Percentage or Points?

This may seem like a minor point, but the implications are not. I have to give the people credit for using the phrase "12% drop" correctly but it doesn't excuse the conclusion.

Sandra Stotsky and Ze'ev Wurman say

"Yet, is it really the case that low-performing high school students would drop out if high school diploma requirements were ratcheted up? That doesn't seem to be the case in Massachusetts, which in 2008 reduced its dropout rate by 12% from the previous year."

You figure this to be an amazing drop worthy of Gates money. Then you follow the link and find that the dropout rate in Massachusetts went from 3.8% to 3.4%. Not only that, but this "proof" of their theory is further shaken by the clauses supplied -- a new way of counting who is which class, new definitions, new SIS system. This is another case of the random lucky bounce being taken as an incontrovertible law of nature that happens to be in your favor ... until the next unlucky bounce demands a new law of nature in your favor.

Did the ratcheting up cause the drop? Can't say. Correlation with two data points is not quite good enough to show cause.