Is this just bad use of stats? The article's premises are based on the regression lines he's put on them, but the're at best very loose. Looking at the second

Is this just bad use of stats? The article's premises are based on the regression lines he's put on them, but the're at best very loose. Looking at the second graph in particular, where the author's point is that the positive skew is unexpected. It seems to me that (1) the line is only just positive anyway; (2) no error bars or survey methodology is evident even though we're talking numbers lower than 0.5 in 100,000; and (3) it's clear that outliers NY and NJ not only form part of the regression, but are skewing it dramatically. I can't do it at the moment but I bet that if you remove those two outliers, the line will tip negative (as expected).