What to do with negative research outcomes (results) of PhD research experiment?

Question

It is an expectation that the PhD would make an original contribution and/or advance knowledge in a given field. I understand this is a universal assumption for this level of study across all universities.

One of my friend's research experiment has not produced a single positive result. This was a science experiment, so it is easy to quantify whether the result is positive or negative.

[It is a bit different in the social sciences, where the outcome (result) would be that either the null hypothesis is supported or rejected (with some analysis on the effect size to make the analysis meaningful in a given context). In other words, the data analysis either supports or does not support the proposition that is being investigated.]

My question is: What should a student do if none of the research outcomes or results are positive?

Simply writing that nothing new was found does not add or advance knowledge other than to just confirm the status quo (which I guess is a form of contribution, but there has to be more than this at this level of research!).

Answer 1

Your friend might not want to hear this, but there's nothing you can do except for start over - with a different experiment. Research fails - and should ! If there isn't a risk of failure, you're not out on the cutting edge doing research.

But in most failures, there's a grain of something to build on ("from the ashes of disaster come the roses of success"). Maybe the student is too demoralized right now to see it, but almost always there's some clue in the failure that leads to a different research question worth asking.

Answer 2

I think there is a case to be done to report negative results, since it gives at least a blueprint of what does not work.

However, as Suresh mentions, usually a PhD is measure on its contribution to expand knowledge. If your friend is already 5 years into his PhD however, I think there is some adviser's fault, since he should have had some kind of insight that this thing was not working and a different course might have been wiser.

Answer 3

A defensible null result (that is being able to definitively say that something isn't there as opposed to not being able to say anything either way) is a result and does advance the frontier of knowledge.

This should be obvious.

If that kind of thing can't be published in your friend's discipline then there is something seriously wrong with the culture of that discipline.

To be sure, null results are generally not sexy and can't expect to get into a first rank journal unless there was a widespread expectation that this was a shoo-in, but it is still real science.

Answer 4

Technically, you have nothing to face failure except doing another experiment as @Suresh said, Consulting supervisor and other acknowledgeable people and carefully looking again to the problem formulation and to your solving method. Here the goal is to identify the error.

More importantly, at least to me, is the non-technical reaction for such failures. I reward myself by relaxing, playing more with the kids, sleeping early; playing some games or sometimes watching movie. I try my best to forget the problem for one or two days..

Answer 5

Two major thoughts:

1. A negative finding is still a finding. Publishing failures is a harder road to publication, but still a valuable contribution to science as a whole.
2. While this might be a little late for your friend, I always advise looking for "fail-safe" research questions for dissertations or other research projects where a student's success depends on a single finding. What I mean by fail-safe is that research questions should be chosen where "Yes" or "No" are both interesting and publishable answers.

Answer 6

I work in the field where publication of negative results is frowned upon, and basically impossible. The result is huge positive bias which severely impacts the ability of the field to properly judge its own advances and perspectives. This field is Artificial intelligence, and it already experienced two "winters" (check wikipedia for AI winter phenomenon). I personally think the third winter is due in two to three years, maybe sooner.

So publishing negative results is crucial to keep out the bias. In AI, every approach just works, if you read the papers (except for the approaches before the one published in the paper, those don't work well, that's why we really need to publish this one). Yes, I am being sarcastic.

On a more serious note, in engineering studies the point of obtaining PhD is inventing a new and better method for something. Of course, the theory goes, if you invent something useless (that it does not work), there is no point in publishing it. However, often people refrain from publishing methods that don't work, but they should, given the general ideas in the field. And that is wrong.

So the main difference is not humanities vs. hard sciences, the main difference is usually engineering (basically hellbent on inventing something new and reporting about it) vs other sciences. For example, the mere fact that some substance does not affect cancerous cells (is therefore useless for curing cancer) can be very important in medicine research, so other people will not waste time examining it further.

Answer 7

This is similar to another question on the topic. There is nothing wrong with publishing null results.

The goal of research is not to have positive results, the goal of research in an academic setting is to be published, read and cited. Having null results is only a minor inhibition on that. You can still get published and have a successful dissertation if:

1. The methodology was sound.
2. The argument that one could have expected your approach to work was clear and agreeable to your audience.
3. The paper was well written and clearly outlays the conclusions and implications for the field that practitioners care about.

Having an interesting and successful thesis helps, no doubt, but it is not the sole issue here. The demands of 1,2,3 are very high.