A while back, I became interested in the quantitaive analysis of polling data. Specifically, out of frustration for how it is normally done. Typically, you will see polls either reported in isolation:
Or, you will see them reported as a moving average, ala FiveThirtyEight style:
I am dissatisfied with both presentations of the data, because they prioritize noise and focus on minutia. There are better statistical techniques for reporting, specifically Bayesian Change Point analysis. This type of analysis is not about wavy lines over time — rather it attemps to identify specific points in time that predict future movements in a time series. I’ve written about this in the past, but just a quick refresher on what has mattered in the polls:
- Both Independents and Democrats saw big drops in Trump approval during the debate about innauguration crowd size.
- Independent approval eroded further in March while Trump was tweeting about “tapps” and the GOP healthcare bill failed
- Republican support dropped further in May when then-FBI Director James Comey was fired.
As I said at the time, this approach gave a much more nuanced look at polling data, because it allows us to mark which events have the greatest impact on Trump’s approval. So, what has changed in the past few months? Here are Trump’s approval ratings as of this writing (mid-March, 2018):
As you can see, Trump’s approval ratings were slipping until they bottomed out in the summer of 2017. There, they remained constant until the end of the year. Starting December, there is a drop in Trump’s disapproval of about 3.7%, and a corresponding increase in his approval of 3.8%. This begs the question — who are these people and what changed their mind?
The first thing to look at is to see if independents have changed their mind:
So, nope, not independents. They’ve stayed consistent since March of last year (either “tapp” tweets or the failure of the Obamacare repeal bill in the House). The change isn’t with Democrats either:
Democrats haven’t moved much at all. The increase in Trump’s approval comes entirely from Republicans:
When Trump fired James Comey in May of 2017, his approval among Republicans dropped by 4%. His disaproval among Republicans ticked up proportionally by about 3.9%. Needless to say, this did damage with his standing among Republicans that held constant for the remainder of 2017. However, starting in December of 2017, Trump’s approval among Republicans ticked up 3.2% from 80% while his disapproval dropped by 2% from 16.3%. This represents a recovery of almost 75% from his firing of James Comey. This begs the question — what happened?
There were two significant events at the end of December, 2017. First, a major overhaul of the US tax code was passed narrowly on partisan lines by Republican majorities in Congress. Second, conservative media outlets trained their fire on the Department of Justice’s investigation by Robert Mueller and his superior, Rod Rosenstein. This can tested by looking at Google trends to see when their respective peaks were reached.
The peak for Rod Rosenstein in the news was reached February 3rd, 2018 — a bit too late for late December/early January improvement in Trump’s approval among Republicans. But, Tax Reform is a different story.
The peak for tax reform almost perfectly coincides with the return of Republicans to the Trump fold in late December.
So, the lesson is clear — the more legislation that Republicans pass, the more they will excite their base. But, they have two significant obstacles to doing this. First, the surprise election of Doug Jones (D-AL) to the US Senate in December of 2017 reduces their majority to 51 — with John McCain’s absence (R-AZ) giving them the barest of majorities at 50 votes. Second, their leadership in Congress has a very modest agenda set for the rest of 2018 — they are afraid to put tough votes in front of their members.
This presents a problem for Republicans. They lost 4% of Republicans with the firing of James Comey, and gained 3% back by passing their tax bill. But the gap between independents widened from -7 points to -19 points when their (ultimately failed) proposal for repealing Obamacare emerged in March of 2017. And legislatively, that is being felt with multiple special election defeats.
Implications for the Mueller Investigation
It’s actually very simple. This doesn’t help Trump in the polls:
This is stuff that at least undoes some of the damage with Repubicans:
So, while Trump managed to win back some of his lost support among Republicans, he hasn’t done anything to win independents back or dampen Democratic opposition (both do not like the tax bill). His improvement in the polls may not tell us much about the midterm elections coming this November.
This also serves as a warning to Trump and Republicans — the Mueller investigation, and any actions by Trump (e.g. firing James Comey) can only hurt them. The more they change the subject, the better off they’ll be.
####Quick Note
The code used to generate these data is available on GitHub. You can also run it in your web browser, even if you don’t git.
