Sanders, Clinton, and Voter Defection

My social media got lit up with posts about how there were enough people who voted for Bernie Sanders in 2016 primaries and who then voted for Donald Trump in the general election to cover the margin separating Clinton from an electoral college win. This, predictably, ignited the usual verbal infighting among my progressive friends who believe Sanders was/is a spawn of Satan, and others who think Clinton had no business running for dog-catcher, much less president. And, as usual, the winner will be the adversaries of both camps.

I will point out something that got rather less attention. At least the Washington Post article noted some grounds for comparison of the current analysis to what went before. The article cites two different survey results of the 2008 election showing Hillary Clinton primary voter rates of defection in the general election to Barack Obama’s opponent, John McCain. Those rates were 25% and 24%. By comparison, the rate of defection of Sanders primary voters to Trump was 10%. That yields some 140% to 150% greater effectiveness on Sanders’ part in preventing defection to other parties than Clinton was able to manage herself in 2008. The notion that Sanders was somehow deficient in his commitment to elect Clinton is belied by that insight, I think.

Wesley R. Elsberry

Falconer. Interdisciplinary researcher: biology and computer science. Data scientist in real estate and econometrics. Blogger. Speaker. Photographer. Husband. Christian. Activist.

2 thoughts on “Sanders, Clinton, and Voter Defection

  • Pingback: Looking Back at Election 2016, Part 2 – The Austringer

  • 2018/10/21 at 1:53 am

    I linked to this post on Facebook when someone was dissing Sen. Sanders. Someone else came up with what they thought was a rebuttal to this post. I’ve removed the name, but figured this is a good place to memorialize a rebuttal of the rebuttal.

    As you know, the number of clinton defectors is not relevant in counting the number of, and effect of, sanders defectors.

    Actually, I don’t know of anywhere that has been established. More on that in a bit.

    In 2016, Clinton did so well that she won the popular vote but she only did well enough to just barely lose. In 2008/12, Obama won handily.

    That is self-defeating: both Sanders and Clinton were losing primary candidates whose primary opponents went on to win the popular vote in the general election. The difference between winning the general election and not is the item at issue. The fact that Obama won despite greater Clinton voter defection and Clinton lost despite lesser Sanders voter defection is not a point in favor of assigning “effect” to Sanders himself in the 2016 election, nor does it appear to be something that can be airily dismissed out of hand if one seriously wants to know “what actually happened”.

    That whole line of argument makes no sense coming from someone who has elsewhere written on the importance of doing statistics correctly. The null hypothesis is obviously that the rate of voter defection is not different from other instances of how those who voted for someone who lost their party’s primary then voted in the general election. Making an assignment of “effect” obviously only can happen after a statistical case is made that we should reject the null hypothesis, and it doesn’t look to me like that is in the offing any time soon. We have a name for accepting the “effect” of things without demonstration of any difference from what would happen by chance; it’s called “superstition”. It’s possible that a full statistical analysis could show that Sanders was ineffective in reducing a typical voter defection rate, but the singular comparable case I have numbers for doesn’t point in that direction. Given that comparable case, it’s an open possibility that there could be an effect the Sanders had, all right, but one going in the direction of being *more* effective in preventing voter defection than is typical. I can’t say the latter is correct based on the limited data, but it certainly is nowhere near demonstrated that there is any particular “effect” to be assigned to Sanders based on numbers or rate of voter defection.

    The comparison is only helpful in the mud slinging part of the discussion, not the “what actually happened” part of the discussion.

    It seems to me that having a historical understanding of the magnitude of primary voter defection is inherently a part of knowing “what actually happened”. I don’t understand why you are so eager to dismiss actual data. Do we expect a primary loser’s endorsement to be 100% effective in getting their voters to support their primary opponent? If not (and the rhetorical nature of the question is thus exposed), then knowing what is a normal rate of defection is a job for statistics. If Sanders’ voters did not defect in greater-than-normal proportion, it would seem that trying to claim that as a reason for the loss, especially as a reason we have any hope of fixing next time around, is more in the line of scapegoating (and thus “mud slinging”) and certainly not in the line of effective analysis.

    Trying to pin any one “cause” as “THE cause” of the 2016 loss is hopeless: the margins were too small to identify any one negative thing as “it”. If the level of voter defection that might have given a win to Clinton is dwelling in cloud-cuckoo land with the unicorns, it seems less than reasonable to assert that, hey, this number here alone supports my contention.

    Your mileage may vary.

    Also, as the studies of the Clinton defection discovered, those were not people mad at the party for dissing Clinton. Those were people who did not want to vote for a black man. Totally different dynamic.

    Does that account for about double the rate of defection? If so, I’d like to see the numbers to support that rather than a simple speculative statement.

    But I should note that there is some data pertaining to the difference between factors of racism and party ID, and it points out a problem with your speculation.

    Have a look at this NYT article. The primary predictor of an Obama voter who then defected to vote for Trump was racism. Not party purity or dissatisfaction. If Sanders’ voters were having the “effect” of interest, why would it be the case that the ideology or party ID factors were the lesser predictors in the overall voting analysis? It looks as if Clinton’s inability to deal successfully with prospective voter racism in 2008 came back to bite her in 2016. Shouldn’t we be paying attention to factors identified as stronger predictors rather than weaker ones, especially if our attention spans are now asserted to be too limited to even assess data and arguments properly?

    But really, are we going to spend the next month relitating the last three elecitons instead of trying to win this midterm? It is a zero sum game. We have a fixed number of hours to be working on this.

    Myself, I can counter bogus argumentation and still support blue candidates

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.