How would publication bias distort the estimated effect size of prototypic antidepressants in the forced swim test?

How would publication bias distort the estimated effect size of prototypic antidepressants in the forced swim test?