I heard a statistician joke years ago that until they learn how to use stats, scientists should only be able to make claims based on plots of the raw data. As the years go on, I find myself kinda thinking it's a fantastic idea for most research papers.

"Sorry sir, we're taking away your multiple regression privileges for more than 2 predictors. Here, play with this:" *hands them a bar plot*

This is actually what I do in pretty much all my papers! The stats are just a double-check of what I know to be there. If it's not obvious without stats, it's not obvious enough IMO.

In Jerome Cornfield's 1974 ASA presidential address entitled 'A Statistician's Apology', he "quoted" Sir William Osler as saying "...medicine will become a science when doctors learn to count".

As someone who has studied how people make biased decisions from visualizations of raw data, NO! There are different biases, not fewer biases.