When I recently read up on DW-NOMINATE, I learned a few things about certain political figures. DW-nominate gives every senator a score based on how left/right their voting behavior is, and sometimes the score does not match the senator’s public image.
In the 115th congress (2017-2019), the leftmost senator was… Elizabeth Warren. Not Bernie Sanders, where did you get that idea? After Elizabeth Warren, is Kamala Harris. Then we have another presidential hopeful, Cory Booker. And finally, we have Bernie Sanders–tied with Tammy Baldwin.
I made a plot! It’s copied directly from Voteview, with labels added for certain senators of interest.
Mostly, I picked out senators who are running for president (Warren, Harris, Booker, Gillibrand), or who we think might run for president (Sanders, Biden, Brown, Klobuchar, Bennet, Merkley). Biden and Clinton haven’t been on the senate for a while, so I added in their older scores (keeping in mind that the Democrats used to be further right as a whole).
I don’t include any presidential hopefuls in the House of Representatives, because they have a separate, incomparable score. I note, however, that Beto O’Rourke, John Delaney, and Tulsi Gabbard are a bit to the right of Democratic median. And if you’re wondering about Alexandria Ocasio-Cortez, she is near the right edge of the Democratic party, and a total outlier in the downward direction. So Ocasio-Cortez isn’t very liberal in the conventional sense, but she is very something.
How skeptical should you be? Oh, be skeptical, be very skeptical. But while we’re at it, let us also be skeptical of the public image of these politicians, which might be based on the whims of social media. Just trusting numbers you don’t understand is a garbage epistemology, but just reacting to what you hear on social media (and ignoring what you don’t hear) is pretty garbage too.
I just want to make a few observations, so that we might be skeptical for the right reasons.
1. DW-NOMINATE is only based on votes in congress. So if you’ve been hearing about Kamala Harris’ record as an attorney general, that’s not included. One may be skeptical of a score that doesn’t include Harris’ full record. On the other hand, one wonders how much her attorney record matters if it doesn’t actually impact her voting behavior on national issues.
2. DW-NOMINATE does not use any information about what people are voting on. Instead it uses information about which senators vote in the same way as each other. So if Elizabeth Warren is to the left everyone else, that means that there are occasionally votes where even a lot of Democrats are voting with the Republicans, but Warren is less likely to vote with the Republicans than anyone else. So maybe Warren isn’t truly to the left of Sanders, it’s just that being super leftist means sometimes agreeing with Republicans? And the truest leftists vote for Trump, of course.
3. There’s a second dimension! Voteview labels this as the “social/racial” dimension, but it’s not like there isn’t any social/racial aspect to the first dimension. Basically, it’s the dimension that used to differentiate Northern and Southern Democrats. But then the Southern Democrats switched parties and nobody really knows what it means anymore. But hey, Jeff Merkley and Alexandria Ocasio-Cortez are way off to one side of that spectrum, and Bernie Sanders is… in that direction? It’s possible that Sanders supporters particularly value this second dimension.
But there’s a reason that the second dimension is portrayed as “squashed” relative to the first. Because the first dimension explains about 83% of the votes, and the second dimension only explains about 2% (perhaps a really important 2%). At least, these are the percentages I’ve heard for DW-NOMINATE in general. Empirically, adding more dimensions doesn’t help. About 15% of the time, senators vote in the opposite way from what they are expected, but they do not do so in an organized way. So you might say Sanders or whoever has a few voting quirks, and perhaps you happen to like those quirks, but those quirks definitely do not constitute some sort of faction among Democrats, they are just quirks.
I want to make clear, the most left candidate isn’t necessarily the “best” candidate, and one may reasonably prefer a candidate that’s more towards the middle. I’m just poking a bit of fun at Bernie Sanders’ super leftist image, which may have been accurate in 2016, but is not clearly the case relative to the 2020 candidates.
Personally, I think the biggest takeaway from all this is that for whatever differences Democrats might have from each other (or from independents like Sanders), they’re pretty small compared to the gap between Democrats and Republicans. Even a brainless algorithm can see that.