Opening presents, watching the ball drop in Times Square, eating roast beast. To these cherished end-of-year traditions, allow me to add one more: gawking at how bad my predictions were. It’s an exercise that both FiveThirtyEight (thanks to our former senior political writer and analyst Harry Enten) and I (at my old blog) like to do to hold ourselves accountable — and hopefully to make us better prognosticators in the future. Without further ado, here are some of the many things I got wrong in 2018.
A question I obsessed over, especially early in the year, was whether the 2018 election results would fall along the fault lines of the 2012 presidential election or the 2016 presidential election. (Case in point: In a March Slack chat where we were asked to pick the best bellwether district for the 2018 election, I chose two — a Romney-Clinton district and an Obama-Trump district.) The tea leaves were a bit murky, but ultimately I expected the country’s 2012 partisanship to win out. That was based on an article I wrote in April that looked at hundreds of special and off-year elections and tested whether Democrats were doing better1 in more-suburban or less-suburban areas. It turned out that Democrats were doing slightly better in “Trumpier,” less-suburban areas, suggesting a possible reversion to the 2012 map.
But after the dust settled on Election Day,2 it turned out that I needn’t have spent so much mental energy on the question. Democrats did well all over, winning 12 of the 13 Romney-Clinton districts and 15 of the 21 Obama-Trump districts, including both of my bellwethers. To the extent that the 2012-2016 question had an answer, I had