PhD student pressured to fabricate data due to bad experiment design

Question

Summary: It took me 3 years to analyze data from one sample (n=1) whereas correct way to perform experiment would be with n>=3. Now my job and funding are in question, since I don't have more time. My supervisor pressures me to fabricate data.

I want to gather all the missing data but that will take another 3 years for same experiment; resulting in failure to timely submit internal evaluation reports, as well as termination of any fellowship. My supervisors are secretly suggesting me to somehow put some imaginary values in at this point, but that feels heartbreaking. So what is the best thing I can do now?

I am a Ph.D. candidate in soil sciences at a small academic university, where I have to do assays about different soil borne microbes such as an antibiotic assay, secondary metabolite, synergy etc. Now before starting my research, I was told to just ask the lab seniors and carry out the experiments. However, I did not have a simple and crisp understanding of biostatistics, and I was not told about what kind of statistics is applicable for my samples. I was tedious about the preparation of chemical reagents and buffers and I kept digging for historically important original papers that lack enough statistical analyses.But I missed that in the future, I will be required to submit statistical analyses that are up to the mark of current trends. I just kept doing simple experiments as shown in previous degree courses, without statistical repeats.

Honestly at this point, I am totally depressed. People all around me are pressuring me that if I do not fill up my dataset with imaginary data, my paper will not be accepted to any journal.

So what should I do in this scenario? I am extremely, severely depressed and go to therapy.

Update to answer questions asked in the comments:

1. Locale: Some Asian country.

2. Journals we choose: Very low impact, paid, costly. They prefer jargon rich language instead of too original language that describes how we reach to conclusion. If I write about a too detailed explanation, it would be double checked by my seniors, and at 'safest', they would be backspaced.

3. Impact of reporting? From word of mouth, perhaps many other people at respectable positions here do (and did at past) this quite casually, and rivalry among them is intense. Except this incident, my supervisors were humanitarian towards me and many other people. Now at this point if I am selected by a rival employer (although that isn't going to happen without the explicit consent of present supervisors), there is no way the rival employers are going to trust a traitor.

4. People are bloodthirsty for job and career (than actually learning), and they are immoral up to the point of stealing my personal stuff, hiding equipment, tearing journal pages to wrap glasswares, mouth pipetting broths risking own life. Period. Nobody is going to witness me. Nobody is going to get my back. Everybody will disagree with me.

I am under such a point where a PhD dropout seems more respectable than a finished PhD.

Update: Based on few more comments and answers:

1. Yes I have considered of personal safety - I took a wait and watch strategy for a while.

2. No, we don't have a statistics office. We have a stat department though, collaborating them might require consent from our department. I will consider this option. Calling the quality checking office is going to instantly identify me, and retaliation.

3. In several other university departments where I studied at past, I have collected a lot of conflicts, again, without written evidence. It is not possible to keep the mobile recorder 24 hour on. Whatever, they are not going to issue me any recommendation letters. For sure.

4. it is unlikely that I will zoom call or network. I have a very little motivation left. I want to choose some rather low paying career such as providing private tuition to children or to teach drawing or guitar or crafting, or to start a career as an animator. This will provide me more creativity and content. I would not require constantly evaluate the moods and intents of others. In future if a new system appears where people spontaneously contribute to research, maybe I will consider that in old age. I am happy with little, and that is enough for me.

5. I would suggest the scientific community to develop some other technique to prevent or detect this type of situations. Make it mandatory to submit all photographic/ videographic evidence for each and every claim. Make it mandatory to submit each mathematical details, procedures to making mixtures, etc. Also publishing behavior can say a lot. Especially a prolific amount of publication may be a red flag for predatory behaviour.

6. System needs a lot other change. Do not mandate Ph. D. for jobs where Ph. D. level skills is not necessary. Also make an outlet for partly incomplete informations to be published, with intellectual properties associated to the students; and unlike peer review by a few referees, let it be reviewed by entire world. let them make comments and collaborations.

Update:

1. I talked to superviser and made it absolutely clear that under no circumstane I will execute that suggestion. They agrees to it, sort of. They tells me to focus on "actual work" of pouring and mixing all those, than neatpicking too much with all those theoretical stuff.

2. I am seeing this type of misconduct as more of a systemic fault than individual fault, because lot of people talk about 'putting some values' quite casually and are achievement-thirsty. I maintain a distance with this kind of people because every time they makes me feel ghosted, ignored, ashamed of not completing the curricula 'as intended', or blameshifts when I try to do anything new.

3. Looks like the Ph.D. memes on social media indeed reflects a lot of unreported truth. Recently I have found a meme regarding n=1 back from 2019 that also indicates somebody else also made a similar blunder.

4. I have also heard, some of our very reputable professors said the ethics curriculum is very "boring"... that reflects they don't know to teach it an insightful way.

5. I met a very senior scholar (who is now in no means doing research or academic job though) said me once that there is credit in filling up with data because it takes "knowing the work better than others" so that "only naive people are caught".

Small update regarding demographics: Our former state education minister, currently one of the supreme power, just caught by detectives for laundering money worth approx 300000 USD (at least) while investigating a huge school teacher recruitment petitions and protests. A similar university like us, known to award him a fake Ph. D. degree as well other professors are hiding their opinion about this education minister's Ph. D. guide, and the responsibility is being pushed solely towards his guide. But we all know the rapport and networking among all those professors. This is the level of corruption in the authority to whom we are expected to "report" something.

Answer 1

1. Never fabricate data under any circumstances. That is the worst form of academic dishonesty there is, and is unforgivable. The scientific process depends on our ability to build on the work of others, and we can't do that if we can't reasonably trust that work. (Honest mistakes happen, as do poor studies. But if a flawed study is written up honestly, we can judge how much we should trust the results based on the experiments described. We can't do that if the paper contains lies.)

2. Report your supervisors immediately. They are terrible people and need to be removed from the scientific community as soon as possible to limit the damage they can do in the future and begin mitigating the damage they've already done. Possible people you could report to include the department chair and the editors of any journals they've published in. Try to get written evidence of the suggestion first, but if you can't, you should still warn the chair so people can keep a closer eye on them in the future. The scientific community is being actively harmed as long as they remain within it.

It sucks that you've found yourself in this situation. If you don't know the chair well, you might want to talk to other professors who you trust (preferably tenured) to get some help dealing with this. Having someone with real clout backing you up could make it a lot easier. One good aspect to come of this: if you successfully uncover people who are fabricating results, you will have made a positive contribution to science by having done so, and will earn the respect of anyone whose respect is worth having.

Answer 2

This sounds more the fault of your advisors than yours for not equipping you with the supervision you need to perform your experiments up to standards.

Do not under any circumstances falsify or misrepresent simulated data as actual data.

Your advisors should help you come up with a solution for moving forward, with the understanding that their own lack of supervision has caused this problem. If their suggestion is to falsify data, well...that tells you everything about their ethics. I would feel that a PhD earned working with those people is worth the same amount as that fake data is. Hopefully you have misunderstood and this is not what they are suggesting. If they are suggesting this, then I have to say they have much more to lose than you do, and if you have documentation of this suggestion and this documentation is shared with the university administration or the public, those supervisors are going to be in quite a lot of trouble.

From a statistical perspective, I do not think the scenario as you describe is as fatal as you make it sound. There may be a standard in the field that samples are collected in triplicate, but nothing about statistics prevents you from analyzing data that are not taken in triplicate. It may be that you do not have enough samples to perform a sufficiently powered null hypothesis test, but performing triplicate sampling doesn't actually give you more samples, it just refines your estimates a bit. I assume you have more than just 1 sample per group total, though; if not, then it seems very unlikely that even 3 would be enough.

Not every project in soil science requires 3 years to complete. Do a version of your investigation that can be done in 1 year. Work with your advisors to figure out what this would look like. Use the work you've already done to make this an efficient 1 year. It's normal in science to have to throw away years of work; often, the time elapsed is about learning the methods and once you know what you're doing it can be performed quickly.

Answer 3

Following up on @BrianKrause 's excellent answer.

Do not present fake data as real.

You might be able to write something useful and interesting with the data you have along with several different invented or simulated data sets that you properly identify as such. Can you discuss the consequences for your work under different assumptions about what the data you do not have might show?

Your advisors might (should?) accept this kind of compromise as an apology to you for failing to instruct you properly from the start.

Answer 4

I'm also going to point out that if it took 3 years to do the experiment, then assuming you're researching something useful then the data is massively valuable and you need to get it written up. Someone else can always spend another 3 years adding to the sample size (or possibly, if you have learnt things which could make it faster and you document them, less time than that), but your "initial" results are still valuable. Just make sure you document the experiment such that its repeatable.

Answer 5

I will answer this from a statistical viewpoint and omit any criticism of your supervisors.

You have n=1 but claim to need n=3. I assume this is a consequence of some statistical tool, which simply cannot be used at all for n=1. I further assume this is due to a frequentistic approach in this tool, i.e. parameters found by this tool are direct functions of the data. With this default approach, it would be indeed impossible to come to statistical conclusions, but by altering your approach, we can remove this constraint.

Specifically, we move from a frequentistic to a bayesian approach of your statistical analysis. For this, we place meaningful prior distributions on each parameter in your statistical analysis. This makes it possible to infer solutions from n=1. Your solutions are then represented as a posterior distribution, which represents the uncertainty in each parameter. This uncertainty will be high for n=1, assuming your prior distribution is broad enough, but the solutions are perfectly valid from a statistical viewpoint.

There are two drawbacks: First, the mathematics can be difficult. It might take you a few weeks or even months to understand this approach and implement it correctly (but this should be better than 3 more years of experiments). Second, picking decent prior distributions can be difficult. By definition, you can pick prior distributions in a way such your solutions will tell whatever you want. This is a bad approach and people will criticize this. Instead, you have to find 'what-makes-sense' in your application and what other people can agree on. This can be very tough, but, as a domain expert, you are in a good position to find such priors. Good luck!

Answer 6

The experience you gained in this 3 years is invaluable and it is written nowhere that everybody has to finish their PhD in 3 years. I am not saying this to downplay you, but jus to state the obvious: unrealistic expectations can destroy the mental wellbeing of a person and it will lead to unethical practices (sorry for stating the obvious, and I know that most likely in your culture all these words are just empty words ... they are empty words even in many of the progressive, western mighty countries).

My supervisors are secretly suggesting me to somehow put some imaginary values in at this point

You have to clearly stand your ground. Explicitly tell your supervisor that you will never produce values that were not measured in the lab. Your supervisor must provide you funding to repeat the experiments. This means fundings for 3 additional years must be secured.

Since this is going to be unlikely, that request must be your "ice-breaker" to fight to get funds for at least 1 year and start the needed sampling for statistical repeatability. Additionally, you are blocked with your supervisor, but you must defend your integrity: data cannot be faked (not even at a Bachelor degree level). You have to fight because he has to find a way for you to publish, or at least to defend your thesis ("how sure are you about your results?" "our interpretation is supported by the experiments performed in the previous years, experiments are being repeated and when results will be available we will publish in an international peer-reviewed paper).

tl;dr

You have a crappy supervisor. Both at the intellectual and moral level. It is not your fault. Either you win their support and bring this PhD to an end, or you have to start it again in a different institution/country. With your experience, re-starting a PhD will be an experience similar to half a post-doc. No person is equal to the other, but no one will judge you because it took you 3 years at an institution and 3 years at a different one to complete your PhD.

Answer 7

1. Quite obviously, most people suggesting answers in this thread are coming from the western culture and severely underestimate the consequences for you in the case they report their advisor at the point when their career is going through such a rough moment.

   Fighting and reporting advisors could be postponed until you are in a safer position. I would suggest this path only if you had enough material for the defense, and they were clearly failing you, which apparently is not the case. Also, in such a situation you would be unlikely to get a reference letter.

2. You should 100% go to therapy independently of how the story evolves. Therapy is a resource, it's a new empty room in your house where you can relax, get support, and maybe even figure out what to do with the rest of your house. Also keep seeking for support besides therapy. Asking a question here was a right step.

3. Honestly ask yourself whether you are still interested in pursuing career in science. If not, there is no point in fighting for PhD. And there is absolutely nothing wrong about such a choice. Do whatever makes you happy, self-content, etc. Taking a break could also be an option.

In what follows I will assume that you want to stay in academia.

4. — What do all great scientists have in common?

   — They all had a great advisor.

   You advisors are clearly trash, from both scientific and humane perspective, and I can hardly imagine that you will benefit from interacting with them.

5. If I could give just one piece of advice to young scholars, it would be "find an advisor who is a decent human, i.e. the one caring about his students". How is this applicable to your situation? One option I can suggest would be restarting / transferring to a different PhD program after you find a new advisor. I know people who did this after 2-3 years of PhD, I did this myself after 5 years, and I have never regretted it.

I don't know if you have any circumstances completely eliminating such an option, so I will assume that you would consider it.

6. Advisor search process tuned to your case:

   • Look up people in your field and find those whose ex students have nice careers, ideally in universities of higher rank than where they did PhD.
   • Once you have a shortlist, contact them directly, asking for a zoom talk asap. You need to explain your whole situation and see if they may be interested in you.
   • Try to talk to professor's current students.

7. At this point, you may feel like you are a loser carrying the burden of bad experience. **But you are also a tough one!

   • You have quite a lot of independent (with little support from you advisors) research experience;
   • You have more knowledge than a fresh PhD candidate;
   • You are clearly passionate about your science, if you are willing to take the risk of devoting to it more of your lifetime after what happened before.

Who would not want to get such a PhD student???

Answer 8

Under no circumstances should you submit fake data. You have data that is useful, that took a long time to collect, and even if your methods were not perfect, you wouldn't want to throw it away.

You could try to learn exactly how bad having one repeat instead of three repeats is. Add this information to your paper. It will likely improve how your data is viewed (you may be able to show that your data is not perfect but still useful) and that information on its own is valuable as well.

Answer 9

Comment: Sorry, I don't have enough points to post a comment. But I'd like to suggest that you should definitely get and read this book...

"Stealing Into Print:   Fraud, Plagiarism, and Misconduct in Scientific Publishing"
https://www.ucpress.edu/book/9780520205130/stealing-into-print

where you'll see that the two most common (among many other) unethical academic publishing practices are (1)fabrication and/or falsification of data, and (2)plagiarism. It also goes into several hundred pages of detail about the issues troubling you. But my (somewhat dim) recollection of that material is that it won't really resolve these issues for you. However, I hope that maybe it'll help you to see that they've been very carefully thought about, documented, and discussed.

P.S. Hmm, I wonder how much of that book is completely honest ( sorry, couldn't resist :)

  E   D   I   T
In reply to the comment "your answer is unclear" below, the op seems torn between the alternatives of (1)behaving in what he feels is an unethical manner, or (2)abandoning his career path. I'm suggesting that, although the book ain't gonna resolve that dilemma for him, it nevertheless provides numerous real-life case studies and in-depth discussions of them. And all that may help the op put his situation in context and in perspective. For some of those cases discussed in the book (as I recall them), it's pretty much similar to the op's "publish or perish" situation. Whereas for others, they're just looking for additional grant money, and don't mind cheating to get it. And lots of cases in some kind of in-between situation. So there's a whole spectrum of ethical-->unethical. And maybe the op can reconcile his "Sophie's Choice" by realizing there are many more people willfully choosing to do many worse things than the choice he's inescapably faced with.

Answer 10

Yes, fake data is anathema to the very essence of science.

However, you are a lone crusader with next to no power struggling to survive in a completely corrupted environment. Noble ethics is not going to help you.

"Virtue is its own reward - it has to be!"

You have my most profound commiserations. Maybe it is because I am getting old, but it seems to me that more and more of academia is sinking into this morass of cut-throat careerism and "who gives a f*ck nobody is going to read this closely anyway"

Answer 11

I've worked in several countries in Asia and I think all the advice about collecting incriminating evidence about your supervisor or reporting them will probably not be helpful in your case. If an issue comes up now or in the future, everyone will work to ensure the survival of a professor and will not think twice about throwing a disposable student under the bus.

All the advice about not ever faking data is gold. You won't be able to live with yourself and if it does come out later there will be a bus waiting for you to get thrown under.

Hang in there; this too shall pass and as @ mavzolej points out you will be stronger and more experienced when it does.

Welcome to Stack Exchange!

There are almost 200 different sites to choose from and one of them is Statistics!

Perhaps this is an opportunity to join a second site and discuss the statistical situation that you are in in more detail. Have a look around that site to see what question asking looks like and then start asking questions.

• Is there any way I can say something of significance with n=1?
• Is there some subset of the work I can do instead of going all the way to n=3?
• Are there alternate analyses I could try? etc.

You can link back to your question here if you want to add some background but mostly just explain your statistical situation.

You can probably find a way to anonymize yourself there (it's a popular site!) and no need to mention soil science or link back here if you don't want to.

Use this as an opportunity to possibly learn some new statistics and to learn more about Stack Exchange - it can really be incredibly helpful in many ways.

Answer 12

Regarding your PhD keep in mind a PhD is a learning experience. What you have to demonstrate are not results, but the ability to investigate. Demonstrating that something is wrong or explaining you did not get the expected results is a valid thesis. Just make extra emphasis in the state of the art and why your approach won't work. This will help some other experiment with similar properties to avoid losing 3 years due to poor experiment design.