Graph of the Day: Trump approval and disapproval trends, 26 August 2017

26 August 2017 (Desdemona Despair) – How low can Trump go, and how long will it take him to get there? One way to gauge public sentiment is to aggregate data from multiple polls and compute an estimate. Fortunately, the good people at FiveThirtyEight are doing this for us at their excellent site, How popular is Donald Trump? Today (26 August 2017), Trump’s approval rating is bumping along the bottom at 36.9%, very close to his lowest rating of 36.6% on 7 August 2017. Where is it likely to go from here? The FiveThirtyEight model is pretty sophisticated, and it predicts a general increase in Approval, and a corresponding decrease in Disapproval.

What do the simple linear trends look like? The FiveThirtyEight folks generously post their trend data for public consumption, so it’s easy to find out.The top graph shows the linear trends for FiveThirtyEight’s Trump Approval and Disapproval estimates. Various other curves (e.g., exponential, logarithmic) don’t improve the fit by much. The current rate of increase in Trump Disapproval is about 0.05% per day. At this rate, Trump Disapproval increases by a full percentage point every 20 days. Extrapolating the trend, the average poll will have Trump at 60% Disapproval sometime in early October 2017.The current rate of decline in Trump Approval is about –0.04% per day, or –0.4% every ten days. At this rate, Trump loses a full percentage point every 25 days. Extrapolating the trend, the average poll will have Trump at 30% approval sometime in late February 2018.In the Disapprove signal, Desdemona’s eyechrometer detects some nonlinearity in the first month or so of data. What do the trends look like for recent polling only? Here are the estimates for the last two months:

Over this time span, Trump’s Approval rating declines more rapidly compared with the full data set, at about –0.05% per day, and his Disapproval increases less rapidly, at 0.04% per day. Extrapolating these trends, Trump Disapproval crosses 60% in November 2017, and Approval falls below 30% in January 2018.Obviously, these curves could be deflected by all sorts of drivers: a sudden event, like Trump going to war with somebody, could cause a step-change in these curves; or a general lessening of his administration’s incompetence could allow a slow return to the mean, like what the FiveThirtyEight model predicts. For example, if Trump doesn’t completely botch the Federal response to Hurricane Harvey, he may be hailed as a success, causing an inflection in the approve/disapprove trends.Also, there’s reason to expect a floor in Trump’s Approval rating. Even Richard Nixon’s freefall from nearly 70% stabilized at around 25%, in what appears to be a smooth exponential decline.

But with 219 data points so far, the trends for Trump are consistent and quite linear, so there’s no reason to expect them to change anytime soon. At some point, Trump’s approval rating will approach a floor asymptotically, but there’s no evidence of that happening yet.