If a researcher conducts 20 hypothesis tests at $\alpha = 0.05$ and the null hypothesis is true in all cases, what is the approximate expected number of 'significant' results due to chance?