Friday, April 29, 2016

Bernie Sanders Performance Improvement Failure: You're Fired


After Tuesday night's primaries, the tone from the media was fairly clear: Bernie Sanders has no chance of winning the election, and is only staying in the race for irrational reasons.  Reports even came out that Bernie would stop actively campaigning, and was laying off most of his campaign staff (which was weird because those are temporary jobs during the primary anyways).

On the other hand, my Bernie Sanders supporting friends kept posting on social media, some of them claiming that Bernie was "still definitely going to win" or that he "has Hillary right where he wants her."  I thought maybe both sides would die down and at the end of the week, and there would be an honest assessment of Bernie's future chances.  Then the candidate posted this morning.

The candidate obviously thinks he still has a chance to "win" from this tweet, but does he?  We'll take another look using our poll performance method.


If you haven't seen my methodology for evaluating Bernie's (or really any losing candidate's) potential to stage a comeback, you can read about it here.  But the short of this method is this:
I use some technical math techniques (a logistic function, generally) to evaluate by what margin a candidate needs to outperform polling in remaining states in order to win.  
The strength of this method is that it allows us to evaluate past performance, current standing, and future polling to get an idea of what kind of comeback the candidate needs to win, and how that compares to their polling.  A reasonable analogy is analyzing a football team that's down by two touchdowns at halftime, and saying that they need to win each remaining quarter by 7 points.  Since our method also includes future polling, we also have a good sense of how they are likely to perform in each quarter, and can calculate a performance differential needed to win.

An additional advantage we bring to the table is that we exclude super delegates, and assume they will follow general preference, though there are different views of super delegate agency.  More details this topic are available here.


From our prior graph, looking at pledged delegates only, Clinton is quite close to the nomination.  In fact, Bernie would need to win 65% of the remaining vote in order to win at this point.

Our chart below shows how Bernie has to outperform Clinton going forward to win.  He has to effectively beat his current polling percentage by 20 points (not spread change, but actual change in his %). The highlighted column is most interesting, (and can be followed going forward) because it shows the vote share he needs in each state to be on pace for victory).


All of this brings up a question: if Bernie is so far behind, then why is he still in the race?  A few thoughts:
  • Platform influence. The most rational theory says that Sanders believes he can impose his policies on the party platform by showing up at the convention with a lot of delegates, even if he has obviously lost the election.  (Moving significantly to the left before the general election may not be the best move for the democrats.)
  • Clinton Indictment.  In this case, Sanders is playing a hedge against the Clinton indictment.  Many of my conservative friends believe that former Secretary of State, Senator, and First Lady Clinton will ACTUALLY BE INDICTED in the next few months and not be able to become president.  Hedging against this by staying in the race has a high-payoff for Bernie, as he loses little by staying in just a few months longer, even if Clinton's indictment is extremely low probability. (I've noted Bernie's lack of incentives by donors to drop out, and the potential moral hazard at play before)
  • Super delegate flipping.  I theorized earlier in the race that Sanders could fair better with super delegates if Trump is the Republican opponent.  The likely Clinton margin would require Sanders to win 70-80% of super delegates, and these delegates haven't started "flipping" yet, making this increasingly unlikely.
  • Sanders people struggle with math, don't realize how obvious the loss is.  (Ok, I'm mostly joking here, but the conversations I've had with Bernie supporters on tax policy lends its self towards this (feel that burn.)


Bernie is out of this race, and outside of two very low probability events (massive super delegate flipping or a Clinton indictment) he is likely to lose.  It seems Bernie likely understands this, and is likely staying in the race to influence the party platform, which is easy for him given is relatively wide and not-dense donor base.

UPDATED 2016-05-04

A lot of questions (and search traffic) this morning asking how significantly Bernie's chances improved after Indiana.  In reality, his chances got a little worse (read: his required future margins of victory of polling increased a bit (now at logit 0.84).  That's because our prior models showed him needing to win 50 delegates in Indiana, where he only won 44.  Here's an update of the chart and required delegates and margins in each contest.

No comments:

Post a Comment