The legislative failure of the GOP’s replacement for Obamacare has been widely reported, obviously, but I remain interested in one bit of polling noted this week by FiveThirtyEight.

The polling firm YouGov estimated the legislation’s unpopularity by congressional district. The bill itself was quite unpopular, it turns out, even in conservative districts, as FiveThirtyEight’s Nate Silver reported.

Thanks to DailyKos Elections, we can also marry the data with President Donald Trump’s vote share in each district.

I’ve been experimenting with maps in D3.js, and I hadn’t yet tried congressional districts. So this seemed like a perfect opportunity, even if thematic maps aren’t particularly useful in this context (because congressional districts vary in size geographically, such maps can be misleading).

Case in point: The national map of congressional districts, with Republicans in red and Democrats in blue . As we all know, Democratic districts tend to be smaller in terms of area and clustered in more densely populated places. So they don’t get a particularly fair representation on a map:

Consider these two treemaps. This first shows members of the U.S. House by party (with some vacancies in gray). Shapes are sized based on the average population of each congressional district: roughly 710,000 people, give or take five percent. The House has 237 Republicans, 193 Democrats and five vacancies. There’s clearly a red majority, but it’s relatively close:

This treemap, however, shows the geographic area in square miles. Now you see the distortion:

OK, you get it. So let’s see how the health care opposition looks on maps.