Distinguishing Association from Causation in Media Headlines
Milo Schield, Robert L. Raymond · 2009
Headlines from 2,000 news stories were analyzed for the presence of 727 keywords indicating an association, a causal connection or something in-between. 71% had such keywords. Of those with such keywords, very few (6%) had keywords clearly indicating causation or association. Most (94%) had “between” keywords: keywords that described an association but had a causal connotation. "Between" keywords included action verb keywords such as ups or cuts (61%), comparison keywords such as more or less (19%), sufficient keywords such as prevent or stop (8%) and temporal or quasi-causal keywords such as after and due to (7%). A content analysis of three statistics textbooks indicates that statisticians may use effect without implying causation. This data may be useful for both statisticians and journalists in trying to understand how the other group distinguishes association from causation.