[Contents]
Copyright © 2018 jsd

2020 Candidate Ideology

1  Some Data

Objective data suggests that the leftiest candidates in the 2020 presidential race are Warren, Harris, Booker, and Sanders – in that order. Biden is in a precarious position, because he is not a lefty, not particularly progressive, and definitely not an Obama clone. Sanders is not nearly as far left as he is cracked up to be, although he is quite progressive on lifestyle and civil-rights issues.

The two-dimensional “dw-nominate” scores1 are reliable, objective information – which is something you don’t see too often in the political world. Figure 1 displays US senator ideology in classic dw-nominate form, with a few enhancements.3

p2020-nom12
Figure 1: US Senator Ideology Map

Here are the numbers for the top ten lefty senators. It is amusing to see that all 4 of the top 4 leftiest senators are running for president:

   nom1nom2p2020
1ElizabethWarren−0.769−0.277running
2KamalaHarris−0.713−0.078running
3CoryBooker−0.607−0.202running
4BernardSanders−0.526−0.371running
5TammyBaldwin−0.511−0.215 
6EdwardMarkey−0.506−0.440 
7MazieHirono−0.499−0.089 
8JeffMerkley−0.466−0.776 
9TomUdall−0.453+0.172 
10KirstenGillibrand−0.439−0.303dropped

Along the same lines, here is the data for all 7 senators who are running for president (plus a couple of ringers):

   nom1nom2p2020
Sen. #1ElizabethWarren−0.769−0.277running
Sen. #2KamalaHarris−0.713−0.078running
Sen. #3CoryBooker−0.607−0.202running
Sen. #4BernardSanders−0.526−0.371running
Sen. #5KirstenGillibrand−0.439−0.303dropped
ExBarackObama−0.363−0.312served
ExJoeBiden−0.321−0.010running
Sen. #6AmyKlobuchar−0.269−0.291running
Sen. #7MichaelBennet−0.212−0.119running

2  Interpreting the Scores

Here is a simple way to think about the political meaning of dw-nominate scores: Primarily, a computer is used to fiddle with the points so that the final placement is a good predictor of how each legislator will vote.1, 2 The algorithm does not know or care about the political meaning of the votes, nor of any particular direction in the resulting plot. The objective at this stage is good prediction, not political meaning. In fact, at this stage the results are invariant with respect to all linear transformations, including rotations.

As a secondary postprocessing step, the results are rotated so that the horizontal axis corresponds to the conventional notion of partisanship. This is the only way in which humans inject political meaning into the process.

The meaning of the nom2 score is not easy to explain. Perhaps the following will be of some use:

*   References

1.
Wikipedia article: “NOMINATE (scaling method)”
https://en.wikipedia.org/wiki/NOMINATE_(scaling_method)
2.
DW-Nominate interactive data viewer
https://voteview.com/congress/house
https://voteview.com/congress/senate
3.
I collected a bunch of data from disparate sources. I cleaned it up and made it available in a number of convenient forms, as described here:
www.av8n.com/politics/data-collection.htm
4.
The Obama and Biden dw-nominate scores are ten years old, but there is strong evidence that a person’s score tends to remain stable over long periods of time. Any change would be imperceptible and irrelevant for present purposes.
[Contents]
Copyright © 2018 jsd