals, such as BioMedical Engineering OnLine allow authors to retain copyright to their work for a slightly higher up-front publishing fee. Access then is free, and you are more likely to be cited. (Three times at the last count.)
More on Open Access Publishing can be found here.
The university of Oxford states: "A conflict of interest arises where the commitments and obligations owed by an individual member of staff or student to the University or to other bodies, for example a funding body, are likely to be compromised, or may appear to be compromised, by:
- that person’s personal gain, or gain to immediate family (or a person with whom the person has a close personal relationship), whether financial or otherwise; or
- the commitments and obligations that person owes to another person or body.
There can be situations in which the appearance of conflict of interest is present even when no conflict actually exists. Thus it is important for all staff and students when evaluating a potential conflict of interest to consider how it might be perceived by others. The duty to declare a possible conflict applies to the perception of the situation rather than the actual existence of a conflict. However, the duty is not infringed if the situation cannot reasonably be regarded as likely to give rise to a conflict of interest. Conflicts of interest may be financial or non-financial or both."
Further information about both types of conflict and Oxford University’s conflict of interest policy can be found here.
Basically, if you know someone involved in a project that you could influence or could influence you, or if a relative or friend is financially connected with your research or publication, you need to declare this. It doesn’t invalidate the study but not declaring this, may cast significant suspicion on the research at a later date. Not declaring interests may even lead to your employment or course of study being terminated, amongst other repercussions.
I’m not against patents. They are designed as a reward for publicly disclosing an invention from which everyone can benefit (particularly when the patent period ends). The incentive for this public disclosure is a limited period of protection from competition. Patents also provide incentives for investors to back research and companies that translate the research into a practical product. As far as I am aware, nothing stops you from developing research using patented material, in the same way you can use any published material - you just can’t make money from it, or let others use it, unless the patents are licensed.
However, patents can limit from where you can receive funding (because of conflicts of interest). Patenting an idea may also disincentivize others to work in this area, diminishing the impact of your research. It’s also an expensive and time-consuming process unless someone else is willing to do all the leg work for you (which is unlikely).
I am against bad patents, or arbitrary patenting of concepts on which the inventor hasn’t spent much time or demonstrated the system works well on more than a few examples.
If you want to patent something you are working on with me, please come and see me to discuss this as early as possible. If you’ve already published it, you can probably still patent it (in the US only up to 12 months from the initial public disclosure for as little as an intial cost of $100). Costs mount over time as you have to file a full patent and probably use lawyers, and then a few years later start paying maintenance costs.
Source code and open source
Please DO NOT post any code you write that is related to research in our department (or elsewhere) in a public repo. Use a private github repo in the lab (not personal) github group.
There are many reasons for not posting code at an early stage:
- You agreed to the T&C’s of working with me here
- You are giving away your (and my) ideas to others so they can beat us to publication and grants. It literally risks our careers.
- It’s a work in progress so it’s likely to be wrong - you are doing a disservice to science
- It makes us look bad and harms our professional reputations
- It hasn’t been reviewed, so it’s probably contravening the licensing of the file, so odds are you’ve opened yourself up to legal action
- You’ve ruined any chance we can patent it. It’s a public disclosure.
Eventually your code will become public, but not until several of us (including me) have reviewed it and I’ve signed off on it.
Most code you find online will carry a license with it. If it doesn’t, it probably isn’t free and you shouldn’t use it. It’s hard to cite it, but it still isn’t yours. Licenses vary in the extent to which you can use, copy or embed them in other code, and often their licenses require you to inherit their licenses (‘copyleft’). As Richard Stallman will tell you, ‘free’ doesn’t mean free as in beer, it means liberated, as in open to scrutiny, which is what science should be. You can copyright your code, and license it to third parties, retaining extensive rights for yourself and others. You can even patent code.
If you work in my research group, your code will probably inherit an open source license and end up in the public domain. If you are unhappy with this idea, let me know up front (N.B. You can still patent the idea for which you have released open source code! If it is after you publicly released the code, then only in the US - where you have 12 months to file a provisional or full patent.).
If you are confused about which free software license to use (and have the liberty to choose), please see this quick look-up chart.
Generally I want to open source code. I think it’s good for everyone and certainly for scientific repeatability. Before posting anything though please check with me. Also check:
- Any other authors who have made ‘substantial’ * contributions (materially or intellectually) to the code are happy too. If they are not, let me know and I’ll adjudicate a reasonable timeline for posting. If you are unsure of authors, please contact me.
- The code doesn’t contain any passwords or PHI (whether you introduced it or not).
- The code has a list of authors, dates and open bugs, limitations, and is commented at each line to show specifically what is yours and when it was done. Notably, each source file should contain a header indicating the author(s), license, date of publication and date of last change (and what the conceptual change was - e.g. ‘minor bug fix to stop buffer overflow in zoom’, or ‘major redesign of annotation interface to allow zooming in x and y’ etc).
- A publication referring to the code is posted on the site where the code is located - probably in a readme file (so this means generally waiting until we’ve published at least one paper on the use of the code).
- Double check with Gari and let him know in an email that all the above steps have been done.
* ‘Substantial’ is open to interpretation, but generally is along the lines of publication authorship. This generally refers to other researcher in my group or collaborators (but generally not people helping with annotation or verbal comments on the design of the interface, such as alpha or beta testers).
Any presentation outside of our group meetings should be considered ’public’. This means you are presenting *our* research (not just yours) and so you should send me your material intended for publication (e.g. slides) well in advance (noting any other contributors prominently in the material). Your research reflects the input of your collaborators and so they should be aware of it in advance, and be given the chance to correct errors or contradictions. You rarely work alone!
You should talk to me before sharing research and results with other groups or individuals. Generally I like to share and work in an ’open’ manner, but sometimes this can be sensitive. However, you may find that I share your research with select colleagues without discussing this with you first. I don’t expect to have to pre-approve every communication I have with colleagues.
A DPhil / PhD is rather like working in a startup, or an apprenticeship. You will do a bit of everything from data collection, to fixing computers and entertaining visitors. Clearly a large part of your time will be spent coding and writing up reports. As your supervisor I will try to meet with you for an hour every other week - sometimes more, sometimes less. I’ll respond to emails almost daily and I’ll expect you to do the same, except during holidays. You can take your holidays when you like, but let me know in advance if possible, just in case there’s an important funder turning up.
As far as academic freedom goes, your independence is extremely important, and you must plot your own course (to a large extent) through your degree. However, your freedom cannot be unlimited for three main reasons:
- Your funders usually have some restrictions, such as a particular focus area, an internship with a company, etc.
- You don’t work alone. You will be working in a group, and so you have to consider the needs of others, making sure your research doesn’t overlap too heavily.
- As you are putting your career in my hands, I am also putting my academic reputation in your hands, In that sense, it’s a 50:50 relationship. If I am giving my time, then I will from time to time (but certainly not always) veto certain paths in your research which I am pretty sure will be fruitless. That’s my responsibility after all.
I don’t have core hours, but I expect you to be in the office on average 40 hours a week (or whatever your contract/agreement states) Monday through Friday.
Try to be around 10am-3pm at least every day. It’s important you communicate with the group for most of the day, are able to share knowledge and absorb it to move quickly as a team. Research groups work as a cascade of knowledge down the gradient of age (and often up). The lifeblood of research is mentorship, and you will absorb lots in the first year or so and then increasingly pay it forward to the incoming cohorts. For that reason I don’t allow remote working, except on the rare occasion.
Even if you are a student, and all your courses are over, you still must be in the research group physically or you are wasting my time and absorbing knowledge without paying it forward to the upcoming cohorts of students. We stand on each other’s shoulders.
If you want to do a rotation with me, make sure your course load is low, so you can spend sufficient time with my team to make it worth the effort (on both sides).
It’s important you make any meeting I schedule and these might be early morning or late afternoon because my collaborators are clinicians. The flexibility goes both ways though ... if you decide you want to come in at lunch time one day, good for you. I want you to work when you feel at your best. Get sleep, work out, get good food, spend time with loved ones. I hate presenteeism. Learn your own optimal times for working (but balance it with the need to communicate with the group and have some core hours on most days). If we do have a meeting though then I expect you to be prompt at meetings.
I have a tight calendar, so if you are 10 minutes late to a meeting it significantly impacts it. I’m pretty relaxed, but it results in a poor impression on my collaborators.
It subconsciously sends the message that you think your time is more important than theirs. I know - I can hear my family laughing at me. I’m terribly guilty of this. I’m working on it myself. I’m sorry about that, but please don’t copy me. I don’t role model time keeping well.
Taking days off with holiday pay or flex time is fine with me but please check in advance and drop it in the calendar so I don’t try to schedule meetings with you or try to ask you a question while you are out.
When you work as part of my team, you represent it. When you email anyone, please try to be formal enough.
Use titles until they request that you don’t or have written back to you (directly) using their first name only in the signature at least 3 times.
Use a signature in your email that correctly represents in which research group you work.
Do not chat, instant/text message or cold call someone until they have responded to the email and indicated this is OK.
If you chat me through any of my devices I’ll block you and be less inclined to answer your emails because I find it intrusive.
I use chat/instant messenger type applications for colleagues (including students) I know well. Please don’t leave me voice messages, as they may take a while to reach me. If you want a meeting, please contact my administrative assistant, and cc me on the email so I know who you are and why you are requesting a meeting. I will then email my assistant back to let her know the priority of the meeting. My assistant will not make a meeting time for you without checking with me first.
I only provide recommendations for students or employees/visitors who have spent a substantial amount of time under my direct supervision, either in class or as part of my group.
If you need a reference, please email me at least 4 weeks in advance to ask me if I can do it. You may have to email me me a few times to receive a response, as my volume of emails means I must triage my funded projects and job responsibilities out to the top of the pile.
Please forgive me if I don’t respond. If you have emailed me three times (5 days apart) and received no response, then I would strongly advise looking for a more responsive human. If I do respond positively, I need several pieces of information from you, described here.
First, please provide a short synopsis of the position for which you are applying with a link to the opportunity on the web.
Brief and succinct emails help me respond quickly. Long emails are triaged to the bottom of the pile, and those that require me to repeat the information below or ask you for more detail (i.e. a short paragraph describing the position, or a reminder of who you are [sorry - I have taught hundreds of students, and interact with thousands of people, so I may not remember you immediately]) will reduce the probability that I respond.
If I respond positively (i.e. I feel I have the knowledge to provide a strong reference and have the time), then I need you to provide me with a draft letter.
Please provide a draft letter which indicates who am I (what makes me so important that my reference means anything), how I know you and our relationship, for how long (if less than a year, #months), specifically what you did, how you performed, how well you ranked in the class/your peers, any results/impact (give very specific examples that illustrate the point), how well you take direction or respond to criticism and mentorship, how well you work in a team and help others out, why you left (if relevant), would I employ you if I had a relevant position, how I view your potential to succeed, any other key attributes/achievements that address the specific needs of the post/chosen career.
If you are a student and want me to talk about your grades and/or publications, please send readable digital versions.
Please add the full name, title and organisation (plus email) to whom the reference should be addressed at the top of the draft (please don’t make me hunt for it in an old email chain).
Provide the actual deadline (date and time in my local time zone) by when I need to submit it.
Aim for 1-2 pages depending on the importance of the position. Brevity is key, but don’t miss any of the points above! Please try to get this to me at least 3 weeks before the deadline in case I’m travelling. It takes about 1-2 hours to craft a good reference even with all this information so I need to budget time.
Don’t worry about getting any details wrong - I’ll definitely be changing them to reflect my realistic views on you. I’m English, so I can’t be hyperbolic and over-effusive in your reference, and I’ll delete any exaggerations, so please be truthful and balanced (but not modest) and it’ll save us all time. I’ll note my cultural bias (to under- rather than over-state) so the reader knows how to normalise your reference against the other ludicrous ‘future world leader’ references they receive).
Please note that you must waive your right to access the reference I write for you. (Most referees will require this, so I suggest you do this for all of your referees.) I’m not willing to spend time haggling over the nuances of a reference. I will most likely have already discussed it with you in detail if you took the initiative to provide me a draft. I am likely to tell you before hand if the reference will be poor, and suggest you find another referee. Of course, you should not request a reference from someone who does not know you well. Try to cultivate and maintain relationships with three or more mentors you know well. They will be key for written support. Try also to ensure that they have supervised you closely for a significant period of time and were in a senior position to you. A current / most recent direct supervisor is important. A mix of direct supervisors and senior thought leaders is very useful. When you contact them for a reference, send them a copy of the last reference they wrote for you (if they shared it), and ask them if they would like you to highlight anything in your CV that directly pertains to the application. If you can’t identify this information yourself then you are not going to fare well in the interview.
I’m not a lawyer, and much of the information in this document might be wrong, particularly in different regions. Laws change, interpretations change and precedents are set. So none of the information on this site should be used without consulting a lawyer and all appropriate authorities. Do your own research, and read the research and thoughts of others.
Also, none of the code you find here is warranted for any purposes. Don’t just download it and run it. Think about it, test it, improve it, and test it again.
I hope that covers it. Feel free to drop by and discuss this with me. If you see me stray from these guidelines, please do call me out; nobody is perfect!