When it comes time to breaking bread on the holidays, just remember that 58% of people in the room is more than half the people in the room. Image via Flickr CC Satya Murthy.
Well, I’ve done you a favor and shielded you from these juicy mathematical and political morsels until after Thanksgiving. A recent NPR/PBSNewshour/Marist poll showed that 58% of people were not looking forward to discussing politics at their holiday table, while 31% said they were looking forward to it. Now, I know this isn’t really how these things work, but my preferred reading of this is the following: if you are one of the 31% who is excited to talk about politics while eating your turkey, look to your left, and look to your right. Neither of those people want to talk to you.
So I’ve spared you from being that guy. But spare no more! I have some mathematical and political news I want to talk about!
Nate Silver, editor-in-chief of FiveThirtyEight finished his 11 part series The Real Story of 2016, examining the misunderstood polling, flawed mathematical modeling, biased human intuition, and journalistic errors that went into the coverage and execution of the 2016 election. I just finally read through the whole thing today and it is an eye-opening interrogation of the intersection of math and media. Silver gives a thorough analysis of what went wrong in the Electoral College math for Clinton and what went right for Trump.
In part VII, Clinton’s Groundgame Didn’t Cost Her The Election, Silver provides some compelling data from after-the-fact regression analysis to point out the unavoidable importance of demographics in election outcomes. We can pick apart the ground strategies and die speculating that if-only-she’d-gone-to-Wisconsin, but Silver makes a pretty clear case that it would have made almost no difference.
In the final installment, The Media Has A Probability Problem, Silver talks about how easy it is for journalists and consumers to misinterpret probabilities. If one candidate has an 85% chance of winning, that just means that the modelers have spun out all likely scenarios and in 85% of possible scenarios the candidate will win. Which means she still loses in 15% of them. And this is totally different from polling at 85. And we mathematicians obviously get that, but I think when people glance at infographics those numbers can be quickly and easily conflated and internalized in the wrong way.
Understanding what numbers mean and why they work is so important and sometimes so difficult.
Since we’re already in the political mode, if you haven’t read it yet, I implore you to go read Cathy O’Neil’s Weapons of Math Destruction. It’s a trove of infuriating examples of data being used and abused to the detriment of democracy and the citizens of the United States. If you can’t see yourself reading an entire book in the next few weeks (I know, it gets hectic, but come on, treat yo self) then at least read O’Neil’s recent New York Times oped about the looming specter of big data and algorithms.
So, I’m sorry I didn’t share this with you sooner so that you could simultaneously dazzle your Aunt TeeTee with your pumpkin pie recipe and all the facts she so desperately craves. The good news is, you now have all of this in your back pocket so that you can be sure to win every conversation you have at every holiday party in the next month.