The article, "Cheating our Children: Suspicious school test scores across the nation," uses large data sets to identify schools across the nation that recorded improbable test scores. I was amazed at the amount of data that was used (and the amount of data that was not included in the analysis, such as small schools or schools that suppressed their data).
I would call this sentence the nut graph:
"A tainted and largely unpoliced universe of untrustworthy test results underlies bold changes in education policy, the findings show. The tougher teacher evaluations many states are rolling out, for instance, place more weight than ever on tests."
What awed me, though, was the amount of statistical analysis that led up to this point. It really emphasizes the fact that writing the story is the last bit of work that takes less time than gathering the information.
I have read about linear regression lines in other classes, but have never touched on it in a journalism skills course. The story, though, was produced by three people, two doing the data analysis and one doing the application. I think it would be very relevant for journalists, and for University of Minnesota students in general, to be required to take classes on data analysis. After reading the article that was written from the data, I have set a goal to try and learn data analysis.