Quote:
Originally Posted by HarryT
Experimental results which don't match what you expected are what lead to advances in science. A classic example is the 1887 Michaelson-Morley experiment, the completely unexpected results of which triggered a revolution in physics. |
I think it's a particular itch for sciences that proceed by means of statistical modelling. A statistically significant effect is observed and then that effect is further explored. However, I wonder whether the statistical significance that is attached to the effect is sometimes an artifact of the research process which excludes anomalous results - or explains them away. A particularly good example is brain research using fMRI scanning. The nice sexy pictures that get in the journal articles that appear to show particular bits of the brain "lighting up", are the result of filtering and manipulating thousands of data elements - if you put someone in a fMRI scanner and ask them to do anything at all their whole brain is activated and "lights up" - and it is the particular statistical manipulations the "reveal" the significant activations. If you run different statistical manipulations on the data you get different "activations".