My former Columbia colleague Matt Kahn sent me this article by Michael Cragg and himself on the political economy of congressional support for legislation intended to mitigate greenhouse gas production:

Stringent regulation for mitigating greenhouse gas emissions will impose different costs across geographical regions. Low-carbon, environmentalist states, such as California, would bear less of the incidence of such regulation than high-carbon Midwestern states. Such anticipated costs are likely to influence Congressional voting patterns. This paper uses several geographical data sets to document that conservative, poor areas have higher per-capita carbon emissions than liberal, richer areas. Representatives from such areas are shown to have much lower probabilities of voting in favor of anti-carbon legislation. In the 111th Congress, the Energy and Commerce Committee consists of members who represent high carbon districts. These geographical facts suggest that the Obama Administration and the Waxman Committee will face distributional challenges in building a majority voting coalition in favor of internalizing the carbon externality.

They make some interesting points, somewhat related to the much-remarked issue that the Democratic-leaning northern and midwestern states tend to pay more in taxes than they get back in government spending, while Republican-leaning sunbelt states are generally net beneficiaries of federal funds. When looked at from this perspective, you can see it’s not so simple as Democrats vs. Republicans. Also, is straight carbon emissions the only story? I see from the map that Michigan has low carbon emissions per capita, but, at least traditionally, the politicians there support heavy industry. I suppose that, nowadays, carbon emissions is much more about extraction than about industrial production.

Cragg and Kahn do an analysis at the congressional district level, which makes a lot of sense. I haven’t looked at income and voting by congressional district, but when you look at it by county, the patterns vary a lot by state. In California, Washington, and Oregon, the richer counties are nowadays the most Democratic. But in Texas and Oklahoma, the pattern goes the other way, with richer counties being more Republican. For example, suburbs of Dallas. So I think you have to be careful about using phrases such as “conservative, poor areas” and “liberal, richer areas.” This pattern fits some parts of the country but not others (a point we made ad nauseum in Red State, Blue State). I think I know what Cragg and Kahn mean by this–they mean that, when they run a regression, both the average liberalness and the average income in the congressional district predicts lower carbon emissions–but you’re just asking for trouble if you blur these concepts.

The other thing I wonder is if Cragg and Kahn have fully accounted for the partisan nature of Congressional voting. To put it bluntly: the Democrats have a majority in both houses of Congress, and so their votes count more than the Republicans’. This should affect their analysis and conclusions. On pages 17-18, they do discuss differences between the parties, but unless I’m missing something (and maybe I am), they’re downplaying the relevant fact that the Democrats are in the driver’s seat.

I also have a few comments about the data display (of course):

Table 1. Remove all percent signs after the first row. This will make everything easier to read. Also use a font so that all 50 states fit on one page. And please, Please, PLEASE, don’t list the states in alphabetical order. (Actually, they use alpha order by two-letter abbreviation, which is even worse. Please spell out the state names. There’s room, and pixels are free.) There are lots of possible orderings you could use. I’d suggest increasing order of per-capita carbon production, but you could do average state income or even Democratic vote share in a recent election. I’d also suggest doing the plant emissions as a graph for each state: they add up to 100% so you could easily use a segmented bar graph. It could take up less space than the existing table!

Tables 2 and 3. A canny display would allow you to put these in a single table. You can save lots of space by removing some of these extra significant digits. If the s.e. is 0.279, then you can call it 0.3 and round the estimate to the nearest decimal place. Etc. Remember: the goal here is to communicate.

Table 6. Make some graphs, show the distributions, not just the averages. Loosen up a little! Show a scatterplot of ideology vs. carbon.

Figures 1-7. Remove the two-letter state abbreviations here. And kill the ugly, ugly legends: explain this in the figure caption. Finally, and most importantly: make these maps much smaller and you can put all 7 of them in one display (yes, it can be done clearly; see, for example, here). It’s just about impossible to figure anything out, flipping forward and backward between all these pages.

And, now, follow up those maps with a couple of scatterplots! Let’s see which states fit the pattern and which don’t.

Appendix 1. Remove the horizontal and vertical lines. Just try it: it’ll make the table easier to read. Also, for heaven’s sake, indicate the party of each congressmember (a little D or R will do), and compress the horizontal space so you can fit in three columns of names. Or simply put the appendix on the web and give a link.