SEPTEMBER 7 — Imagine spending months, perhaps even years, testing an idea. The experiment is carefully designed, samples are collected, and the data are painstakingly analysed. Expectations are understandably high. Perhaps the study will identify a promising new treatment, uncover a risk factor for disease, or confirm a relationship that scientists have long suspected.

Then the results come in. There is no significant difference. The treatment does not perform better than expected. The relationship you thought would appear in the data simply is not there. Has the research failed?

Not necessarily. In fact, such a result may be one of the most useful pieces of information that the research produces.

Scientific research usually begins with a question. Is drug A more effective than the existing treatment? Does a particular factor increase the risk of developing a disease? Are two biological processes related? When a well-designed study does not find sufficient evidence to support the original hypothesis, the outcome may be described as a negative or null result. Unfortunately, the word "negative" is easily mistaken for "failure". Scientifically, however, these are very different things.

Part of the problem is that we are naturally drawn to stories of scientific success. A new drug treats a previously difficult disease, scientists develop a technology that could save lives, or researchers discover a biological mechanism that helps explain a longstanding medical mystery. These findings are exciting, and understandably so. They attract public attention and make compelling headlines. Within academia, statistically significant or apparently novel findings have also historically been easier to publish than studies reporting that an expected effect was not found.

This creates what researchers call publication bias. Related to it is the "file drawer problem", where studies producing inconclusive, negative or null findings remain unpublished and effectively disappear from the scientific record. The consequences may be more serious than simply leaving some research unread.

Imagine that ten research groups independently test the same compound. Nine find no meaningful effect, but their findings are never published. One group reports a positive result, and that study appears in a journal. Anyone reading the published literature would see only that one positive result. The compound might therefore appear far more promising than the total body of evidence actually suggests.

This is why null results matter. Science depends not only on what we discover, but also on having as complete a picture of the evidence as possible. Knowing that an approach did not produce the expected result can itself help determine where research should go next.

It can also save considerable time, money and resources. If one laboratory reports through a rigorous study that a particular compound does not produce the expected effect, other researchers can take that evidence into account before investing resources in pursuing the same path. This is especially important when research is supported by public funds. Every ringgit spent repeating an avoidable dead end is a ringgit that could potentially have been used to investigate another promising question.

The stakes can be even higher in medical research. Finding that a treatment does not provide the expected benefit is important information. So is discovering that it performs no better than an existing treatment, or that its side effects outweigh its potential benefits. Such findings can help researchers and clinicians decide which approaches deserve further investigation and which assumptions need to be reconsidered. Knowing what does not work, in other words, helps narrow the search for what might.

Sometimes, however, an unexpected result does more than close one avenue of investigation. It opens another. Science does not progress only through experiments that confirm what researchers hoped to find. Some of its most interesting questions emerge precisely when the data refuse to cooperate with expectations.

Why did the expected relationship not appear? Was the original assumption incomplete? Is another factor influencing the outcome? Could the underlying biological mechanism be more complicated than previously thought? Questions like these can lead to new hypotheses and entirely different directions of research.

This is also why we should be careful about what we call a failed experiment. A genuinely poor experiment may fail because its design is flawed, its methods are inappropriate, its sample is inadequate, or its data are unreliable. That is very different from a rigorous experiment producing a result that happens not to support the original hypothesis. The more useful questions are whether the research question was sound, whether the study was properly designed, whether appropriate methods were used, and whether the data were analysed and reported transparently. If those standards are met, an unexpected answer is still an answer.

The scientific community has become increasingly aware of the problems created when positive findings receive disproportionate attention. One response has been greater emphasis on practices such as preregistration, where researchers specify their hypotheses, methods and analysis plans before examining the results. Such practices can improve transparency and reduce the temptation, whether conscious or otherwise, to adjust analyses after the fact simply to produce a more interesting or statistically significant finding.

But improving research practice is only part of the solution. Publication culture also has a role to play. The value of research should not depend solely on whether its findings can be labelled "positive" or "negative". The quality of the research question, the strength of the study design, methodological rigour, transparency and reliability of the evidence are ultimately more important.

The public, too, may benefit from a better understanding of how science actually progresses. Scientific discovery rarely follows a straight line from question to breakthrough. Behind one successful medicine may be dozens of compounds that did not work. Behind an accepted scientific explanation may be hundreds of experiments that challenged earlier assumptions and forced researchers to refine their understanding. Those apparently unremarkable findings are not separate from scientific progress. They are part of it.

Perhaps, then, we should be more careful with the word "failure" in research. A well-conducted experiment that does not find the expected relationship or effect is not necessarily wasted effort. It adds another piece to our collective understanding, sometimes by showing us where not to look and sometimes by forcing us to ask a better question.

A scientist's contribution is not always to prove that something is true. Sometimes it is to demonstrate, carefully and convincingly, that something we expected to happen does not. In science, discovering that we have taken the wrong path can be an important step towards finding the right one.

* Dr Manorenjitha Malar Sivanathan is a research officer at Pusat Kanser Tun Abdullah Ahmad Badawi (PKTAAB), Universiti Sains Malaysia (USM).

** This is the personal opinion of the writer or publication and does not necessarily represent the views of Malay Mail.