The trouble with scientific research
03 August 2003
As far back as the early 1600s Francis Bacon, an English philosopher, questioned the certainty of (scientific) knowledge. How can we be sure our knowledge is completely reliable he asked? Until that point the tradition of early Greek philosophers gave pre-eminence to logical deduction, but for Bacon this was not enough. He identified both observation of nature and formal experimentation as necessary to test a hypothesis properly. This concept still holds true today, scientific theories are examined from a number of different angles and must be reliably repeated before they become the accepted wisdom to the day.
Many scientific studies are based on observations of specific populations. This form of research is known as epidemiology and has been used for centuries mainly in the area of public health. Early work focused on the link between the plague and poor sanitation. A typical modern epidemiological study might observe potential links between cancer or heart disease, and diet, lifestyle or genetics.
However, observed associations can only highlight which aspects of diet or lifestyle warrant further study, they do not necessarily imply cause and effect. For example if a study observed that vegetarians live longer than meat eaters, it is not possible to determine from this information whether it is mainly down to diet or the result of other lifestyle choices. In a world where almost everything is influenced by many different factors, sometimes it is difficult to pick out what is really important. The epidemiological study is a starting point for further vigorous experiments, which must examine each possibility in turn. In this way it becomes possible to distinguish between real causal effects and spurious ones.
The wood from the trees
Scientists are trained to understand the limitations of scientific research and in the past any new scientific findings would be hotly debated and rigorously tested amongst scientists BEFORE the public were informed. But in today’s climate of instant information, this useful step is often overlooked, and the public are frequently presented with ‘new’ findings before the peer-review process has been completed (or even started). As everyone has to eat, nutrition is an area of science that people can relate to and will often take to heart. For non-scientists deciphering a scientific study is not an easy business. How do you determine the importance or relevance of the study? Should you make changes to your food choices? Here are a few pointers to help identify the wood from the trees:
- Dietary changes should not be made on the basis of one study, especially if the study is the first of its kind or has startling results. Many more studies will be needed to confirm or refute the findings first.
- Check out which population was studied. If it was young females, the results may not apply to older people or men. Similarly animal studies do not necessarily relate to humans.
- Associations or correlations do not prove a cause or effect. More research is always needed to check if the association really exists.
Another area where there is great potential for confusion is in the level of risk or benefits reported by a study. Perhaps a study might find consuming a particular supplement halves the risk of contracting a disease. This sounds very convincing but if the actual risk of contracting the disease is extremely rare in the first place this result is less startling. Conversely if the disease is common, halving the risk would be of great importance.
The scientific method
Science is not cut and dried, the latest scientific findings should be considered as part of a dialogue between scientists. Differing opinions between research findings just highlight that science is a dynamic, evolving process.
Dunbar R (1995) The Trouble with Science. Faber and Faber. London ISBN 0-571- 17448-5s