The flip side of this is that there is almost always a very small effect, even if you are testing a crazy hypothesis (there are very weak correlations between a

The flip side of this is that there is almost always a very small effect, even if you are testing a crazy hypothesis (there are very weak correlations between all sorts of things). So you can often get a ‘significant’ result just by using a huge sample, even though the effect size is too small to matter practically.