I think that was his point. The test doesn't consider any placebo effects, whereas placebo effects may have an actual effect in real life.