Philosopher Karl Popper suggested that it is impossible to prove a scientific theory true by means of induction, because no amount of evidence assures us that contrary evidence will not be found. Instead, Karl Popper proposed that proper science is accomplished by deduction. Deduction involves the process of falsification. Falsification is a particular specialized aspect of hypothesis testing. It involves stating some output from theory in specific and then finding contrary cases using experiments or observations. The methodology proposed by Popper is commonly known as the hypothetico-deductive method.
Popper's version of scientific method first begins with the postulation of a hypothesis. A hypothesis is an educated guess or a theory that explains some phenomenon. The researcher then tries to prove or test this scientific theory false through prediction or experimentation (Figure 1). A prediction is a forecast or extrapolation from the current state of the system of interest. Predictions are most useful if they can go beyond simple forecast. An experiment is a controlled investigation designed to evaluate the outcomes of causal manipulations on some system of interest.
To get a better understanding of the hypothetico-deductive method, we can examine the following geographic phenomena. In the brackish tidal marshes of the Pacific Coast of British Columbia and Washington, we find that the plants in these communities spatially arrange themselves in zones that are defined by elevation. Near the shoreline, plant communities are dominated primarily by a single species known as Scirpus americanus. At higher elevations on the tidal marsh Scirpus americanus disappears and a species called Carex lyngbyei becomes widespread. The following hypothesis has been postulated to explain this unique phenomenon:
The distribution of Scirpus americanus and Carex lyngbyei is controlled by their tolerances to the frequency of tidal flooding. Scirpus americanus is more tolerant of tidal flooding than Carex lyngbyei and as a result it occupies lower elevations on the tidal marsh. However, Scirpus americanus cannot survive in the zone occupied by Carex lyngbyei because not enough flooding occurs. Likewise, Carex lyngbyei is less tolerant of tidal flooding than Scirpus americanus and as a result it occupies higher elevations on the tidal marsh. Carex lyngbyei cannot survive in the zone occupied by Scirpus americanus because too much flooding occurs.
According to Popper, to test this theory a scientist would now have to prove it false. As discussed above this can be done in two general ways: 1) predictive analysis; or 2) by way of experimental manipulation. Each of these methods has been applied to this problem and the results are described below.
If the theory is correct, we should find that in any tidal marsh plant community that contains Scirpus americanus and Carex lyngbyei that the spatial distribution of these two species should be similar in all cases. This is indeed true. However, there could be some other causal factor, besides flooding frequency, that may be responsible for these unique spatial patterns.
If the two species are transplanted into the zones of the other they should not be able to survive. An actual transplant experiment found that Scirpus americanus can actually grow in the zone occupied by Carex lyngbyei, while Carex lyngbyei could also grow at lower Scirpus sites. However, this growth became less vigorous as the elevation became lower and at a certain elevation it could not grow at all. These results falsify the postulated theory. So the theory must be modified based on the results and tested again.
The process of testing theories in science is endless. Part of this problem is related to the complexity of nature. Any one phenomenon in nature is influenced by numerous factors each having its particular cause and effect. For this reason, one positive test result is not conclusive proof that the phenomenon under study is explained. However, some tests are better than others and provide us with stronger confirmation. These tests usually allow for the isolation of the phenomena from the effects of causal factors. Manipulative experiments tend to be better than tests based on prediction in this respect.