Bad Graph, Disturbing Meaning

I have my reservations about Hillary Clinton, but her being a liar—any more than the average politician—is not one of them. It's hard not to think some of her coverage isn't rooted in sexism (especially the incessant male whining about her voice).

But even with that suspicion as a starting point, I was still surprised by this graph, which I saw on Mother Jones, though it's drawn from a study by the Shorenstein Center on Media, Politics and Public Policy.


The graph is badly rendered, in my opinion (more on that in a minute), but what it means is even worse: While Clinton's coverage was more likely to be about issues than the other candidates (29% vs. 12% or less for the others), the issue coverage she got was 84% negative, while Trump's issue coverage was only 43% negative. As Kevin Drum at Mother Jones put it,

Hillary wasn't just savaged on her tone or her clothing or her poll numbers. She was savaged on the issues, the one place where practically everyone agrees she's strong and knowledgeable. Even if you disagree with her—and that isn't supposed to affect media coverage—she knows what she's talking about.... Even the non-scandal portion of Clinton’s issue coverage—what she was saying on trade, jobs, foreign policy, and the like—was reported more negatively than positively. Clinton was the only one of the major candidates whose policy platform generated an unfavorable balance of news coverage.

Not surprisingly, that kind of coverage shows up in this Shorenstein-study graph of 2015 Clinton coverage by month:


Note all of the negative numbers. There's only one month with a positive number.

Bernie Sanders' 2015 graph looks like this:


As Kevin Drum said, no wonder Clinton doesn't like talking to the media.

Oh, and here's why I think that original graph is bad: the red bar is supposed to represent a subset of the blue bar (the percent of issue coverage that's negative). But instead, it's external to the blue bar and overwhelms it. At first glance, the graph looks like Clinton got more coverage than the other candidates (rather than more of a certain kind of coverage), and that the red to blue height ratio is consistently about 3-to-1 for all of the candidates, so that seems kind of fair.

When what it really means is this:



Or if you treat the tone as a separate data point, you get a graph like this, which really shows how startling the difference is in how Clinton is covered: