How convincing is a 90% Result?

Scientists can be subject to a con- firmation bias. For instance, louis pasteur was involved in a major debate with other scientists about whether organisms could spontane- ously generate. The other scientists argued that the appearance of bac- teria in apparently sterilized organic material was evidence for sponta- neous generation of life. pasteur performed many experiments trying to disprove this, and 90% of his experiments failed, but he chose to publish only the successful experi- ment, claiming that the results of the rest were due to experimental
errors (geison, 1995). Scientists frequently question their experimen- tal results if those results seem to contradict established theory. For instance, if one dropped a rock from a 100-m tower and timed its fall as 1 s, it would be wise not to conclude that acceleration due to gravity was 200 m (using the formula distance 5 ½ 3 acceleration 3 time 2) rather than the established value of approximately 10 m on earth. Almost certainly, something was wrong in the measurements and the experiment needs to be re- peated. on the other hand, the pasteur case does seem rather extreme, ignoring 90% of the ex- perimental results on a question that was much debated at the time. In this case, however, he turned out to be right.