R for Political Data Science Week 6: Just How Liberal Are the 2020 Democratic Candidates?

Many 2020 hopefuls have signed on to the Green New Deal. What exactly does that mean, and how liberal are they?

By G. Elliott Morris

On Feb. 08, 2019  in R for Political Data US Politics R-Posts



This is part of a series of short posts about politics that seeks to show how we use data science to learn more about the real world. Follow along here.


Several of the prominent 2020 Democratic hopefuls this took a big step this week in endorsing a Senate bill for the so-called Green New Deal (GND). The proposed bill text included a wide range of policy proposals, some more feasible than not. From funding renewable energy initiatives to expanding high-speed rail to providing a universal basic income and retrofitting every building in America, the GND is rightly considered to be one of the most ambitious liberal policy proposals in Congress today. The endorsement from the 2020 leaders has provided one crucial data point for those wanting to argue that the Democratic party is sprinting to the left and embracing socialism. The National Journal has run a piece titled “Democrats Are Boosting Trump’s Reelection Prospect” and Axios chimed in with “Trump’s lifeline: Democrats’ socialism surge”.

While it is true that the Green New Deal lies far to the left of where Democrats are today — and is not as feasible as some would like to believe, either politically or from a policy perspective — I think the conventional wisdom has floated too far in characterizing this year’s crop of Democratic candidates as excessively more liberal than those that have recently come before them. Bernie Sanders is the obvious point in this case, but the argument goes further than than that. I assert that (A) Far-left liberal ideas have been floating around in the Democratic Party for quite some time now and (B) there are several 2020 candidates that don’t sign on to the policies in the Green New Deal. Further, the cosponsorship of Sen. Markey’s GND legislation seems to be primarily driven by support for green climate policies and renewable energy, as demonstrated by Sen. Gillibrand’s comments on her cosponsorship:

“‘Climate change is real, it threatens us, and the evidence is now irrefutable that if we don’t act immediately to stop it, then our land, our water, our air, and our lives will all be upended in potentially catastrophic ways. There will be no going back. I urge my colleagues: Rise to this challenge, prevent the catastrophe, and pass a Green New Deal that protects and strengthens our country in this new, uncertain era,’ said Senator Gillibrand. ‘We can end the climate change crisis, we can dramatically modernize our economy, and we can create countless new jobs across the entire country that can’t be shipped overseas – but we can only do it if Congress seizes this opportunity and acts now, instead of wasting more time arguing about whether or not the problem is even real. We cannot wait another day. I urge all of my colleagues to fight with me for a Green New Deal that puts Americans to work to solve this extraordinary challenge.’”

I’ll lean on ideological scores calculated by OnTheIssues.org to provide some evidence that suggests that the whole distribution of 2020 Democratic candidates’ ideologies is not that different from the positions of those who ran in 2008 or 2016. You can find the data and code for this on GitHub.

Here are the median scores along two dimensions of policies, ‘economic’ policies and ‘social’ ones, for the past three presidential cycles with a Democratic primary:

# You need: library(tidyverse); library(ggalt); library(ggthemes)

# read in file of candidate ideologies
cand_ideo <- read_csv("../../data_no_export/post/2019_02_08_how-liberal-2020-dems/2019_02_08_how-liberal-2020-dems.csv")

# scale to -1,1
cand_ideo <- cand_ideo %>%
  mutate(economic = rescale(economic,c(-1,1),c(0,1)),
         social = rescale(social,c(-1,1),c(0,1)))

# summary stats
cand_ideo %>%
  group_by(year) %>% 
  summarise(social = median(social),
            economic = median(economic)) %>%
  kable()
year social economic
2008 -0.6 -0.8
2016 -0.6 -0.6
2020 -0.6 -0.7
options("stringsAsFactors" = TRUE) # has to be done for the polygon shading to work properly

# plot
gg <- ggplot(cand_ideo, aes(x = economic, y = social, col=as.factor(year),fill=as.factor(year))) +
  geom_vline(xintercept = 0,linetype=2) +
  geom_hline(yintercept = 0,linetype=2) +
  geom_point(size=2) +
  coord_cartesian(xlim=c(-1, 1),
                  ylim=c(-1, 1)) +
  stat_bkde2d(bandwidth = c(0.12, 0.12),aes(alpha = (..level../1)*0.5), geom = "polygon",show.legend = F) +
  scale_color_wsj() +
  scale_fill_wsj() +
  guides("color"=guide_legend(title="Year"),
        "fill"="none")  +
  labs(title="2020 Democrats Look Ideologically Similar to Their Predecessors",
       subtitle="Coverage painting the 2020 candidates as record-setting liberals is unwarranted. (Each point is a candidate.)",
       x="Position on Economic Issues\n(-1=Liberal, 1=Conservative)",
       y="Position on Social Issues\n(-1=Liberal, 1=Conservative)",
       caption="Source: OnTheIssues.org")

We can show this a little better by plotting the scores on an x-y coordinate plane and showing their density, where each point represents the social-economic scores of a candidate:

preview(gg,themearg = theme(legend.position = c(0.8,0.5)))

Looking at the two-dimensional density plot above, you can see that the 2020 candidates’ ideologies aren’t significantly different than those who came before them. I think this justifies that, at the least, the take popular among pundits right now that Democrats are more liberal than ever before is not justified by the data. Speaking of which, I’ll offer a few caveats worth mentioning. First, the scores for Julian Castro (represented by the point closest to (0,0) in the chart) likely are not as accurate as they could be. This is because the OnTheIssues profile for Castro is incomplete. Second, the scores are being calculated for the policies that the candidates support today, which are more liberal than the positions they had a decade ago, particularly along the social dimension. Even if we shift all their positions about 20% to the right, however, the ideological variety of the 2020 candidates makes them similarly located as the recalculated 2008 candidates. If we did this again for the economic dimension, however, we might notice slightly larger shifts leftward over time.

In other words, this is not the final issue on the subject, but even if we adjust the ideological scores per our qualitative reasoning, the distribution of candidates’ ideologies today is not insanely more liberal, certainly not enough to say, in Axios’ words, that the Democrats are “flirting with socialism in ways they carefully and clearly ran away from in the past” — because (A) their support for GND policies is mostly for funding renewable energy and better green regulations, not UBI etc., (B) the party has had rather liberal candidates run for president before, and (C) there are still quite a few moderates in the running.







comments powered by Disqus