I’ve been playing around with some presidential approval data, inspired by Sides and Vavreck’s analysis in The Gamble. They predicted Obama’s approval ratings over the course of his two terms, using data from Truman to Bush II. The underlying question they ask is whether Obama’s actual approval at various points was higher or lower than what the model they design would predict. Sides graciously shared the data with me, and I’ve used them in my courses.
Using a simple model that predicts quarterly approval as a function of current economic factors, a president’s time in office (new president vs. late second-termer), and the presence or absence of determined positive or negative events, I can compare Trump’s actual approval against the model’s prediction for him. I show that below for Trump’s first five quarters in office.
To date, Trump is more unpopular than the statistical model expects, but that has changed in the last quarter. In the first three months of 2018, Trump’s approval ratings have moved upwards to match more directly what we might expect for a different president situated currently in this economic environment. This is despite having some rough patches this past quarter, such as the Fire and Fury book, the Conor Lamb loss in PA-18, and the continued swirl of the Mueller investigation. (I also give him a bonafide positive event in the quarter with the announcement of the North Korea talks, first made public in March.)