Originally Posted by brian1115
You clearly can't test for all individuals and people all react differently so it could go either way.
Thus the problem with labeling social sciences as hard sciences. They aren't like, say, physics, where you always have universal laws like gravitation. Even assuming you managed to perform The Perfect Psychology Experiment (true random sampling that has an equal chance of pulling from literally every culture, age, etc. across humanity, create a true experiment rather than a quasi-experiment that many psychologists use for ethical reasons, had an airtight experimental process that introduced no type I or type II error into the system at all, and goooooood luck
in all of these regards), your statistical model would (at best) only be able to account for 2 or 3 stdev of the population. People are individuals, so there are always going to be outliers, and like Morgan says in the discussion of his study, outliers are hard to interpret. The usual response that researchers take is to either remove them from the data set or "refit" them to the set.