Sanders versus Clinton supporters in the American National Election Studies data

Were it not for Trump, the great drama of the 2016 election would have been the primary contest between Hillary Clinton and Bernie Sanders.  Sanders generally fit the mold of a “leftist protest candidate”, but was far more successful than previous such candidates have been.  In this post, I will examine the 2016 American National Election Studies data, hoping to find clues that explain why.

Popular explanations come in two main flavors:

  • Those that emphasize Hillary Clinton’s problems:
    • Clinton was the second least popular candidate ever to run for president, next to Donald Trump.
    • Clinton was under FBI investigation, rightly or wrongly.
    • Clinton was targeted in several ways by Russian hackers.
    • Clinton accepted money from Wall Street banks.
    • Clinton was an insider.
    • Clinton was unlikable.
    • Some voters were unwilling to vote for a woman.
    • Clinton was especially unpopular in the Midwest.
  • Those that emphasize Sanders’ appeal:
    • The Democratic base is friendlier toward socialism than it used to be.
    • The Democratic base is friendlier toward single payer health care than it used to be.
    • The Democratic base is less friendly to free trade than it used to be.
    • Sanders was an outsider.
    • Sanders was authentic in a way that appealed to younger voters.
    • Sanders was especially popular in the Midwest.

Unfortunately, the ANES data doesn’t help us quite as much with these questions as it did with the Obama/Trump questions.  It asks quite a bit about immigration and racism, and very little about Russia, e-mail servers, FBI investigations, socialism, single payer, or opinions of Sanders, for that matter.  I tested models combining demographic factors and political opinions, and found the following (remember that this is a regression; all effects are measured in terms of differences assuming all other variables are artificially held constant.):

  • Sanders voters were much younger on average than Clinton voters.  This was the strongest single effect.
  • Sanders voters were much less enthusiastic about Obamacare than Clinton voters.  This was almost as strong as the effect of age.
  • Sanders voters were less likely to be black than Clinton voters.  However, they showed no strong differences on policy issues related to race – for example, believing that police treat blacks worse than whites.
  • Sanders voters were slightly more likely than Clinton voters to prefer more government involvement in health care, to believe that the government is run for the benefit of a few rather than for the benefit of all, and to believe that defense spending should be cut.  These effects were only slightly less important than race.
  • Sanders voters were slightly more likely than Clinton voters to live in the Midwest and to believe that the United States would be better off with a non-interventionist foreign policy.
  • Several effects hovered on the edge of significance.  Sanders voters may have been…
    • …slightly more likely to be male.  They did not, however, have significantly different views on feminist issues than Clinton voters.
    • …slightly less enthusiastic about free trade.
    • …slightly more likely to favor strict banking regulations.
    • …slightly more likely to believe the income gap between rich and poor has increased greatly in the past twenty years.

The overall R-square was about 0.25, slightly weaker than for the Obama/Trump model, even though the Obama/Trump model was kept “artificially weak” by excluding race (the Obama/Trump model including race has a whopping R-square of 0.59.)  These are the things I found surprising:

  • Age was an enormous factor that simply can’t be explained away in terms of any policy questions on the survey.  Young people either found Sanders very appealing, Clinton very unappealing, or both.  Race also played a major role, again in a way that couldn’t be explained by policy issues.  Living in the Midwest also played a small-but-significant role.
  • The effect size of Obamacare surprised me, especially that it was more important than wanting more government involvement in healthcare in general, a likely proxy for single payer.  It played a major role in the Obama/Trump model, as well.  It seems likely that, for many people, Obamacare basically defines the mainstream of the Democratic party.
  • Issues related to militarism seemed relatively important, despite receiving very little discussion during the campaign.  On this particular issue, Sanders may have been a poor vehicle for his voters’ preferences, given that he seemed basically unprepared to discuss foreign policy.
  • Sanders’ platform other than health care seems to have played a relatively small role.  The effects of trade and banking regulations were barely significant, and neither the minimum wage nor education spending showed up at all – although no survey question specifically asked about “free college.”
  • The overall impact of demographics seemed roughly equal to the overall impact of policies and attitudes.
  • Some things that didn’t seem to matter:
    • Education or income.
    • Specific policy issues or attitudes related to race or sex.
    • Overall, self-reported liberalism or conservatism.
    • Perceived liberalism or conservatism of the Democratic Party; Sanders voters perceived themselves as being almost exactly as liberal as the Democratic party overall.
    • Any other measures of “trusting the system”, other than believing the government is run for the benefit of the few.
  • I intentionally excluded certain questions from the analysis; for example, the “feeling thermometer” for Hillary Clinton was extremely significant, far more important than any other factor, but as I explained in my posts about Obama/Trump voters, this type of direct question is in a certain sense “too good” – it crowds out more specific, interesting questions about attitudes and policy.  Sanders voters also had less positive feelings about the Democratic Party and, to a lesser extent, about Barack Obama.

Overall, I declare this analysis less successful than my analysis of Obama/Trump voters – I found some things that are interesting, but I don’t think I found “the answers.”  The were a few crucial questions that could have been on the survey but were left unasked:

  • How did voters feel about Clinton’s e-mail travails?
  • How did they feel about her speaking fees?
  • How warmly did voters feel toward Sanders?
  • How did they feel about free, government-funded college?
  • How did they feel about socialism?
  • How do they feel about “political insiders”?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s