The way they are interpreting their data here is highly suspect. First off, the data points are pretty much in a circular cloud, and so the association must be

The way they are interpreting their data here is highly suspect. First off, the data points are pretty much in a circular cloud, and so the association must be super weak. What bothers me even more is the way they arbitrarily divided the top and bottom halves of their data set into two separate data sets. This is highly unusual. Doesn’t the way you draw the line introduce some kind of bias? I think you cannot draw any concrete conclusions from this data.