We live in an age shaped by scientific research. Medical practice, for example, changes a bit each year because of new discoveries in the laboratory or in drug trials. We have come to expect progress in a variety of technical fields, and science often lives up to our hopes for it.
But science can also falter. One of the challenges for non-scientists – whom I call “normal people” – must address is how to interpret new scientific studies. Which ones contain valuable information that should influence our activities or government policies? Which can be put on the back burner of our minds, awaiting further evidence?
The matter is both important and sometimes quite practical. Scientific studies claim to address many things that truly matter. Should you be taking a statin drug? Is global climate warming? What is causing the deaths of so many honeybees? What is the best way to try to lose weight?
Recently the prestigious journal Nature ran a piece about what non-scientists need to know when they hear about the results of scientific studies. The point is not to make everyone into a scientist, but to sketch some of the basic limits of scientific work so that the general public can better interpret the results of technical research.
The Nature piece featured 20 concepts to be borne in mind when hearing about the conclusions of scientific research. I cannot go through all 20 ideas here, but I will give you a sampling of some of those I think most important.
Chance can cause substantial variation. Scientists spend their days looking for patterns in data and in the natural world. We scientists are always trying to answer the basic question, what is the cause of patterns embedded in the world around us? But when we evaluate data, we must bear in mind that sometimes the world changes more due to chance than due to some specific cause. This means the general public sometimes needs to be patient and await confirmation of results from other studies.
Bigger sample size is generally better. It may cost more to have a large sample size in a study, but bigger is usually better in terms of the reliability of results. A drug trail involving only a dozen people is unlikely to be as valid as one involving 600 people. This is particularly important in fields like medicine where there are substantial variations between subjects.
Measurements are not exact. It is common in science to report a measurement plus an estimate of the error involved in making that measurement. Thus, a scientist does not say an object is 8.5 inches wide, but 8.5 inches wide, give-or-take an eighth of an inch. We do this because if the measurement being reported is a small value, it may be swamped by the error possible in the measurement. The example the Nature piece gave for this idea is the kind of report you may hear on the news, something like, “The economy grew by 0.13 percent last month.” That number is so small and the error involved in such matters is so substantial, there is a chance the economy may actually have shrunk.
Identifying two patterns does not necessary mean one is caused by the other. It is easy to ascribe meaning to patterns we see in the world around us. But just because we can see two patterns, it doe not mean one causes the other. It is possible that both of the patterns identified in a study are caused by a third factor, sometimes called a confounding variable.
Scientists are human. Scientists are people. We do our best, but that does not make us perfect. Scientists have several reasons to try to promote the work that has been done, quite apart from whatever merit it may have. Scientists want to have successful careers and that means promoting results obtained in the lab or field. For some scientists, professional status really matters, and for most scientists today, further funding is an issue always kept in mind.
It is important for the general public to bear in mind some of the limits of science. Technical research is still the best way we have of understanding the natural world, an approach that brings us astonishing advances every few years. But a scientist – and science itself – is not perfect.