Monthly Archives: January 2020

How concerned should we be about the Wuhan coronavirus (2019-nCov)?

By Kent R. Kroeger (, January 29, 2020)

Data current through 27 January 2020

Data current through 27 January 2020

The Fandom Menace: More media-created myth than reality

By Kent R. Kroeger (, January 24, 2020)

My popular vote forecast for the 2020 presidential election

By Kent R. Kroeger (, January 8, 2020)

Occasionally, I pocket my ego and read comments left by readers of my various essays and articles.

Two essays in particular, Trump’s market share problem and Trump should deny his enemies their ultimate victory, attracted an unusually high percentage of negative responses, as both essays seemed to annoy strong partisans on both ends of the political spectrum.

In both essays I argued Trump’s probability of reelection is significantly reduced for three reasons: (1) A shrinking Republican base, (2) a shrinking pool of swing voters who could potentially alter the outcome on Election Day, and (3) Trump’s stubbornly low (almost invariant) job approval rating.

Many readers vehemently disagreed with my conclusion.

Donald Trump supporters disagreed with my pessimism about his chances of being reelected, while Trump-haters suggested I am a Republican troll trying to promote complacency among Democrats and thereby ensure his reelection.

However, as those who know me can attest, I possess no strong affinity towards any party or ideology.

For me, Trump is a market researcher’s dream challenge. How does a president so deeply disliked by nearly half of the U.S. adult population get reelected?

It is not going to be easy.

In the polling era, only one president — Barack Obama — has been reelected with a job performance rating as low as Trump’s is less than one year out from Election Day. At this juncture in Obama’s first term, his Gallup job approval stood at 42 percent (see Figure 1 below), compared to 45 percent for Trump in Gallup’s December 2–15, 2019 survey.

Figure 1: Comparing 1st-Term Job Approval for Trump, Obama & G. W. Bush

Source: The Gallup Organization

However, there is a major difference in Trump’s situation. Obama was above or near 50 percent job approval for the first year of his presidency. Trump has never sniffed 50 percent approval, his highest approval being 46 percent approval, according to Gallup. A strong, though not conclusive, piece of evidence that Trump may have a maximum approval threshold around 46 percent.

Is 46 percent approval enough to win reelection?

According to one of my anonymous readers, 46 percent approval would make Trump very competitive in the next election. For evidence, he/she directed me to a popular vote econometric model created by political scientists Michael Lewis-Beck and Charles Tien (see Figure 2). Their two-variable model predicted Hillary Clinton would win 51.1 percent of the two-party popular vote in 2016 — which was exactly her two-party vote share.

Figure 2: The Lewis-Beck-Tien Presidential Vote Share Model

Source: Lewis-Beck, M. S. & Tien, C. (2016). The political economy model: 2016 US election forecastsPS: Political Science & Politics, 49(4), 661–663.

There is no guarantee the Lewis-Beck-Tien (LBT) model’s bulls eye accuracy in 2016 will be repeated in 2020. In every election, one model is always going to have the most accurate prediction, either by chance or model validity.

Nonetheless, the LBT model has predicted the winner in all presidential elections since 1948 (with the exceptions of 1960, 1968, and 1976).

The LBT model also has the virtue of simplicity — just two variables: Presidential approval in July of the election year and GNP growth in the first two quarters of the election year.

Assuming current conditions remain constant through Election day, what does the LBT model predict for 2020? The LBT model predicts Trump will win 49.4 percent of the two-party presidential vote. Even if we assume no economic growth in the first two quarters of 2020, the LBT model predicts Trump will receive 49.2 percent of the two-party vote if he maintains 45 percent job approval through Election Day.

Accepting the predictive validity of the LBT model — and its failure to predict the 1976 winner (Jimmy Carter) gives me significant pause in this regard — Trump will be competitive in 2020, all else equal.


Given this evidence, therefore, do I retract my previous conclusion that Trump has little chance of winning reelection in 2020?

Yes and no.

Even before considering the possibility of another Electoral College victory and popular vote loss by Trump, there are other econometric models using presidential approval and the economy — besides the LBT model — to predict presidential popular vote outcomes (descriptions of the most widely cited models are found here). And, currently, these models predict Trump will receive between 45 and 50 percent of the two-party popular vote.

So, yes, I was wrong to assert Trump has a meager chance of winning reelection. In reality, he has more than a punchers chance.

Yet, I stand by my other conclusion that Trump has little room for popularity growth. The Donald Trump we see today — in terms of job approval — is likely the same one we will see in November 2020.

As I stated in my February 2019 essay, “Donald Trump, the candidate, has a market share problem. His support has very little upside and — as of now — has a maximum potential market (vote) share of just 45 percent among eligible voters.”

Given a shrinking political center amidst a highly partisan electorate combined with a Republican base also in demographic decline, Trump has a market share growth problem that is only getting worse. There simply are not enough likely and potential Trump voters to ensure his reelection. Which is way many pundits and forecasters, not just me, have questioned Trump’s base-focused electoral strategy.

Among them, political blogger Kevin Drum, wonders if the Republican Party is prepared for the day of reckoning should it continue Trump’s base-focused strategy in 2020.

In 2012 I thought that the Republican Party had gone as far as it could to get votes from the white working class. There was just no more blood to be squeezed from that particular turnip. But I was wrong — barely. In 2016 they did what no one could have predicted by nominating a guy who was an all but open racist. And it worked, buying them just a few more votes than it lost them. Once again, they gained a few years.

But it can’t last forever. The Republican Party has already gone much further down the road of lashing itself to the cause of white racial resentment than I would have guessed possible. How much longer can it last? Like most bubbles, longer than you think. But someday the bubble will burst. The big question is, what will happen next?

Trump’s limited market growth potential is evident in the December 2019 poll, which found 47 percent of U.S. adults possess a ‘very unfavorable’ opinion of Trump (see Figure 3). No other major politician comes close to the intensity of Trump’s unfavorability. The next closest is House Speaker Nancy Pelosi at 38 percent of U.S. adults having a ‘very unfavorable’ view of her.

Conversely, 30 percent of U.S. adults view Trump ‘very favorably’ — more than any other major U.S. politician. The next closest political figure in this regard is Vice President Mike Pence at 22 percent.

It is this large, intense pro-Trump base that fills his pep rallies and gives the media pundits the impression that the incumbent president will be hard to beat in 2020.

“Look at crowd sizes and enthusiasm as a better metric than opinion polls which oversample Democrats,” writes Denver physician and conservative writer Brian Joondeph, MD.

Unfortunately for Dr. Joondeph, the evidence does not support his idea of using crowd size as a proxy for electoral viability. And the YouGov survey plainly shows us why: Trump’s loyal base (30% of the population) is swamped in size by the number of Americans that can’t stand him (47%).

What the YouGov survey also tells us is that most Americans have strong feelings about Trump, whether they be positive or negative. Only 7 percent of U.S. adults lack an opinion on Trump.

Figure 3: Favorability Ratings of Selected Political Figures (December 2019)

Source: The Economist/YouGov Poll; Conducted Dec. 28–31, 2019 with 1,500 US adults; Margin of Error = ±2.7%

At first blush, the 47%-Very-Unfavorable figure does not bode well for Trump’s reelection chances. My analysis of the 2016 American National Election Study found that three percent of eligible voters with a ‘very unfavorable’ view of Trump in October 2016 voted for him anyway (see Figure 4). That some Trump-dislikers did this was apparently driven by their even greater distaste for Hillary Clinton, particularly on their perception of each candidate’s honesty.

Figure 4: Favorability Ratings of Selected Political Figures (December 2019)

Data source: American National Election Study (2016); analysis by Kent R. Kroeger; data are weighted

Even if we subtract the 1 to 2 percent of Trump-dislikers who will still vote for him (3% of 47%), this puts him in a deep hole as he tries to pull off another Election Day miracle: Trump starts with 45 percent of eligible voters most likely unwilling to consider his candidacy. In other words, Trump has little margin for error to work with heading into the 2020 campaign.

But virtually the same argument could be made from the perspective of the eventual Democratic Party nominee. In a highly-polarized electorate where almost 80 percent of adults have already made up their minds about Trump — and half of all eligible voters find our political system so irrelevant to their lives they don’t even bother to vote — a small percentage of swing voters can tip the balance.

Figure 5 (below) is a back-of-the-envelope exercise that shows how close the presidential popular vote could easily be in 2020 — despite the substantial favorability disadvantage Trump takes into the contest.

Using Trump’s favorability distribution from the December 2019 Economist/YouGov Survey, along with voting ratios estimated from the 2016 American National Election Study (ANES), I generated a two-party popular vote forecast for 2020.

And the winner will be: The eventual Democrat nominee…51% to 49%. This forecast is almost identical to the LBT econometric model forecast detailed earlier in this essay.

Figure 5: 2020 Presidential Vote Forecast

This forecast should not be surprising. American presidential elections tend to be relatively close. And despite my earlier musings on the dire situation facing Trump and the Republicans, I am now convinced Trump has a legitimate chance of getting reelected.

  • K.R.K.

As always, please send suggestions and complaints to:

The Democratic Party’s impeachment strategy isn’t as futile as it appears

By Kent R. Kroeger (, January 6, 2020)