Let me first address the elephant in the room.

Arizona was a catastrophe. Thankfully, the controversy has picked up enough media attention that many of you already know what happened. For those of you that don’t, this article touches on some of the issues, though I don’t agree with everything that he says.  I have been aware of several instances of election fraud (though these were through manipulation of votes on electronic voting machines) in this election cycle already through the incredible work of this statistician named Beth Clarkson, but have largely remained silent on the issue because the instances thus far haven’t altered the results so much that the candidate that should’ve won lost. Not to mention, anyone that speaks out against perceived electoral injustices is immediately deemed a sore loser and totally discredited.

I encourage you to read through Beth’s work. She has received a great deal of media attention over the past couple of years and is actively working to improve the electoral process. I know many of you will disagree, but I stand by my Arizona projection and believe that if the election had been conducted in a normal, reasonable way, Hillary would’ve lost or came very close to losing. I have honestly lost a lot of sleep over this, and I can only hope that none of us witness anything like that again. Like many of you, I just want a fair election.

Now, for the elections today. Here are the numbers:

Screen Shot 2016-03-26 at 12.48.37 AM

Bernie Sanders should win Alaska, Hawaii, and Washington, largely for three reasons:

  • Extremely low populations of African Americans, 1.6-3.6%, among the nations lowest
  • All three are caucuses
  • Hillary Clinton has an unusually low proportion of Facebook likes in all three states, 17-19%, which is among her worst

With all this being said, there is once again the question of how a particular ethnic group will vote, but this time it is with respect to Hawaii. Hawaii has a large population of Asians, native Hawaiians, and Pacific Islanders, unlike any state we have seen thus far. These groups could be predisposed to favor Hillary Clinton, but the null hypothesis that I must currently accept is that they aren’t. I have tested the effect of Asian population size  on previous results specifically for the sake of Hawaii after a friend suggested that I do, but it was very statistically insignificant, with a p-value of ~0.8 and actually a positive coefficient for Bernie vote share at that. Regardless, Hillary Clinton won the Northern Mariana Islands as well as American Samoa, so perhaps it is the case that in locales with Asian majorities, the dynamic changes. Hawaii is a politically unique state in many other ways, so it will be interesting to see if this estimate holds true.

Also, I want to sincerely thank everyone for the outpouring of support. I received countless emails and messages after Tuesday’s elections, even immediately after the initial Arizona results made me look like a complete moron. To all of you that I haven’t yet been able to respond to personally, I apologize for the delay but I will get to you!. I have no agenda, and I’m not doing anything remarkable, though I’m flattered by those that suggest as much. I just want to perform solid regression analysis and statistical work to give you all the most accurate electoral projections (without using polls!).





    • I think it’s too early, but based on the google trend data and the (old) FB data from 538, Wisconsin is looking very very good for Sanders.


  1. Tyler

    2 points

    It would be interesting to see your demographic analysis plotted against @home internet access (either including or instead of %demo). This may reveal a parallel between the demographics rather than the inverse relationship done by %demo.

    Congratulations on your work this election cycle. It is interesting to see people looking for how to poll in a post landline society. I hope this is rewarding for you at present and into the future.



  2. There is every chance that difficulties in voting and having to wait a long time actually affect Clinton’s vote more severely than sanders vote. Older people are less likely to have the ability or even the health to wait for hours.


    • You’re forgetting that older voters tend to vote via absentee ballot and the majority of younger voters do not. The election day issues disproportionately affected those voters who did not vote via absentee ballot.

      Liked by 1 person

  3. Hey Tyler, Congrats on the accurate predictions. Haters might say you were a bit off but these projections are by far the closest I’ve seen by ANYONE. And considering that you’ve done all this with no polling and using only publicly available data (correct me if i’m wrong) makes it that much more impressive. I’m now convinced that going forward that old methods of polling will be obsolete soon (if they aren’t already).

    Keep up the great work. Looking forward to more predictions for your more refined model.

    Liked by 1 person

    • Also on a side note.

      Did you factor in the Arizona result which might have reduced Bernie`s projections? or do you believe that the offset was due to the process being a caucus rounding of delegates?


  4. She won by 72,000 votes and you think voter suppression is the reason she won Arizona..? That means even if he somehow won 80% of the suppressed votes (which is unlikely), there would need to be 120,000 people who tried to vote and weren’t able to. Do you realize how absurd that sounds?

    Overall, turnout in the primaries has been down since 2008. So if there were 120,000 disenfranchised voters, turnout would actually be up 20% in Arizona in 2016 compared to 2008. And that’s pretty much a lower bound because once again, he almost definitely wouldn’t have gotten 80% of the suppressed votes.

    In actuality, the problem (like I mentioned in your original predictions) is that you way overfit your model. You can’t just throw in dozens of variables with n = 30 and expect OLS to work. Even if you are using data at the congressional district level (and increasing n by a significant margin), you are still likely overfitting the model.

    I guess I don’t know if you’re using OLS, but that’s definitely what it sounds like based on what I’ve read. If that’s the case, OLS probably isn’t even the right framework. You should be using GLS (especially if using congressional districts) because of the high correlation among observations. State-level wouldn’t be as big of a deal, but the observations are still definitely correlated.


    • So say it was a more “reasonable” 45,000 voters that were suppressed (based on comparisons to other recent primary states turnout, 2008 turnout rate, the expected amount of voters based on early voting, anecdotal evidence, the known facts like number of polling stations, etc.) and Sanders won 70% of them (which is I believe what the actual rate was among election day voters)-

      That would mean that the difference between C and S would have gone down 18,000 to 54,000 or so. That’s a pretty conservative estimate and comes to the population equivalent of about 2 delegates. As a Sanders supporter, I want those 2 delegates back (a net 4 delegate difference).

      And that’s discounting the possibility that, for whatever reason, this underestimates the suppressed turnout, which you seem very comfortable doing, without very much evidence. It wouldn’t be shocking if it was substantially higher because of factors that haven’t been taken into account, given that this election cycle has been, if nothing else, a little less predictable than most.


    • You do know voter suppression was only one part of the outrage? There was lots of evidence that pointed toward election fraud. People’s party ID’s got changed to independent, republican, or librarian.

      Liked by 1 person

      • Yes I get that and it still isn’t going to make up a gap of 72,000 votes… And yes, those extra delegates would be nice, but even if Bernie gets 2 additional delegates for a 4 delegate swing, he’s still down 225 delegates or so!

        And to SS, you can believe that Sanders got 70% of election day votes, but believing it doesn’t make it true. There is absolutely no evidence to suggest that. Furthermore, if you look at where these votes were suppressed, you’ll see they tended to be in heavily Hispanic areas who favored Clinton in Arizona… So essentially, I’m not buying anything you’re saying.

        Last point, in case it isn’t clear, if there is disenfranchisement in Arizona, that falls on the GOP, not Hillary. She had nothing to do with it.


  5. Great job on your projections, man! Much better to underproject than to overproject. Washington was almost dead on too! Especially impressive considering these are all caucuses.

    Liked by 1 person

  6. Looks like you nailed it Tyler – actuals are 81 in AK, 72 in WA, and probably about 65 in HI – 2 of 3 right on, the other’s just crazy. Still really think it would be good to build more testable metrics into this and not be so dependent on untestable Facebook data. In business you have to be able to explain why things work, or why they blow up. You’ll eventually have to go in that direction and understanding the dynamics of those metrics now, even if you’re just testing them on the side, would be immensely useful.

    Liked by 1 person

  7. I just back from my caucus here in SW Washington. The 70% Bernie looks pretty accurate. Our district awarded 5 delegates to Bernie & 1 to Hillary. Keep up the good work, Tyler.

    Liked by 1 person

  8. I hope Bernie does even better than your predictions, to make up the fraud he faced in Arizona. I’m hoping Hawaii is also 70-30 because Tulsi Gabbard endorsed and campaigned for Sanders. That’s gotta count for something.

    Liked by 1 person

  9. I think your model may have predicted day-of voting in Arizona pretty accurately. According to the election results website of Arizona’s largest county, ( Bernie won 60.8% of votes cast on March 22. Problem is, at least in this county, about 85% of total votes were cast by absentee ballot. Many of these ballots may have been sent in weeks before you ran your projections.

    It is interesting that such a large portion of total votes in Arizona are cast by absentee ballot. Is there any other state where early voting constitutes the majority of total votes?


      • Great job! We’ve been following your predictions.
        As a geek I’m curious if you wrote a program or are you using statistics software?

        You should be aware: Here in Oregon we do mail-in ballots. Everyone receives their ballot in the mail at the same time in April and they’re due on Election Day in May. Will this require a larger interval for your prediction?


      • I only use Microsoft Excel and Stata, though you could do mostly everything I’m doing with just Microsoft Excel.

        Interesting on the Oregon thing. That will definitely help Hillary. Working on incorporating early voting into my model but there are some difficulties associated with it.


  10. Hey, Tyler

    Have you considered the effect of early voting?
    A snip from a recent piece on Huffpo-

    “About Hillary Clinton winning Arizona: she certainly did. She won early voting, which started weeks before Election Day — well before Sanders had started advertising or campaigning in the state — by 61.5% to 36.1%…
    So what actually happened on Election Day? She lost the live voting 52% to 48%, with thousands of provisional ballots yet to be counted-“

    The author refers to himself as a ‘hard data’ guy – and whether it ends up helping you or not, it’s worth reading!
    Thanks for what you’re doing.

    Liked by 1 person

  11. I really don’t like any narrative that excludes Blacks from being Bernie supporters. I think it’s Clinton propaganda. Considering all the reports of voter disenfranchisement, suppression and corporate media lies, I don’t think it’s fair. I do think it’s divisive. I don’t trust their data.


  12. I’m sorry, but this is really getting embarrassing. That “article” you posted was utter nonsense. The author literally tried to argue that voter fraud occurred because independents weren’t allowed to vote in a CLOSED primary. The article also argues that more than a MILLION Sanders voters were turned away. If that were true than the 2016 primary would have had three times more voters than the 2008 primary. Give me a f’ing break.

    Your projection for Arizona was wrong and you should take responsibility. Instead, you assume that:

    1) At least 70,000 voters were being kept from voting.
    2) Nearly ALL of them would have voted for Sanders.

    Even the most optimistic estimates had Sanders basically tied with Clinton among voters who cast their ballot on Tuesday. How can you possibly assume that he would have overtaken her lead with those margins? Its just astounding that you would claim that Sanders lost over 70,000 votes due to fraud. Thats obviously nonsense.

    You should stop working backwards from the assumption that your model was correct. Just take a step back and relax for a second. You are now endorsing conspiracy theories in order to support your projections.


    • Regardless of what other flaws there may be with that article, your point about ‘independents shouldn’t have been able to vote because it’s a closed primary’… the problem is that a lot of people who had newly registered or updated their registrations as Democrats in time, and even some people who had been registered as Democrats for years, showed up on election day in Arizona only to be told that the computer record now showed them as anything other than a Democrat. For example, I saw someone who said she had been a registered Democrat since 2010 but was now mysteriously coming up in the AZ voter registration database as a Libertarian.


    • It’s not a conspiracy theory with all the evidence that has been presented. I think your just uninformed or naive. There has been so much evidence as to some type of fraud. He is just giving a theory based on scientific evidence. The man talks in science while you talk with no evidence to back anything up.


    • Agree. This is not smart data analysis. Even the assumption that Bernie Sanders supporters were disproportionately disenfranchised in Arizona defies logic. I made GOTV calls in Arizona all day Tuesday. The people who told me that they had been unable to wait out the lines were ALL older voters. With one exception, they were all Hillary supporters. I spoke to one couple who had waited in line for over an hour *two different times* on Tuesday, only to finally give up in exhaustion each time.
      Obviously, my anecdotes are not statistically valid data. But they are evidence of the logical problem inherent in insisting that ONLY or primarily Sanders supporters were kept from voting.
      Moreover, if you look at Sanders’s success in caucus states, you can clearly see that the youth vote is able to turn up and wait out a longer voting process than are Clinton’s older, more diverse supporters.
      The voting problems in AZ were atrocious. So were similar problems in NC, FL, and numerous other states this cycle – almost all states with Republican controlled Boards of Election. The lesson to take from this is not that Sanders’s voters exclusively are being disenfranchised. It is definitely not “voter fraud.” The lesson here is that Democrats are going to have to work VERY HARD in the fall to make sure their voters get to vote! This affects all of us!!


    • Yeah I feel like Richard’s article was extremely biased, and I wish that he had been more objective. Also, the language he used when talking about the independent thing made it seem like he was angry that independents weren’t allowed to vote in a closed primary- which is ridiculous, so I agree with you 100% that that is a ludicrous argument. However, I interpreted it as a reference to the database system switching voter affiliation though, which has been widely reported.


    • Turnout in Arizona per current updated results is 48%. 455,190 ballots cast for 948,983 Registered voters. In 2008, 455,635 ballots were cast, so basically the same number lol.

      This is beyond embarrassing. How do people take this seriously? Seriously… when I interview someone I google them. If I found this I would NEVER hire him. He really needs to wake up from dreamland.


    • Craig,
      You haven’t been doing any research on this topic. That much is clear, because if you knew what the hell you were talking about you would know that Tyler was right, and that only massive voter suppresion and a “computer glitch” turned away a huge percentage of first time voters, most of who were Berniecrats. There’s already a couple investigations happening, so pretty much everybody, excluding you and HRC acknowledge that some bad things happened in AZ, that DID effect the outcome.


    • Is using the word “conspiracy theory” your way of delegitimizing people’s concerns? When it comes to the “computer glitch” or the Sanders’ database hack, IF something intentional was done, it does not need a conspiracy (using the traditional sense, not in the loaded-word sense), all it takes is one person with an agenda (or the desire to do it for the LULZ) who has access to a computer.

      Liked by 1 person

  13. Hello Tyler

    I’ve been following your work lately and plan to follow in your footsteps in 2020/2024.

    I think Washington state will be more of a landslide. 75-80% for Bernie having looked at the age demographic data for each state otherwise I think you’ll be fairly correct in your predictions. How much weight do you add to the age groups per state?

    I’m also intrigued as to how much weight caucuses vs open primaries vs closed primaries get? Obviously you won’t give up every detail but I was hoping you’d enlighten me just a little.

    Thanks and keep up the good work

    Kind regards from the UK



    • I think people are blowing this out of proportion, Bernie has a strong organization in Washington and they’ve been pushing the affidavit forms so I think the split will be a lot less dramatic than other states


  14. So glad you’re still updating your predictions. I felt like your Arizona prediction would have been spot on without all the election fraud taking place. These numbers look very good for Bernie although I’m hoping to outperform those predictions. Keep doing good work, Tyler! And thanks again!


    • Turnout is at 48% with current updated results and Bernie is still behind 57% to 41%.

      His model was absolutely wrong beyond comprehension since he was 99.8% certain Bernie would get 51% and he didn;t come close.

      You people dont realize there are only so many registered voters in the pool, and there is never 100% turnout, let alone 75%


  15. I can almost assure you that Hillary wining American Samoa and CNMI had little to do with the demographics of the Islands. I was part of the small call team for CNMI. We started making calls only two days before the caucus, but we wished we had started earlier. We talked with many people who had no idea who Bernie was. We talked with people who had no idea they could even vote. In fact, there was very little publicised information about the caucuses on the islands (I believe this was the first year the CNMI held a democratic caucus). On the smaller islands (Tinian and Rota), we talked with people who thought the caucus was an event for Hillary only, so they weren’t going to show up. My sense in that in the Pacific territories, there is such a disconnect from mainland politics that typically only party officials and their friends/family turn up to these events–and as we’ve seen, party officials tend to back the establishment candidates.

    Liked by 3 people

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 )

Facebook photo

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

Connecting to %s