A scatterplot claims club emails strongly increase meeting attendance. It includes only weeks when at least one email was sent and omits six weeks with zero emails and normal attendance. Which flaw would be most important to address before using the visual as evidence?