Hacker News Reader

Also has issues of random chance causing these differences. How many different positions are there that have the chance of a 80% effect?

They're using a Benjamini–Hochberg correction (alpha=0.05) to account for multiple comparisons (see Table 2).