Maybe it's as simple as this : Start a new research. Work with 1-sigma significance (~2% error rate allowed). Then investigate the favorite brand of soup of th

Maybe it's as simple as this : Start a new research. Work with 1-sigma significance (~2% error rate allowed). Then investigate the favorite brand of soup of the brothers of cured cancer patients. Make sure to research about 50 brands. Huge news ! If you get your brother to eat soup brand 23 you have a significantly higher chance to survive cancer ! Huge article in every newspaper. What happened here ? 2% error rate.