We hadn't planned on using this graph. But I tweeted it, as I do sometimes with things I don't plan to use, and people were intrigued. So this tweet is a lie:

Because here's the graph, in a post. Who knew?

It's a riff off of analysis from Cook Political Report, assessing which 2014 House candidates did better or worse than expected. Cook produces a metric called the Partisan Voting Index, which evaluates every district based on how likely it is to vote Republican or Democratic. It makes sense, then, to compare the two: Figure out how big the margin of victory was for a candidate vs. how partisan that district is assessed to be.

In our chart, which includes every House race in which the candidate was opposed, you can see that the PVI generally lines up with the actual results. The more Republican the district, the bigger the Republican win and same for the Democrats. There are some weird outliers, though. Those two Texas districts that are more Republican than they voted. Or the Florida seat of Rep. David Jolly (R), which elected him overwhelmingly in part because they'd just elected him for the first time in a very competitive special election in March. And Utah's Mia Love (R), named by Cook as the top Republican underachiever for only barely winning in a strongly Republican area. (We included Mike Honda (D-Calif.) here because he should have been expected to have won more handily. But in California's top-two system, he was running against another Democrat.)

So there you have it. From popular demand — or as close to "popular" as cryptic charts can get — this graph made it onto the Web site of The Washington Post. It's an inspiration to poorly labeled graphs the world over.