Early Returns

Taking the Rational View of Polls

Jonathan Bernstein's morning links.

Chill out.

Source: Desk/AFP/Getty Images)

Did I mention I was cranky about polling? I'm cranky about polling, or at least the way some have been talking about the polls. 

First of all, please stop saying that the 2016 election "proved" polls are irrelevant. The election did no such thing. What the polls got wrong were a handful of states; the national polls in the general election did quite well, and even with some state polls off -- as they will be at times -- Nate Silver had correctly been beating the drums about Donald Trump's electoral college advantage for weeks before the election. In the primaries, which are harder to poll, the predictive power of surveys was pretty normal -- a few spectacular misses, which isn't unusual, but plenty of accurate forecasts. 

Polls will sometimes be wrong, and certainly have to be interpreted properly. A few predictor programs were far too confident in a Hillary Clinton electoral college win, but that seems to have been a problem with those prediction systems, not the underlying polling. And even those weren't so far off that we would say polls were irrelevant. Polling and interpreting polls remains a complicated topic, but there's simply no reason to toss everything out based on 2016 results.

Second of all, can we please, please, please stop overreacting to single survey surges and drops? Far too many people read far too much into the single Gallup 2-percentage-point Trump approval drop reported Monday, a drop that was reversed on Tuesday. Polling averages, polling averages, polling averages. Look: I enjoy polling trivia as much as anyone, and it's fine, really, that people noted that the Gallup approval tracker for Trump hit a new low on Monday. Placed in proper context, that's a worthwhile story. But anyone acting as if Trump really became less popular over the weekend based on that result was Doing It Wrong. C'mon, folks, we know better than this.

And third? The polling averages are great. They're also, as you'll notice if you click through, all slightly different. Those differences spring from different designs, and the underlying decisions are, to be frank, fairly arbitrary. That's fine; each one can justify the way that they do it, and we're much better off with all of them. But the point here is we really don't know exactly where Trump's approval (or anything else) is on any particular day, and therefore ascribing causal relationships for relatively small fluctuations is, in my view, just not sustainable. That is, we can say that Trump's approval is down about 5 percentage points over the last three months, but whether that drop was a steady one or one punctuated by one or more cliffs is hard to know, and therefore knowing which event or events was responsible is also hard to know. 

1. Ashley Jardina at the Monkey Cage on the politics of white identity

2. Also at the Monkey Cage, James Fearon on trust and the North Korea situation

3. Elaine Kamarck at Brookings on Trump as a weak president

4. Dan Drezner on running government like a business

5. Richard Skinner at Mischiefs of Faction assesses the potential Republican challengers for the 2020 presidential nomination. 

6. John Harwood on Trump's moral sense

7. The bright spot of the Republican fight against Trumpism has been among several conservative writers, who have not only criticized Trump (easy!) but also taken a serious look at where the movement has gone wrong (much harder). Here's Jonah Goldberg at the National Review with one critique

8. My Bloomberg View colleague Francis Wilkinson makes the liberal case for President Pence

9. And Josh Kraushaar makes the case that 2020 is an opportunity for a third-party presidential run. Alas, I have to say: plausible! Well, at least somewhat. I think Kraushaar overestimates both the chances that Democrats will nominate a truly fringe character and the chances that the Democratic candidate's ideology and policy positions will matter all that much. Both are possible, but unlikely. He's correct that failed or unpopular presidents up for re-election appear to be what cause significant third-party runs (see 1948; 1968, even though Lyndon Johnson eventually dropped out; 1980; and 1992). But the same dynamics that shake loose some partisans and make them more open to third-party candidates also tend to make the out-party candidate look better. Thus Republicans were able to elect Ronald Reagan in 1980 even though he might have been too conservative to be elected in some years; against Jimmy Carter, everyone was going to look good. But a third-party win? It's not technically impossible, but it remains extremely unlikely. 

Get Early Returns every morning in your inbox. Click here to subscribe.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

    To contact the author of this story:
    Jonathan Bernstein at jbernstein62@bloomberg.net

    To contact the editor responsible for this story:
    Brooke Sample at bsample1@bloomberg.net

    Before it's here, it's on the Bloomberg Terminal.
    LEARN MORE
    Comments