The Economist - USA (2020-09-05)

22 United States The EconomistSeptember 5th 2020

2 there to “domestic terror”. Earlier, when

asked about the white teenager charged

with killing two protesters and injuring an-

other in Kenosha, who has become a cause

célèbre among some conservative media

hosts, the president suggested he had acted

in self-defence.

Promises to restore law and order might

seem more persuasive coming from an in-

surgent outsider than from an incumbent

president. When Mr Trump took office in

2017, he pledged “I alone can fix it,” and

promised an end to “American carnage” in

his inaugural address. Now he argues that

the explosion of urban violence under his

administration is proof of Democratic mis-

management and the need for him to have

a second term. Some accept the argument.

According to a poll released by YouGov on

August 28th, 32% of adults believe re-elect-

ing Mr Trump would make the country saf-

er (though a plurality, 43%, think he would

make it less safe).

Can such a strategy work? Some com-

mentators, perhaps scarred by the last con-

test where Hillary Clinton seemed to be the

heavy favourite throughout, already spy a

polling bounce for Mr Trump. At least in the

national polls, the purported bounce looks

quite dainty. His vote share has increased

by one percentage point, from 45% on June

17th to 46% in The Economist’s average of

presidential-election polls. Some of that

bump came in the week after the party con-

vention—when incumbents usually enjoy

a real, if ephemeral, boost in support dur-

ing the post-coronation bliss.

Previous racial unrest has not been es-

pecially kind to the president. According to

our maths, Mr Trump experienced the

worst three weeks of the campaign after

George Floyd was killed by police in Minne-

sota on May 25th. His deficit in the polls

ballooned then from seven points to 11

points in a relatively short span of time.

It is still too early to determine how

large Mr Trump’s national bounce will ulti-

mately be, or whether it will subside. So far,

the president’s average support over the

course of the election has been both low

and the most stable on record. And since

the national election will not be decided by

the popular vote but by the electoral col-

lege, if he ekes out a victory again this time,

it is likely to come by a similar route.

At the state level, our election model-

ling finds Mr Trump’s position in the Mid-

west has improved relative to his June low.

Whereas he was down 11 points in Michigan

two months ago, he is down just six now; in

Wisconsin, an 11-point hole is now a six-

point one; and the president is down just

five points in Pennsylvania, up from nine

earlier this year. These are still significant

deficits, but the trend is certainly looking

up for Republicans.

Mr Franklin, the Wisconsin pollster, is

also cautious. Older voters are dismayed by

the president’s handling of covid-19. Many

disapprove of his approach to race rela-

tions, although they still trust him on the

economy. For weeks, polls have shown

public support for Black Lives Matter to be

slipping. In Wisconsin, approval peaked at

61% in June before falling to 48% in August.

It may be even lower now. So far, however,

Mr Franklin has seen no evidence of that

shift affecting overall support for either Mr

Biden or Mr Trump.

The foot soldiers feel the strain. “We’re

ground zero, we’re under the gun,” says

Terry Dittrich, the Republican chairman in

neighbouring Waukesha County—the bell-

wether county for election-night pundits.

Should Mr Trump fall short here, “the rest

of the state has to make up for that”. Mr Dit-

trich admits there has been a wobble of

late. The Republican Party typically expects

to scoop 65-70% of votes in his county. In

the midterms, in 2018, the Democrats’ vic-

torious candidate for Senate, Tammy Bald-

win—an openly gay war hero—scooped a

heady 38% support in Waukesha, points

out Matt Mareno, the Democrats’ party

chairman in the county. If Mr Biden could

match her result, then he predicts “it’s

game over” in the state.

So far, Republicans have cheered the

new suburban strategy. “The more chaos

and anarchy and vandalism and violence

reigns, the better it is for the very clear

choice on who’s best on public safety and

law and order,” said Kellyanne Conway, a

close adviser to the president, in a televi-

sion interview. Mr Dittrich also applauds

the new approach. He says he has received

a surge of texts, “general feedback” and

(unspecified) polling evidence in the past

week to suggest voters are startled by the

urban violence and are flocking back to Mr

Trump. Kathy Broghammer, the Republi-

can party chairwoman in Ozaukee County,

is even more enthusiastic, pointing to a

surge in interest among Republican volun-

teers because of “the threatening feeling

we see in Kenosha”. “We love our law and

order,” she says. “People are saying, by golly

I’m gonna roar.” 7

T

here’s a reasonthattheperiodfrom

Labor Day to election day is consid-

ered the home straight in a presidential

contest. Robert Erikson and Christopher

Wlezien, two political scientists at Co-

lumbia University and the University of

Texas at Austin, have studied the history

of American election-polling since 1952

and found that the leader in the polls one

week after the second party convention

has always won the popular vote. Two

candidates have trailed in the polls

around Labour Day, then won in the

electoral college: George W. Bush in 2000

and Donald Trump in 2016. In both cases

the polls were much tighter than they are

now.JoeBidentherefore has past perfor-

mance on his side.

Mr Trump’s odds have improved in

betting markets, suggesting that many

think the combination of the Republican

National Convention and the protests in

Portland, Kenosha and elsewhere are

working in his favour. That in turn sug-

gests they expect a lot more volatility in

the polls than has ever been seen before.

Going back to 1948, how wildly polls

have changed in the first seven months

of the election cycle helps explain how

much they vary in the last three. The

standard deviation of the average of

national polls—a measure of how much

they jump around from day to day—over

the first two-thirds of the election cycle

has explained 50% of the variance in the

last third (a perfect relationship would be

100%). Polls in the last third of the cam-

paign have tended to be about 40% as

volatile as the first two.

Mr Trump’s poll numbers have been

low and stable. Mr Biden’s numbers have

moved with a historically low standard-

deviation of 0.9 percentage points so far.

Mr Trump is a known quantity. Few

voters are yet to make up their minds

about him and political polarisation has

decreased the share of true swing voters

in the electorate. A bet on volatility in the

polls is therefore highly contrarian.

History is on the side of the herd.

Shortingvolatility

Polls and nerves

WASHINGTON, DC

How much do polls move after Labour Day?

VIX for politics

United States, average error between Democratic

vote margin in polls and final election results

Percentage points

Sources:WillJenningsandChristopherWlezien;TheEconomist

12

9

6

3

0

200 150 100 50 0

Days until election day

2000-16

1 948-2016