I didn't mean we shouldn't engage with data at all. However there are so many possible confounding factors in this type of measurement that we should "take it with a lot of salt."