Next time you’re perusing your favourite newspaper or news website, it’s quite likely you’ll come across a headline announcing a new scientific discovery, perhaps saying something like “New drug found to reduce tumour growth in lung cancer patients”. This headline seems simple enough, but don’t be fooled, in order to generate it several scientists, several years and a whole lot of blood, sweat and tears (sometimes literally) will have been involved. What makes it all so complicated you may ask? Well to answer this question I have created, for your entertainment and enjoyment, some (very generalised) flow-charts of a scientist’s life, which I hope will provide you with a small window on our world:
#1 Experimentation:
All research starts out with a hypothesis (an idea). Normally, after months of reading around a topic, you build on existing knowledge by proposing your own question. This question may stem from something someone else has discovered, or perhaps a hunch or suspicion you have based on your own previous research. For example, you may propose that “daily intake of a new experimental drug will significantly reduce growth of tumours in lung cancer patients”. The difficult part is working out how to prove this! You must find the best way to design an experiment to answer the question, preferably in the quickest and cheapest way possible.
Designing the experiment is arguably the trickiest part of scientific research. It is important that you can prove, beyond reasonable doubt, that the factor you are testing is the sole cause of the effect you are seeing. Taking our cancer-drug hypothesis above, you will need to prove that the experimental drug is the main factor which is causing the reduction in tumour growth. In order to prove this, you need to find a relevant ‘control’ for your experiments. A control experiment largely mimics the actual experiment with the exception of the factor you are testing. To test that a new experimental drug works against lung cancer, you will need to use two groups of individuals (humans, or perhaps experimental animals depending on what stage of development the drug is in) with the same type of lung cancer. The first group will have the experimental drug given to them and the effect it has on the size of their tumours will be monitored. This group is the “experimental group”. However, the data from this group cannot stand alone; you need something to compare it to. This is where your control group (also known in human trials as the “placebo group”) comes in. The placebo group will usually also receive a pill, but it will be inactive, such as a sugar pill. You can then compare the rate of tumour shrinkage between the experimental and placebo groups. If the placebo group also experiences a reduction of tumour growth, then it shows that the effect is not due to your drug, but rather something else.
The selection of the experimental and placebo group is immensely important; you must ensure that the only real difference between the groups is whether or not they receive the drug. For example, if your experimental group are all females but your control group are males, you will not be able to say whether it was the drug or the effect of gender which caused your results. This is much easier to achieve in laboratory settings than in clinical studies (carried out with human subjects) since in a lab you can easily control for factors such as diet and lifestyle in both groups.
Ensuring your experiment works and is properly controlled can often be a serious headache and can take up a significant amount of time. However, once you’ve got your experiment working, then it’s time to get the results. If you’re lucky, the results will prove your hypothesis and you can move on to the next stage. However, there’s also a chance it’ll disprove your hypothesis. This means that you’ll need to generate a new hypothesis and start the whole process again.
So you’ve got your experiment to work, it’s well controlled and it proves your hypothesis. Now you’re good to go, right? Actually, no. Science rarely gives a definitive answer, and what’s more likely is that getting the answer to your original question will produce more questions which need further investigation. You may have noticed several infinite loops in the flow charts (if you follow the “no” answers anyway) – it is sometimes arbitrary when you decide to break the loop and publish your data in a ‘journal’ for the rest of the scientific community to see.
#2 Publishing:
There are several hundred different journals covering all specialities, from general science (maths, physics, biology, chemistry, engineering etc.) to incredibly specific areas such as Alzheimer’s disease. These journals publish “papers” which have been submitted by a group of scientists detailing their research, any relevant data required to back up their claims and explanations of how their work is relevant to the wider community. The important thing about publishing data is that you have a coherent story which is interesting to other people, especially fellow scientists.
So what happens when you finally decide to publish? Firstly, there is a significant advantage in having positive data (results which prove your original hypothesis). A big problem with scientific research is that negative data (results which disprove a hypothesis) are much harder to publish. A positive result is generally regarded as more interesting and a journal is more likely to accept it.
Another thing about publishing is that the data you present has to be of the highest quality. This may mean repeating experiments until the data produced is both convincing and aesthetically pleasing. This could take months to do well. Once you’ve got your data, then you can write your paper, in the style of your intended journal. This again can be a lengthy process.
So you’ve got convincing positive data which you’ve written up as a paper but now what do you do? Well, It depends on which journal you submit to but generally the paper will first be scrutinised by the journal’s editors. This is where they decide whether it’s interesting enough to publish. This can depend on the journal – top-level publications (also known as “high impact journals”), will only accept the highest quality most interesting stories and are quite likely to reject research if it’s not interesting enough. If your work not deemed good enough for your chosen journal the paper is ‘bounced back’ to you and you’ll have to rewrite it for another journal with a lower “impact factor”, or perform more experiments to help back up or round off your story.
If the editors accept your data, they’ll send it for review. This means giving it to (usually) three other scientists who work in a similar field. They thoroughly check the data and ensure that it makes sense. At this point, they may give the green light to publish, but they could also ask for additional experiments or data to add to the story. However, they could also decide that it isn’t interesting or convincing enough to be published in this particular journal. You can then attempt to submit it to a different journal, or you may have to scrap your whole idea and come up with a new hypothesis.
The pay off for all of this time and frustration is finally having your work published and available to the wider scientific community. Fellow scientists around the world will now be able to see what you’ve been doing and what you’ve achieved, making all the blood, sweat and tears worth it. Hopefully your work will make a recognisable contribution towards your field – even small or seemly insignificant discoveries can turn out to be very important later on.
So you’ve published … now what? You guessed it: time for a new hypothesis. Prepare to re-enter the loop and start the whole business over again.
Post by: Louise Walker
Where is the flow for negative results? Where are negative results published?
As this is a general flow chart, I followed the General Rule of Publishing – You are far, far more likely to get published if you have a positive result. It’s a big problem in research (especially Biology) that negative results are rarely published, unless they disprove something which was thought to be true. However, there are some journals emerging which specialise in publishing negative data (for example the Journal of Negative Results in Biomedicine) so maybe one day it will become easier to publish results when they are negative!
Journal of Negative Results in Biomedicine link:
Perhaps this is a fine point, but the box labeled “does it prove your hypothesis” should be labeled “do the results support your hypothesis”
At least this is typical for most scientific hypotheses- an unanticipated result can disprove it, while a predicted result supports it.
We always look for support and for lack of outcomes that are inconsistent with the hypothesis. One experimental result never constitutes a positive proof, but one experimental result can certainly disprove a hypothesis.