Podiatry Arena members do not see these ads

Dear all

The question I want to ask has its genisis in this weeks publication of New Scientist. Some Swans are Grey P44. 10/ May / 08

In 1934 Karl Popper wrote a book, which later in 1959 he translated from German into English. The book is titled The Logic of Scientific Discovery. This book has been described as one of the most important scientific books of the twentieth century.

The esssence of this book is that science can be defined as the theory that is falsifiable. In other words it has the possibility of being proved wrong by experimetation. My understanding of the term 'falsifiable' is that, if a theory is able to be proved wrong therefore then there must be some method of observing the changes or effects proposed. Having the capacity to be proved wrong or falsified does not necessarily mean it will only that it has the possibility. So even if you do the experiment thousands of times and find it proved this still only proves it for the population that you have observed and there may be some as yet unobserved section of the entire population that can disprove the theory. So in this case after you have proved the theory many times the probability that is true is very high but never 100%.

The classic case is the theory that because the as the single white swan observed is white then all white swans are white. Clearly this has very low probability. But if we observe all the swans that we can find, and we assume that this is the whole population, then the probaility that all swans are white is very high. It is still possible however that somewhere there is a black or non white swan somewhere in the world that will disprove the theory.

So, in biomechanics there are almost always many confounding variables that make observation of cause and effect unreliable, assuming that mostly we want to attribute causation and not simply correlation to our work.

We have to make reasonable assumptions about many parameters and can often only attribute correlation and not causation to an intervention that we use experimentally or clinically. Direct mechanical cause and effect may be confounded by many things such as placebo effect, phsycological factors, changes in CNS activity, observer error or bias (intentional or accidental) etc.

In this case the theory may not be falsifiable because we have not a 100% reliable way of observing the cause and effect. So even if we falsify the theory we may have done so under a false premise or because of data confounding but unconnected variables.

Can we always say then that our theories are 'falsifiable' and if not how much leeway are we allowed before we enter the realms of fiction and snake oil? Is there a definite line or is there a large grey area?

Is the best that we can do is that we ensure our experimental methodology tests the theory in such a way that it has a high probability of being falsifiable?

Robert Mathews of the New Scientist (Science reader at Aston Uiversity Birmingham UK ) writes quoting Lawrence Kruass " I have fallen into the trap of applying falsifiability criteria to decide whether something is worth publishing" ---- "I have decided not to write papers because I thought they (the theories) would never be falsifiable and yet they turned out to be so".

Do we and should we employ such stringent protocols to our own work and claims or would this leave many avenues (though potentially useful) unexplored?

Does this lead us back to snake oil country?

All the best Dave

Tweet

Share