I have recently been developing a statistical model for the sake of predicting the electoral outcomes of the Democratic primaries. I have used the first three primaries, Iowa, New Hampshire, and Nevada, to gather the relevant data for the model. This model is experimental in the sense that I am using data that has been aggregated from social media to make my estimates. To my knowledge this has never been done before, but after this election season we will be able to determine the viability of social media political sentiment as a proxy for broader public political sentiment.

There are many elements and controls to any good statistical model, and at this point there is not enough variance in some of the variables that I would like to include; nor is there enough observations to truly call anything statistically significant. However, the model will continue to get more and more accurate as the primary season progresses. Regardless, I am very confident in the predicted Super Tuesday (and beyond) outcomes at this point. Here are the estimates:

Massachusetts and Oklahoma are italicized and bold because I have determined they are too close to call. By slightly changing one of the underlying assumptions in my model, a different winner results. I do believe that Bernie will win Massachusetts and Hillary will win Oklahoma, but that is really no more than conjecture. I hope you all find this interesting. For anyone with a background in statistics and/or econometrics that would like to know more about the fundamentals of my model, just shoot me an email and I would be happy to discuss it with you.

-Tyler