This is in response to a recent article on FiveThirtyEight.com entitled “Bernie Sanders Is Even Less Competitive Than He Appears”. Which for some strange reason was also posted on Facebook as “Bernie Sanders is even further behind in votes than he is in delegates”.
The analysis is certainly not of the quality that I would have expected from FiveThirtyEight. It appears that the conclusion was reached before the analysis was performed. The analysis that appears in the article is flawed to the point is is difficult to come to any other conclusion. It certainly was not an article intending to present the results of unbiased analysis.
The entire point of the article seems to be to rail against the argument made by Sanders supporters that superdelegates should conform to the “will of the people” and switch their support to Bernie Sanders if he reaches 50% of pledged delegates.
The article argues that the “will of the people” is more appropriately measured in terms of raw votes rather than Pledged delegates. It tries to prove that, even if Sanders reached 50% of pledged delegates, Hillary would still have such a significant lead in popular vote that there is no chance the superdelegates would switch to Sanders.
A reasonable analysis will demonstrate that if Bernie Sanders reaches 50% of pledged delegates, he will also have earned approximately 50% of the raw votes. It is important to realize that this still is no easy task as it still requires that he wins 57% of the pledged delegates in the remaining primaries and caucuses.
In the event that the pledged delegate count is close between Clinton and Sanders, then a number of reasonable (as well as unreasonable) arguments will be made for each of the candidates. One of these arguments that will certainly be the results of the actual raw vote. Other arguments include how each candidate has performed in recent primary results in comparison to the early primary results, how each candidate performs in the critical swing states, the honesty and trustworthiness of each candidate and, of course, which candidate has the best chance of beating the republican candidate in the general election. Of course there also the potential issue of an actual or pending indictment.
In any event, the comparison of raw votes will be a meaningful, but not definitive, piece of information. Unfortunately, the analysis performed by 538.com is defective in a number of areas. From the onset, and throughout the article, the numbers reported are incorrect, and often significantly so.
The article states that ”Including caucus results, Clinton leads Sanders by almost 2.4 million raw votes, 9.4 million to just more than 7 million” which equates to 42% of the raw vote for Sanders. However, those numbers which come directly from The Green Papers are slightly understated as they actually exclude Iowa, Maine, Nevada, and Washington. Based on Wikipedia, including those caucuses, changes the results to 9.6 million for Clinton and 7.3 million for Sanders and increases Sanders percentage of raw votes to 43.3% and reduces the vote differential to 2.3 million.
Coincidently, correcting the various errors but still using the same methodology as 538.com happens to arrive at results which are not dissimilar to the results they presented. However, it is the methodology that is most flawed. Bernie Sanders wins primaries and caucuses where he can get large numbers of voters to turnout. If he is going to successfully achieve over 57% of the remaining pledged delegates, he certainly isn’t going to achieve that if he gets low turnout. The methodology applied assumes that although Sanders has to win 57% of the remaining vote, he will be able to accomplish that by getting only average turnout.
The key element in the analysis is the rate at which voter turnout has declined in 2016 from 2008 levels. The article correctly recognizes that there are inherent differences in the vote turnout between caucuses and primaries.
However, Sanders only wins where voter turnout is high, so if the intent is to determine the share of the votes assuming Sanders wins at least 57% of the remaining pledged delegates, he will need to win or come close in each of the remaining states. The table below shows decrease in voter turnout that 538com used, what the corrected numbers would be, as well as the results split between the caucuses and primaries that Sanders won and lost1. Correcting for their calculation errors alone would result in a larger decrease in voter turnout than they utilized. As you can see from the table, there is a significant difference in the decrease in voter turnout since 2008 in states where Sanders has won vs those he lost.
Decrease in Voter Turnout Rate
|
538.com
|
Corrected
|
Sanders Lost
|
Sanders won
|
Caucuses
|
-15.9%
|
-18.0%
|
-33.9%
|
-6.5%
|
Primaries
|
-30.0%
|
-32.0%
|
-36.6%
|
-11.6%
|
Revising the results using the decrease in turnout rates where Sanders won, and correcting for the error in the initial raw votes, increases the expected number of votes from the remaining primaries and caucuses from 12.1 million to 15.2 million. At this level, Sanders would have to get 57.5% of the vote to reach Clintons popular vote, rather than 60% as stated in the 538.com article.
These same revisions project that Sanders would get 57.1% of the remaining popular vote. The total votes for Sanders would then be only 138,000 less than projected for Clinton. Out of a total of 32 million total votes, this is well within any margin of error in the projections, and is reasonably consistent with the need for Sanders to get 57% of pledged delegates.
Unquestionably, Sanders still has an uphill climb, and the results for the upcoming NY primary will be critical. You must remember that at this point less than 60% of the pledged delegates have been awarded. It would be great if he wins, but no matter what he has to have a decent showing. Any shortfall in NY will only make the remaining requirement that much more difficult. It isn’t only the issue of the delegates, momentum is extremely important for both candidates as we finish the last half of the primaries.
For further information regarding Sanders road to the nomination see Delegates for Dummies.
Footnotes
1 Like the analysis from 538.com, this analysis excluded Utah and Washington, which switched contest types between 2008 and 2016, Michigan, where Barack Obama was not on the ballot in 2008 and Wyoming which had not yet occurred at the time of the analysis. It also exclude New Mexico which also switched contest types between 2008 and 2016, as well as Hawaii because the 2008 raw votes were not readily available.
. ^
2
This analysis excludes Puerto Rico. ^